New Scientist - 29.08.2026.pdf

《新科学家》周刊 2026年8月29日

量子理论终于有了合理的解释

……但前提是我们必须在现实的织面中戳出漏洞

更多智能成本。

MG

全新 7 座 MGS9 PHEV

£0 首付

0.9% 代表年利率 (APR)

- 全景天窗 - 最高 62 英里纯电续航里程* - 12.3" HD 双屏

名爵汽车英国有限公司（MG Motor UK Ltd）是 Compliance Guys 有限公司的委任代表，后者经英国金融行为监管局（FCA）授权并监管。我们是信贷经纪人而非贷款机构。本金融方案由 MG Motor Financial Services RH1 1SR 提供。我们为您介绍的经销商将获得佣金。融资需视信用状况和收入而定。18岁以上。适用条款与条件。优惠截止日期为 30.09.26。

*提供的数据是在测试条件下计算得出的。这些数据仅用于比较目的，可能无法反映实际驾驶结果。实际数据可能会根据天气条件、驾驶风格、车辆载荷以及注册后安装的配件等各种因素而有所不同。

探索更多

封面内容

30 量子理论终于有了合理的解释……但前提是我们必须在现实的织面中戳出漏洞

第 271 卷 第 3610 期 封面图片：艾科·奥贾拉 5 全球最高龄人群的非凡免疫细胞

7 冰川融化如何加剧欧洲热浪

36 为什么现在是重新思考长期使用抗抑郁药的时候了

26 次世代香蕉 24 吉姆·阿尔-哈利利探讨时间 40 有史以来最危险的算法

30 特写 “一个世纪以来，我们对量子物理的思考是否一直都是错误的？”

新闻

4 暗物质的新线索 奇异的伽马射线信号可能揭开暗物质的秘密

6 温度升高 全球变暖程度可能比我们想象的高出 25%

13 有父之风，有女之姿 利用雄性小鼠创建健康的雌性克隆体

观点

19 我们的上限 格雷厄姆·劳顿探讨人类可能的最高寿命

20 空白时代 安娜莉·纽维茨论失去文化历史的风险

22 光圈 年度野生动物摄影大赛的惊艳快照

24 本周之书 对时间的全面审视

29 读者来信 异质宇宙引发的疑问

9 沉睡的神经元 我们终于找到了产生睡眠驱动力的原因

特写

30 在现实中戳洞 如果我们将无理数从量子力学中移除，会发生什么？

36 抗抑郁药陷阱 长期使用 SSRI 后的戒断症状似乎比我们想象的要严重得多

40 最危险的算法 试图证明量子计算机可行的人

封底页

45 益智游戏 尝试我们的填字游戏、谜题和快速测验

46 近乎最后的定论 为什么人类如此轻易地将虚构内容融入生活？

47 Tom Gauld 为《新科学家》创作 漫画家眼中的世界

48 反馈 深入探讨未来量子互联网上的音乐创作

48 Twisteddoodles 为《新科学家》创作 描绘生活轻松的一面

你又没听到她那个好笑的笑话。

现在你欠她一个冰淇淋。

预约一次免费听力测试。* 为了你的孙辈。

通过 0% 月付计划分摊费用**

访问 bootshearingcare.com 或致电 0345 200 7052 $^{†}$

何时停止？

三个星系团中一个奇特的伽马射线信号可能会揭开暗物质的秘密——或者它可能揭示了一种更奇异的现象，Karmela Padavic-Callaghan 发现

一个异常的伽马射线信号可能是迄今为止最直接的暗物质证据。然而，该信号也可能是望远镜异常的结果，或者是由某种比暗物质更奇怪的事物产生的。

暗物质构成了我们宇宙大部分的骨架，例如驱动着星系如何聚集。它的质量也远超所有可见物质，但几十年来，研究人员一直对它究竟是什么感到困惑。

美国国家航空航天局（NASA）戈达德太空飞行中心 / A. SIMONNET，索诺马州立大学

问题的一部分在于，他们目前无法直接观测到暗物质。相反，他们只能记录它对周围环境产生的影响。一个更直接的调查方向是探测暗物质在相互碰撞并湮灭时产生的粒子，一些理论认为这种情况应该会发生。现在，中国广西大学的梁云峰（Yun-Feng Liang）及其同事发现了一个可能是该过程产生的伽马射线信号。

他们通过分析费米伽马射线太空望远镜（FGST）15.5 年的数据，专门研究室女座、天炉座和蛇夫座星系团后发现了该信号。梁云峰表示，研究人员选择这些星系团是因为它们被认为包含巨大的暗物质晕。该团队在那里发现的伽马射线信号形状像一个尖峰或一条线，类似于仅有一种颜色的光爆发，一些理论预测这是暗物质自我湮灭的结果。

“发现一条尖锐的伽马射线线谱将是证明暗物质粒子存在并揭示其在粒子物理学中性质的终极‘冒烟枪’（确凿）证据，”团队成员、中国科学院的范义忠（Yi-Zhong Fan）说道。基于他们对数据的统计

一次奇怪的伽马射线爆发能成为暗物质的证据吗？

分析，研究人员估计，该信号是随机宇宙噪声或纯属巧合模式的可能性低于 1 / 10,000（《Physical Review Letters》，doi.org / rkp4）。

然而，在团队宣布一项确定性的发现之前，仍有几个问题需要解决。在伽马射线中探测暗物质特征在历史上一直很困难，因为信号往往很弱，且容易与望远镜误差混淆。例如，2012 年一个极具前景的伽马射线信号最终被证明正是如此。同样来自中国科学院并参与这项新研究的沈照强（Zhao-Qiang Shen）表示，团队成员进行了大量测试以确认信号的真实性，但仪器误差的可能性依然存在。

此外，虽然在观察这三个星系团时新信号很清晰，但在我们银河系中心附近——那里暗物质密集，因此也应该在自我湮灭并产生伽马射线——该信号似乎消失了。

“该信号在遥远的星系团中亮起，但在我们自己的宇宙后院却保持沉默，这一事实非常奇特，”梁云峰说。“如果这确实来自暗物质，它表明这些粒子的相互作用方式必须比

“该信号是随机宇宙噪声或巧合的可能性低于 1 / 10,000”

传统理论预测的要复杂得多。”或者，该信号可能来自某种更为罕见或新颖的现象，例如来自奇异磁化中子星的超快粒子风。

迫使重新思考

“我认为这个结果很有趣，但在将其解读为暗物质证据时，我会非常谨慎，”英国利物浦大学的尤里·斯米尔诺夫（Juri Smirnov）表示。他认为，仅观察一个星系团而非将三个星系团结合起来观察，会增加该信号是噪声产物或随机模式的可能性。他还担心，为了产生这种远离我们但不在我们星系中心附近的伽马射线峰值，暗物质过程必须变得多么不寻常。“这使得标准的暗物质解读面临相当大的压力，”他说。

该团队希望未来的望远镜任务能让情况变得更加清晰，例如几年前提出的超大面积伽马射线空间望远镜，它可以以更好的分辨率收集更多伽马射线。范（Fan）表示，到 2040 年，FGST 的数据集也将增加一倍，这可能进一步阐明这一奇异伽马射线峰值的本质。

“如果未来的望远镜确认该信号是真实的，这将代表一次历史性的突破。它将立即揭示难以捉摸的暗物质粒子的质量，并为物理学家构建下一代基本粒子理论提供一个具体目标，”申（Shen）说。“与此同时，它将迫使我们彻底重写关于暗物质的标准教科书模型。”

传染病

致命的新兴蜱传病毒正在美国传播

爱丽丝·克莱恩（Alice Klein）

两种蜱传病毒——心脏地带病毒（Heartland virus）和波本病毒（Bourbon virus）——似乎正在美国传播。内布拉斯加州已成为第九个在当地蜱虫中检测到其中一种或两种病毒的州。

心脏地带病毒最早于 2012 年报道，波本病毒于 2015 年报道，两者均是在被蜱虫叮咬后出现严重疾病或死亡的人员身上发现的。此后，研究发现它们由孤星蜱（Amblyomma americanum）传播，该蜱虫因雌性背部有一个白色的、大致呈星形的斑点而得名。

从历史上看，孤星蜱分布在美国东南部各州，但由于气候变化和重新造林等因素，它们一直在向北和向西迁移。

在密苏里州、堪萨斯州、伊利诺伊州、阿拉巴马州、佐治亚州、纽约州、宾夕法尼亚州和弗吉尼亚州的孤星蜱中已发现心脏地带病毒、波本病毒或两者兼有。这些蜱虫最近也在内布拉斯加州定居，这促使内布拉斯加医学中心（University of Nebraska Medical Center）的约瑟夫·福弗（Joseph Fauver）及其团队研究这些蜱虫是否随之带来了病毒。

该团队分析了内布拉斯加州卫生与人类服务部在 2022 年至 2026 年期间从该州多个地点采集的 7000 多只孤星蜱。在来自萨皮县（Sarpy county）的蜱虫中检测到了两种病毒，而在理查森县（Richardson county）的蜱虫中发现了心脏地带病毒，这两个县均位于内布拉斯加州的东南部（medRvix, doi.org / rkp8）。

目前，内布拉斯加州尚无已知人员因蜱虫叮咬而感染这些病毒。

福弗表示，对这些病毒进行监测至关重要，这样人们才能采取预防措施保护自己。“显然，你总是希望避免被蜱虫叮咬，但如果你知道自己处于这些病毒循环传播的热点区域，你可能需要格外小心，”他说。

衰老

超百岁老人拥有高度警觉的免疫系统

詹姆斯·伍德福德 (James Woodford)

地球上最高龄人群的免疫系统可能并非缓慢衰退，而是进入了超负荷运转状态，产生能够对抗疾病的特化细胞。

一项小型研究表明，年龄超过 110 岁的超百岁老人体内某种特定免疫细胞的水平较高。这可能是为了应对老年时期面临的挑战（例如更高的癌症风险）而增强的，而这或许正是驱动他们极长寿命的原因。

“我认为令人惊叹的是，即使在经历了超过一个世纪的生活之后，免疫系统可能仍然保留着适应能力，”日本大阪大学的桥本浩介 (Kousuke Hashimoto) 表示。“这些极高龄个体的免疫系统并非简单地随年龄增长而衰退，而是可能保留了适应随年龄增长而累积的持续挑战的能力。”

在之前的研究中，研究人员将这种免疫细胞——一种被称为 CD4 细胞毒性 T 淋巴细胞 (CD4 CTLs) 的 T 细胞——确定为超百岁老人的一个独特标志。在这次新研究中，他们希望量化这些能够识别并杀死癌症等威胁的细胞如何随年龄增长而变化。

渡边千津 (Chitetsu Watanabe) 是少数几位寿命超过 110 岁的人之一

他们招募了 8 名 70 至 99 岁的人，10 名 100 至 109 岁的人，以及另外 10 名 110 岁及以上的人。桥本表示，虽然人数看起来很少，

“即使在超过一个世纪之后，免疫系统可能仍然保留着适应能力”

但根据日本的国家人口普查，该国仅有大约 150 名超百岁老人，“因此 10 人已经是一个相当大的数字”。

血液样本显示，随着各组年龄的增长，CD4 CTLs 在总 T 细胞计数中所占的百分比从 4% 增加到近 10%，而在最高龄人群中则增加到 17% 以上 (Cell Reports, doi.org / hcf7rj)。

研究人员还发现，这些免疫细胞没有表现出 T 细胞耗竭的迹象（即杀死某些细胞，如癌细胞或被病毒感染细胞的能力下降）。“我们的研究结果表明，[CD4 CTLs] 可能在极高龄阶段有助于癌症的免疫监视，”桥本说。

随着年龄增长，免疫系统面临的挑战——如癌症等异常细胞和重新激活的病毒——变得更加普遍。

“我们认为 CD4 CTLs 可能是为了应对这些挑战而扩增的。如果我们能理解为什么这些细胞在某些极高龄人群中会扩增，以及它们攻击的目标是什么，我们最终可能会找到在其他人身上复制这种有益免疫反应的方法，”桥本说。

去年 9 月，西班牙巴塞罗那圣保罗研究中心的马内尔·埃斯特耶尔 (Manel Esteller) 对玛丽亚·布兰亚斯·莫雷拉 (María Branyas Morera) 进行了详细研究，她在 2024 年去世前是世界上最高龄的女性，享年 117 岁 168 天。埃斯特耶尔表示，桥本的研究表明，超百岁老人体内富集了攻击免疫系统威胁以及可能包括癌细胞的免疫细胞。

“玛丽亚经历过多次感染，但能够抵御所有感染，”他说。“她是西班牙已知感染过 covid-19 病毒且无症状的最高龄人士。”

THE ASAHISHIMBUNVIA GETTY IMAGES

气候变化

我们低估了全球变暖

关于全球升温幅度的官方预测已经足够糟糕——而迈克尔·勒佩吉（Michael Le Page）发现，现在的情况似乎表明全球升温可能会比这些预测高出 25%

一些预测在给定二氧化碳水平下会出现更高升温幅度的气候模型此前被认为是不切实际的，但一种评估模型的新方法表明，这些模型是准确的。

“对于给定的排放轨迹，我们预计升温将高出 25%，”瑞士苏黎世联邦理工学院（ETH Zurich）的格尔加纳·古列娃（Gergana Gyuleva）表示。

这一结论具有巨大的影响。例如，一个名为气候行动追踪器（Climate Action Tracker）的项目目前预测，如果各国实施现有政策，到 2100 年，全球平均地表温度将比工业化前水平上升 2.6℃。而古列娃团队的结果表明，我们实际上正朝着 3.25℃ 迈进。

全球升温多少取决于我们向大气中排放多少二氧化碳（$CO_{2}$），以及地球对这些 $CO_{2}$ 的反应如何。不同的气候模型在预测给定 $CO_{2}$ 水平下的升温幅度时存在差异。例如，一些模型认为，$CO_{2}$ 浓度翻倍将导致短期内全球平均地表温度比工业化前水平增加约 $1.6^{\circ}$ C。而其他模型则认为升温可能高达 $3^{\circ}$ C。

“这个范围很大，而且至关重要，”古列娃说，“所以，关键在于找出哪些模型是正确的。”

此前，这一工作是通过使用气候模型模拟近期过去的情况来完成的。模型被输入过去一个世纪左右的温室气体排放数据，然后将其预测的气候反应与实际发生的情况进行比较。

在政府间气候变化专门委员会（IPCC）的上一份报告中，许多此类模型模拟的升温幅度超过了实际发生的情况，因此被剔除。基于剩余的模型，该委员会的上一份报告得出结论：$CO_{2}$ 浓度翻倍将在短期内导致 1.2℃ 到 2.4℃ 的升温，最佳估计值为 1.8℃。

但地球气候存在很大的波动性。例如，在拉尼娜（La Niña）气候模式期间，来自深海的较冷海水在海洋中扩散，导致地表温度较低。

GRAHAM HUNT / ALAMY

气候变化正使热浪变得更加频繁、剧烈且持续时间更长

因此，古列娃及其同事对历史气候记录进行了过滤，试图消除由拉尼娜等现象引起的波动影响。

他们的结果表明，自然波动限制了 1981 年至 2014 年间的温度上升——这一时期包括 21 世纪第一个十年中所谓的全球变暖停滞期。“正是 1981 年到 2014, 这个在之前论文中被使用的时间段，出现了最强烈的向下偏差，”她说，“这真的是运气很糟糕。”

ANTHONY DEVLIN / BLOOMBERG / GETTY IMAGES

仅对这一偏差进行修正，就会导致对 $\mathrm{CO}_{2}$ 翻倍所预期升温幅度的估计值升高。但古列娃的团队还使用一种新方法评估了模型。几颗卫星上的仪器一直在测量进入大气的太阳短波辐射量以及发射出的长波辐射量。进入大气与离开大气的热能之差，是衡量全球变暖最根本的指标。

因此，模型在世界变暖过程中对入射和出射辐射趋势的预测能力，可以被视为比其对地表

“与辐射观测趋势最匹配的模型预测的变暖幅度要大得多”

地表温度。但由于进行此类评估所需的卫星测量直到 2001 年才开始，这种评估方式直到最近才成为可能。

该团队发现，与短波和长波辐射观测趋势最匹配的模型预测的变暖幅度要大得多。基于在 2 种评估中表现均最佳的模型，该团队得出结论：二氧化碳（$CO_{2}$）浓度的翻倍将导致气温上升 1.9℃ 至 2.6℃ 之间，最佳估计值为 2.25℃（载于《地球的未来》Earth's Future, doi.org / rkn8）。

“在我们可以完全确信之前，我们的结果需要更多来自不同来源的证据来证实，”久列娃（Gyuleva）表示。

“这似乎是一个可靠的方法，”挪威奥斯陆大学的特鲁德·斯托雷尔夫莫（Trude Storelvmo）说道。

“在我看来，过滤掉自然变率是一个有价值的进步，”北卡罗来纳州杜克大学的德鲁·辛德尔（Drew Shindell）表示，尽管他尚未被说服短波和长波趋势是评估模型的良好方式。

“我认为这是一个合理的结论和一种可行的方法，”悉尼新南威尔士大学的史蒂文·舍伍德（Steven Sherwood）说道。

但舍伍德表示，这些发现是基于气候模型的，而没有任何一个模型能很好地还原关键现象。“为了更好地完成这项工作，我们确实需要更好的气候模型。另一方面，无论采用何种方法，过去几年剧烈的气温上升都将导致更高的估计范围。”

在气候学术语中，久列娃团队所估计的被称为“瞬时气候响应”。这只是三种气候敏感性类型之一，它们在涉及的时间尺度上有所不同。

在二氧化碳（$CO_{2}$）翻倍后的几十年里，各种反馈机制将开始起作用并导致进一步变暖，例如海洋的持续升温。在这些中期反馈起作用后的响应被称为“平衡气候敏感性”。去年有研究表明，短波和长波辐射的趋势表明，平衡气候敏感性高于之前的估计值。

此外还存在需要数百年或更长时间的极慢反馈，例如冰盖的融化。对这种“地球系统敏感性”——即在数千年时间跨度内对二氧化碳（$CO_{2}$）翻倍的全面响应——的估计值高达 $7^{\circ}C$ 的变暖。

然而，如果通过碳移除降低大气中的二氧化碳（$CO_{2}$）水平，一些长期后果将会被避免。

格陵兰岛融冰正加剧欧洲热浪

Alec Luhn

格陵兰冰盖的融化正导致欧洲出现更极端的酷热，并加剧了今年以来当地经历的热浪。

一项正在进行同行评审的研究发现，大量冰冷融水涌入北大西洋，正在改变海洋温度并使喷流（jet stream）发生偏移，从而有利于西欧和地中海地区出现热浪。随着格陵兰冰层流失的加速，欧洲遭受的热浪频率和强度可能会超过目前气候模型的预测。

该研究的作者之一、英国国家海洋学中心的 Marilena Oltmanns 表示，这种新发现的“大气桥”在今年夏天前所未有热浪中发挥了重要作用。

英国南极调查局

“欧洲升温更快。这是因为喷流发生了偏移，”她说道。“在未来，我们会认为[今年的热浪]根本不算热浪；反而觉得挺舒服的。”

全球气温的升高增加了热浪的发生。但主导天气模式的变化是否也在起作用，目前仍存在争议。

研究发现，在过去的 30 年里，高压脊（在天气图上看起来像一个倒 U 型的大面积高气压区）在欧洲上空形成的频率更高，导致天气晴朗且极端酷热。这些高压脊是由喷流（环绕北半球的一带强风）的波动引起的。当大致呈直线流动的喷流在西欧附近向北转向时，它允许来自中大西洋和撒哈拉沙漠的暖空气涌入已形成的高压脊中。

研究人员表示，高压脊的形成可以通过一系列气候事件追溯到格陵兰岛。他们发现，在出现大规模高压脊年份之前的夏天，冰盖流出的融水量平均增加了 810 亿加元 吨（Research Square, doi.org / rkxx）。这些冰冷的融水在秋季和冬季扩散至北大西洋时，会降低海面温度。

“在未来，我们会认为今年的热浪根本不算热浪；反而觉得挺舒服的”

其低盐度抑制了与更咸、更温暖的深层海水的混合，从而进一步冷却表层。

这些脉冲式的融水冷却现象叠加在“冷 blob”（北大西洋海洋温度的长期下降）之上，一些科学家将后者归咎于大西洋经向翻转环流（AMOC）系统电流的减弱。两者共同在北大西洋的冷水区与更南侧的温暖表层水之间创造了一个剧烈的温度锋面。

格陵兰岛的融水正在使喷流发生偏移

随着暖气团在锋面南侧上升，冷气团在北侧下降，它们开始旋转并产生气旋。这使得喷流的波动更加剧烈。喷流倾向于在这些气旋的南侧蜿蜒，然后再次向北跨过西欧，从而在当地产生持久的高压脊和热浪。

英国雷丁大学的 John Methven 认为，格陵兰岛的融水可能在迫使喷流蜿蜒。但 AMOC 的减弱或北大西洋上空大气环流的变化也可能导致这种偏移。

“在我看来，这是薄弱环节，冷 blob 与格陵兰岛融化现象之间的联系尚未建立，可能还有其他因素在起作用，”他说道。

癌症简报

癌症疫苗离我们还有多远？

迈克尔·勒佩吉（Michael Le Page）发现，一种个性化 mRNA 癌症疫苗在针对皮肤癌的最终阶段试验中取得了积极结果，这为该方法带来了希望。

由于每种癌症都不同，几十年来，研究人员一直致力于开发个性化癌症疫苗，旨在使免疫系统能够识别并攻击每个个体特有的癌症。此前的结果好坏参半，但现在，一种个性化癌症疫苗首次在最终阶段（即 III 期）试验中取得了积极结果。

Moderna 在 19 August 宣布，对于那些接受了手术以切除严重

“针对某些类型的肺癌、肾癌和膀胱癌的试验已经在进行中”

皮肤癌（称为黑色素瘤）的患者，如果他们在接受另一种免疫疗法的同时接种个性化 mRNA 癌症疫苗，那么与仅接受该免疫疗法相比，他们在癌症未复发或未扩散到身体其他部位的情况下生存的时间更长。

人们希望这种个性化疫苗能适用于许多其他类型的癌症。针对某些类型的肺癌、肾癌和膀胱癌的试验已经在进行中。BioNTech 公司也在进行个性化 mRNA 癌症疫苗的人体试验。

什么是癌症疫苗？

癌症是由细胞积累突变引起的，直到它们失控并开始无限制地增殖。细胞表面突出的蛋白质发生突变，也会使癌细胞看起来与其它细胞略有不同。这使得我们的免疫系统能够在我们毫无察觉的情况下识别并杀死许多癌症。但有些癌细胞能逃脱监测——癌症疫苗的理念就是教会免疫系统去识别它们。

这些疫苗是如何制造的？对于名为 intismeran autogene 的 Moderna 疫苗，首先会对患者的癌症样本进行测序。然后将其与健康的血细胞进行对比，以识别癌细胞表面具有独特突变的蛋白质。最多选择 34 种，随后制造编码这些蛋白质的 mRNA 并将其放入疫苗中。

接下来的步骤是什么？

癌症疫苗被注射到肌肉中。肌肉细胞随后会暂时产生这些癌症蛋白质，从而触发针对它们的免疫反应。这有助于治疗，因为免疫系统会“记住”这种反应，这样在日后遇到癌细胞时就能做好准备。

为什么要使用 mRNA 疫苗技术？虽然可以使用其他类型的疫苗，但 mRNA 更快且成本更低。制造基于蛋白质的疫苗需要使用活细胞，这非常耗时且需要进行许多安全检查。对于

旨在使免疫系统识别并攻击癌症的癌症疫苗正在研发中

癌症患者来说，疫苗创建得越快越好。

在试验中，该疫苗与哪种免疫疗法结合使用了？这是一种所谓的 anti-PD-1 疗法，由一种名为 pembrolizumab（Keytruda）的抗体组成。许多癌症通过竖起一个“不要攻击我”的标志来进化，从而逃避免疫攻击。免疫细胞借助其表面一种名为 PD-1 的蛋白质来“读取”这个标志。

阻断 PD-1 会使“不要攻击我”的标志对免疫细胞不可见，因此无论如何它们都会发起攻击。但是，anti-PD-1 疗法只有在免疫系统已经在尝试攻击癌症时才有效，因此将其与癌症疫苗结合以诱发这种攻击具有优势。

联合治疗的效果如何？

Moderna 尚未告知我们具体结果。该公司仅宣布，III 期临床试验的结果显示，生存时间在“统计学上具有显著意义且在临床上具有实质性的改善”。

该试验涉及 1137 人，他们在接受切除黑色素瘤的手术后，被随机

ADAM GLANZMAN / BLOOMBERGVIAGETTYIMAGES

分配接受疫苗和帕博利珠单抗（pembrolizumab）的联合治疗，或仅接受帕博利珠单抗治疗。我们已知的是，在之前一项涉及 157 名黑色素瘤患者、为期五年的试验中，与单独使用帕博利珠单抗相比，疫苗与帕博利珠单抗的联合疗法将复发风险降低了 50% 且将扩散至身体其他部位的风险降低了 60%。

这种疗法何时能投入使用？

接受《金融时报》采访的分析师表示，该疗法可能会在明年获得批准。但由于这是一种个体化治疗，扩大生产规模将具有挑战性。

目前其定价尚不明确。路透社 1 月份的一份报告指出，每次治疗的费用约为 $200,000。

这是否适用于所有癌症？ 人们希望它能适用于多种癌症，但我们需要更多的 III 期临床试验结果才能确定。

目前是否已有获批的癌症疫苗？

不像 intismeran autogene 这样。有些资料会告诉你，目前有三种获批的癌症治疗疫苗：Provenge、BCG 和 T-VEC。但 Provenge 是一种涉及从人体中提取免疫细胞、使其接触前列腺癌蛋白然后重新植入的疗法。BCG 是结核疫苗，恰好能帮助治疗某些膀胱癌。而 T-VEC 是一种经过基因工程改造、用于杀死癌细胞的活体单纯疱疹病毒。

HPV 和肝炎疫苗怎么说？

这些疫苗是通过预防可能导致癌症的病毒感染，而不是直接针对癌症，这是一种更佳的方法。

神经科学

我们现在可能知道哪些神经元决定了何时该睡觉

克里斯·西姆斯 (Chris Simms)

研究人员在小鼠大脑中发现了特定的神经元群体，这些神经元在啮齿动物长时间清醒后会被激活，从而驱动它们进入睡眠状态。如果人类体内也存在类似的系统，那么针对该系统进行干预可能有助于改善某些睡眠障碍的治疗。

在大脑内部，有多个神经回路调节睡眠与觉醒，其中包括下丘脑中的神经元，它们像交换机一样控制这一过程。

其中一个或多个回路可能解释了为什么我们清醒的时间越长，闭上眼睛的冲动就越不可抗拒。但究竟是什么驱动了这种冲动，目前仍是一个谜。

为了揭开谜底，瑞士巴塞尔大学的威尔·周 (Will Joo) 及其同事对比了小鼠在正常睡眠-觉醒周期、睡眠剥夺和恢复性睡眠期间的大脑激活模式。他们通过追踪一个名为 Fos 的基因在细胞中产生蛋白质的数量来实现这一点，该基因在神经元激活过程中会被开启。

“我们发现大脑的非常特定区域与觉醒状态相关，”周表示。这使得研究人员能够识别出活动与清醒时长相关的区域。

在其中一个区域——正中缝核 (median raphe nucleus) 中，他们发现了两个截然不同的神经元群体，随着动物清醒时间的增加，这些神经元的激活程度也随之增加，并在睡眠开始后再次恢复平静。这些群体分别是 GABA 能神经元和血清素能神经元，之所以如此命名，是因为它们分别响应的化学信使分子是 $\gamma$-氨基丁酸 (GABA) 和血清素。

“这是一项卓越的研究。

是什么驱动我们产生疲劳感并入睡？

INDRANIL ADITYA / NURPHOTO VIA GETTY IMAGES

全脑 Fos 绘图是一项真正的杰作，”威斯康星大学麦迪逊分校的基亚拉·西雷利 (Chiara Cirelli) 表示。

为了观察这些神经元群体变得活跃仅仅是因为小鼠清醒时间长，还是具有更深层的功能，周及其同事随后利用病毒将基因包传递给神经元，以激活或抑制其作用。

当两组神经元都被人工激活时，小鼠的睡眠时间是对照组小鼠的两到三倍，并且在非快速眼动 (non-REM) 慢波睡眠中花费的时间更多——这种方式类似于人类在长时间清醒后所经历的深度恢复性睡眠。

“它不仅增加了动物睡眠的时间，还提高了睡眠的质量，”周说。

当这些神经元被抑制时，小鼠的睡眠时间比对照组减少了约 70%——每天额外清醒约 6.5 小时——且没有表现出小鼠睡眠剥夺通常会带来的更多类焦虑行为。然而，缺乏睡眠导致约 17% 的小鼠死亡 ($\textit{Nature}$, doi.org / rkqf)。

团队成员、同样来自巴塞尔大学的亚历克斯·希尔 (Alex Schier) 表示，激活和抑制神经元产生的强烈效果意味着它们不仅仅是在发出动物已清醒的信号。

“这些神经元对于促进睡眠至关重要，并且可能是产生睡眠驱动力的回路中的关键部分”

“它们对于促进睡眠至关重要，并且可能是产生睡眠驱动力的神经回路中的关键组成部分。”

马里兰州约翰·霍普金斯大学的马克·吴 (Mark Wu) 表示，这项研究“增加了一系列日益增多的研究成果，证明存在由觉醒激活但却促进睡眠的独立睡眠回路”。他的研究已经表明，大脑中一个名为丘脑 reuniens 核 (thalamic nucleus reuniens) 的区域中的神经元能够追踪我们需要的睡眠量与实际获得睡眠量之间的差距。

目前尚不清楚人类体内是否存在相同的睡眠驱动神经元群体，但研究人员表示，如果确实存在，一旦我们了解这些细胞如何与其他脑区通信，该发现可能会为某些与睡眠相关的疾病开辟新的治疗途径。

这些来自正中缝核的神经元投射与视前下丘脑相连，已知这些区域共同协作以控制睡眠-觉醒周期。

“这条主要神经通路对睡眠的影响非常强，但正如我们在开始操纵神经回路以来所学到的那样，每条通路都提供细微的差别。大脑有许多种方式来唤醒和入睡，”加利福尼亚州斯坦福大学的路易斯·德·莱塞亚（Luis de Lecea）说道。

例如，在果蝇实验中已有迹象表明，大脑中为细胞提供能量的线粒体可能在睡眠的启动中起着至关重要的作用。

人工智能

为什么 AI 水印无法奏效

AI 公司已开始对其模型的输出内容添加水印，以提高透明度并应对虚假信息——但马修·斯帕克斯（Matthew Sparkes）报告称，这种方法存在局限性

随着 聊天机器人的兴起，你可能经常在阅读一段文字时好奇，它是由人类还是机器编写的。现在，为了响应欧盟的立法，许多 公司正在为其模型生成的文本、图像和视频添加水印，目的正是为了让你能够分辨。人们希望这些水印能帮助遏制虚假信息的传播，并提高 使用的透明度，但专家指出，至少在文本方面，水印方案不太可能奏效。

“水印并不能突然解决欧盟可能希望解决的问题”

水印要求是欧盟《人工智能法案》（AI Act）的一部分，并于 8 月 2 日起生效。ChatGPT 的开发商 OpenAI 已经在对图像和音频进行打水印，并表示计划在未来增加文本水印。Claude 的背后的公司 Anthropic 已经宣布，未来的模型将对文本打水印。《人工智能法案》为已经部署的模型提供了一段宽限期，但所有模型必须在今年 12 月 2 日之前提供水印功能。

FG TRADE / GETTY IMAGES

AI 安全公司 NOPE 的詹姆斯·帕多尔西（James Padolsey）表示，虽然图像和视频水印由于由密集且多层的数据组成而更加可靠，但文本水印的实施将始终更加困难。

一种文本水印的方法是对词汇的选择应用一种数学上可检测的模式。

OpenAI 已经在对使用 ChatGPT 生成的图像打水印。大型语言模型（如 ChatGPT）通过附加统计学上最有可能出现的下一个词来创建句子。通过改变这种词汇选择方式——例如在整个句子中重复地交替选择最可能的下一个词和第二个最可能的词——模型可以在不改变文本整体含义的情况下嵌入一种模式。根据《人工智能法案》，公司将被要求提供能够搜索这些模式的工具，以便人们识别 生成的文本。

但帕多尔西指出，这些水印模式很容易通过轻微的编辑而被删除。他说：“这没什么用，而且人们尝试使用它的方式将会是错误的。”

为了证明他的观点，帕多尔西创建了一个名为 declaude 的在线工具，该工具接收 生成的文本并对其进行轻微修改，使得现有的 检测器不再起作用。他说，同样的过程也应该适用于带水印的文本。

帕多尔西表示，即使没有这类工具，一些

你认为你能分辨出文本是否是由 生成的吗？

不遵守欧盟规则的开源 模型也将始终可用，因为这些模型不受任何一家科技公司的控制。他说：“任何恶意行为者都会直接使用那些模型。而且它们运行起来可能更便宜、更简单。如果你是一个卑劣的人或恶意行为者，并且试图大规模生成这些内容，这并不难实现。”

欧盟委员会的一位发言人表示：“这些规则将帮助人们识别他们何时在与 交互，或者内容何时是由 生成或修改的。标记和检测方案的对抗鲁棒性必须在抵御恶意行为方面进行评估，例如针对标记的复制、删除、重新生成和修改攻击。”

虽然帕多尔西担心水印容易被删除，但其他人则担心“误报”——即文本被错误地判定为 生成，或者因为某个模型在作品创作中仅扮演了极小的角色而被判定为 生成。英国雷丁大学的彼得·斯卡夫（Peter Scarfe）表示，许多公司向他的部门推销产品，声称能够帮助发现学生的抄袭行为（尽管使用的是除水印以外的其他方法），但同时在小字条款中注明，他们的系统不应被用于最终证明或惩罚抄袭事件，因为它们并非万无一失。

并非万能药

“如果系统显示这段文本有 67.8% 的概率是以某种方式由 AI 生成的，作为一名教育者，我该如何应对？我真的不确定，”斯卡夫（Scarfe）说道。“这些系统运行得并不完美，正因如此，它们的用途有限。”

水印检测工具的运作方式与之类似，它们可能会将仅仅是由 AI 处理过，而非完全由 生成的文本标记出来。

SHUTTERSTOCK / ROKASTENYS

学生们可能会陷入困境，即便他们只是在提交前请求模型扫描其手写作品中的语法错误。

“水印可能仅能制造足够的阻力，从而抑制某些恶意使用 的行为”

斯卡夫表示， 可用于多种目的，其本身并不传播误导信息，因此，一项统一的水印政策可能是一个过于简单粗暴的手段；但他也指出， 可能会出错并虚构事实，且可被用于快速传播谎言——例如驱动社交媒体上的虚假信息机器人。

“由于这些技术的出现如此之快，且加速发展如此之快，生活在各个领域的人们在某种程度上都处于被动地位，”他说。“但我认为（水印）并非万能药。我不认为它能突然解决欧盟可能希望它解决的问题。”

但或许水印能制造足够的阻力，从而抑制某些恶意使用 的行为，例如大规模创建推送虚假信息的社交媒体帖子。俄亥俄州克利夫兰市凯斯西储大学（Case Western Reserve University）的埃尔曼·艾戴（Erman Ayday）表示，即便移除水印只需极小的努力，但随着时间的推移，这能使某些类型的滥用变得不切实际。

“只要花足够的时间，任何人都可以篡改水印，但当你花这么多时间在上面时，也许你当初直接自己生成内容会更好，对吧？”艾戴说道。

人类进化

人类真的进化成仅在白天活跃吗？

迈克尔·马歇尔（Michael Marshall）

人类通常被认为是昼行性的，即主要在白天活动，但来自狩猎采集者的资料表明，他们在夜间也相当活跃。

《失眠之猿》（The Sleepless Ape）的作者、加拿大多伦多大学的大卫·萨姆森（David Samson）表示，这一发现意味着人类的睡眠模式更接近于某些狐猴物种。人类可能并非主要在白天活动、在夜间休息，而是“昼夜活动性”（cathemeral）的，这意味着他们将活动分布在每个 24 小时周期内。“关键点在于，白天和夜晚的活动都是行为库中的常规组成部分，”萨姆森说道。

为了查明人类是昼行性还是昼夜活动性，萨姆森汇总了从人类收集的数据，并将其与非人类

哈扎社区的成员在白天和夜晚均有活动。

灵长类动物的记录进行对比：包括七种狐猴和一个种类的长臂猿。在所有案例中，活动记录均使用腕式加速度计完成。由于许多现代人住在永久性房屋中且可以使用电能，萨姆森使用了来自哈扎人（Hadza）的加速度计记录：他们是生活在坦桑尼亚的狩猎采集者，通常在户外或由树枝和草搭建的遮蔽处睡眠，没有人工光源。

哈扎人在白天的平均活动时间为每小时 52 分钟，夜间为每小时 24 分钟。这与具有强烈昼行性的长臂猿形成对比，后者在夜间的活动时间每小时仅为 2 分钟。人类则更接近于昼夜活动性的狐猴，后者在白天的活动时间为每小时 45 到 48 分钟，夜间为每小时 18 到 21 分钟（《人类进化杂志》，doi.org / rkn6）。

萨姆森还构建了一个灵长类动物进化的数学模型，以观察所有灵长类动物是否都在向相同的最佳睡眠和活动模式进化。他发现人类与其他灵长类动物并不匹配。

灵活的节律

“人类的睡眠-觉醒活动在进化过程中所承受的选择压力，可能与其他灵长类动物有所不同，”他说道。

萨姆森认为，人类已经进化到能够违背自身的内部昼夜节律。虽然我们的生物钟促使我们在白天活动、在夜晚睡眠，但我们的社会和技术允许我们采取不同的方式。他表示，通过利用遮蔽所（Shelter）、炉火（Hearth）、环境准备（Environmental preparation）、光线（Light）和哨兵（Lookouts）——萨姆森将这些统称为“SHELL”——我们可以在夜晚保持活跃。“这种‘SHELL’并没有抹除我们的生物特性，它缓冲了我们在异常时间活动或睡眠所带来的一部分生态成本。”

KENNETH WHITTEN / UNIVERSAL IMAGES GROUP VIA GETTY

“人类在睡眠和活动时间的模式上表现得非常灵活，”加拿大卡尔加里大学的凯瑟琳·莱因哈特（Kathleen Reinhardt）说道，但她认为我们基本上仍然是昼行性动物。

她表示，“间歇活动”（cathemeral）的定义比单纯在夜间有活动证据要严格得多。“我们讨论的是更有意义的行为，例如觅食和迁徙，并且这类行为在两个时段都发生，而不仅仅是在光亮和黑暗中检测到移动，”她说。她认为，加速度计的数据不足以确定夜间活动的性质。

健康

新发现的免疫枢纽可能有助于维持大脑健康

卡丽莎·王 (Carissa Wong)

颅骨中的免疫细胞枢纽可能有助于保护我们的大脑免受疾病侵害。这项基于小鼠大脑研究和人类遗传数据的发现意味着，利用药物针对这些枢纽，可能会为免疫细胞失控的疾病（如脑癌）带来新的治疗方法。

“这是在理解大脑免疫反应方面迈出的重要一步，”芬兰奥卢大学的维萨·基维涅米 (Vesa Kiviniemi) 表示，他没有参与这项研究。他认为，除了适用于癌症之外，“它还与理解感染、炎症性脑病、多发性硬化症 [以及] 神经退行性疾病等问题相关”。

如果淋巴系统中出现了威胁信号（如肿瘤碎片），被称为 T 细胞和 B 细胞的免疫细胞会被激活并在大脑中发挥作用。现在，

密苏里州圣路易斯华盛顿大学的乔纳森·基普尼斯 (Jonathan Kipnis) 及其同事发现了这些细胞被激活的另一种方式。

通过对小鼠颅骨进行成像和分析，该团队发现 B 细胞和 T 细胞与一种被称为抗原递呈细胞的免疫细胞共同聚集在颅骨后方的免疫枢纽中。“这些枢纽此前从未被描述过，”基维涅米说。这些枢纽类似于淋巴结，抗原递呈细胞在其中将威胁暴露给 T 细胞和 B 细胞。

“免疫枢纽有助于产生抗癌免疫反应”

该团队认为，人类体内也存在这些免疫枢纽。这一结论基于先前研究中从人类颅骨收集的基因活性数据，这些数据表明 T 细胞在该部位被激活并帮助激活了 B 细胞。基维涅米表示：“这表明人类体内也存在同样的情况”，尽管还需要进一步分析尸体颅骨的研究来确认这一点。

为了探索这些枢纽是否能启动保护性免疫反应，研究人员将癌细胞注入小鼠大脑。随后，他们向一半小鼠的头皮下方注射了一种实验性药物，该药物可干扰颅骨中 B 细胞和 T 细胞的激活。其作用机制是阻断抗原递呈细胞上的一种名为 CD40L 的蛋白质，该蛋白质有助于激活这些免疫细胞。其余的小鼠接受了生理盐水注射。

在注射肿瘤后，接受药物治疗的小鼠平均生存时间约为 25 天，而安慰剂组的小鼠生存时间约为 30 天。基维涅米表示，这表明免疫枢纽有助于产生抗癌免疫反应。

在另一项实验中，一组小鼠接受了同样的肿瘤注射，但这次，其中一半小鼠接受了三种能够增强其颅骨中 B 细胞和 T 细胞激活的药物。这些小鼠的生存时间比接受安慰剂注射的小鼠长了约 10 天（《自然》杂志，doi.org / rkxw）。

基维涅米说，如果人类体内也存在相同的免疫枢纽，那么针对它们可能会为许多大脑相关疾病带来新的疗法。“我们可以研究如何唤醒并加强这些 [枢纽]，”他说。

环境

用塑料瓶制成的曲奇饼干有望减少浪费

马修·斯帕克斯 (Matthew Sparkes)

通过基因工程酵母，塑料瓶和农业废弃物已被转化为香草味的曲奇饼干。

来自南伊利诺伊大学卡本代尔分校的桑迪亚·贾亚塞卡拉 (Sandhya Jayasekara) 及其同事在制作这些曲奇饼干时，首先将废旧塑料瓶以及丢弃的甜玉米植株茎叶通过一个名为“氧化水热溶解”的过程进行处理。该过程利用热量将材料分解为微生物可以消化的成分。

随后，这些成分被喂给经过 CRISPR 基因编辑技术改造的酵母。这使得酵母能够将这些成分转化为多种有用成分，包括蛋白质、脂肪和酸，以及香草风味剂和 $\beta$-胡萝卜素（人体可将其转化为维生素 A）。最后，在混合物中加入纤维、淀粉和甜味剂，然后将面团通过 3D 打印机挤出成任何所需的形状。由此产生的曲奇饼干被命名为 $\mu$ Bites（读作 microbites）。

研究人员在芝加哥举行的美国化学学会 (American Chemical Society) 秋季会议上展示了这项工作，他们渴望在大学批准实验后尽快亲自尝试 $\mu$ Bites。但他们坚信这些曲奇饼干既安全又美味。

“我认为它有一种令人愉悦且吸引人的香气，”贾亚塞卡拉说。“我们还没有品尝过，但它闻起来很香——就像真正的曲奇饼干一样。”

该团队希望 $\mu$ Bites 能有助于解决塑料污染问题，但并非所有人都对此深信不疑。伦敦帝国理工学院 (Imperial College London) 的杰森·哈莱特 (Jason Hallett) 表示，这项研究有趣且具有启发性，但极有可能在实际应用上是一条死路。“我们每年产生 4 亿 吨塑料垃圾。你不可能把它们全部变成曲奇饼干，”他说。

雄性小鼠的雌性克隆体

一项研究发现，一种基因编辑工具可以将雄性小鼠胚胎可靠地转化为健康的雌性——克里斯·西姆斯 (Chris Simms) 指出，这一成就对于物种保护可能具有重要价值。

通过一种消除 Y 染色体的程序，雄性小鼠胚胎已被转化为雌性。该技术依赖于 CRISPR 基因编辑，并已被用于创建健康且具有生育能力的雌性和雄性小鼠，两者均为父亲的克隆体。

这种方法可能在保护那些濒临灭绝、仅剩一只雄性或极少数雄性的动物时变得有用。

在进行有性生殖的动物中，雄性或雌性的缺失可能意味着一个种群或物种的终结。然而，某些物种可以转向无性生殖，例如通过孤雌生殖，在这种情况下，包括蜥蜴和鲨鱼在内的动物的未受精卵可以发育成为新的动物。这会产生与亲本完全相同的克隆后代，且性别相同。

少数动物，如冲绳碎石虾虎鱼 (Trimma okinawae)，可以通过改变性别来确保有性生殖能够继续。受这些色彩斑斓的海鱼启发，日本山梨大学的石内隆 (Takashi Ishiuchi) 及其同事想出了一种改变雄性克隆胚胎性别的方法，从而使个体之间能够进行有性生殖——尽管这些个体之间具有极近的亲缘关系。

在一些克隆雄性小鼠的尝试中，当决定雄性的 Y 性染色体被删除，使样本仅剩一条 X 染色体（称为 XO）而非 XY 对时，偶然产生了一只雌性。

石内及其同事开发了一种名为 Y-CUT 的基于 CRISPR 的工具，用以可靠地消除

一只雄性小鼠与其雄性和雌性克隆体相遇 雄性小鼠胚胎中的 Y 染色体。

首先，他们将一只雄性小鼠血细胞中的 DNA 插入到已经去除自身 DNA 的小鼠卵细胞中——这种技术被称为体细胞核移植。在受精后立即。

“克隆体具有生育能力，且其后代看起来很健康”

他们用 Y-CUT 处理了其中一些细胞，并将它们转移到代孕雌性小鼠体内。

所有在使用 Y-CUT 后出生的小鼠均为雌性克隆体——除了缺失 Y 染色体外，它们在遗传上与父亲完全相同。而其他未经过 Y-CUT 处理的后代则全部为雄性克隆体（bioRxiv, doi.org / rkjv）。

“我们称之为双性别克隆，”日本筑波 RIKEN 生物资源研究中心的团队成员正悟（Shogo Matoba）说道。

石内（Ishiuchi）表示，让父亲小鼠与它的克隆体见面是一个非常特殊的时刻。“他能看到雌性版本的自己，这简直就像科幻小说一样。”

这些克隆体具有生育能力，并且当它们交配时，产生了看起来健康的后代。“这意味着有性生殖是从单个雄性基因组开始启动的，”正悟说，“这些后代到目前为止已经存活了一年多，没有明显的缺陷。”

石内希望 Y-CUT 技术能够帮助复活那些极度濒危或已灭绝、且在细胞库中仅存冷冻组织的物种。

“如果这项技术在小鼠以外的物种中取得成功，那将具有极高的价值，”同样来自山梨大学但未参与该研究的若山照彦（Teruhiko Wakayama）表示。

保护潜力

野生动物保护组织 Revive & Restore 的本·诺瓦克（Ben Novak）持有类似观点，该组织一直参与包括普氏野马在内的濒危物种克隆工作。他说：“他们正在为未来的染色体操纵奠定基础，这在许多保护案例中可能会变得至关重要。”

诺瓦克同意该技术可能对来自濒危或灭绝物种的冷冻组织样本有用，因为在这些情况下，我们可能只有一两个个体的样本。

此前的一项研究曾利用雄性小鼠胚胎干细胞系产生雄性和雌性克隆体，但其使用的方法不太可能在其他物种中奏效；与之不同的是，体细胞核移植技术已成功应用于多种哺乳动物，且适用于活体或存储的供体材料。

一个问题是，虽然 XO 雌性小鼠健康且具有生育能力，但许多其他物种并非如此。

“Y 缺失的后代在人类和马类中没有生育能力，因此在多种野生动物物种中可能也是这种情况，”美国动物园和保护中心组织 SEZARC 的琳达·彭福尔德（Linda Penfold）说道。她还表示，生产 Y 缺失动物可能会引发伦理担忧。

石内承认这一点，但指出有研究表明，一种名为 Reversine 的药物可以让 XO 细胞获得第二个 X 染色体，因此有可能将此方法与 Y-CUT 结合，使更多物种的雌性后代具有生育能力。

彭福尔德还警告说，利用单个动物的克隆体进行繁殖会产生遗传问题。她说：“这会导致缺乏遗传多样性，而多样性对于种群的健康至关重要。”

诺瓦克对此持更积极的态度。他说：“与完全没有有性生殖相比，通过单个个体进行有性生殖肯定要好，”但他补充道，我们应该尽一切努力，以确保永远不必使用此类工具。

正悟、石内 等（2026）

彭福尔德表示，保护栖息地和减少狩猎在保护物种方面更为有效。

环境

旨在避免凝结尾迹的试验即将启动

空中交通管制员将使用 AI 生成的预测，建议飞越大西洋的飞机调整飞行高度，以努力减轻凝结尾迹对气候的影响，Madeleine Cuff 报道

今年晚些时候飞越北大西洋的飞机将被重新规划航线，以避免产生凝结尾迹。这是一项首创的试验，旨在解决整个空域内凝结尾迹带来的全球变暖影响。

从 11 月开始，管理北大西洋东侧 Shanwick 空域的空中交通管制员将建议飞机调整高度，以避免产生新的凝结尾迹。管制员将

“避免凝结尾迹带来的气候效益，足以抵消因改变飞行路径而增加的任何燃料消耗”

使用来自 Google 的 AI 生成的大气条件预测，来预测凝结尾迹将在何处形成。

参与该项目的研究组织 Contrails.org 的 Marc Shapiro 在 17 August 一次新闻简报会上表示：“我们可以预测哪些区域将形成持久的、导致气候变暖的凝结尾迹，并能引导飞机绕过这些区域。”

当喷气发动机排放的烟尘颗粒触发冰粒形成时，就会产生凝结尾迹，从而留下可维持数小时的凝结轨迹。在白天，这些持久的凝结尾迹有助于将太阳辐射反射回太空，从而产生冷却效果。但它们同时也会将热量困在下方，导致变暖。总体而言，凝结尾迹被认为占航空业气候影响的约三分之一。

Shanwick 空域是凝结尾迹的高发区，也是跨越北大西洋飞机的繁忙区域。模型显示，该空域约占全球凝结尾迹变暖影响的 5 per cent，且大部分影响发生在 11 月至 2 月的冬季。

这项耗资 £500 万 的“蓝天行动”（Operation Blue Skies）项目将在变暖效应最显著的傍晚和夜间重新规划航班航线。进入 Shanwick 空域的飞行员将被指示将高度调整最多 2000 英尺（约 600 米），以避开容易形成凝结尾迹的大气条件。该建议将按照空中交通管制员引导飞行员避开湍流或其他天气条件的现有程序发布。

WIRESTOCK, INC. / ALAMY

该试验将在 2026-2027 和 2027-2028, 的两个为期四个月的冬季试验期间运行，每个冬季季节大约有 20 到 40 个测试日。预计每年在试验时段内将有约 10,000 架次航班通过该空域，其中数百架次需要重新规划航线，以产生具有统计学意义的分析数据集。来自剑桥大学和伦敦帝国理工学院的研究人员将分析此次试验的数据。

避免

当喷气发动机排放的烟尘导致冰粒形成时，就会产生凝结尾迹

凝结尾迹带来的气候效益，足以抵消因改变飞行路径而增加的任何燃料消耗，Google 英国公司的 Paul Hodgson 在简报会上表示。重新规划航线的航班在飞越该空域时，每架飞机可能需要额外燃烧 100 公斤燃料，从而导致额外排放数百公斤的 $CO_{2}$。相比之下，他表示，“这些航班（尤其是处于该时段的航班）产生的凝结尾迹影响是以吨甚至数十吨来衡量的”。

AI 可助力设计更高效的飞机机翼

避免凝结尾迹只是利用 AI 减轻航空旅行对气候影响的方法之一（见主报道）。 智能体还设计出了一种降低飞机机翼模型摩擦力的方法。

华盛顿大学的史蒂文·布伦顿（Steven Brunton）及其同事创建了一个名为 HydroGym 的平台，在该平台中，许多 智能体可以通过调整流体在虚拟物体上的流动方式，来降低物体所受的阻力。这些 智能体首先使用相对简单的计算机模拟进行训练，随后通过一种被称为“强化学习”的试错法来处理该任务。

研究团队发现， 智能体能够将从计算机模拟中简单示例里学到的经验，应用于解决更复杂的场景——即使在无法获得这些复杂场景的计算机模拟的情况下也是如此（《自然》杂志，doi.org / rkjs）。

例如，在学习如何控制平面通道中湍流的流动后，这些智能体解决了控制围绕弯曲的三维飞机机翼模型的流体问题，成功将机翼与流体之间造成能量浪费的摩擦力降低了 38%。

“即使是极小的改进，其在财务和生态方面的影响也是深远的，”布伦顿说道。卡梅拉·帕达维奇-卡拉汉（Karmela Padavic-Callaghan）

该项目是在谷歌 凝结尾迹预测之前的试验基础上开展的，这些预测已被用于在起飞前为航空公司的飞行计划以及空中交通调度员设定的航线提供信息。这项工作表明，重新规划飞机航线可以将凝结尾迹的形成量减少 50 到 60%。

但这是首次在整个空域范围内使用凝结尾迹预测，航空公司将无法选择退出。“这就是我们所说的空域规模试验，”霍奇森（Hodgson）说道。

物理学

关于丢失中微子的 30 年之谜可能已得到解决

实验中一些缺失的中微子困扰了物理学家 30 多年，但一项新的理论研究表明，这最终可能并非谜团。这一出人意料的结果可能并非暗示存在一种新型粒子，而仅仅是因为计算上的疏忽，这将重新确认粒子物理学定律，而非对其发起挑战。

中微子是像幽灵一样难以捉摸的粒子，极难检测，并且可以在其三种类型之间自发振荡。自 20 世纪 90 年代以来，几项使用镓基探测器的实验发现，中微子的数量比预期少了约 20%，这引发了关于部分中微子已转化为一种新型“惰性”中微子的观点，这种中微子与物质的相互作用更少，因此避开了检测。如果情况确实如此，那么列举了所有现有粒子和力的粒子物理学标准模型将不得不进行修正。

意大利卡利亚里大学的马特奥·卡德杜（Matteo Cadeddu）及其同事现在提出，这样剧烈的干预可能

“多次使用镓基探测器的测试发现，中微子数量比预期少了 20%”

可能并非必要。他们的计算表明，问题可能出在实验的数学建模上。

在实验中，当一个镓原子被中微子击中时，它会转化为一个锗原子和一个电子。为了计算镓基探测器中的中微子数量，研究人员测量锗与电子之间的相互作用，然后进行反向推算。

直到现在，他们一直习惯性地假设电子和中微子的量子波函数（在数学上对它们的属性和行为进行编码）在发生转化的原子核范围内是不变的。这正是 Cadeddu 及其同事关注的细节。

令他们惊讶的是，他们发现只要考虑到原子核的某些特性，放弃这一假设就可以解释此前报道的 20% 中微子缺失问题（《物理评论快报》，doi.org / rkgq）。“我们不需要增加新粒子或新的相互作用，”Cadeddu 表示。

“这是一个值得进一步研究的有趣线索，”德国美因茨大学的 Joachim Kopp 表示。在他看来，试图扩展标准模型以适配实验结果需要“极其古怪且经过精细调整的理论模型”，因此这项新分析更具前景。但他表示，必须独立测量或计算镓和锗的核结构细节，才能完全解开中微子的谜团。

密歇根州立大学的 Ante Ravlić 表示，这项新研究是朝着正确方向迈出的一步。“我们必须等待基于微观核模型的更复杂计算。然而，这项工作确实证明了在解决该异常现象的过程中，更严谨地处理电子和中微子波函数具有明显的影响。”

Ravlić 预计核物理学家将开展相当多的后续研究，尤其是因为中微子不仅用于测试标准模型，而且对于大质量恒星爆炸等天体物理过程也至关重要。■ KPC

衰老

高龄女性通过举重获得显著益处

50至80岁的女性在学会硬拉 50 公斤后，变得更强壮、更苗条且行动能力更强。

在这项由台湾台北大学的 Juezhen Zhang 领导的研究中，32 名年龄在 52 至 84 岁之间的绝经后女性被随机分配参加力量训练计划或维持其通常的生活方式。

力量训练组的成员学习了如何使用六角杠铃（一种环绕身体且两侧带有配重的六边形框架）。在 24 周的时间里，这些女性每周进行两次由监督指导的训练，其中包括多组硬拉。

在研究过程中，力量训练组的女性减少了腰腹部的脂肪并增加了精瘦肌肉量，同时减轻了大腿和臀部的脂肪。她们的力量也得到了提升：能够连续五次硬拉的最大重量从平均 45 公斤增加到 58 公斤。

到研究结束时，这些女性在闭眼站立时身体晃动减少，且在 6 分钟内能行走更远的距离，这表明其平衡能力和行动能力有所提高。相比之下，被要求维持通常生活方式的人员没有看到任何改善（Experimental Gerontology, doi.org / rkgp）。

大多数针对老年人的力量训练计划都涉及使用健身器材。但 Zhang 的团队认为，举起自由重量（如杠铃和哑铃）可能更有益，因为这需要更多肌肉的激活与协调，包括对现实生活任务至关重要的核心稳定肌。

澳大利亚珀斯库廷大学的 Elissa Burton 表示，有兴趣首次进行力量训练的高龄女性应该在合格教练的指导下开始，以确保她们正确执行练习且不会受伤。Alice Klein

评论 神经多样性

暗示 ADHD 是个神话是伪科学且危险的 ADHD 是一种真实的疾病。卡罗琳·威廉姆斯表示，一部缺乏事实依据的纪录片如果暗示 是一个现代的社会构建，可能会造成伤害

如果我接受第四频道纪录片《伟大的 神话？》（The Great Myth?）的前提，那么我多年来一直生活在一种错误的信念之中。我被诊断出的所谓 （注意缺陷多动障碍）显然是一个随意的概念，是由大众共识发明并由《新科学家》等出版物推广的。

英国国家医疗服务体系（NHS）医生马克斯·彭伯顿（Max Pemberton）在 8 月 18 日于英国播出的一部纪录片中提出了这一观点。在随后的几天里，许多文章和专栏作家评论了为什么这一观点如此令人反感。虽然我对此有自己的看法——请相信我，我也感到被冒犯了——但最紧迫需要解决的问题是该节目暗示 可能并非一种真实疾病的观点。

其中一项主张是， 是一种新疾病，是为了强行将“方榫”塞进现代生活这个“圆孔”而臆造出来的。这根本不是事实。关于我们现在称之为 的早期医学记录可以追溯到 1798 年，当时来自英国皇家内科医师学会的亚历山大·克里奇顿（Alexander Crichton）描述了一些无法长时间集中注意力以完成任务的人，称他们患有“坐立不安症”（the fidgets）。

但本月发表在《柳叶刀》（The Lancet）上的一项分析结果

从那时起，这种疾病在医学上被多次描述和重新命名（谁能忘记 20 世纪初的“儿童道德控制异常缺陷”？）。虽然彭伯顿正确地指出近年来诊断率飙升，但英国的诊断率仍低于全球范围内的流行率估算值。对大规模人群症状进行筛查的研究一致显示，普通人群中 的患病率在儿童中约为 5%，在成年人中约为 2 到 3%。

MINNOW PRODUCTIONS

发现英格兰整体人口中仅有 1% 被诊断出患有此病。即使在最容易被诊断出的群体——男孩和年轻男性中，记录的患病率也分别仅为 2.9% 和 3.6%。如果非要说的话， 在这片土地上是被低估诊断的。

接着是关于 是社会构建而非神经发育状况的主张。这是彭伯顿在与伦敦国王学院的神经科学家卡蒂亚·鲁比亚（Katya Rubia）交谈后思考的问题。鲁比亚表示，虽然在群体层面有证据表明 患者与神经典型人群的大脑存在差异，但无法通过大脑扫描来诊断 。

当然，她是正确的。你无法通过大脑扫描诊断 。大多数（如果不是全部）精神病学和神经发育状况也是如此。其中大多数疾病与 一样，是通过评估症状及其对个人日常功能能力的影响来诊断的。“我的采访持续了数小时，而他们挑选了那些符合他们将 呈现为神话之目的片段，”鲁比亚告诉《新科学家》杂志。“这种

“我们有证据表明，无论是基于药物还是其他方式的治疗，都能提高生活质量和自尊心”

马克斯·彭伯顿， 一名国家医疗服务体系 医生，主持了该节目

像这样强推其将 视为虚构实体的节目是不负责任的，并将产生不利影响。”

彭伯顿随后尝试通过在线评估寻求 诊断，尽管他认为自己可能没有此病。在现实中， 的诊断是通过一个漫长的过程完成的，包括多份问卷和一次访谈。要获得诊断， 特征必须在两个或更多场景中引起显著问题，例如在工作中或在处理账单时。如果它不影响你的生活、心理健康或人际关系，你就不符合 中第二个 D（disorder，障碍）的定义。

执行诊断的人员身份并未公开。我们所知道的仅是彭伯顿显然患有 ADHD，并被开具了一份被描述为“基本上就是可卡因”的药物处方。

到此为止，关于药物影响大脑（尤其是儿童大脑）的恐慌宣传真正开始了。诚然，有一种观点认为我们需要更多关于 ADHD 长期用药效果的数据。我们目前还没有这些数据，部分原因是尽管有人声称这些药物具有成瘾性，但众所周知，ADHD 患者很难长期坚持服药。

而我们确实拥有的证据是，无论是否采用药物治疗，治疗都能提高生活质量和自尊心；而未诊断且未治疗的 ADHD 会增加心理和生理健康受损、药物滥用以及犯罪的风险。

希望所有这些都能给梅森的父母带来些许安慰。梅森是纪录片中的一个小男孩，他被停止用药，转而尝试瑜伽、自然漫步和零屏幕时间。这似乎起到了一定作用，但他的学业和行为却有所下降，因此他的父母无奈地让他恢复用药。

纪录片将此描绘成一次失败。但真正的失败在于节目制作方选择了惊悚手段而非科学准确的报道，这让梅森以及我们所有人都失望了。

Channel 4 拒绝置评，彭伯顿也未回应置评请求。

卡罗琳·威廉姆斯（Caroline Williams）是一位专注于神经科学、心理学和健康的科学作家。

每个阶段 都有储蓄

我们无法向您承诺绝对的宁静，但我们可以帮您获得内心的安宁。

ybs.co.uk / savings 欢迎光临我们的分行

10.6k+ 条评论

Which? 推荐供应商

★★★★★

Trustpilot

约克郡建筑协会 (YORKSHIRE BUILDING SOCIETY)

始于 1864 年

约克郡建筑协会是建筑协会的成员，获得审慎监管局的授权，并受金融行为监管局 (FCA) 和审慎监管局的监管。约克郡建筑协会已登记在金融服务登记册中，注册号为 106085。总部地址：Yorkshire House, Yorkshire Drive, Bradford BD5 8LJ. ybs.co.uk

YBM 13894 07 08 26

超过 50% 的配送现在采用了更少的包装。

29 30

$\frac{4}{2025}$

或者完全不添加包装。

亚马逊 (amazon)

格雷厄姆·劳顿 (Graham Lawton) 是《新科学家》杂志的顾问。他是《不能抱怨：日常小病的惊人科学》(Mustn't Grumble: The surprising science of everyday ailments) 一书的作者。您可以通过 @grahamlawton 关注他。

格雷厄姆的这一周

我正在阅读 伊恩·莱斯利 (Ian Leslie) 所著的《约翰与保罗：歌曲中的爱情故事》(John & Paul: A love story in songs)，这是一部关于列侬和麦卡特尼的精妙传记。

我正在观看 我妻子买了一台新缝纫机，所以我们正盯着《英国缝纫大赛》(The Great British Sewing Bee) 看。

我正在忙于 写作与半退休生活的常规结合。

我们的上限：人类究竟能活多久？最近的一项研究认为寿命可能接近两个世纪——但格雷厄姆·劳顿认为有理由对此持怀疑态度

几年前，我写过一篇令人难忘且颇具争议的故事，关于让·卡尔芒（Jeanne Calment），她当时——以及现在——仍是官方认可的世界长寿纪录保持者。她出生于 1875 年，于 1997 年去世，享年 122 岁 164 天。但根据一些质疑者的说法，以让·卡尔芒之名去世的女性实际上是她的女儿伊冯娜（Yvonne），后者在让于 1934 年去世时欺诈性地冒用了母亲的身份。伊冯娜比让小 23 岁。这依然是一个高寿的人生，但并非纪录级的。

此事至今尚未解决，因此，目前卡尔芒仍然是生物老年学中最重要的数据点之一。当我们询问“一个人能活多久？”时，答案是“至少 122 岁 164 天”——前提是让·卡尔芒确实活了那么久。

一个人是否能活得更久长期以来一直困扰着科学家。这是一个重要的问题，因为如果我们想要实现寿命延长，我们需要知道目标是什么。但这出乎意料地难以回答。卡尔芒是一个离群值：没有其他人达到 120 岁。寿命第二长的人是田中坚（Kane Tanaka），他活到了 119 岁 107 天，而目前世界上最年长的人刚刚满 117 岁。除了少数特例，

“在过去的一个世纪里，医学突破稳步增加了人们的预期寿命”

然而，每个人都同意存在一个自然极限。

但这个极限在哪里？处理这个问题有两种主要方法。一种是研究经过验证的超级百岁老人的人口统计数据。一项经常被引用的此类分析得出的估计值约为 115 岁，但并未排除有人能活到 125 岁的可能性。另一项研究则得出结论

官方记录显示，让·卡尔芒（图中为 1996 年照片）活到了 122 岁 164 天

认为卡尔芒的纪录很可能很快会被打破。

另一种方法是通过计算机建模。一个模型分析了 85 岁及以上成年人红细胞和白细胞计数的短期波动，研究人员将其转化为衡量他们对终末期年龄相关疾病抵抗力的指标。在某个时间点，这种抵抗力会完全崩溃，标志着不可逆转的临界点。研究人员得出结论，这因人而异，通常在 120 岁左右开始出现，而硬性上限为 150 岁。

但这些研究没有考虑到的一点是医学未来的进步。在过去的一个世纪里，医学突破稳步增加了人们的预期寿命，（在更好的营养和卫生条件的帮助下）推动长寿纪录不断上移。

如果这种趋势继续下去会怎样？为了找出答案，来自俄罗斯莫斯科的斯科尔科沃科学技术学院和人工智能研究学院的一个团队进行了一次思想实验。他们说，想象一下，如果我们能够治愈所有理论上可治愈的衰老原因。唯一无法修复的是细胞中 DNA 的损伤，这种损伤随着年龄的增长而累积。在这种情况下，人类生命的上限将飙升至 194 岁。

我敢大胆预言，这在我有生之年不会发生，而且我想指出，衰老的原因尚未被完全理解。即便如此，这表明仍有很大的提升空间。

但回到现实世界。为什么最高年龄的估计值差异如此之大？在报道让·卡尔芒的故事时，我采访了一位名叫索尔·纽曼（Saul Newman）的长寿研究员，他当时在堪培拉的澳大利亚国立大学工作。他刚刚发表了一篇论文，表明纸质记录中少量的错误或欺诈性条目就能显著扭曲长寿数据。

纽曼目前就职于牛津大学，他刚刚出版了一本书，名为《病态：揭秘现代长寿科学》（Morbid: Debunking modern longevity science），揭露了那些错误和欺诈行为令人震惊的程度。例如在希腊，人们发现大约 70% 的百岁老人实际上已经死亡，但在纸面上被维持“生存”状态，以便其亲属能够继续领取养老金。历史上，人们出于许多其他原因虚报年龄，比如为了参军（或逃避兵役）、结婚或找工作。文书错误也屡见不鲜，尤其是那些如今的百岁老人和超百岁老人出生的年代。

如果用于估算人类寿命上限的输入人口数据充斥着虚报的年龄，那么输出结果可能同样被夸大了。建模研究中使用的人员也被标记了年龄，而这些年龄可能是错误的。

在研究卡尔芒的故事时，我还读到一篇 2011 年的论文，该论文发现，在出生记录严谨且可追溯 150 年以上的国家中，没有一个人被证明活过了 115 岁。我认为，这是我们目前能得到的最佳估算值。一百年加上 15 年。我已经走了一半的路程。

SIPA / SHUTTERSTOCK

虚空时代 当我们将艺术作品输入 AI 模型时，我们面临着将文化历史输给自动生成作品的风险。安娜莉·纽维茨表示，我们必须保存人类的写作，以避免创造出这个虚空。

在 800 年后，一群学生正坐在一场专门探讨一个被称为“虚空时代”时期的研究生研讨会上。他们读了几本书，观看了一些曾经被称为“电影”的二维低分辨率媒体。几名学生甚至组装了设备来体验“电子游戏”。有人举起了手。

“我不明白为什么这个时代的保存媒体如此之少。所有东西都是碎片化的，而且我们在课上读到的大部分内容都是根据概率重建的。是因为原子战争吗？还是火之年？”

“没有任何战争或全球山火能造成这种结果，”教授严肃地回答道，“那个时代的人们将他们的文化喂给了大语言模型，然后丢弃了原始文本。我们现在剩下的只有古代算法的输出结果。我们必须从中反向推导当时人们在做什么。”

“那么，我们永远无法知道我们在这一课程中所学习的任何内容，是否真的是当时人们所知道、所看到或所相信的？”

“没错。这就是为什么我们称之为虚空时代的原因。”

这听起来很荒诞，但这就是我们现在正在为自己构建的未来。当我们把书籍、电影、游戏和艺术喂给生成式 AI 模型时，我们面临着用自动生成作品的大杂烩取代文化历史的风险，而这些作品对于那些渴望了解我们当时想法的后代来说毫无意义。这就像是我们故意在青铜时代历史的核心地带重新创造那个悲剧性的虚空——在那个时代，人类改变了他们的文明，却几乎没有留下任何书面叙述来解释发生了什么以及为什么发生。

作为一名科学新闻和科学幻想小说作家，我一直对这样一个想法感到震惊：使用大语言模型（LLM）竟然可以替代彼此之间写作、为彼此写作以及关于彼此写作的体验。写作、创造文化的意义，在于传达我对这个世界独特的观察方式。这是我写给你的、写给未来的信，是我记录的一只无毛类人猿在 21 世纪初这个非常特定的 era（时代）、在目前被称为旧金山的城市区域中的经历。这是一份人类的记录，是写给其他人类的。

ROSLAN RAHMAN / AFP VIA GETTY IMAGES

回到 2020 年，当时我因为完全不知道如何在这种全球大流行病中生存而感到恐慌，于是我翻开了萨缪尔·皮普斯（Samuel Pepys）那本有着 350 年历史的日记。在 1666 年，他是一个普通的英国人，正努力在瘟疫中生存下来。皮普斯担心全球政治和生命意义等宏大议题，但他也记录了在逃离伦敦时如何藏起一轮奶酪，以及走在那些门上被刷有红色隔离十字架的街道上是什么感觉。阅读他的文字让我感到宽慰，因为这就像是他跨越几个世纪向我伸出手，安慰我并非孤身一人。

当然，当我写关于时间旅行者和机器人的虚构故事时，我记录的并非真实生活的精确体验。但我依然在捕捉我这个时代的焦虑与希望。有些事情在虚构作品中可以说，但在新闻报道中不能说；我可以反映社会上的模棱两可之处和幻想，以及那些比事实更能揭示我们自身特质的偏见。这就是为什么我们要保存像乔治·艾略特（George Eliot）和 H. G. 威尔斯（H. G. Wells）这类作家的作品。他们捕捉了 19 世纪的抱负与执念，揭示了人类在面对迅速工业化的世界时是如何反应的。

我不反对在代码开发中使用生成式算法，也不反对将其作为分析大数据的辅助工具。但我们对人类写作的需求与对技术的需求一样迫切。如果我们真正想要理解宇宙，我们需要人类世界的记录，同样需要物理世界的记录。当我们把所有的写作内容都喂给 AI 时，我们面临的风险不仅仅是失去个别文学作品，我们还会失去在现在、过去和未来中彼此之间的联系。

安娜莉·纽维茨（Annalee Newitz）是一位科学记者和作家。其最新著作是《Automatic Noodle》。他们是《STARLOG》杂志的主编。您可以通过 @annaleen 关注他们，其个人网站为 techsploitation.com

安娜莉的一周

我正在阅读 记者厄普顿·辛克莱 1907 年创作的科幻末日喜剧《千年》（The Millennium）

我正在观看 大量关于气动管的视频

我正在忙于 整理我的藏书，将它们放入我那漂亮的新书架中

构建低碳农业未来的概念农场

在整个英国，农民在适应气候变化和削减排放的同时，面临着越来越大的粮食生产压力。但没有任何一项单一技术能够解决所有这些挑战。相反，农业的未来可能在于将多种创新相结合——并证明它们在真实的商业农场中能够协同工作。

这就是“特易购（Tesco）低碳概念农场”背后的理念。在这里，可持续技术正在为超市货架提供食物的实际耕作地中进行测试和试验。

特易购可持续农业与渔业负责人纳塔莉·史密斯（Natalie Smith）表示，我们需要多种技术方案至关重要。“每个农场和耕作系统都不同，因此没有任何一项单一技术能够独立实现向更具韧性的低碳农业转型。”

该项目与一家蔬菜农场和一家畜牧农场合作，采取“全农场方法”。特易购及其供应商和农民正与研究人员合作，将低碳肥料、无泥炭生长介质、精准作物监测和替代燃料等技术相结合，以观察它们在整个作物轮作周期中的协同表现。

初步结果令人鼓舞。在林肯郡的农产品和耕地概念农场中，特易购的马铃薯供应商 Branston 在使用循环经济低碳肥料、最低耕作以及由氢处理植物油驱动的农用车辆后，记录到碳排放量减少了 $50\%$，同时保持了产量和质量。随后，约 500 吨作物在特易购门店售出。西兰花的试验也达到了

“每个农场和耕作系统都不同，因此没有任何一项单一技术能够独立实现向更具韧性的低碳农业转型” 纳塔莉·史密斯（Natalie Smith）， 特易购可持续农业与渔业负责人

传统作物的水平，而其他试验则将燃料使用量降低了约 60%。

对于史密斯来说，证明这些技术在日常耕作中有效与创新本身一样重要。她说：“农民需要实际的证据和可控的风险水平。他们可以看到这些技术在与其自身规模相当的业务中运行，并查看实际的数字和数据。”

概念农场带来的经验（无论是积极的还是消极的）将在特易购的供应商网络中共享——该零售商是英国农业最大的客户——从而帮助农民在采用减排技术的同时，维持高效且盈利的业务。正如史密斯所言：“特易购低碳概念农场正在帮助证明，粮食安全与脱碳可以并行不悖。”

在 2026 年《新科学家》现场展（New Scientist Live 2026）的“食品与农业之未来”专题中了解更多信息 10 – 12 October / Excel London

专题合作伙伴：

ROTHAMSTED RESEARCH

MAP OF AG

□·BASF

UKCEH

TESCO

JENS CULLMANN / 年度野生动物摄影大赛 ANDREA AND MACEOGRAMMATICO / 年度野生动物摄影大赛 GIL WIZEN / 年度野生动物摄影大赛

快拍！

年度野生动物摄影大赛 自然历史博物馆

从带着嗜血“搭车客”的蝰蛇，到险些错过午餐的牛蛙，这些入围今年年度野生动物摄影大赛的照片，让我们得以窥见动物王国中上演的戏剧性瞬间。

最左侧的《反击》（Biting Back）是由安德里亚和马塞奥·格拉马蒂科（Andrea and Maceo Grammatico）拍摄的一张极具威慑力的照片，在比赛的“动物肖像”类别中获得高度赞赏。这对双胞胎表示，直到在回顾照片时，他们才发现一只蚊子正从尖头蝮（中美洲最危险的蛇类）的头部吸血。

在延斯·库尔曼（Jens Cullmann）的《瞬间逃脱》（Split-Second Escape）中出现了一次惊险的擦肩而过，该作品在“行为：两栖动物与爬行动物”类别中获得高度赞赏，位于左侧上方。这一瞬间几乎像这只青蛙的猎物——一只非洲巴布蓝蝶——一样短暂；更令人惊叹的是，库尔曼在一个月前发现了蝌蚪，再次回来时发现它们已变成了幼牛蛙。

隐藏而非狩猎是吉尔·威森（Gil Wizen）拍摄的地衣龙螽斯（位于左侧下方）的特长。威森的这张名为《隐龙》（Hidden Dragon）的照片在“行为：无脊椎动物”类别中获得高度赞赏，捕捉到了这只伪装近乎完美、体长15毫米且长满尖刺的昆虫——这是一种仅分布在厄瓜多尔安第斯山脉的稀有物种。

本次大赛的总冠军将于10月13日公布。年度野生动物摄影大赛由伦敦的自然历史博物馆组织，该馆将于10月16日至2027年7月11日期间举办这些作品的展览。

索菲·法罗

穿越时间

约书亚·豪格戈表示，如果你想大致了解关于时间的所有知识，那么这本优秀且内容极其充实的书就适合你。

书籍 《On Time》 吉姆·阿尔-哈利利 普林斯顿大学出版社 9月1日

我想知道你对时间持有怎样的看法？例如，你是否认为过去存在，而未来不存在？或许，如果被追问，你会说现在是真实的，但过去已经消逝，未来尚未到来。或者，你可能像阿尔伯特·爱因斯坦一样，认为所有时刻都同样真实，而我们只是在其中穿行。

如果你喜欢思考这类深奥的问题，那么物理学家兼广播员吉姆·阿尔-哈利利的最新著作可能适合你。不过，先给你一个提醒。《On Time: The physics that makes the universe tick》（时间：驱动宇宙运转的物理学）是一本如果你想大致了解关于时间的所有知识就应该阅读的书。但如果你只是想初步接触这个主题，那么有更简单的入门选择。

阿尔-哈利利首先概述了思考时间所涉及的许多交织在一起的线索，并将它们分为两类。首先是“显现时间”（manifest time）的问题，即我们所体验到的时间。这包括诸如：为什么我们感觉到时间的流动？为什么这种流动像河流一样感觉是连续的？以及过去和未来之间究竟有什么区别？

其次，我们面临的是“物理时间”的问题，这涉及物理学各个领域对时间的不同解读，以及试图解决这些解读所产生的矛盾的努力。例如，阿尔-哈利利写道，物理学中的某些方程将时间视为一个标签，而爱因斯坦的广义

“广义相对论告诉我们时间是一个维度，而热力学将其描绘成一支箭头。那么，时间究竟是什么？”

相对论告诉我们它是一个维度，而热力学将其描绘成一支箭头。那么，时间究竟是什么？我们如何调和所有这些矛盾的观点？物理时间的难题能否帮助我们解决显现时间的难题？

我发现书中理清这些问题的开篇部分极其有用。作为一名《新科学家》杂志的编辑，我在我的，呃，时间里，读过很多关于时间之谜的内容，但阿尔-哈利利以极高的技巧和清晰度将这一切阐述得淋漓尽致。

由此，本书揭示了我们对时间的认知，并循序渐进地处理各种问题。如果这听起来涵盖的内容过多，那么事实确实如此。而这正是我想要提醒注意的地方。一些关于物理学的科普书籍试图用诗意的神秘感来诱导读者，并且

关于时间的另外三项推荐

《寻找现在》（In Search of Now），作者：乔·马钱特（Jo Marchant）

前《新科学家》员工乔·马钱特探讨了当下时刻的矛盾之处。当下是唯一存在的东西，但所谓的“现在”究竟是什么？在本书中，她借鉴了广泛的领域以寻求答案。

《星际穿越》（Interstellar），导演：克里斯托弗·诺兰（Christopher Nolan）

如果你真的想体验震撼，可以尝试这部史诗级的科幻电影。一位父亲前往一个超大质量黑洞，在那里，时间膨胀意味着他的衰老速度比留在地球上的女儿慢。接下来的情节则更加离奇……

《时间的秩序》（The Order of Time），作者：卡洛·罗韦利（Carlo Rovelli）

这本书比吉姆·阿尔-哈利利的书更具诗意（且篇幅更短）。在书中，卡洛·罗韦利认为时间并非现实的基本属性，而是当我们从特定视角观察时产生的一种涌现现象。

现实的壮丽，让你在这一切的惊叹中随之起舞。这里确实有这种感觉——毫无疑问，这些都是深奥的问题。但阿尔-哈利利以出色的科学讲解能力著称，他在这里的重点是解释，而非娱乐和叙事驱动。他极力追求清晰、公正且有条理。有些人会很喜欢这一点（事实上，我大部分时间也是如此），但我仍然觉得这本书约 350 页的内容中，有些部分可以删减，以使故事推进得更有节奏感。

除此之外，书中有很多段落展现了阿尔-哈利利令人印象深刻的清晰思维。比如他在讨论时间与“变化”哪个更基本时，提出了一个简单且极具说服力的论点。他写道：“当然，如果我们没有体验到变化，我们就无法感知到时间的流逝。但是……为了让我们感知到某样东西没有在变化，我们需要感知到背景中时间的流逝，这样才能说某样东西在不同时间点保持不变。”

书中还提出了一些新观点，我们稍后会讨论，但其大部分价值在于阿尔-哈利利对过去 100 年物理学教会我们的关于时间的知识所做的阐释。

例如，广义相对论的一个推论是，不存在一个普适的“现在”，因为如果不同的观察者移动速度极快，他们对事件顺序的认知可能会产生分歧。这虽然令人眩晕，但阿尔-哈利利的描述是我读过的最清晰的表述之一。

他非常坦率地向我们提供了所有事实和相互竞争的论点，同时也给出了他的观点。例如，他认为我们对时间流逝的感知是一种错觉。此外，

ROGER VIOLLET VIA GETTY IMAGES

他并不准备认同物理学家中日益流行的观点，即时间本身是自然的一种涌现属性（比如温度），而非现实的一个基本组成部分（比如量子场）。

在物理学文献中，人们经常读到时间的方向源自热力学第二定律，该定律指出事物随时间推移趋于更加无序。但阿尔-哈利利向我们展示了这种解释或许过于简单，他认为时间之箭可能源自他所称的唯一真正不可逆的过程，即量子系统在受到环境干扰而失去其量子特性时的“退相干”。

对于所谓的“过去假设”，他也有一种新鲜且有趣的思考方式，但我担心他的论点过于复杂，无法在此概括。

令人欣喜的是，阿尔-哈利利还在书末加入了一个他称之为“奖励章节”的部分。

“令人欣喜的是，吉姆·阿尔-哈利利加入了一个‘奖励章节’，探讨时间旅行是否可能”

物理学家将日常的时间体验描述为“显现时间”

这本书探讨了时间旅行是否可能。他将这本书定位为“大餐后一份美味但略显奢侈的甜点”，并巧妙地执行了这一构思，在探讨这个狂野话题的趣味性与保持其作为一名严肃物理学家的严谨性之间取得了平衡。

除此之外，他还涵盖了著名的“祖父悖论”——即一名时间旅行者回到过去杀死了自己的祖父，从而引发关于他们最初如何能出生的疑问。但他深入且详尽地探讨了该悖论的几种变体。同样，他比大多数撰写时间旅行的人挖掘得更深，帮助我们以一种细致入微的方式去思考这个问题。

这本书涵盖的内容如此之多，以至于很难用一句话来概括——而且我怀疑每位读者从中获得的启发都将有所不同。但有一点是可以肯定的：如果你在寻找关于时间本质的简单答案，你在这里找不到——因为根本没有简单答案。

这里有一个想法。如果这其实没关系呢？在我早年从事物理报道的记者生涯中，我倾向于认为这门学科的任务就是解释现实。这没错，但肯定存在极限。在《On Time》的开头部分，阿尔-哈利利坦诚地告诉读者：“也许宇宙中有些方面是我们永远无法解释的，或者事实证明它们不需要解释，因为它们就那样存在着。”

读完这本书后，关于如何思考时间，我确实了解得多得多了。但我们能否完全理解它？相比翻开第一页之前，我现在对此更加怀疑。

Karmela Padavic-Callaghan 记者 纽约

在古希腊神话中，九位缪斯女神帮助人类在科学、文学和艺术领域达到顶峰。

在《发明之母：从希腊缪斯女神看创造力历史》（Mothers of Invention: A history of creativity from the Greek muses）一书中，民俗学家兼天体物理学家莫伊亚·麦克蒂尔（Moiya McTier）赋予了她们每一个人新的声音，让她们各自讲述几个世纪以来关于创造力和天才的故事。

通过历史缪斯克利俄（Clio）之口，麦克蒂尔提出了关于什么成就了一个天才、谁有权将天才之作归功于自己，以及这些观念在人工智能时代如何演变的问题。克利俄认为，人类在创造力方面存在问题，随后她开始讲述她每位姐妹的故事，试图解释我们是如何走到今天这一步的。

接下来的内容是一场在艺术和科学不同领域之间巧妙展开的畅游，其中穿插了以该领域守护缪斯之口写成的日记条目。这使得麦克蒂尔在展现自身创造力的同时，也能获得极大的乐趣。

香蕉是否即将面临崩溃？

我们热爱香蕉，但如果没有严肃的技术努力，这种作物将在全球范围内面临崩溃。詹姆斯·伍德福德（James Woodford）分享了一篇充满激情的恳求

SHUTTERSTOCK / MEANDERING MOMENTS

书籍

香蕉的未来

詹姆斯·戴尔，MHP出版社

十多年前，我和家人搬到了澳大利亚北昆士兰的一个热带城市汤斯维尔，在那里住了几年。可以说，这座城市拥有世界上最令人愉悦的冬季气候之一：连续几个月每天 25℃ (77°F)，天空万里无云，夜晚清爽。

对我而言，一个令我大开眼界的事实是，即使在郊区的后院里，也能种植极其丰富的蔬菜水果。我们种植并收获了木瓜，以及如此之多的冬季番茄，以至于我们不得不将后者一桶又一桶地冷冻起来。

但真正的多产之王位于我们四分之一英亩土地的后方：一群由朋友赠送的吸芽种植而成的香蕉树。

这些香蕉树迅速长成了高耸的巨兽，随后开始结出绝对巨大且奇特得惊人的糖香蕉串，在我伫立欣赏时，它们似乎还在不断长大。

我收获的第一串香蕉规模宏大。它大到我有一张照片，照片中的那串香蕉让当时 6 岁的女儿显得十分矮小。

从那时起，我就对香蕉产生了浓厚的兴趣，因此也被一本新书吸引，那就是由布里斯班昆士兰科技大学领先的香蕉研究员詹姆斯·戴尔（James Dale）所著的《香蕉的未来》（The Future of Bananas）。

这是一本短小精悍的书，我的阅读方式就像吃香蕉一样——在一次坐下阅读中，几口就把它“吃”完了。

几十年来，关于全球通用品种“卡文迪什”（Cavendish，学名 Musa acuminata）所面临威胁的警报一直在鸣响，包括在《新科学家》杂志上。

在世界上数百种奇特而美妙的香蕉品种中，卡文迪什可能不是最美味的，但它是 Musa 属中最普遍的——皮厚且口感好，外观美观且价格低廉。

戴尔回顾了卡文迪什迷人的历史，但就像所有曾经存在过的香蕉品种一样，真菌疾病一直紧随其后。为了应对包括镰刀菌（Fusarium）等多种致命真菌所展开的“猫鼠游戏”，人类花费了几个世纪的时间，试图帮助这种水果跑赢其病原体对手。

农民和科学家为保证香蕉供应所付出的巨大努力，证明了我们将其视为全球重要商品的重视程度——

香蕉种植园提供了全球热爱的水果

以及我们多么希望保护这种世界上最受欢迎的水果之一。

然而，现在看来，卡文迪什香蕉——由于无籽、克隆，因此无法通过自身繁殖来摆脱危险——如果没有严重的科学助力，将无法再跑赢 TR4 菌株的真菌疾病。

“卡文迪什的末日即将到来，”戴尔写道。他表示，目前没有合适的抗病菌株能够替代该品种，必须创造出一种能够应对气候变化和疾病所带来所有威胁的新型香蕉。

2024 年，戴尔及其同事获得了澳大利亚和新西兰监管机构的批准，商业化发布一种基因改造的卡文迪什香蕉，QCAV-4, 他们对其进行了工程化处理，使其能够抵抗 TR4 菌株。

但戴尔认为，现在我们需要走得更远，因为传统方法太慢，无法防止全球香蕉作物最终发生灾难性的崩溃。

这并非一个浪漫的田园故事，而是一个详细的、有时带有技术性的恳求，旨在让人们接受科学农业。讲述者是一位充满激情的专家，他对比香蕉的了解或许超过了任何其他人。

这本书的核心是一个充满激情的恳求，希望克服公众对基因工程的普遍不信任，利用技术不仅是为了拯救卡文迪什，而且是为了将其他香蕉品种——尝试一下红色和蓝色，以及它们不同的口味和健康益处——带到我们的水果盘中。

WESTEND61 GMBH / ALAMY

一个阴郁的未来

艾莉森·弗鲁德表示，对于即将登上大银幕的后末日小说《狗星》，我们的反应褒贬不一

艾莉森·弗鲁德（Alison Flood）是《新科学家》杂志的图书与文化负责人，并负责运营每月一次的《新科学家》读书会。

书籍 《狗星》（The Dog Stars） 彼得·海勒（Peter Heller） 著

我们九月份在读的书

《新科学家》读书会从七月份阅读的遥远未来的太空歌剧——克莱尔·诺斯（Claire North）的《缓慢之神》（Slow Gods），转向了八月份地球上一个近在咫尺的未来。彼得·海勒的《狗星》设定在一次流感大流行抹去了地球大部分人口之后，讲述了一名孤独的飞行员希格（Hig）、他的爱犬贾斯珀（Jasper）以及他们那位有些麻烦的邻居班格利（Bangley）的故事。如果这听起来像是彼得在借鉴 2020 年的真实事件，那他并没有：他在新冠疫情之前就写好了这本书，并且（正如他在为读书会写的一篇随笔中所述）在六年前世界在我们周围陷入停摆时，发现生活在模仿艺术，这让他感到极其诡异。

我必须先承认：我是彼得·海勒的忠实粉丝，但不知为何我之前没读过《狗星》。所以，当我发现雷德利·斯科特（Ridley Scott）即将将其改编成电影时，我迫不及待地想和我们的读书会一起阅读它。

我对自己的喜爱并不感到惊讶。这部小说结合了我所有最喜欢的东西：荒野生存、灾后世界、优美的文字——以及一只非常棒的狗。在彼得的随笔中，我非常好奇地了解到他是如何“冷启动”创作这部小说的——没有初步研究，没有计划：“就好像希格在十月之夜的营火另一侧，风吹动着火焰，在告诉我几年前他身上发生了什么。我在一种极度亢奋的状态下写作。我不断地被发生的事情所惊讶、震撼和摧毁。我的眼泪会滴在咖啡馆的桌子上。”在我们的采访中，彼得还告诉我他是如何决定影响其世界的末日性质的，以及看到他的角色出现在大银幕上是什么感觉——剧透预警：他流泪了。

“我喜欢即使在看似不可能的环境中依然存在希望的方式”

读书会成员的反应则更为分化。凯伦·沃伦（Karen Warren）认为“前半部分有点乏味”，但她“在接近结尾时爱不释手”。她在我们的 Discord 频道上写道：“我喜欢即使在看似不可能的环境中依然存在希望的方式。”

文字风格非常独特——跳跃且不连贯，反映了希格的心境。读者们就此原因展开了一场有趣的讨论。“我的理解是他与班格利生活在一起时失去了一些表达能力：没有太多的真正对话，没有太多的新鲜感，充满了沉默和重复，”warsow_sun 写道。“他的经历中暗示过发烧，这是否影响了他的语言，或者仅仅是因为‘行动派男人’试图效仿海明威，将描述尽可能地精简？”Alan_P 疑惑道。

总的来说，Alan 并不喜欢这本书。“我相信很多人会喜欢这本书——但对我来说，它对人性的看法太阴暗且不现实，”他写道。“让我困扰的是，这些故事将这样一种观点自然化：生存的唯一方式就是成为那种武装起来的、不可挑战的家长式人物，而正是这种人物在最初助长了暴力。”

与此同时，Dee55 则在思考科幻小说的本质。对我而言，科幻小说是一个宽广且包容的领域，涵盖了从像《狗星》这样科学元素较少的作品，到金·斯坦利·罗宾逊（Kim Stanley Robinson）的《红色火星》（Red Mars）这类硬科幻。不过，下个月，定义科幻小说将不再是读书会需要担心的事情……

您可以在 newscientist.com / bookclubextras 找到彼得·海勒（Peter Heller）的文章和访谈。

时空的边缘——钱达·普雷斯科德-温斯坦（Chanda Prescod-Weinstein）

九月对许多人来说是返校季，我期待着通过本月的阅读来提升我的大脑：天体物理学家钱达·普雷斯科德-温斯坦的《时空的边缘》（The Edge of Space-Time）。

这本书的副标题（相当精彩地）定为《粒子、诗歌与宇宙梦幻布吉》（Particles, poetry and the cosmic dream boogie），在这本书中，这位《新科学家》杂志的专栏作家致力于让理论物理学中较为抽象的概念变得更容易理解，并利用流行文化来帮助她解释时空的量子特性。

读书会成员可以阅读该书的节选，以及钱达撰写的一篇关于物理学重大问题为何如此重要的文章。这些内容将于 28 August 在 newscientist.com / bookclubextras 上线。我还将在九月中旬与她进行对话——请加入我们的 Discord 在线社区 bit.ly / 4ewuwyo，分享您的问题并讨论这本书。阅读愉快！

加入《新科学家》免费读书会 在 newscientist.com / bookclub 注册

NewScientist Live

最高可享 44% 优惠（对比当日票价）

2026年10月10日 – 11日 / 学校日：10月12日

伦敦 EXCEL 中心 + 在线参与

顶尖演讲嘉宾

Chris and Xand van Tulleken / Maddie Moate / Helen Sharman Alice Roberts / Tim Spector / Chris Lintott / Giles Yeo / Turi King / 吉姆·阿尔-哈利利

震撼人心的演讲

如何构建终身大脑健康 / 合成基因组 / 恐龙还活着！如果你的 AI 具有意识，你如何得知？ / 寻找地球之外的生命 / 如何生长一个超大质量黑洞 / 21 世纪的大脑 / 儿童如何在 AI 时代茁壮成长

展区体验

量子飞跃：探索量子比特 / 参观量子城市 / 利用核聚变能 会见人形机器人 / 飞行之未来 / 自然界的隐藏英雄 未来医院 / VR 过山车 / 高科技农业 / 肠道漫步

立即购票：

荣誉支持单位

Yakult

THE FUSION CLUSTER

FARMERS WEEKLY

伦敦国王学院 (KING'S College LONDON)

医学研究基金会 (Medical Research Foundation)

Technicians We make the difference

THE iPAPER

编辑精选

火星上有生命？只有样本返回任务才能告诉我们

8月1日，第 26 页

来自美国弗吉尼亚州威廉斯堡的 Ed Prior：

你们对天体生物学家 Steven Benner 书籍的评论非常出色，该书重新分析了 1977 年海盗号探测器的测量数据，这些数据似乎发现了火星生命的生物学证据。这让我想起了当时的争议，以及航天领域许多人似乎不愿再次尝试此类项目的心态。在随后的几年里，NASA 曾两次宣布火星任务，旨在将样本带回地球以检测任何生物学迹象，但遗憾的是这两次任务都被取消了。

Benner 详细研究了数据并得出结论，火星“再次焕发生机”。我希望 2031 年中国的样本返回任务能够成功（尽管我更希望 NASA 在 2031 年之前再次尝试），这样我们才能最终且谨慎地——在不让未知生物分子逃逸到我们大气层中的情况下——回答这个问题：“火星上有生命吗？”

无论答案如何，我深感遗憾的是，海盗号任务背后许多才华横溢的工程师、科学家和技术人员，包括 Gil Levin 和 Gerry Soffen，已不在人世，无法看到这个问题的解决。

肯尼斯·克雷克的短暂生命充满了才华

8月8日，第 30 页

来自英国伦敦的 Peter Brooker：

在你们关于开发能够理解周围世界的 AI 的文章中，提到了认知科学、建模和风险分析的先驱之一肯尼斯·克雷克 (Kenneth Craik)，这太棒了。在仅持续几年的职业生涯中，他做出了巨大的贡献。考虑到他的研究工作，克雷克的去世既悲剧又平凡，且带有讽刺意味。当时他正骑车行驶在英国剑桥的国王游行道 (King's Parade) 上，一辆静止车辆的乘客打开车门，将他撞向了一辆卡车，导致其受致命伤。

气候危机需要全球经济变革

社论，8月8日

来自英国北约克郡皮克林的 Peter Brown：

气候变化的根本原因无疑是已经占据主导地位的市场体系。它创造了现代文明，但没有考虑到地球的资源、能源开发带来的破坏，以及产品生命周期结束后的去向。我们必须修改我们的经济模式，使其造福人类和我们的星球。市场是一个全球系统，因此需要全球性的努力来改变它。我们需要对所有破坏性活动设置抑制机制，并为减轻不利影响以及尝试扭转破坏提供补贴。

异质宇宙引发的疑问

8月8日，第19页

来自英国利物浦的约翰·华莱士（John Wallace）：我很喜欢利亚·克雷恩（Leah Crane）对宇宙学原理的概述——即认为我们的宇宙在极大规模上观察时大致是均匀的。尤其是她承认自己并不真正相信这一点：“我们可以接受某件事在接近真相的程度上足以产生实用价值，同时深知它并非完全正确。”

我有时在想（或许有人能对此予以澄清），在计算宇宙年龄时是否也使用了类似的假设。我认为，在足够大的尺度上观察，我们将宇宙想象成类似于一个由超星团组成的3D网络，大部分物质集中在其中，

因此引力在减缓时间的流逝，而其间则存在巨大的空洞。在几乎没有物质的地方，时间运行得会更快。当宇宙学家声称大爆炸发生在约138亿年前时，他们使用的是谁的时间？

50岁及以后应该如何饮食？

8月1日，第30页

来自英国伦敦的戴维·乔克尔森（David Jockelson）：在关于中年以后饮食的报道中，乔安·曼森（JoAnn Manson）表示，“你不能通过给不健康的生活方式加上多种维生素来获得最佳健康”。但她自己的COSMOS-Web试验发现了多种维生素带来的显著益处，同一期（第18页）格雷厄姆·劳顿（Graham Lawton）的文章也证实了这一点。

然而，参与者是自选的，他们既对自己的健康足够感兴趣而愿意成为志愿者，又能承诺参加一项长期的试验。他们可能已经在摄入相当不错的饮食。因此，真正的问题在于，饮食习惯糟糕的人是否会比试验中那些营养相对充足的群体从多种维生素中获益更多。

来自英国伦敦的罗宾·雅各布（Robin Jacob）：“按年龄饮食”中的见解在威廉·莎士比亚的戏剧《无事生非》中就已预见：“人年轻时喜食之物，年老时则不能忍受”。

石器时代的鱼类运输充满潜在惊喜

7月25日，第7页

来自加拿大维多利亚大学的 T. E. Reimchen： 有报告称，在现今的挪威地区，古代狩猎采集者将鱼类运输到

想要联系我们？

请将信件发送至 letters@newscientist.com；条款详见 newscientist.com / letters。寄至 New Scientist, 9 Derry Street, London, W8 5HY 的信件将会延迟收到。

没有鱼的山区湖泊中，从而创造了夏季捕鱼的机会，这些报告很有趣且很重要。但这并不是鱼类到达那里的唯一解释。

在加拿大西部的海达瓜伊群岛（Haida Gwaii archipelago），一种名为红喉潜鸟的鸟类在许多没有鱼的湖泊中筑巢，它们在相邻的海洋栖息地为幼鸟觅食，并将鱼（其中一些仍然活着）带回湖中喂养。这种情况在鸟类出巢前的50天预备期内每天都会发生，每对筑巢的鸟类在每个繁殖季节会转移500到1000条鱼。

挪威恰好也有大量的红喉潜鸟，因此这些湖泊中的鱼类定植是通过这些食鱼鸟类的转移而实现的，这并非不可能。我怀疑人类运输是更可能的解释，但这些潜鸟非常普遍的鱼类转移行为是不容忽视的。通过对周围鱼类种群使用基因组技术，或许可以查明鱼类究竟是如何进入湖泊的。

来自英国伦敦的格雷厄姆·帕金森（Graham Parkinson）：

在读到关于石器时代挪威鳟鱼养殖场的报道后，我兴奋地意识到，我从鱼贩那里买鱼回家的路程，就是一个当代“人类介导的鱼类易位”案例。我原以为我只是在买晚餐。

记录在案

猪眼在死后最多 10 小时仍能在体外保持活性（7月18日，第11页）。查尔斯·达尔文的著作《物种起源》出版于 1859 年（8月15日，第15页）。关于新电池技术章节的第一封信是由克里斯托弗·杰索普（Christopher Jessop）撰写的（信函，8月15日）。

戳破现实的漏洞

无理数长期以来一直是量子力学的基础。卡梅拉·帕达维奇-卡拉汉（Karmela Padavic-Callaghan）发现，移除它们可能会消除量子力学最诡异的特性

确切的故事版本虽有争议，但无论哪个版本，结局都十分凄惨。一名男子发现了一个新的数学真理，并为此失去了生命。在每一个版本的叙述中，古希腊哲学家梅塔庞图姆的希帕索斯（Hippasus of Metapontum）都淹死在海中，要么是被神惩罚，要么是被同僚惩罚。他的罪名是什么？在大多数讲述中，是因为他发现有些数字无法用精确的分数来表示，其小数位在不重复的情况下无限延伸。这些无理数——例如 2 的平方根或 pi——极大地扰乱了希帕索斯的同僚，即具有教团色彩的毕达哥拉斯学派，他们可能因为希帕索斯揭露了这些数字而将其扔进大海。

2000 多年之后，无理数被嵌入到量子力学的数学基础之中。撰写该理论早期公式的物理学家们不必面对波涛汹涌的大海，但他们的同僚称他们的工作令人震惊，近乎异端。一个世纪以来，我们已经与无理数达成了和解，尽管包含它们的量子领域极其诡异，其中存在着如果我们不观察它们可能就不存在的粒子，或者能将宇宙两端的两个粒子联系在一起的纠缠——阿尔伯特·爱因斯坦将其称为“幽灵般的超距作用”。

但一位物理学家表示，情况不必如此。“没有人对[量子力学]中发生的事情完全满意，但大多数物理学家会说：‘听着，我使用这个理论是因为它能预测正确的结果’，”牛津大学的蒂姆·帕尔默（Tim Palmer）说道。在一项大胆的新提案中，他起草了一种摆脱这种不满的方法：将导致希帕索斯丧命的无理数驱逐出去。

帕尔默认为，这个新理论将不再有任何古怪之处——纠缠及其幽灵特性将不复存在，允许两种矛盾可能性同时存在的叠加态之奇异性也将消失。至关重要的是，他的有理量子力学（RaQM）将给量子计算机的能力设定一个根本性的限制，这意味着它不仅可以在短期内进行测试，还可能危及一个价值十亿美元的产业。事实上，如果帕尔默的提案正确，量子计算机最终将无法运行著名的秀尔算法（Shor's algorithm）——这是一种用于分解极大数、能够破解广泛使用的加密技术的量子方案——这将打破自 1990 年代以来激励并证明其开发合理性的数十年梦想。

消除诡谲

“这无疑会对量子计算产生重大影响，但与它对物理学产生的影响相比则显得微不足道，而后者将成为自量子力学诞生以来最大的革命，”澳大利亚格里菲斯大学的霍华德·怀斯曼（Howard Wiseman）说道。

需要明确的是，帕尔默并非毕达哥拉斯追随者。就其本身而言，他对无理数没有异议。在他看来，量子力学故事中真正的数学反派是约翰·冯·诺依曼（John von Neumann），因为是他制定了量子力学第一个严谨的框架。量子力学课程至今仍在教授冯·诺依曼的数学理论，首先从希尔伯特空间（Hilbert space）开始，这是所有量子活动发生的数学领域。

为了分析电子等量子客体的行为，或预测其未来的行为，物理学家会为其分配一个存在于希尔伯特空间中的数学对象。在一个三维希尔伯特空间中，你可以将其想象成一个箭头，指向由分别代表深度、高度和长度的三个数字所指定的点。你在希尔伯特空间中使用的坐标系取决于你想要测量的内容——寻找电子的自旋或位置需要不同的坐标系。一旦选定一个，你就可以画一个箭头，并利用其长度来计算测量的概率。传统上，希尔伯特空间中任何东西都是允许的：欢迎各种各样的箭头。这是因为希尔伯特空间是连续的。在这方面它类似于数轴——放大任何一个点，你总能发现更多的数字。但帕尔默认为，这种对所有数字的包容正是量子力学问题的根源。

在 RaQM 中，并非所有箭头都被允许，这在它们所占据的抽象空间中留下了数学裂缝。任何长度平方为无理数的箭头都被绝对禁止。这就像你在数轴上寻找 2 的平方根，放大它通常所在的 1 和 1.5 之间的空间，却发现那里只有一个缺口。“自然界厌恶连续体，”帕尔默说道。

常规的量子力学取得了惊人的成功，经受住了一个世纪严苛的实验测试。重要的是，帕尔默表示 RaQM 与这些发现是一致的。 的筛网如此之细，以至于在所有这些实验中，它会被误认为是传统量子力学中那个无孔的实心容器。但接受 所带来的好处是惊人且多方面的——并且具有深远的影响。

首先， 不会受到传统量子力学所带来的怪异现象的困扰。为了证明这一点，帕尔默将重点放在贝尔测试（Bell test）上。该测试以物理学家约翰·斯图尔特·贝尔（John Stewart Bell）的名字命名，是一个关于通过量子纠缠连接的粒子的著名实验。想象一对对重复产生的纠缠粒子，每对中的一个粒子被发送给名为爱丽丝（Alice）的实验员，而其配对粒子则发送给远方的实验员鲍勃（Bob）。

爱丽丝和鲍勃独立选择测量其粒子的某项属性。爱丽丝可能会检查她的第一个粒子相对于垂直方向是“自旋向上”还是“自旋向下”。对于下一个粒子，她可能会测量相对于水平方向的自旋，依此类推。与此同时，鲍勃也在独立选择如何测量他的粒子，但两人之间没有沟通。只有在实验结束时，他们才会将各自的数据代入贝尔在 20 世纪 60 年代推导出的所谓不等式方程中，该方程用于测试爱丽丝和鲍勃测量结果之间的相关性。

有些相关性可能是偶然产生的，但贝尔确定了一个阈值，用以证明粒子是以非定域方式相关联的，在某种程度上跨越长距离相互影响。数十年的实验（包括那些获得 2022 年诺贝尔物理学奖的实验）一致发现，量子效应在极远距离上依然存在。但如何理解这种明显的非定域连接仍然是一个开放性问题。

量子纠缠是量子力学最奇异的特性之一

标准量子力学的一些解释简单地接受了非定域性。另一些解释则走得更远，认为某种物理定律预先决定了测量结果——表观上的相关性并非源于某种“幽灵般”的现象，而是对爱丽丝（Alice）和鲍勃（Bob）自由意志的限制。根据某些解释，贝尔测试中最奇异的部分在于，粒子仅拥有我们所测量的属性——如果爱丽丝测量垂直自旋，那么在她观察之前，她的粒子在水平方向上的自旋并没有一个确定的值。帕尔默（Palmer）的理论则打破了这一切。

在 RaQM 中，自由意志依然完好，但仅仅是某些测量无法实现。放弃连续体而采用一种更像网格的图景，似乎使得一些被禁止的测量掉进了现实的数学裂缝中。

帕尔默表示，对贝尔测试的仔细分析表明，这些缺失的测量是那些在反事实情况下产生的——也就是说，它们是爱丽丝本可以进行但实际上没有进行的测量。“对贝尔测试的整个解释彻底地、彻底地、彻底地改变了。你不再需要任何非定域性，也不再需要那种奇怪的‘幽灵般的远距离作用’。”在他看来，仔细重新思考哪些量子态在理论上是可能的，可以解决各种古怪的量子场景，包括埃尔温·薛定谔（Erwin Schrödinger）著名的思想实验——其中一只猫在未被观察时处于一种生与死两种状态的奇异混合之中。

在 RaQM 中放弃无理数和希尔伯特空间的连续体，对于量子计算也具有实际意义。在量子计算机内部，信息被编码在被称为量子比特（qubits）的对象的量子态中。为了在量子计算机中操纵信息，你需要让量子比特相互作用或使它们纠缠。量子计算的这些步骤随后可用于运行所有传统计算机都无法处理的计算。

重要的是，数学家已经证明，只有极少数问题只能由量子计算机解决。

“这将是自量子力学诞生以来物理学领域最大的革命”

其中一个算法尤为突出：由数学家彼得·肖尔（Peter Shor）在 1994. 年首次开发的用于分解大质数的算法。由于大多数现代加密密钥都基于质因数分解的数学原理，一台能够运行肖尔算法的量子计算机将成为一台威力惊人的解密机器。这就是为什么全球的政府、公司和研究人员都在为“Q日”（Q-Day）做准备——在那一天，量子计算机将击溃目前保护我们数字通信和交易的所有防御体系。

然而，研究人员的最佳估计表明，这种设备需要大约 500,000 个量子比特，比目前最大的量子计算机多出数百倍。目前的量子比特还存在噪声且易出错，因此不足以运行复杂的计算。构建一台能够运行肖尔算法的量子计算机是一项艰巨的挑战，人们正投入巨大的资源来克服这一难题。但如果我们的物理现实与有理量子力学（RaQM）相符，那么这一切都将是徒劳。

在 RaQM 中，量子比特的信息容量是有限的，因为它们的量子状态遵循一种新的数学限制。如果将量子比特量子状态中可以编码的最大信息量表示为由 0 和 1 组成的字符串，那么这个比特串将极其漫长，但仍然是有限的。

LEONARDO GERZON / NURPHOTO VIA GETTY IMAGES

如果有理量子力学正确，它将限制量子计算机的威力

帕尔默（Palmer）估计了 RaQM 所允许的最大长度，这个数字绝对巨大：$2^{400}$，比宇宙中所有原子的估计数量还要大 10 billion 亿亿倍。这个数字也可能让一台拥有 400 个以上完美无错量子比特的量子计算机——这仍然比我们现在拥有的多得多——陷入困境。运行像肖尔算法这样基于传统量子力学的算法，需要能够包含比具有 $2^{400}$ 个条目的比特串更多信息的量子状态。由于这超过了 RaQM 中可能的最大字符串长度，帕尔默预测这种量子计算机将会失败。

因此，将不会有大数被分解，也不会有加密被破解。但如果未来的量子计算机确实能做到这一点，RaQM 将化为乌有。“如果他们能分解某个大数，那么我的理论就是错的，我很乐意放弃并去做其他事情，”帕尔默说。但他对此充满信心。他说，像贝尔测试（Bell test）这类长期谜团的解决，让他确信自己走在正确的道路上。

这些巨大的赌注使得人们必须对 RaQM 这一解释世界量子基础的理论进行仔细审查。两个问题变得不可回避。第一个问题是，为什么希尔伯特空间（Hilbert space）会被替换成某种从一开始就不是连续的东西。

密歇根州立大学的斯蒂芬·徐（Stephen Hsu）表示，物理学家可能习惯于处理连续而非离散的理论——即平滑的空间而非像筛子一样的空间——但在数学家和信息论专家之间，这一点存在更多争议。他说，有非常有力的证据表明，依赖于连续统的数学方面并不像建立在有限离散部分之上的数学那样严谨。

这是因为连续数字不可避免地会导致无穷大。徐表示，一些数学家认为无穷大是对其学科合理基础的威胁，而且关于无穷大是否能具有物理意义也存在争议。“如果我需要无限个不重复的数字来表示这个 [无理] 数……我不确定什么样的公理系统实际上包含这个数字，以及你如何生成它，因为你无法通过有限的步骤来生成它。”

第二个问题是，为什么是无理数而非其他集合，会成为掉进帕尔默（Palmer）数学空间缝隙中的对象。尽管帕尔默专注于移除无理数，但他并非凭空地想出离散化这个想法。2005, Hsu及其同事认为，如果时空是离散的，那么没有任何实验能够排除类似离散希尔伯特空间（discrete Hilbert space）这种可能性的存在。他们对此类空间的分析使他们得出了与帕尔默相似的结论，包括量子计算机将永远不具备运行秀尔算法（Shor's algorithm）的能力。

现实检验

这增加了 RaQM 理论的分量，但同时也带来了诸多注意事项。仅仅因为不能排除量子理论存在一种网格状数学空间的可能性，并不意味着它一定就是这个样子。那么时空又如何呢？它真的可能是离散的吗？这里需要记住的关键量是一个被称为“普朗克长度”的数值。我们所有最优秀的物理理论在面对比这个长度更小的物体时都会失效——它比质子还要小约 1000 亿加元 billion 倍——这导致一些物理学家认为，普朗克长度就像是宇宙的像素大小。

这种情况是否属实，直到物理学家制定出一套能成功将量子物理（微观万物的法则）与引力（主宰宏观宇宙的法则）结合起来的理论之前，都将保持不明确。“时空的离散化仅仅是一个假设，因为我们还没有一套量子引力理论，”怀斯曼（Wiseman）说道。

但帕尔默（Palmer）认为，在决定量子客体现实状态的数学空间中，引力应当发挥作用。他将这一假设直接构建在 RaQM 之中。量子客体（如量子比特）具有脆弱性，也就是说它们很容易失去其量子特性。一个处于两种互斥状态混合中的量子比特，如果被单独放置，最终将坍缩为其中一种状态，从而失去其独特的量子特性。这种坍缩背后的确切机制存在激烈争议，但在 20 世纪 80 年代，匈牙利维格纳物理研究中心的拉约什·迪奥西（Lajos Diósi）和牛津大学的罗杰·彭罗斯（Roger Penrose）提出，这可能是由引力引起的。帕尔默将这一观点应用于他具有有限信息容量的量子比特，而那

詹姆斯·马奇（James March）的画作唤起了人们对希尔伯特空间的想象，那是量子波函数的抽象家园

“一个世纪以来，我们对量子物理的思考是否完全错了？”

400 个完美量子比特的关键数量出现了，这激发了他关于量子计算极限的假设。“这是一个合理的假设。也许是一个并非所有人都会认同的大胆假设，但这是一个理性的论点，”瑞士日内瓦大学的尼古拉斯·吉辛（Nicolas Gisin）表示。

对于吉辛来说，引入引力并没有让 RaQM 变得更清晰，反而提出了更多问题。例如，他指出，量子比特可以用无质量粒子组成，而引力仅影响有质量的物体。他也不相信 RaQM 能将量子物理从贝尔测试所揭示的非定域“幽灵般”特性中完全拯救出来。在吉辛看来，如果未来量子计算机和秀尔算法（Shor's algorithm）的实验证实量子理论需要的是离散空间而非连续的希尔伯特空间，那么还需要进一步的研究来证明，其间隙恰好是标准量子力学之怪异之处可以“掉入”的孔洞。

怀斯曼（Wiseman）表示， 若要取代量子力学，必须解决比量子比特多得多的问题。而目前它还没有做到。他说，应该有一套系统，让 能够让我们解释所有可能存在的事物，包括宏观物体。这将是对我们思考现实方式的一次真正重塑。总的来说，怀斯曼认为帕尔默（Palmer）的提议过于激进。

西班牙塞维利亚大学的阿丹·卡贝洛（Adán Cabello）同样对此感到担忧。传统的量子力学建立在方程式和公式之上，而这些公式仅在物理学家所熟知且钟爱的希尔伯特空间中才有效。例如，量子客体的动量和位置必须满足一种非常特定的关系，卡贝洛认为，一旦量子态不再能由无限的比特串表示，这种关系就会崩溃。

“这种关系具有许多经过实验验证的结果，包括海森堡不确定性原理和原子的量子化能级，所有这些都经过了极高精度的测试，”卡贝洛说。如果有人明天就造出一台能通过运行秀尔算法来测试 的量子计算机，卡贝洛会赌标准量子力学胜出。

就帕尔默而言，他并不担心其理论的基本规则或公理。他表示，与拥有多个基本公理的传统量子理论不同， 只有一个公理，其他一切都由此衍生。他说，这是 的另一个巨大优势。

尽管存在这种分歧，卡贝洛仍支持帕尔默的工作。“有无数个理论看起来近似于量子理论，但它们并不成熟，且无法做出可进行实验验证的预测，”卡贝洛说。 的前景在于它提出了一项具体的实验测试，而且巧合的是，相关的设备已经在构建中。卡贝洛认为，一些更复杂的贝尔测试实验或许也能表明，量子物理究竟需要物理学家捍卫了一个世纪的希尔伯特空间，还是实际上遵循 的“有孔”规则。

吉辛也有类似的复杂心情。他毫不避讳地称帕尔默的一些想法很狂野，且不会赌 能通过实验测试。但他说：“我认为这是优秀的物理学，因为它是可以测试的。”

今年 6 月，就在帕尔默概述 的论文发表几个月后，量子计算公司 QuEra 宣布计划在 2029. 年前构建一台拥有 1000 个以上无误差量子比特的设备。其他几家公司也提出了类似的时间表，构建一台真正强大的量子计算机的竞赛正在升温。在终点线处，等待着一个最根本问题的答案：一个世纪以来，我们对量子物理的思考是否完全错了？

欲了解更多关于量子计算和秀尔算法的内容，请参阅第 40 页

卡梅拉·帕达维奇-卡拉汉（Karmela Padavic-Callaghan）是《新科学家》杂志的记者，专注于物理学、材料科学和量子技术。

《新科学家》 订阅

半价优惠：一年仅需 £99

在这个夏天释放你的好奇心，通过无限量的数字访问权限，随时随地探索最新的科学理念。

立即订阅：newscientist.com / summersale

欲通过电话订购，请致电 +44 (0) 330 333 9470，并报号 22393。这些订阅为自动续订，可随时取消。我们在首次付款后提供 14 天的冷静期。优惠将于 2026 年 10 月 06 日结束。

作为一名工作过度医学院学生，马克·霍罗维茨（Mark Horowitz）在 20 年前开始服用一种抗抑郁药，以减轻他对未来的焦虑感。这种名为艾司西酞普兰（escitalopram）的药片起到了作用，在接下来的 10 年里，他每天服用该药，并未对此过多思考。在读到一篇暗示抗抑郁药的效果往往会随时间而减弱的文章后，他决定停止用药，并在几个月内小心地逐渐减量。即便如此，他说，他的生活还是“爆炸了”。

他开始出现恐慌发作、恐惧、眩晕和失眠。周围的环境感觉像“梦境”一样，仿佛他正在失去与现实的联系。这种戒断反应比促使他开始服药的那种虽然困难但可控的抑郁症要糟糕得多。霍罗维茨最终恢复了用药——并非因为他认为药物在帮助他，而是因为他发现戒断症状实在难以忍受。

他并非孤例。艾司西酞普兰和其他选择性血清素再摄取抑制剂（SSRIs）是常见的药物，被宣传为一种副作用极小且能缓解抑郁症的安全方式。然而，越来越多的人开始公开讲述在尝试减量或停药时遇到的问题。这些担忧引发了更深层的问题：我们是否在抗抑郁药的无害性方面受到了误导，并且过度依赖它们来治疗抑郁、焦虑和其他心理健康挑战。

霍罗维茨现在是英格兰国家医疗服务体系（NHS）的一名精神病学研究员，也是反对当前抗抑郁药治疗方法的主要发声者。包括他自己在内的一系列近期研究支持了他的经历，而这些日益增多的证据终于产生了影响，改变了我们治疗抑郁症的方式。

在 SSRIs 被批准用于医疗用途的近四十年里，它们已深深植根于日常生活中。根据 1 月份发布的一项人口普查数据分析，美国现在有超过 16% 的人服用抗抑郁药（其中大多数是 ），英国约有 15% 的居民以及 14% 的澳大利亚人服用此类药物。

并非第一代抗抑郁药，但它们之所以成为重磅药物，正是因为它们比其他获批治疗抑郁症的药物更安全且副作用更少。早期的抗抑郁药（如三环类药物）虽然治疗效果显著，但伴随着复杂风险，如心律不齐和震颤，且在高剂量下可能产生毒性。相比之下， 引起的副作用较轻，包括恶心、睡眠障碍和性功能障碍。由于风险较低，它们迅速成为了治疗抑郁和焦虑的首选方案。

支持使用这些药物是有充分理由的。去年一项针对英国近 20,000 名服用 人员的调查发现，近四分之三的人认为这些药物有用，尽管这一数字可能因安慰剂效应而有所虚高。尽管如此，一项针对 670,000 多名抑郁症患者的 2024 年分析发现，这些药物具有轻微到中度的疗效。

陷阱：迈克·马里亚尼（Mike Mariani）发现，长期使用 后的戒断症状似乎比我们想象的要严重得多。

“一名男子发现，戒断反应比促使他服药的抑郁症要糟糕得多”

长期服用之困

但由于 SSRI 被认为比早期的抗抑郁药安全得多，临床医生和服用者在监测其使用情况方面，并不像对待前代药物那样严格。结果，SSRI 在过去几十年中被广泛开具，但关于如何或何时重新评估处方的建议却寥寥无几。2022 年的一项研究审查了包括美国、英国、澳大利亚和加拿大在内的各国国家卫生当局发布的临床实践指南，发现没有一项指南就如何减量或管理戒断症状提供具体指令。

这意味着，当人们开始服用抗抑郁药时，他们往往会长期服用——无论他们是否打算如此。去年发表的一项研究显示，美国人目前服用 SSRI 的中位时长为 5 年，其中超过四分之一的人服用时间长达十年或更久。

然而，这种长期使用尚未得到彻底研究。抗抑郁药临床试验的平均时长仅为 8 周，这意味着我们对其副作用的临床认识也仅限于这一时间范围。“不幸的是，仍有很多问题尚未解决，”德克萨斯 A&M 大学精神科医生兼研究员萨尔瓦多·金霍安（Salvador Guinjoan）表示，目前还没有强有力的证据表明长期使用如何影响疗效和戒断症状。一些研究人员认为，服用药物的时间越长，出现副作用的风险就越大。

“服用抗抑郁药的时间越长，出现戒断症状的可能性就越大，症状就越严重，停止用药就越困难，”伦敦大学学院研究抗抑郁药的临床心理学家约书亚·巴克曼（Joshua Buckman）说道。

像霍罗维茨这样的人发现，当他们尝试减量时，症状可能会非常剧烈。这些心理困扰经常被误认为是原病症的体现——这种现象

WESTEND61 GMBH / ALAMY

“服用抗抑郁药的时间越长，停止用药就越困难”

被称为复发或再发——并被用作恢复用药或增加剂量的理由。“医生对戒断反应的忽视往往导致患者被‘煤气灯操纵’（gaslighting），或将症状误诊为复发，从而导致治疗时间延长，”哈佛大学神经科学家、美国国家精神卫生研究所前任主任史蒂夫·海曼（Steve Hyman）表示。

霍罗维茨知道自己的戒断症状并非抑郁症复发。毕竟，在服用该药之前，他从未经历过失眠、恐慌发作或感觉自己生活在梦境中——而这些正是他在停止服用艾司西酞普兰后经历的症状。

现在有越来越多的证据可以证实他的经历。去年，霍罗维茨共同撰写了一项研究，调查了 300 多名停止服用（或尝试停止服用）抗抑郁药的人员，发现近五分之四的人报告了某种类型的戒断副作用，其中 45% 的人经历了中度或严重的困难。该研究重点关注戒断症状的身体体征，如头痛、头晕、电击感（通常被称为“脑电击”）和眩晕，以将戒断症状与抑郁症复发区分开来。

“关于这些药物引起中度至严重戒断反应的普遍程度，存在广泛不同的观点，”英国布里斯托大学精神科医生大卫·凯斯勒（David Kessler）表示，但许多人认为这种情况比此前承认的“要普遍得多”。

被忽视的风险

一项 2024 年的分析研究了 16,000 多名停止服用各种抗抑郁药的人员，其中包括 SSRI、SNRI（血清素-去甲肾上腺素再摄取抑制剂，其用途与 SSRI 类似）以及三环类等早期药物。研究人员估计，大约 15% 的人出现了戒断症状，其中两种 SNRI 药物最常引起此类影响，约 1 in 35 的人经历了严重症状。

由 Horowitz 共同撰写并于同年发表的另一篇论文对 1148 名尝试停止服用抗抑郁药的人员进行了调查，发现 43% 的人戒断症状持续了至少一年，而 1 in 4 的人无法停止用药。

除了戒断副作用外，新兴研究还表明抗抑郁药还伴随着其他风险。

运动有助于缓解轻度抑郁症状

今年早些时候，哥本哈根大学医院的研究人员发表了一项研究，揭示了长期使用抗抑郁药与心脏意外停止导致猝死之间存在强相关性。服用抗抑郁药至少六年的个体，发生致命性心脏事件的风险增加了 74%。

此前未知的风险不仅限于心脏健康。一篇发表于 2023 年的论文发现，被开具 处方的 70 岁和 80 岁个体经历了“显著更快的全球认知能力下降”。同样，一项 2025 年的研究发现，服用抗抑郁药的痴呆症患者认知能力下降速度更快。“有时我们在开始用药时会考虑很多，但我们没有足够地思考在某个时间点停止用药的问题，”参与该研究的瑞典卡罗林斯卡学院神经学家 Sara Garcia Ptacek 表示。

医生偶尔会开具 以改善老年人的认知能力，并且有证据表明，通过缓解抑郁症状，这些药物确实有所帮助。但加拿大安大略省麦克马斯特大学的心理学家 Paul Andrews 认为，风险超过了收益。他表示， 在短期内可能有益，但长期使用则表明“更多的是负面影响或增加的风险”。

2023 年的研究还发现，服用 的人在脑神经元中积累了更多的 tau 蛋白缠结——这是阿尔茨海默病的一个标志——尽管目前尚不清楚这些缠结是由 的使用驱动的，还是与此相关的行为症状促使医生开具了抗抑郁药。

多伦多大学的心理学家 Marta Maslej 表示，这些研究并未明确证明认知能力下降和痴呆是由抗抑郁药引起的，而非由抑郁症本身引起。“这类因素在研究中确实很难控制，”她说。但这些研究依然强化了一种日益增长的观点，即 在被广泛接纳和开具时，对其所带来危险的审查不足。

并非所有人都认同抗抑郁药被过度开具，或者确信戒断症状会引起严重的副作用。例如，Guinjoan 表示，不开具抗抑郁药而让抑郁症得不到治疗的风险，超过了戒断症状的潜在影响。伦敦国王学院心理医学讲师 Rebecca Strawbridge 表示，她两种情况都见过：“有些人对抗抑郁药有非常负面的体验，而另一些人则看到它们带来了深远的积极改变，甚至挽救了生命。”

所有这些新研究都给目前正在服用 SSRI 的人们带来了疑问。他们是否应该尝试停止服用抗抑郁药？如果需要，应该如何操作？凯斯勒（Kessler）和巴克曼（Buckman）都强调，任何考虑这样做的人必须先咨询医生。在霍罗维茨（Horowitz）为了缓解戒断症状而重新服用 SSRI 后，他转向在线支持小组寻求关于如何逐渐减量的指导。他最终成功完全脱离了药物，但这花费的时间远比美国官方指南建议的要长。“美国的指南说你可以在大约六周内停止服用抗抑郁药，”他说，“而我花了大约六年时间。”

JAAP ARRIENS / NURPHOTO VIA GETTY IMAGES

此后，他创建了 Outro Health，这是一家旨在帮助人们安全减量停止服用抗抑郁药的诊所，并且他还开发了一种新的、更系统化的减量方法，他在英国皇家精神科医师学会的官方

阿尔茨海默病患者（下图）有时会被开具 SSRI 以改善认知能力，但这可能会增加大脑中 tau 蛋白缠结的数量

“我们在开始用药时会考虑很多，但在停止用药时考虑得不够”

指南以及首本关于停止用药的临床手册。他写道，人们应该以自己能够耐受的速度逐渐减量。如果开始出现戒断症状，他们应该维持在该剂量直到症状消退，或者如果症状难以忍受，甚至可以稍微增加剂量。他说，为了实现如此缓慢的减量，需要使用液体版本的 SSRI 而非传统的药片，以便能够以微小的增量精确地降低剂量。

这项工作为澳大利亚的医生以及英国的国民医疗服务体系（NHS）提供了指导，而后者曾将抗抑郁药的戒断症状描述为轻微且短期的。今年早些时候，美国临床精神药理学协会（American Society of Clinical Psychopharmacology）也发布了关于停止处方的建议，建议临床医生定期重新评估精神类药物处方的价值，“至少每年一次”。

随着越来越多的精神科医生质疑长期使用 SSRI 的合理性，英国的医疗服务机构正更多地尝试替代方案。运动和谈话疗法缓解抑郁症的速率与抗抑郁药相当，且没有任何药物副作用。英格兰和威尔士关于治疗抑郁症的新指南中包含了一份针对该病较轻症状的“一线治疗”清单，其中包括认知行为疗法、正念和冥想。

总体而言，精神医学领域这一萌芽状态的转变并非是要放弃 。有些人，例如患有强迫症或危及生命的抑郁症的人，可能需要服用 数年之久。而对于情境性抑郁症患者， 在短期内通常是有益的。但对戒断症状影响的新认识，旨在将 重新定义为更广泛、更多样化治疗方案的一部分。

在霍罗维茨（Horowitz）看来，这种转变的关键部分将来自于人们开始在不预设药物是唯一答案的情况下，去应对挑战性的生活境遇。他说，很多人需要帮助，但这并不总是意味着需要药物：“你需要时间，需要支持，需要治疗。我们需要开始允许医生和患者共同超越药物。”

迈克·马里亚尼（Mike Mariani）是《那些没能杀死我们的使我们更强大：在悲剧与创伤之后我们成为了谁》（What Doesn't Kill Us Makes Us: Who we become after tragedy and trauma）一书的作者

竞速运行全球最危险算法的工程师

量子计算工程师克雷格·吉德尼（Craig Gidney）的先驱性工作可能会给我们所有人带来巨大的影响。他告诉卡梅拉·帕达维奇-卡拉汉（Karmela Padavic-Callaghan），为什么他正专注于一种可能会终结我们认知中的加密技术的算法。

几年前，当我刚开始报道量子计算时，我对这些复杂且具有潜在突破性的机器及其开发领导者的运作方式还很不了解。从那时起，我很幸运地与其中许多人进行了交流。但有一个人一直让我难以接触——直到现在。

似乎所有与量子计算相关的人都知道克雷格·吉德尼这个名字。我还注意到，近年来他参与撰写的任何研究都引起了轰动。然而，身在加利福尼亚州谷歌量子人工智能（Google Quantum AI）部门的吉德尼没有写过书，不面向大众发表华丽的主旨演讲，也很少接受媒体采访。他少数几个数字化足迹之一就是他那个相当专业的技术博客。

不过，驱动我想与吉德尼交谈的不仅仅是好奇心。他正在研究一种在量子计算机上运行可能成为全球最危险算法的方法。这使他成为了一个极具影响力的人物，因为如果且当这种算法在真实机器上运行时，可能会引发一场被称为“Q日”（Q-Day）的危机——届时，从电子邮件到银行账户的所有加密系统都将可能被突然破解。

要理解吉德尼的故事，首先需要了解加密技术。如今，我们的许多私密数据都由加密系统保护，除非解决一个困难的数学问题，否则这些系统无法被破解。一种常见的选择是对极大的数字进行质因数分解。另一种则涉及被称为椭圆曲线的抽象数学对象。传统计算机难以解决这些问题，因此我们的数据和存款保持安全。但在 1994 年，麻省理工学院的理论计算机科学家彼得·秀尔（Peter Shor）发现，一台足够强大的量子计算机将能出类拔萃地应对这些挑战。因此，通往“Q日”的倒计时，

实际上就是量子计算机成功运行秀尔算法的倒计时。

在 20 世纪 90 年代，“Q日”看起来更像是遥远的科幻小说而非现实威胁，因为量子计算机本身还不存在。这些机器的动力源自支配其基本构建模块——量子比特（qubits）的量子效应。一台真正强大的量子计算机需要数百万个量子比特，而这在很长一段时间内似乎都遥不可及。但形势正在发生变化。过去几年中，几家公司和研究团队取得了显著进展，这引发了一个问题：秀尔的想法在现实中多久能实现？

正是吉德尼的工作让一些人认为，这个时间点可能近得令人担忧。几年来，他一直带头研究如何让秀尔算法在规模较小的量子计算机上运行——即我们在不久的将来更有可能构建出的机器。如果说秀尔提供的是一个理想

情况下的配方，那么吉德尼的工作就类似于在探索如何用尽可能少的食材清单使其生效。

因此，当我终于获得与吉德尼采访的机会时，我很想揭开这位神秘量子巫师的真相。在一个平凡的周一下午，他从加利福尼亚州的一间酒店房间登录视频通话，他当时正在参加一个会议。当发现房间里竟然神秘地没有椅子时，他直接坐在了地板上——我们的对话开始了。

超越炒作

不过，在与本人交谈后可以清楚地看到，释放混乱并非他的真实目标。事实上，他告诉我他的目标有两个：一是证明量子计算机可以运行，二是告知密码学家如何让我们在面对这些计算机时保持安全。

吉德尼（Gidney）对量子计算机的着迷不仅仅在于机器本身。大多数物理学家认为世界在本质上是量子的，其最基本的构建模块是量子粒子和场。吉德尼将构建量子计算机视为一种尝试，旨在挖掘现实的底层结构。“在最基础的层面上，整个宇宙都是由量子信息组成的，而我们目前无法操纵它，”他说道。

当涉及到由椅子、砖块和烤箱等组成的宏观人类世界时，我们可以购买各种设备来获取关于它们几乎所有的信息：温度计、相机、直尺等等。那么，是否可以拥有所有这些设备的量子版本？这就是驱动吉德尼的动力。“我只是想让量子计算机运行起来，”他说。从这个角度来看，他对 Shor 算法的关注是一种手段，而非目的本身。这仅仅是证明量子计算机值得如此炒作的最清晰、最直接的路径。

“我只是想让量子计算机运行起来”

纽约时报 / REDUX / EYEVINE

Q-Day 是量子计算机能够破解通用加密、使私密数据陷入危险的时刻 PATRICK LUX / GETTY IMAGES

关于量子计算机究竟将如何发挥作用的问题仍存在争议。一个经常被提及的可能性是，它们在模拟新候选药物分子方面可能表现卓越，但目前还没有人能证明如何以一种可靠且明确的方式，使其在性能上超越所有传统的模拟方法。对于其他无数应用——从优化航空公司航班计划到发现新型可持续材料——情况也是如此。

但 Shor 算法是个例外。除非量子力学定律是错误的，否则毫无疑问，只有量子计算机才能运行 Shor 算法。“我非常喜欢将其作为‘量子性’的证明。如果你用量子计算机实现了它，那么量子计算机确实有效这一点就无可争议了，”吉德尼说。此外，它与某些可能对化学计算等领域有用的算法具有共同特征。因此，吉德尼希望，如果能攻克 Shor 算法的实现，将产生真正的连锁反应，使量子计算机在处理其他任务时变得更有用。

当然，Shor 算法也有阴暗面，这引发了关于研究该算法者责任的问题。但在讨论这个问题之前，我想知道吉德尼的秘诀是什么。他以在使量子算法更好地适配现实世界量子计算硬件方面具有卓越推动力而闻名。他是如何做到的？

他告诉我，量子计算最初并不在他的职业规划之中。2017年，他加入谷歌（Google）时是一名软件工程师，而非量子计算研究员。但他对量子算法已经产生了好奇，并开始自学这一科目。“没有人教我，”吉德尼说，“在加入量子计算团队之前，我一直在研读教科书。”

量子领域

在交谈中，我意识到他对量子世界的心理模型与我在大学学习量子理论时接触到的截然不同。在我的认知中，量子领域由粒子和场组成，虽然它们确实很诡异，但仍然存在于三维空间中。而吉德尼（Gidney）则从量子信息的角度思考，并且似乎能自在地在抽象的高维空间中穿梭，那里是量子比特之间相互作用的数学表示法所在地。“量子比特是我理解量子物理的基础，”他说。

当涉及到运行 Shor 算法——或者事实上任何量子算法时，关键在于量子电路，即对各种量子比特状态进行顺序操作的方案。你的电路优化得越好，用相同数量的量子比特能完成的工作就越多。这正是吉德尼所擅长的。他告诉我，在早期，他自学了如何将量子电路的工作原理描绘成彩色三维图表，现在他每天花费大量时间沉浸在这个抽象的

世界中。量子计算最紧迫的挑战之一是如何捕捉设备不可避免地产生的错误。对于吉德尼来说，答案就隐藏在这些他已产生强烈直觉的图表中。他说，有时图表中的颜色排列看起来就很不协调。“我认为量子计算一直被认为极其困难，这种名声并非完全不实——其中确实有一些非常违背直觉的东西。”

那么，研究将我们带到真正运行 Shor 算法的距离有多近了？2018年，在吉德尼加入 Google Quantum AI 量子计算团队约一年后，当时最可靠的估计表明，研究人员需要一台拥有超过 1 亿 个量子比特的量子计算机才能实现这一目标。但在今年 3 月，吉德尼及其同事基于他们的电路优化发布了一项新估计，称实际上仅需 500,000 个量子比特的阵列即可实现。这虽然仍远大于当今最先进的量子计算机——迄今为止创建的最大量子计算阵列仅有 6000 多个量子比特，且仍然非常容易出错——但这依然是 200 倍的提升。这一发现足以令人震惊并促使人们采取行动。消息传出后，量子网络安全公司需求大增，其中一家公司收到的企业咨询量增加了十倍，这些企业都希望提高自身的量子安全性。

关于吉德尼及其同事的那项估计，“我很开心我们似乎成功说服了安全社区尽早采取行动，而不是拖到以后。”

量子计算机使用量子比特（qubits）在芯片中存储信息 GOOGLE QUANTUM AI

出于安全考虑，由于该算法与加密相关，他们故意省略了量子电路配置的一些细节。然而，仅在三个月内，法国雷恩大学 Inria 中心的安德烈·施罗滕洛赫（André Schrottenloher）就填补了大部分隐藏的细节。尽管他的解决方案与吉德尼及其同事的并不完全相同，但这凸显了他们所做工作的兴趣点和重要性。这本并非旨在发起挑战，但它激励了其他研究人员。吉德尼说：“正因为没有告诉人们电路的具体情况，这篇论文引起了关注，人们意识到‘哦，实际上，优化量子电路是非常可能的’。我认为这挺令人愉快的。”

安全至上

这一切让我想到了我想问吉德尼（Gidney）的或许是最重要的问题：他是否担心自己的研究成果如果落入心怀叵测之人之手会造成巨大的危害？吉德尼表示，他承认自己的工作增加了全球企业和政府在面对潜在量子威胁时，开始改变其加密方法的压力。但他也认为这并非坏事。“我很高兴我们似乎成功说服了安全社区尽早采取行动，而不是拖到以后，”他说，“去年，我还有点担心人们似乎在很大程度上忽视了这个问题。”

他不愿推测“Q日”（Q-Day）何时到来，但在他看来，他目前所做的工作应该有助于减轻其影响。事实上，我们确实拥有在Q日之后保护自己的方法。在美国，国家标准与技术研究院（NIST）在2017年开始测试“后量子”加密算法，并在2022年由其研究人员公布了几个有望成为安全可靠加密新标准的算法。吉德尼表示，他的工作有助于探索我们还有多少时间来实施这些保护措施，而这一工程过程将既昂贵又困难。“我们不希望出现这样一种情况，即人们低估了这些（实施保护所需的）措施有多昂贵，导致无法及时完成，”吉德尼说。

不过总体而言，吉德尼对量子计算的未来持乐观态度。他在纠错和电路方面的工作不仅仅是为了 Shor 算法，更是为了让量子计算机达到能够良好运行并完成有用工作的程度。在他的博客中，他写道，他预计量子纠错将出现巨大的飞跃，这将提高量子计算机运行复杂程序的可靠性。他谈到这件事时的语气充满了坚定且乐观的情绪。

但当我问他 10 或 15 年后是否仍将致力于优化量子电路时，他的回答让我感到意外。“我不知道，”他说。让他犹豫的是人工智能以及像 ChatGPT 这样的大语言模型（LLMs）的飞速发展。“也许在两到五年后，你给它们一个需要优化的电路，它们能做出我想不到的操作，而且我想到的所有方案它们也都能想到。但我们还没到那个阶段，”他说。

在与吉德尼交谈之前，我对他的工作可能给隐私带来的负面影响感到好奇。交谈结束后，我意识到他身上并没有什么神秘或阴暗之处。克雷格·吉德尼（Craig Gidney）并不是一个笼罩在新兴量子计算产业之上的危险人物，他只是对量子信息深感兴趣——并且非常擅长思考这个问题。

尽管如此，无论他是否有意，他都处于技术收益与风险之间一个微妙的交汇点。然而，在面对 Q 日时，他所挥舞的警示旗帜无疑能帮助我们避免灾难。

让您的孩子在学校领先一步

√ 一本面向 7–14 岁儿童的精彩月刊

√ 帮助他们爱上阅读与学习

√ 惊人的事实、照片以及远离屏幕的乐趣

√ 涵盖他们所有感兴趣的话题，从动物、太空和自然到艺术、体育、科学和发明

√ 谜题、测验、游戏、活动和笑话

√ 荣获包括年度儿童杂志在内的 10 项大奖 立即订阅，起价 £14.99

SAVE 54%

离奇但真实...

赠送价值 £27* 的免费书籍

隐喻填字游戏 #195 出题人：Trurl

涂鸦区

下周预告：快速填字游戏

快速问答 #367 由索菲·法罗（Sophie Farrow）出题

1 红光匮乏可能会给我们的身体带来严重影响，但细胞中的哪些结构被认为能介导红光的有益效果？

2 树轮中碳-14 的大规模激增是所谓“宫家事件”（Miyake events）的证据，这类事件在数千年间多次发生。是什么导致了这些事件？

3 严格节食或跳餐会触发哪种激素的激增？

4 天花疫苗是第一款研发成功的疫苗，但天花是在哪一年被宣布根除的？

5 大西洋经向翻转环流（AMOC）将海水从哪个海湾向北输送？

横向（填字游戏）

1 从导演计划中掉落的一个 (4) 3 准备好的猎鸟，加热平底锅 (8) 9 在Burnham最终进入第一轮后，为前首相争取发言权而奋斗 (7) 10 被撞击的旧容器 (1-4) 11 与标准DNA片段（大部分）一致 (8) 12 六对五——需要防御！ (4) 14 极其聪明，且像 10 或 11 那样？ (6) 16 那个女人出现了两次，第二次喝醉了？天哪！ (6) 18 协同行动的广播，并经历沉浸式体验 (4) 19 出题者的最爱（在我离开之前，患有感染） (8) 22 辅助的一次性工艺被推翻 (5) 23 触发美国大学反应元素的撤回 (7) 24 提供一流合金的官员 (3,5) 25 传递信息的更狂野方式 (4)

纵向（填字游戏）

1 在打烊时间之后，也许是秋天，像青少年一样落下 (13) 2 生活，摄入氧气等 (2,3) 4 亨利，一个非常卓越的人，以循环系统发现而闻名 (6) 5 冉冉升起之星的重要开端 (4) 6 失去电阻后由爆炸反应产生的有毒物质 (7) 7 告诉 RI 教师关于化合物的事 (13) 8 在恶魔学中真诚地询问 (7) 13 绝望地希望 9 号是那个在灰烬之后繁荣的人 (7) 15 恒定的，错过了周期？ (7) 17 一些模因是艾萨克抛出的 (6) 20 在家中被占用后部署 (2,3) 21 据报道，削减导致手弱 (4)

Plus Side 由 Puzzlr 创建

在下面的网格中圈出数字以使用它们，或将其划掉以剔除。圈出的数字之和必须同时等于每行末尾的目标值和每列下方的目标值。

2	5	6	8	3	13
4	7	3	9	8	7
7	2	6	1	2	17
7	3	8	9	4	19
9	1	9	5	6	9
25	12	6	17	5

本周谜题答案见第 47 页

虚构的创造

为什么人类如此轻易地将虚构内容融入生活，并且对角色产生情感，仿佛他们是现实生活中亲近的人一样？

菲奥娜·利琴斯卡 (Fiona Lyczynska) 英国，汉普郡，塔德利 我认为这是由我们的社交和部落本性，以及我们的情感和实际需求所驱动的。

人类具有建立准社会关系（parasocial bonds）的能力，例如与领导者、英雄、媒体影响力人物，以及人工智能和虚构角色建立的关系。在 20 世纪 90 年代，研究人员发现，通过文本与计算机互动的成年人会将其视为具有“感情”一样。如果我们检测到足够多的人类互动特征，就很容易产生情感反应。

JUPITERIMAGES / GETTY IMAGES

此外，这可能迎合了我们的情感或欲望，或者从另一方面来看，可能是为了让我们顺从或合作。传播的想法可以将我们团结起来，并在某些情况下赋予我们作为物种的优势，例如促进我们生存或帮助我们逃离危险的技术创新。如果一位富有魅力的领导者能将我们团结起来，或者我们觉得自己的

“如果我们检测到足够多的人类互动特征，就很容易产生情感反应”

需求将被某人或某物满足，那么短期的社交焦虑感就会减轻——就此而言这对每个人都有好处，尽管这种焦虑很容易被重新激活。

关于伟大的想法，通常只有极少数人能提出更好的见解。大多数人，即使在当今时代，可能仍然依赖社交纽带、等级制度和社会传染来决定信任谁以及采纳什么，而不是在自行研究之后才做出决定。

本周新问题

气味幻觉：存在视觉幻觉和听觉幻觉。那么是否存在嗅觉幻觉？贾斯汀·贝克，澳大利亚墨尔本

不要旋转：什么样的力作用于流体粒子，使其在管道中开始旋转或呈螺旋状向下移动，而不是平稳流动？菲尔·莫里，通过电子邮件

以及使用科学方法。

虚构作品可能不是真实的，但它通常足以刺激我们的情感。当我们能与某个角色产生共鸣时，这可以是对自我的肯定；我们可能会在其中看到自己的影子，从而给我们在现实世界中采取行动的希望和勇气。与混乱现实生活中关系的压力和不平等相比，这相对安全。因此，即使我们意识到它们是虚构的，我们也经常乐于接受那些对我们具有隐喻真理的虚构故事。

理查德·辛德 英国北约克郡，查佩尔·哈德尔西

一段时间以来，我一直在思考一个相关问题：我如何在教学中利用这种效应，通过使用叙事来帮助我的学生在现实世界的语境中更好地

理解该学科？

安格斯·弗莱彻在其同名书籍中解释了“故事思维”（叙事认知）以及它如何成为生命本身的基础。西方学术传统倾向于将逻辑认知推崇为主要的思维模式，这可以追溯到希腊哲学家（始于苏格拉底）。弗莱彻认为这种观点具有误导性，因为“叙事是我们体验世界的方式”。

安尼尔·塞斯关于“受控幻觉”的概念吸引了我这个计算机科学家的脑细胞。大脑似乎并非在每一刻都从零开始构建意识，而是

想要向我们发送问题或答案？请发送电子邮件至 lastword@newscientist.com。问题应关于日常科学现象。完整条款与条件请参阅 newscientist.com / lw-terms

嗅觉幻觉是否存在，即你被误导而闻到了错误的气味？

维持一个现实的预测模型，仅在传入的感官信息与预期不符时才对其进行更新。这是解决感官带宽有限问题的一个优雅方案，让我想起了 MPEG 压缩在流媒体视频中的工作方式！

我意识到这些想法可能是相互关联的——如果我们每个人都生活在一个基于内部世界模型、并随新信息接收而更新的个人“故事”（自我意识）中，那么故事或许也像预测编解码器一样工作？当然，当输入内容与我们的预期不匹配时，这个过程可能会产生混乱，例如视觉幻觉。

当我们阅读虚构故事时，它们能与我们的内部世界模型产生共鸣，触发情感反应。作为能够应用心理理论的社会动物，我们可以进入他人的内心，并从他们的视角想象世界。我们的脑部在处理虚构角色和真实人物时，似乎使用了许多相同的认知和情感过程，我认为这就是我们能够建立情感连接的原因。

最后一段话我留给弗莱彻：“叙事智能是经验科学、医学、工程学、技术创新和生理韧性的引擎。”

珍·惠特尼

美国加利福尼亚州圣迭戈

孤独流行病是一个答案。也许 covid-19 疫情也与之有关。社交媒体和娱乐产业创造了一个完美的虚构世界，产生了不切实际的期望，因此，真实的人类，嗯，太令人失望了。于是，由人类最初创建、现在也由 AI 创建的虚构角色登场，拯救了局面。

我们选择那些令我们愉悦的角色，摒弃那些我们认为有缺陷的角色。没有痛苦的分离，没有泪水，没有愧疚；只需点击 X 即可关闭。这有什么不喜欢的呢？

祝你健康

为什么我们在睡觉时不会打喷嚏？

Herman D'Hondt

澳大利亚，悉尼

这个说法是错误的，至少对我而言：我经常在睡觉时打喷嚏。不过，喷嚏会立刻把我惊醒。这种情况大约每周发生一次，但每晚只会发生一次，通常是在我入睡后 10 到 15 分钟左右。然后我只需要擤一下鼻子，翻个身，几分钟后就又睡着了。

“我见过我的猫在没醒过来的情况下打喷嚏，但当然，猫是非常沉睡的睡眠者”

Carol Stevenson

英国，伦敦 在花粉热季节，我经常被喷嚏惊醒，所以这一定是在我睡着时开始的。此外，我见过我的猫在没醒过来的情况下打喷嚏，但当然，猫是非常沉睡的睡眠者。

所以我认为我们在睡觉时是可以打喷嚏的，如果它没有把我们惊醒，那么我们就意识不到。

前瞻规划

是否有任何植物和生物可以被带到月球，并且能够在，比如说 25 年后生存、繁殖并成为可行的食物来源？（续）

John Davies

英国，兰开夏郡，哈弗布雷克斯 之前的一个回答（Robin Maguire，7 月 4 日）提到了沃德箱（Wardian case），但没有提到有史以来最大的此类装置——生物圈 2 号（Biosphere 2）。

如今，生物圈 2 号类似于英国的伊甸项目（Eden Project），但它最初是一个包含 8 名人类的密封环境。两次尝试实现封闭生态系统的努力都失败了，因为植物和动物死亡，氧气不足，且人类之间发生了争吵。

虽然沃德箱是将物种移植到异域环境的一种成功手段，但将其应用于更大规模的环境还需要更多研究。

Eric Kvaalen

法国，莱萨尔勒鲁瓦

法国，莱萨尔勒鲁瓦 Robin Maguire 说：“月球相当寒冷，且月球的一天大约长达一个月，因此需要地暖和部分时间的日光灯”。实际上，在阳光照射期间，月球温度可达 120℃ (248°F) 左右。但当然，由于没有大气层，温室内的温度仅取决于其自身的能量平衡，而非周围环境的温度。然而，如果想法是在南极或北极附近种植植物，那么无论何时都没有足够的阳光用于加热或光合作用，因为太阳距离地平线最高不超过 1.5 度，或者处于地平线以下。

4 Harvey, 5 Anus, 6 Acronite, 7 Tetrachloride, 8 Implore, 13 Phoenix, 15 Nonstop, 17 Emesis, 20 in use, 21 Pair

横向 1 Plot, 3 Phesants, 9 Hammer, 10 U-boat, 11 Parallel, 12 Viva, 14 Binary, 16 Sheesh, 18 Sink, 19 Impetigo, 22 Extra, 23 Stimuli, 24 Top Brass, 25 Gene

195

隐喻填字游戏

5 墨西哥湾

08614

3 皮质醇

2 太阳表面

1 线粒体

快速测验 #367

Solution Plus Side

量子即兴

流行音乐的历史在一定程度上就是一部技术史。从乔治·博尚（George Beauchamp）发明电吉他，到那个在 Dubstep 音乐中一度无处不在的“摆动低音”（wobble bass）的创造者，我们聆听的内容一直由富有创造力的工程师们塑造。

因此，“反馈”（Feedback）专栏在收到记者卡梅拉·帕达维奇-卡拉汉（Karmela Padavic-Callaghan）关于“在未来的量子互联网上创作音乐”的线索时感到非常好奇。卡梅拉将我们的注意力引导至一篇发表于 5 August 的论文，题为《传送游戏：用于交互式音乐的多智能体系统中的量子传送》。该论文发表在一家标题宏大且包罗万象的科学期刊《国际并行、涌现与分布式系统杂志》（International Journal of Parallel, Emergent and Distributed Systems）上。

研究人员在文中介绍了一种“具有量子音乐智能体的交互式音乐系统，这些智能体通过向彼此传送量子态来进行通信”。简单来说，你演奏一段音乐短句或作品，录音被交给一组量子计算机。然后，它们以某种量子方式对其进行处理，并输出该录音的修改版本。

“反馈”并不打算假装能理解量子计算机具体在做什么。论文告知我们，“旋律和节奏行为使用单量子比特概率幅度调制（SQPAM）编码为量子态，并通过量子相位估计（QPE）进行结构化”，并且最多三个智能体“被组合在单个量子电路中，通过量子传送进行定向通信”。

出于好奇想知道是谁想出了这个主意，“反馈”查询了第一作者。爱德华多·米兰达（Eduardo Miranda）是一位作曲家兼计算机科学家，他曾利用从脑机接口到黏菌，以及最近的量子计算机等各种手段创作音乐。2024, 年，他发行了一张名为《QUBISM》的专辑，完全由量子计算机创作。

有故事想告诉“反馈”吗？请发送至 feedback@newscientist.com 或邮寄至 New Scientist, 9 Derry Street, London, W8 5HY（邮寄件的处理时间会有所延迟）。

Twisteddoodles 为《新科学家》杂志绘制

“反馈”听了几首曲目，感到有些困惑。开篇曲 Lumina Grooves 1 围绕着欢快的类木琴音符构建：听起来像是你在电梯里会听到的音乐，或者可能是关于海洋无脊椎动物的野生动物纪录片的背景音乐。相比之下，《Spinnings》则使用了怪异的机械呻吟声，听起来像是《银翼杀手 2049》电影原声带的剪辑片段。

我们非常想知道新论文中的量子音乐智能体创作出了什么，但该研究没有包含任何音频，所以我们不得而知。不过，“反馈”注意到，作者开发这些智能体是为了支持“让人想起自由爵士即兴演奏的、模棱两可且具有变革性的交互”，并且他们的系统是一个“容忍发散和模糊性”的系统。借用已故伟大人物汤姆·莱勒（Tom Lehrer）的一句话，我们认为这意味着这些量子智能体时不时会弹错一个音符。

优质水

乔纳森·斯托皮（Jonathan Stoppi）在加拿大开车旅行时，发现并拍摄了一辆当地企业的货车。车上在为水做广告，但并非普通的水。这种水提供“卓越的水合作用”。起初，我们以为这意味着它比其他水更“湿”，但随后的宣传语则更进一步。

事实证明，这种水比其他水更好，因为它具有“碱性”、“抗氧化”且“富氢”。这让“反馈”（Feedback）栏目组感到晕头转向，因为酸的定义特征就是释放大量的氢离子（也称为质子）。因此，我们不明白这种水如何能既富氢又是碱性的。我们同样不明白水如何能成为一种抗氧化剂。

幸运的是，我们发现了一条可以追踪的线索。乔纳森照片中的车辆上有一个 URL，反馈栏目组勇敢地将其输入浏览器。它将我们引导至一个关于“Kangen ®”的网站，该产品除了看似富氢和具备上述所有特性外，还具有“负 ORP”——这显然是“口服首选”的。深入浏览该网站后，我们了解到“这种电解还原的富氢水有助于使您的身体恢复到更碱性的状态，这对健康最为理想”。

我们还被告知，“Kangen ® 可能会给植物带来新鲜感和生命力”。反馈栏目组在想，这是否是有史以来承载压力最大的一个“可能”一词。

随后，一个模糊的记忆在脑海中闪过。在翻阅过去的专栏后，我们意识到早在 2011 年，我们就写过关于所谓“碱性水”的健康益处，在 2012 年再次探讨了该话题，并在今年早些时候（3 月 28 日）深入研究了“六角水”的健康益处。这件事已从我们的意识中淡出，但看来，在那些渴望向你推销昂贵水的商人心中，它并未消失。

质量保证

在 12 August 的信函中，我们提到了一起发生在“中风协会”（Stroke Association）慈善机构身上的尴尬事件。社交媒体上的一个“建议关注”框截断了他们的名称，由于空间不足以容纳“ociation”部分，从而产生了一个令人尴尬的双词短语。

读者皮特·劳埃德（Pete Lloyd）由此想起了他工作单位的一次重组，当时他的一个朋友成为了质量保证经理（quality assurance manager）。然而，组织架构图没有足够的空间来显示这三个如此长的单词，因此人力资源团队通过删除“urance”和“ager”对其进行了缩写。“我觉得必须告诉他，这样他才能要求人力资源部门更新图表，”皮特说道。

New Scientist Discovery Tours

100+ science-led tours in 50+ countries

11 January 2027 / 11 days

科学、野生动物与宇宙——探索印度的另一面

参加一次独一无二的沉浸式保护区狩猎之旅，深入印度未被驯服的荒野之心

准备好步入印度的野生之心。在塔多巴（Tadoba）和彭奇（Pench）国家公园深处，通过 4x4 越野车、自行车和徒步方式，寻找行踪诡秘的孟加拉虎、豹、黑豹和野狗。在专业自然主义者的指导下，了解旨在保护这些不可思议物种的保护工作。

随后，天空将成为主角。在空间科学高级讲师、天文学专家马丁·格里菲斯（Martin Griffiths）的指导下，在德里和那格浦尔的历史天文台解锁数世纪的天文遗产，通过引人入胜的讲座和研讨会，解码宇宙背后的仪器与物理学，随后在包括印度首个暗夜公园在内的引导式夜间观星活动中，观察银河的璀璨景象。

11 days. 白天看虎，夜晚看星系。这将是你从未体验过的科学与狩猎之旅。

难忘体验：

) 参观由马哈拉贾·贾伊·辛格二世（Maharaja Jai Singh II）建造的 18 世纪德里简塔·曼塔（Jantar Mantar）天文台

) 探索尼赫鲁天文馆（Nehru Planetarium）和拉曼科学中心（Raman Science Centre）——印度最著名的天文学教育中心

探索塔多巴（Tadoba）和彭奇（Pench）国家公园的茂密森林、隐藏小径和丰富的生态系统

Key 2 Number of nights --- Included flight ● Key site visit

任何人 都不应该在 缺乏正确支持的情况下 面对癌症。

但现在，太多的人正处于这种境地。你能否获得这种支持，可能取决于你居住的地方、你的收入，或者仅仅取决于你是谁。一个人被诊断的速度、他们接受的护理，甚至他们的生存机会都可能因此受到影响。

这不公平。

今天就向麦基伦（Macmillan）捐款，您可以帮助我们的福利权利顾问在财务担忧演变成危机之前及时介入，帮助我们的伙伴提供来自真正理解者的每周一对一支持，以及我们的支持热线在一年中的每一天提供专业的指导和信息。

您愿意捐赠 £10，以确保没有人会在缺乏所需护理的情况下面对癌症吗？

癌症是不公平的。但癌症护理应该是公平的。请立即在 macmillan.org.uk / news 捐款。

扫描此二维码立即捐款

New Scientist WEEKLY 29 August 2026

QUANTUM THEORY FINALLY MAKES SENSE

...but only if we poke holes in the fabric of reality

MORE SMART COST.

MG

THE ALL NEW 7 SEAT MGS9 PHEV

£0 DEPOSIT

0.9% APR REPRESENTATIVE

- PANORAMIC ROOF - UP TO 62 MILES EV RANGE* - 12.3" HD DUAL SCREENS

MG Motor UK Ltd is an Appointed Representative of The Compliance Guys Ltd which is authorised and regulated by the Financial Conduct Authority. We are a credit broker not a lender. This finance offer is provided by MG Motor Financial Services RH1 1SR. Dealerships we introduce you to will receive commission. Finance subject to status & income. 18+. Terms and Conditions apply. Offer ends 30.09.26.

*Figures provided are calculated under test conditions. These figures are for comparative purposes only and may not reflect real-life driving results. Actual figures may vary depending on various factors such as weather conditions, driving styles, vehicle load, and accessories fitted post-registration.

DISCOVER MORE

On the cover

30 Quantum theory finally makes sense ... but only if we poke holes in the fabric of reality

Vol 271 No 3610 Cover image: Eiko Ojala
5 The extraordinary immune cells of the world's oldest people

7 How ice melt intensified European heatwaves

36 Why it is time to rethink long-term antidepressant use

26 Next-gen bananas 24 Jim Al-Khalili takes on time 40 The most dangerous algorithm ever

30 Features "Have we been thinking about quantum physics all wrong for a century?"

News

4 Fresh light on dark matter Peculiar gamma-ray signal may unlock dark matter's secrets

6 Heating up The world may warm by 25 per cent more than we thought

13 Like father, like daughter Healthy female clones created from a male mouse

Views

19 Our upper limit How long could we possibly live, asks Graham Lawton

20 The Void Ages Annalee Newitz on the risk of losing our cultural history

22 Aperture Stunning snaps from the Wildlife Photographer of the Year competition

24 Book of the week A comprehensive look at time

29 Letters The questions raised by a heterogeneous cosmos

9 Sleepy neurons We've finally found what creates the drive for shuteye

Features

30 Poking holes in reality What would happen if we removed irrational numbers from quantum mechanics?

36 The antidepressant trap Withdrawal symptoms following long-term SSRI use appear far more serious than we thought

40 Most dangerous algorithm The man trying to prove that quantum computers can work

The back pages

45 Puzzles Try our crossword, puzzle and quick quiz

46 Almost the last word Why do humans so readily adopt fiction into our lives?

47 Tom Gauld for New Scientist A cartoonist's take on the world

48 Feedback A delve into music-making on the future quantum internet

48 Twisteddoodles for New Scientist Picturing the lighter side of life

You didn't hear her hilarious joke, again.

Now you owe her an ice cream.

Book a free hearing test.* Do it for your grandchild.

Spread the cost with 0% monthly payment plans**

Visit bootshearingcare.com or call 0345 200 7052 $^{†}$

When to stop?

Why the healthcare industry isn't built for coming off medication

MODERN medicine has an impressive array of drugs in its arsenal. If you are unwell, there is a very good chance your doctor will prescribe a pill to get you back to health. What our medical system is less good at, however, is stopping medication.

Take antidepressants. The drugs aren't effective for everyone, but they work well for some. Once the medication does its job, though, or the life circumstances that brought on depression grow easier – what then? We don't yet have a good answer.

Because antidepressants are relatively safe drugs, they can often be dispensed for months or years on end, without much consideration about the long-term implications, as we discuss on page 36. But thousands of people struggle to stop taking them, with little help from doctors.

The problem starts long before doctors write out the prescription, though. Clinical trials are typically short-term, lasting only a matter of weeks, and there are few financial rewards or even resources for those who want to study long-term use or when medication should be stopped. This

"Few industries want to help people stop using their products, but healthcare should"

lack of evidence then informs clinical guidance – or rather, its absence.

Antidepressants are just one example. The opioid crisis was spawned, in part, by a flippant attitude to prescribing drugs, and a failure to regulate how often physicians should review patients' doses. Medications from beta-blockers to acid-reducing stomach drugs, though beneficial for heart or stomach issues in the short term, can cause serious long-term problems. When such issues do come up, physicians have a tendency to prescribe more medications to help combat any side effects – a problem known as a “prescription cascade”, which can be especially difficult for people managing health issues later in life.

Few industries are invested in getting people to stop using their products, but healthcare should be an exception. We need to pay attention to the problems that arise when people stop taking medication, and develop detailed guidance and support, so that when it is the right time to make a change, deprescribing is a valid option, too.

PUBLISHING & COMMERCIAL

Display advertising Tel +44 (0)203 615 6456 Email displayads@newscientist.com Sales director Claudia Nicoletti Partnership growth lead Joe Teal Agency partner Daniel Sinclair

New Scientist Events Tel +44 (0)203 615 6554 Email live@newscientist.com Sales director Jacqui McCarron Sales manager Maureen Ignacio Head of product delivery Henry Gomm Events and projects manager Georgia Hill Events and projects executive Ciara Lucas

New Scientist Discovery Tours Email tours@newscientist.com Director Kevin Currie Head of travel marketing Ginny Smith Head of travel product Lara Paxton Travel product manager Pip Orchard

Chief customer officer Claire Overstall
Head of campaign marketing James Nicholson
Subscription acquisitions manager Molly Hayden
Campaign marketing coordinator Charlotte Weeks
Head of customer experience Emma Robinson
Engagement marketing manager Kelly Spillane
Senior marketing executive Sarah Fabian
Head of CRM & audience data Rachael Dunderdale
Analytics manager Gary Parung
Junior data analyst Joseph Cole
CRM manager Ffion Evans
CRM executive Kornal Nadeem

Technology Director of engineering Michael Ilett Principal engineer Phil John Senior engineers Amardeep Sian, Maria Garrido, Matt Watson QA engineer Kaveri Daitkar

Head of product Remy Becher Product manager Caitlin Jenkins Product manager Thomas Reader Head of product design Irmak Sunal Lutkin Digital marketing designer Katarina Pollu Brand marketing designer Laura Spanu Senior product designer Sophie Majecki

Partnerships Consultant editor Justin Mullins Editorial director B2B Edward Glanville Head of strategic partnerships David Allard

NewScientist

Chief executive Roland Agambar Chief operating officer Jonas Hermans Chief financial officer Depak Patel Executive assistant Lorraine Lodge Finance & operations Head of finance Charlotte Lion HR business partner Tinka Bleijenberg

CONTACT US newscientist.com/contact

General & media enquiries UK 9 Derry Street, London, W8 5HY Australia 58 Gipps Street, Collingwood, Victoria 3066 US 51 Astor Place, 4th Floor, New York, NY 10003

UK Newsstand Marketforce UK Ltd Email mfcommunications@futurenet.com

Syndication Tribune Content Agency Tel +44 (0)20 7588 7588 Email tca-articlesales@tribpub.com

Subscriptions View the latest offers newscientist.com/subscription Tel +44 (0)330 333 9470 Email subscriptions@newscientist.com Post New Scientist, Rockwood House, Perrymount Road, Haywards Heath, West Sussex RH16 3DH

© 2026 New Scientist Ltd, England. New Scientist is published weekly by New Scientist Ltd. ISSN 0262 4079. New Scientist (Online) ISSN 2059 5387. Registered at the Post Office as a newspaper and printed in England by Warners Midlands Plc

EDITORIAL Editor Catherine de Lange Executive editor Timothy Revell Managing editor Penny Sarchet

News News editor Jacob Aron Assistant news editors Alexandra Thompson, Sam Wong Reporters (UK) Madeleine Cuff, Michael Le Page, Alec Luhn, Matthew Sparkes, Carissa Wong (Aus) Alice Klein, James Woodford

Digital Head of audience Matt Hambly Podcast editor Rowan Hooper Head of editorial video David Stock Senior Social Media Manager Isabel Baldwin

Features Head of features Claudia Canavan Deputy head of features Joshua Howgego Editors Abigail Beall, Alison George, Olivia Goldhill, Jacklin Kwan, Thomas Lewton

Culture and Community Head of books and culture Alison Flood Senior culture editor Liz Else

Magazine Magazine editor Eleanor Parsons Assistant magazine editor Sophie Berdugo

Subeditors Chief subeditor Kelsey Hayes Bethan Ackerley, Tom Campbell Trainee subeditor Lucy Wall

Design Art editor Ryan Wills Joe Hetzel, Phoebe Watts

Picture desk Picture editor Tim Boddy Assistant picture editor Jenny Quiggin

Production Head of publishing operations Joanne Keogh

New Scientist US Senior editor Chelsea Whyte Features editor Grace Wade Subeditor Alexis Wnuk Audience editor Gerardo Bandera Reporters Leah Crane, Karmela Padavic-Callaghan

Space

Most direct sign of dark matter yet

Dark matter's secrets could be unlocked by a peculiar gamma-ray signal in three galaxy clusters – or it may reveal a more exotic phenomenon, finds Karmela Padavic-Callaghan

AN UNUSUAL gamma-ray signal could be the most direct evidence of dark matter yet. However, the signal could also be the result of a telescope anomaly or produced by something even stranger than dark matter.

Dark matter forms the bones of much of our universe, driving how galaxies cluster, for example. It also far outweighs all the visible matter, yet for decades researchers have puzzled over what exactly it is.

NASA'S GODDARD SPACE FLIGHT CENTER/A. SIMONNET, SONOMA STATE UNIVERSITY

Part of the problem is that they can't currently observe dark matter directly. Instead, they can only register its effect on its surroundings. A more direct line of investigation would be to detect particles produced when dark matter collides and annihilates with itself, which some theories suggest ought to happen. Now, Yun-Feng Liang at Guangxi University in China and his colleagues have identified a gamma-ray signal that may be the result of this process.

They found the signal after analysing 15.5 years of data from the Fermi Gamma-ray Space Telescope (FGST), looking at the Virgo, Fornax and Ophiuchus galaxy clusters specifically. The researchers chose these galaxy clusters because they are thought to contain large halos of dark matter, says Liang. The gamma-ray signal the team found there was shaped like a spike or a line, analogous to burst of light of only one colour, which some theories predict is the result of dark matter annihilating itself.

“Discovering a sharp gamma-ray line would be the ultimate ‘smoking gun’ evidence proving the existence of dark-matter particles and revealing their properties in particle physics,” says team member Yi-Zhong Fan at the Chinese Academy of Sciences. Based on their statistical

Could a strange burst of gamma rays be evidence of dark matter?

analysis of the data, the researchers estimated that there is less than a 1-in-10,000 chance that this signal is random cosmic noise or a purely coincidental pattern (Physical Review Letters, doi.org/rkp4).

Yet, several questions remain before the team can declare a definitive discovery. Detecting dark-matter signatures in gamma rays has historically been difficult because the signals tend to be weak and can be confused with telescope errors. One promising gamma-ray signal from 2012, for example, proved to be exactly that. Zhao-Qiang Shen, also at the Chinese Academy of Sciences, who worked on the new study, says the team members performed a large number of tests to confirm the veracity of their signal, but some possibility of instrument error remains.

Moreover, while the new signal is clear when looking at the three galaxy clusters, it seems to vanish closer to the centre of our galaxy, where dark matter is dense, so it should also be annihilating itself and creating gamma rays.

"The fact that the signal lights up in distant clusters but remains silent in our own cosmic backyard

"There is less than a 1-in-10,000 chance that this signal is random cosmic noise or coincidental"

conventional theories predict." Alternatively, the signal could be coming from some even more rare or novel phenomenon, such as ultra-fast particle winds from exotic magnetized neutron stars.

is deeply peculiar," says Liang. "If this is truly from dark matter, it suggests the particles must interact in a much more sophisticated way than

Forcing a rethink

"I find the result intriguing, but I would be quite cautious about interpreting it as evidence for dark matter," says Juri Smirnov at Liverpool University in the UK. He says that looking at only one galaxy cluster instead of all three together increases the chances that the signal is a product of noise or a random pattern. He is also worried about how unconventional a dark-matter process would have to be to produce this gamma-ray spike far away from us but not close to the centre of our galaxy. "This puts the standard dark-matter interpretation under considerable tension," he says.

The team hopes that future telescope missions will add clarity to the situation, for instance, the Very Large Area Gamma-ray Space Telescope, which was proposed several years ago and could collect more gamma rays with better resolution. By 2040, FGST will also have doubled its dataset, which could further elucidate the nature of the odd gamma-ray spike, says Fan.

"If future telescopes confirm that this signal is genuine, it would represent a historic breakthrough. It would immediately reveal the mass of the elusive dark-matter particle and give physicists a concrete target to build the next generation of fundamental particle theories," says Shen. "At the same time, it would force us to thoroughly rewrite our standard textbook models of dark matter."

Infectious diseases

Deadly emerging tick viruses are spreading in the US

Alice Klein

TWO tick-borne viruses – Heartland virus and Bourbon virus – appear to be spreading in the US. Nebraska has become the ninth state to detect one or both viruses in local ticks.

Heartland virus was first reported in 2012 and Bourbon virus in 2015 after they were discovered in people who had become severely unwell or died after being bitten by ticks. Since then, they have been found to be transmitted by lone star ticks (Amblyomma americanum), named for the white, vaguely star-shaped spot on the female's back.

Historically, lone star ticks were found in south-eastern states of the US, but they have been moving north and west due to factors like climate change and reforestation.

Heartland virus or Bourbon virus, or both, have been found in lone star ticks in Missouri, Kansas, Illinois, Alabama, Georgia, New York, Pennsylvania and Virginia. These ticks also recently became established in Nebraska, leading Joseph Fauver at the University of Nebraska Medical Center and his team to investigate whether the ticks have brought the viruses with them.

The team analysed more than 7000 lone star ticks collected by the Nebraska Department of Health and Human Services from multiple locations across the state between 2022 and 2026. Both viruses were detected in ticks from Sarpy county, and Heartland virus was found in ticks from Richardson county, both in the south-east of Nebraska (medRvix, doi.org/rkp8).

At this point, no one in Nebraska is known to have contracted these viruses from tick bites.

Monitoring of the viruses is crucial so people can take precautions to protect themselves, says Fauver. "Obviously, you always want to avoid tick bites, but if you know you're in a hotspot area where these viruses are circulating, you might want to be extra cautious," he says.

Ageing

Supercentenarians have hypervigilant immune systems

James Woodford

THE immune systems of the oldest people on earth may not slowly wind down, but instead go into overdrive, producing specialised cells that fight off ill health.

A small study suggests that supercentenarians, who are older than 110, have high levels of a particular immune cell. This may be ramped up in response to the challenges faced in old age, such as a higher risk of cancer, which could be what's driving their extreme longevity.

"I find it remarkable that even after more than a century of life, the immune system may still retain the ability to adapt," says Kousuke Hashimoto at the University of Osaka, Japan. "Rather than simply declining with age, the immune system in these exceptionally old individuals may retain the ability to adapt to the persistent challenges that accumulate with age."

In previous work, the researchers identified this immune cell – a type of T-cell called CD4 cytotoxic T lymphocytes (CD4 CTLs) – as being a distinctive marker of supercentenarians. For the new study, they wanted to quantify how these cells – which identify and kill threats such as cancer – change as we age.

Chitetsu Watanabe is one of the few people to have lived beyond 110

They recruited eight people aged 70 to 99, 10 people aged 100 to 109, and an additional 10 people aged 110 and older. It may seem like a small number of individuals, says Hashimoto,

"Even after more than a century, the immune system may still retain the ability to adapt"

but according to Japan's national census, there are only around 150 supercentenarians in the country, "so 10 is a substantial number".

Blood samples revealed that as the groups aged, their percentage of CD4 CTLs as a proportion of their total T-cell count increased from 4 per cent to nearly 10 per cent, and then to more than 17 per cent among the oldest people (Cell Reports, doi.org/hcf7rj).

The researchers also found that these immune cells didn't show signs of T-cell exhaustion, a reduced ability to kill certain cells, such as cancerous ones or those infected by viruses. "Our findings suggest that [CD4 CTLs] may contribute to cancer immunosurveillance in extreme old age," says Hashimoto.

As we age, challenges to the immune system – such as abnormal cells like cancer and reactivated viruses – become more common.

“We think CD4 CTLs may expand in response to these challenges and if we can understand why these cells expand in some very old people and what they attack, we may eventually be able to find ways to reproduce this beneficial immune response in others,” says Hashimoto.

Last September, Manel Esteller at the Sant Pau Research Institute in Barcelona, Spain, conducted a detailed study of María Branyas Morera, who was the world's oldest living woman before she died in 2024, aged 117 years and 168 days. Hashimoto's research shows that supercentenarians are "enriched" in immune cells that attack threats to the immune system and probably also cancer cells, says Esteller.

“María underwent many infections but was able to defend against all of them,” he says. “She was the oldest [known] person in Spain that had the covid-19 virus and was asymptomatic.”

THE ASAHISHIMBUNVIA GETTY IMAGES

Climate change

We've underestimated global warming

Official projections for how much the world will warm are already catastrophic – and now it appears the world may warm by 25 per cent more than they suggest, finds Michael Le Page

CLIMATE models projecting higher warming for a given level of carbon dioxide have been discounted as unrealistic, but a new way of assessing models suggests they are accurate.

“For a given emission trajectory, we expect warming to be 25 per cent higher,” says Gergana Gyuleva at ETH Zurich in Switzerland.

The conclusion has enormous implications. For instance, a project called the Climate Action Tracker currently projects that average global surface temperatures will rise by 2.6°C above pre-industrial levels by 2100 if countries implement existing policies. Gyuleva's team's results suggest we are instead heading for 3.25°C.

How much the world will warm depends on how much more $CO_{2}$ we put into the atmosphere and how the planet responds to that $CO_{2}$ . Different climate models vary in how much warming they project for a given $CO_{2}$ level. For instance, some models suggest that a doubling of $CO_{2}$ will result in the average global surface

temperature increasing by around $1.6^{\circ}$ C above pre-industrial levels in the short term. Others suggest it could be as high as $3^{\circ}$ C.

"The range is big, and it really matters," says Gyuleva. "So, the point is to find out which ones are right."

Previously, this was done by using climate models to simulate the recent past. The models are fed data on greenhouse gas emissions over the past century or so, and their projected climate response is compared with what actually happened.

For the last report by the Intergovernmental Panel on Climate Change, a lot of these models simulated more warming than had happened, so were discounted. Based on the remaining models, the panel's last report concluded that a doubling of $CO_{2}$ would result in between 1.2°C and 2.4°C of warming in the short term, with a best estimate of 1.8°C.

But there is lots of variability in Earth's climate. For instance, during the La Niña climate pattern, cooler seawater from deeper down spreads across the ocean, resulting in cooler surface

GRAHAM HUNT/ALAMY

Climate change is making heatwaves more frequent, intense and longer lasting

temperatures. So, Gyuleva and her colleagues have filtered the historical climate record to try to remove the effect of variability due to phenomena like La Niña.

Their results suggest that natural variability limited temperature rises between 1981 and 2014 – a period that includes the so-called global warming hiatus in the first decade of the 21st century. “It’s really exactly this period 1981 to 2014, which was used in the previous papers, where you see the strongest bias down,” she says. “It was really bad luck.”

ANTHONY DEVLIN/BLOOMBERG/GETTY IMAGES

Correcting for this bias alone leads to higher estimates of the warming expected from a doubling of $\mathrm{CO}_{2}$ . But Gyuleva's team also assessed models using a new approach. Instruments on several satellites have been measuring how much short-wave radiation from the sun enters the atmosphere and how much long-wave radiation is emitted. The difference between how much heat energy enters the atmosphere and how much leaves is the most fundamental measure of global warming.

So, how well models project the trends in incoming and outgoing radiation as the world warms can be seen as a better measure of their performance than their projections of surface

"The models that best match the observed trends in radiation project much greater warming"

temperatures. But because the satellite measurements required to do this only began in 2001, this kind of assessment has only recently become possible.

The team found the models that best match the observed trends in short-wave and long-wave radiation project much greater warming. Based on the models that perform best at both kinds of assessments, the team concluded that a doubling of $CO_{2}$ will lead to a rise of between 1.9°C and 2.6°C, with a best estimate of 2.25°C (Earth's Future, doi.org/rkn8).

"Our results need to be confirmed by more evidence from different sources before we can be fully confident," says Gyuleva.

"This seems to be a sound approach," says Trude Storelvmo at the University of Oslo in Norway.

"In my opinion, the filtering out of natural variability is a valuable step forward," says Drew Shindell at Duke University in North Carolina, although he isn't yet convinced that the short-wave and long-wave trends are a good way to assess models.

"I think it's a plausible conclusion and a reasonable approach," says Steven Sherwood at the University of New South Wales in Sydney.

But the findings are based on climate models, and none of them reproduces the key phenomena very well, says Sherwood. “To do a better job of this, we really need better climate models. On the other hand, the steep temperature rises of the last few years are going to lead to higher estimated ranges regardless of method.”

In climate jargon, what Gyuleva's team has estimated is known as the transient climate response. It is just one of three types of climate sensitivity, which vary in terms of the timescales involved.

In the decades after a doubling of $CO_{2}$ , various feedbacks will kick in that lead to further warming, such as the continued warming of the oceans. The response after these medium-term feedbacks have kicked in is known as the equilibrium climate sensitivity. Last year, it was shown that the trends in short-wave and long-wave radiation suggest the equilibrium climate sensitivity is higher than previous estimates.

There are also very slow feedbacks that take centuries or more, such as the melting of ice sheets. Estimates of this earth system sensitivity – the full response to a doubling of $CO_{2}$ over many millennia – are as high as $7^{\circ}C$ of warming.

However, if atmospheric $CO_{2}$ levels were lowered by carbon removal, some of the longer-term consequences would be avoided.

Melted ice from Greenland is intensifying European heatwaves

Alec Luhn

THE melting of the Greenland ice sheet is causing more extreme heat in Europe, contributing to the heatwaves experienced there so far this year.

The surge of cold meltwater into the North Atlantic is altering ocean temperatures and shifting the jet stream to favour heatwaves in western Europe and the Mediterranean, a study undergoing peer review has found. As Greenland ice loss accelerates, Europe may suffer more frequent and intense heatwaves than climate models currently predict.

This newly discovered “atmospheric bridge” played a major role in the unprecedented heatwaves this summer, according to Marilena Oltmanns at the UK’s National Oceanography Centre, who is one of the study’s authors.

BRITISH AN TARCIC SURVEY

“Europe heats up faster. This is because of the jet stream shift,” she says. “In the future we will think this [year’s heatwave] was not a heatwave; this was nice.”

The rise in global temperature has increased heatwaves. But it is still debated whether changes in prevailing weather patterns are also contributing.

In the past 30 years, high-pressure ridges – large areas of elevated air pressure that look like an upside-down U on weather maps – have been forming more frequently over Europe, the study finds, leading to sunny skies and extreme heat. These ridges are caused by waviness in the jet stream, the band of strong winds that circles the northern hemisphere. When the roughly straight-flowing jet stream heads northward around western Europe, it allows warm air from the central Atlantic and Sahara desert to pour into the ridge that has been created.

The researchers say the formation of the ridges can be traced back to Greenland through a cascade of climate events. In the summer before years of extensive ridging, an average of 81 billion tonnes more meltwater rushed off the ice sheet, they found (Research Square, doi.org/rkxx). The cold meltwater lowers sea surface temperatures as it spreads

"In the future we will think this year's heatwave was not a heatwave; this was nice"

through the North Atlantic in the autumn and winter. Its freshness inhibits mixing with saltier, warmer deep layers, further cooling the surface.

These pulses of meltwater cooling come on top of the “cold blob”, a long-term decline in ocean temperatures in the North Atlantic that some scientists blame on the weakening of the Atlantic Meridional Overturning Circulation (AMOC) system of currents. Together, they create a sharp temperature front

Meltwater from Greenland is shifting the jet stream

between the cold waters and the warmer surface waters further south in the Atlantic.

As warm air masses rise to the south and cold masses descend to the north of the front, they begin to swirl and generate cyclones. This makes the jet stream wavier. It tends to meander southward of these cyclones, then back northward over western Europe, creating the persistent, high-pressure ridges and heatwaves there.

Meltwater from Greenland could be forcing the jet stream to meander, according to John Methven at the University of Reading, UK. But the weakening of the AMOC or changes in atmospheric circulation over the North Atlantic could also be causing this shift.

“That’s the weak link as far as I’m concerned, the link between the cold blob and whatever’s melting in Greenland is not established, and there might be other things going on,” he says.

Briefing Cancer

How close are cancer vaccines?

A personalised mRNA cancer vaccine has produced positive results in a final-stage trial involving skin cancer, raising hopes for the approach, discovers Michael Le Page

BECAUSE every cancer is different, researchers have for decades been working on personalised cancer vaccines designed to make the immune system recognise and attack cancers specific to each individual. The results have been mixed, but now a personalised cancer vaccine has produced positive results in a final-stage, or phase III, trial for the first time.

Moderna announced on 19 August that people who had had surgery to remove severe

"Trials are already underway for certain kinds of lung, kidney and bladder cancer"

skin cancers, called melanomas, survived longer without the cancer returning or spreading to other parts of the body if they were given a personalised mRNA cancer vaccine in conjunction with another immunotherapy, compared with receiving that immunotherapy alone.

The hope is that this personalised vaccine will work for many other types of cancer. Trials are already underway for certain kinds of lung, kidney and bladder cancer. The company BioNTech is also testing personalised mRNA cancer vaccines in human trials.

What is a cancer vaccine?

Cancers are caused by cells accumulating mutations to the point where they go rogue and start multiplying out of control. Mutations in proteins that protrude from the surface of cells also make cancer cells “look” a bit different to other cells. This allows our immune system to recognise and kill many cancers without us ever knowing about it. But some slip through the net – the idea of cancer vaccines is to teach the immune system to recognise them

How are these vaccines made? For the Moderna vaccine, called intismeran autogene, a sample of a person's cancer is sequenced. It is then compared with their healthy blood cells to identify surface proteins on the cancerous cells that have distinctive mutations. Up to 34 are selected, and mRNAs coding for those proteins are made and put in the vaccine.

What happens next?

The cancer vaccine is injected into muscles. Cells in the muscles then temporarily produce the cancer proteins, triggering an immune response to them. This supports treatment because the immune system “remembers” this response and is then primed if it later comes across cancer cells.

Why use mRNA vaccine technology? It is possible to use other kinds of vaccines, but mRNA is faster and less costly. Making protein-based vaccines requires using living cells, which is time-consuming and requires many safety checks. For people with

Cancer vaccines that make the immune system recognise and attack cancers are being made

cancer, the faster a vaccine can be created, the better.

What immunotherapy was the vaccine combined with for the trial? It is a so-called anti-PD-1 therapy consisting of an antibody called pembrolizumab (Keytruda). Many cancers evolve to evade immune attack by putting up a “don’t attack me” sign. Immune cells “read” this sign with the help of a protein on their surface called PD-1.

Blocking PD-1 makes the “don’t attack me” sign invisible to immune cells, so they attack regardless. But, anti-PD-1 therapies work only if the immune system is already trying to attack a cancer, hence the advantage of combining them with a cancer vaccine to provoke that attack.

How well did the combined treatment work?

Moderna hasn't told us yet. The company has announced only that the results of the phase III trial "demonstrate statistically significant and clinically meaningful improvements" in survival times.

The trial involved 1137 people who, after surgery to remove melanomas, were randomly

ADAM GLANZMAN/BLOOMBERGVIAGETTYIMAGES

assigned to get both the vaccine and pembrolizumab, or just pembrolizumab. What we do know is that in an earlier trial involving 157 people with melanoma over five years, the combination of the vaccine and pembrolizumab cut the risk of recurrence by 50 per cent and the risk of spread to other parts of the body by 60 per cent, compared with pembrolizumab alone.

When will this treatment become available?

Analysts speaking to the Financial Times say it could be approved next year. But because it is an individualised treatment, scaling up production will be tricky.

It also isn't clear yet what the pricing will be. A January report by Reuters suggested it would cost around \$200,000 per treatment.

Will this work for all cancers? The hope is it will work for many cancers, but we need more phase III trial results to be sure.

Are there any approved cancer vaccines already?

Not like intismeran autogene. Some sources will tell you there are three approved vaccines for treating cancer: Provenge, BCG and T-VEC. But Provenge is a treatment that involves removing immune cells from a person's body, exposing them to a prostate cancer protein, then replacing them. BCG is the tuberculosis vaccine, which just happens to help treat some bladder cancers. And T-VEC is a live herpes simplex virus engineered to kill cancer cells.

What about the HPV and hepatitis vaccines?

These vaccines prevent infections by viruses that can cause cancers, rather than targeting cancers directly, which is an even better approach.

Neuroscience

We may now know which neurons dictate when it is time to go to sleep

Chris Simms

RESEARCHERS have identified specific groups of neurons in the brains of mice that are activated after the rodents have been awake for a long time, driving them to sleep. If a similar system exists in humans, it could be targeted to help improve treatments for some sleep disorders.

Inside the brain, several neural circuits regulate sleep and wakefulness, including neurons in the hypothalamus that act like a switchboard to control the process.

One or more of these circuits might be involved in why it is that, the longer we are awake, the more irresistible the urge to close our eyes becomes. But what actually drives this impulse is a mystery.

To find out, Will Joo at the University of Basel in Switzerland and his colleagues have compared brain activation patterns in mice during normal sleep-wake cycles, sleep deprivation and recovery sleep. They did this by tracing how much protein is produced in cells by a gene called Fos, which is turned on when neurons are in the process of activating.

“We saw that very specific parts of the brain were correlated with wakefulness,” says Joo. This allowed the researchers to identify areas in which activity correlated with time spent awake.

Within one of these regions, the median raphe nucleus, they found two distinct neuronal populations that were increasingly activated the longer the animals stayed awake and that got quieter again after sleep started. These populations are the GABAergic and serotonergic neurons, so named because the chemical messenger molecules they respond to are gamma-aminobutyric acid (GABA) and serotonin, respectively.

"This is an outstanding study.

What drives the urge for us to feel tired and fall asleep?

INDRANIL ADITYA/NURPHOTO VIA GETTY IMAGES

The whole-brain Fos mapping is a real tour de force," says Chiara Cirelli at the University of Wisconsin–Madison.

To see whether these neuronal populations became more active simply because the mice had been awake for a long time or whether they do something more, Joo and his colleagues then used viruses to deliver a genetic package to the neurons to either activate or inhibit their action.

When both sets of neurons were artificially activated, the mice slept for twice or three times as long as control mice and spent more time in non-REM slow-wave sleep – in a way that resembles the deeper recovery sleep humans have after we have been up for a long time.

"Not only does it increase the amount of time the animal spends sleeping, but it also increases the quality of the sleep," says Joo.

When the neurons were inhibited, the mice slept about 70 per cent less than control mice – spending around 6.5 extra hours awake a day – without showing more of the anxiety-like

behaviours that usually come with sleep deprivation for mice. However, the lack of sleep was lethal in about 17 per cent of the mice (Nature, doi.org/rkqf).

The strong effect of activating and inhibiting the neurons means they don't just signal that an animal has been awake, says team member Alex Schier, also at the University of Basel.

"These neurons are crucial to promote sleep, and may be key parts of the circuitry that generates sleep drive"

"They are crucial to promote sleep, and may be key components of the neural circuitry that generates sleep drive."

The study “adds to a growing body of work showing that there are distinct sleep circuits that are activated by wakefulness, yet promote sleep”, says Mark Wu at Johns Hopkins University in Maryland, whose research has shown that neurons in a brain region called the thalamic nucleus reuniens track the gap between how much sleep we need and how much we actually get.

It isn't known if the same sleep-driving neuron populations exist in humans, but if they do, the findings could open doors to new therapies for some sleep-related conditions – once we know how the cells are communicating with other brain regions, say the study researchers.

Projections from these neurons in the median raphe nucleus link to the preoptic hypothalamus, and these areas are known to work together to control the sleep-wake cycle.

"The effect of this major neuronal path on sleep is very strong, but, as we have learned since we started manipulating neuronal circuits, each path provides nuances. The brain has many ways to wake up and fall asleep," says Luis de Lecea at Stanford University in California.

For example, there are hints from experiments in fruit flies that mitochondria, which power cells, in the brain may play a crucial role in the onset of sleep.

Artificial intelligence

Why AI watermarks won't work

AI companies have begun watermarking their models' outputs to improve transparency and tackle misinformation – but the approach has limitations, reports Matthew Sparkes

AS AI chatbots have taken off, you have probably found yourself wondering whether a piece of text you are reading was written by a human or a machine. Now, in response to European Union legislation, many AI firms are adding watermarks to text, images and video created by their models with the aim of letting you do just that. The hope is that these watermarks will help stem the spread of misinformation and bring transparency to the use of AI, but experts point out that, at least when it comes to text, watermarking is unlikely to work.

"Watermarking isn't going to suddenly solve what the EU maybe are hoping it's going to solve"

The requirement to watermark is part of the EU's AI Act, and came into force on 2 August. OpenAI, the maker of ChatGPT, is already watermarking images and audio, and says it plans to add text watermarking in future. Anthropic, the firm behind Claude, has announced that future models will watermark text. The AI Act allows a grace period for models that have already been deployed, but all models must provide watermarking by 2 December this year.

FG TRADE/GETTY IMAGES

James Padolsey at NOPE, an AI safety firm, says that while image and video watermarks are more reliable because they are composed of dense and multi-layered data, watermarks in text will always be more difficult to implement.

One approach to watermarking text is to apply a mathematically detectable pattern to the choice

OpenAI is already watermarking images made using ChatGPT of words. Large language models like ChatGPT create sentences by appending whichever word is statistically most likely to come next. By varying this selection of words, perhaps by choosing alternately the most likely next word and then the second, in repetition throughout a sentence, models can embed a pattern without changing the overall meaning of the text. Under the AI Act, companies will be required to offer tools that can look for these patterns, allowing people to identify AI-generated text.

But Padolsey points out that these watermarking patterns are susceptible to deletion by even light edits. "It's not useful, and the way that people are going to try and use it will be incorrect," he says.

To drive home his point, Padolsey has created an online tool called declaude, which takes AI-generated text and lightly changes it in such a way that existing AI detectors no longer work. The same process should work on watermarked text, he says.

Even without such tools, Padolsey says that some open-source AI models that

Do you think you can tell when text has been generated using AI?

don't abide by the EU's rules will always be available, because these aren't controlled by any one tech firm. "Any bad actors will just use those. And they'll probably be cheaper and easier to run," he says. "If you're a nasty individual or a bad actor and you're trying to generate this stuff en masse, it's not hard to do."

"These rules will help people recognise when they are interacting with AI or when content has been generated or altered by AI," says a spokesperson for the European Commission. "The adversarial robustness of marking and detection solutions must be assessed in terms of resilience to malicious behaviour, such as copying, removal, regeneration and modification attacks on the markings."

While Padolsey's concern is that the watermarks are easily removed, others are worried about false positives – when text is found to be AI-generated in error, or because a model had even a minor role in producing a piece of work. Peter Scarfe at the University of Reading, UK, says lots of companies have pitched products to his department claiming to be able to help spot student plagiarism, albeit using other methods than watermarking, while also including small print that says their systems shouldn't be used to conclusively prove or punish incidents of plagiarism because they aren't infallible.

Not a magic bullet

"If it says there's a 67.8 per cent chance that this text was generated in some way by AI, as an educator, how would I act upon that? I'm not really sure," says Scarfe. "These systems don't work perfectly, and because of that, they're of limited use."

Watermark-detection tools will work in a similar way and could flag text that is simply processed

SHUTTERSTOCK/ROKASTENYS

by AI, not just text entirely generated by it. Students could find themselves in hot water even if they just asked a model to scan their manually written work for grammatical mistakes prior to submission.

"Watermarks may just create enough friction that they will put a brake on some nefarious uses of AI"

Scarfe says that AI can be used for a host of purposes and doesn't inherently spread misinformation, so perhaps a blanket policy on watermarking is a blunt instrument, but points out that it can make mistakes and hallucinate facts, and can also be used to disseminate lies quickly – such as by powering disinformation bots on social media.

“Because these technologies came in so quickly, and they’ve accelerated so quickly, everyone in every area of life is kind of on the back foot,” he says. “[But] I don’t think [watermarking] is a magic bullet. I don’t think it’s going to suddenly solve what the EU maybe are hoping it’s going to solve.”

But perhaps watermarks may just create enough friction that they will put a brake on some nefarious uses of AI, such as mass creation of social media posts pushing disinformation. Erman Ayday at Case Western Reserve University in Cleveland, Ohio, says that even if removing a watermark takes a trivial effort, this can, over time, make certain kinds of misuse impractical.

"Anybody can tamper with the watermark if they spend enough time on it, but then, when you spend enough time on it, then maybe you should have just generated the content yourself in the first place, right?" says Ayday.

Human evolution

Did humans really evolve to be active only in the day?

Michael Marshall

HUMANS are generally considered diurnal, or mainly active during the day, but data from hunter-gatherers suggests they are fairly active at night.

This finding means that the human sleep pattern is closer to that of some lemur species, says David Samson at the University of Toronto in Canada, author of The Sleepless Ape. Instead of being primarily active in the day and restful at night, humans may be “cathemeral”, meaning they spread their activities throughout each 24-hour period. “The key point is that day and night activity are both regular parts of the behavioural repertoire,” says Samson.

To find out if humans are diurnal or cathemeral, Samson compiled data collected from humans and compared it with recordings from non-human

Members of the Hadza community were active during the day and night primates: seven species of lemur and one species of gibbon. In all cases, activity was recorded using wrist-mounted accelerometers. Because many humans now sleep in permanent houses and have access to electricity, Samson used accelerometer recordings from Hadza people: hunter-gatherers living in Tanzania who generally sleep in the open or in shelters made from branches and grass, without artificial light.

The Hadza were active for an average of 52 minutes per hour during the day and 24 minutes per hour during the night. This contrasted with the strongly diurnal gibbons, who were active for only 2 minutes per hour at night. The humans were closer to the cathemeral lemurs, who were active for 45 to 48 minutes per hour in daytime and 18 to 21 minutes per hour at night (Journal of Human Evolution, doi.org/rkn6).

Samson also built a mathematical model of primate evolution, to see whether all primates were evolving towards the same optimal pattern of sleep and activity. He found that humans didn't match other primates.

Flexible rhythms

“Human sleep-wake activity may have evolved under somewhat different selective pressures than in other primates,” he says.

Samson argues that humans have evolved to defy our internal circadian rhythms. While our body clocks encourage us to be active in the day and asleep at night, our societies and technologies allow us to do otherwise. By using shelter, hearth, environmental preparation, light and lookouts – which Samson groups under the acronym SHELL – we can remain active at night, he says. “That SHELL did not erase our biology. It buffered us from some of the ecological costs of being active or asleep at unusual times.”

KENNETH WHITTEN/UNIVERSAL IMAGES GROUP VIA GETTY

“Humans show very flexible patterns in when we sleep and when we’re active,” says Kathleen Reinhardt at the University of Calgary in Canada, but she thinks we are still primarily diurnal.

The definition of cathemeral is stricter than simply evidence of activity at night, she says. "We're talking about more meaningful behaviours such as feeding and travelling, and those kinds of behaviours occurring in both periods, not just simply detecting movement in the light and the dark," she says. The accelerometer data isn't enough to determine the nature of nighttime activity, she says.

Health

Newly discovered immune hubs may keep our brain healthy

Carissa Wong

STEVE GSCHMEISSNER/SCIENCE PHOTO LIBRARY

HUBS of immune cells in the skull may help to defend our brain from ill health. The discovery, based on a study into the brains of mice and human genetic data, implies that targeting these hubs with drugs could bring new treatments for conditions where immune cells go rogue, such as brain cancer.

“It’s an important step forward in understanding the brain’s immune response,” says Vesa Kiviniemi at the University of Oulu in Finland, who wasn’t involved in the study. As well as being applicable for cancer, “it’s relevant for understanding things like infections, inflammatory brain diseases, multiple sclerosis [and] neurodegenerative diseases”, he says.

Immune cells called T-cells and B-cells are activated to act in the brain if they are presented with signs of threats, like fragments of tumours, in the lymphatic system. Now,

Jonathan Kipnis at Washington University in St. Louis, Missouri, and his colleagues have uncovered another way these cells are activated.

By imaging and analysing the skulls of mice, the team found that B- and T-cells cluster together with immune cells

"The immune hubs help to generate an anti-cancer immune response"

called antigen-presenting cells in immune hubs at the back of the skull. “These haven’t been described before,” says Kiviniemi. These hubs resemble lymph nodes, where antigen-presenting cells expose threats to T- and B-cells.

The team thinks these immune hubs are also in people. This is based on gene activity data collected from human skulls in prior studies, which suggest that T-cells were

T-cells (green) act when shown threats like brain cancer cells (purple)

activated and helped activate B-cells in this part of the body. "It indicates the same is present in humans," says Kiviniemi, although further studies analysing the skulls of cadavers are needed to confirm this.

To explore whether these hubs launch protective immune responses, the researchers injected cancer cells into the brains of mice. They then injected half the mice beneath the scalp with an experimental drug that disrupts the activation of B- and T-cells in the skull. This works by blocking a protein called CD40L on antigen-presenting cells, which helps them activate these immune cells. The remaining mice received saline injections.

The mice that received the drug went on to live for about 25 days, on average, after the tumour injection, whereas those in the placebo group lived about 30 days. This suggests the immune hubs help to generate an anti-cancer immune response, says Kiviniemi.

In another experiment, a group of mice was given the same tumour injection, but this time, half received three drugs that enhanced the activation of B- and T-cells in their skull. These mice survived for about 10 days longer than others that got placebo injections (Nature, doi.org/rkxw).

If the same immune hubs also exist in people, targeting them could bring new therapies for many brain-related conditions, says Kiviniemi. "We could figure out how to awaken and strengthen these [hubs]," he says.

Environment

Cookies made from plastic bottles could reduce waste

Matthew Sparkes

PLASTIC bottles and agricultural waste have been converted into vanilla-flavoured cookies via genetically engineered yeast.

Sandhya Jayasekara and her colleagues from Southern Illinois University Carbondale created the cookies by first putting used plastic bottles and discarded sweetcorn plant stalks and leaves through a process called oxidative hydrothermal dissolution. This uses heat to break down the materials into components that microbes can digest.

Those components are then fed to yeast that has been genetically altered using CRISPR gene-editing technology. This allows the yeast to convert the components into a variety of useful ingredients, including proteins, fats and acids, as well as vanilla flavouring and beta-carotene, which the body can convert into vitamin A. Finally, fibre, starch and sweetener are added to the mix, and the paste is extruded through a 3D printer into any desired shape. The resulting cookies have been dubbed $\mu$ Bites, pronounced microbites.

The researchers, who presented the work at the fall meeting of the American Chemical Society in Chicago, are eager to try $\mu$ Bites for themselves as soon as their university approves the experiment. But they are confident that the cookies are both safe and tasty.

"I think it has a pleasant, appealing aroma," says Jayasekara. "We've not tasted it, but it smells good – like a real cookie."

The team hopes μBites may help with plastic pollution, but not everyone is convinced. Jason Hallett at Imperial College London says the research is fun and interesting, but also, in all likelihood, a practical dead end. "We produce 400 million tonnes a year of plastic waste. You're not going to turn it all into cookies," he says.

Female clones of a male mouse

A gene-editing tool can reliably turn male mouse embryos into healthy females – a feat that could be valuable for conservation, finds Chris Simms

MALE mouse embryos have been turned into females by a procedure that involves eliminating the Y chromosome. The technique relies on CRISPR gene editing and has been used to create healthy and fertile female and male mice, both of which are clones of the father.

The method may become useful for conserving animals that are so close to extinction they are down to just one male or only a few.

In animals that reproduce sexually, running out of males or females can mean the end of the line for a population or species. However, some species can turn to asexual reproduction, such as by parthenogenesis, in which unfertilised eggs of animals including lizards and sharks can develop into a new animal. This produces offspring that are identical clones of the parent and are the same sex.

A few animals, like Okinawa rubble gobies (Trimma okinawae), can change sex to ensure sexual reproduction can continue. Inspired by these colourful marine fish, Takashi Ishiuchi at the University of Yamanashi, Japan, and his colleagues came up with a method to change the sex of a male clone embryo, so sexual reproduction between individuals could take place – albeit exceedingly closely related ones.

Some attempts at cloning male mice have, by chance, produced a female when the male-determining Y sex chromosome was deleted, leaving the specimen with just one X chromosome (known as XO), rather than an XY pair.

Ishiuchi and his colleagues developed a CRISPR-based tool called Y-CUT to reliably eliminate

A male mouse meets his male and female clones the Y chromosome from male mouse embryos.

First, they inserted DNA from the blood cells of a male mouse into mouse egg cells that had already had their own DNA removed – a technique called somatic cell nuclear transfer. Immediately after fertilisation,

"The clones were fertile, and they produced offspring that seemed healthy"

they treated some of these cells with Y-CUT and transferred them to surrogate female mice.

All of the mice born after the use of Y-CUT were female clones – genetically identical to the father apart from lacking the Y chromosome. Other offspring, which hadn't undergone Y-CUT, were all male clones (bioRxiv, doi.org/rkjv).

"We call this dual-sex cloning," says team member Shogo Matoba at the RIKEN BioResource Research Center in Tsukuba, Japan.

Introducing the father mouse to his clones was a very special moment, says Ishiuchi. "He could see the female version

of himself, which is like sci-fi."

The clones were fertile, and when they mated together, they produced offspring that seemed healthy. “This means sexual reproduction was initiated from a single male genome,” says Matoba. “The kids have survived over a year so far with no clear defects.”

Ishiuchi hopes that the Y-CUT technique could help revive species that are highly endangered or extinct, and for which only frozen tissue remains in cell banks.

"It would be incredibly valuable if this were successful in species other than mice," says Teruhiko Wakayama, also at the University of Yamanashi, who wasn't involved in the study.

Conservation potential

Ben Novak at wildlife conservation organisation Revive & Restore, which has been involved in cloning efforts with endangered species including Przewalski's horse, thinks similarly. "They are laying the foundations for a future of manipulating chromosomes that is probably going to be important in many cases for conservation," he says.

Novak agrees the technique may be useful for frozen tissue samples from endangered or extinct species, where we have samples from only one or two individuals.

A previous study generated both male and female clones from a male mouse embryonic stem cell line, but it used a method that is unlikely to work in other species, unlike somatic cell nuclear transfer, which has been successfully applied to many mammal species and works with living or stored donor material.

One issue is that, though an XO female mouse is healthy and fertile, that isn't the case in many other species.

“Y-deleted offspring are not fertile in humans and horses, and therefore this is likely to be the case in multiple wildlife species,” says Linda Penfold of SEZARC, a group of zoos and conservation centres based in the US. She also says there could be ethical concerns with producing Y-deleted animals.

Ishiuchi accepts this, but notes that there is research showing that a drug called Reversine can make XO cells gain a second X chromosome, so it might be possible to do this in combination with Y-CUT to make resulting females fertile in a wider range of species.

Penfold also warns that breeding from clones of a single animal would create genetic issues. “There is a lack of genetic diversity, which is critical for healthy populations,” she says.

Novak is more positive about that aspect. “Sexual reproduction with one individual is certainly better than no sexual reproduction,” he says, but he adds that we should do everything we can to never have to use such tools.

SHOGO MATOBA, TAKASHIISHIUCHI ET AL. (2026)

Habitat protection and reducing hunting are more effective at conserving species, says Penfold.

Environment

Trial to avoid contrails set to take off

Air traffic controllers will use AI-generated forecasts to advise planes flying over the Atlantic to adjust their altitude in an effort to reduce the climate impact of contrails, reports Madeleine Cuff

PLANES flying across the North Atlantic Ocean later this year will be re-routed to avoid causing condensation trails, in a first-of-a-kind trial to address the global warming impact of contrails across an entire airspace.

Starting in November, air traffic controllers managing the Shanwick airspace on the eastern side of the North Atlantic will advise planes to adjust their altitude to avoid creating fresh contrails. The controllers will be

"The climate benefit of avoiding contrails is worth any extra fuel burn from shifting flight paths"

using AI-generated forecasts of atmospheric conditions from Google that predict where contrails will form.

“We can predict the regions that are going to form persistent climate-warming contrails, and we can re-route planes around them,” Marc Shapiro at Contrails.org, a research organisation that is participating in the project, said during a press briefing on 17 August.

Contrails are formed when soot particles emitted by jet engines

trigger the formation of ice particles, leaving a condensation trail that can last for hours. During the day, these persistent contrails help to reflect solar radiation back to space, which has a cooling effect. But they also trap heat below them, causing warming. Overall, contrails are thought to account for around one-third of aviation's climate impact.

The Shanwick airspace is a contrail hotspot and a busy area for planes crossing the North Atlantic. Modelling suggests the airspace is responsible for about 5 per cent of global contrail warming, and most of this effect occurs during the winter season between November and February.

The £5 million Operation Blue Skies project will re-route flights in the evening and night, which have the biggest warming effect. Pilots entering the Shanwick airspace will be instructed to adjust their altitude by up to 2000 feet (around 600 metres) to avoid atmospheric conditions where contrail formation is likely. The advice will be issued under the same procedures already used by air traffic controllers to help pilots avoid turbulence or other weather conditions.

WIRESTOCK, INC./ALAMY

The trial will run during two four-month winter trials over 2026-2027 and 2027-2028, with approximately 20 to 40 test days during each winter season. Around 10,000 flights are expected to pass through the airspace during trial hours each year, and hundreds will need to be re-routed to produce a statistically significant dataset for analysis. Researchers from the University of Cambridge and Imperial College London will analyse the data from the trial.

The climate benefit of avoiding

Contrails occur when soot emitted by jet engines causes ice particles to form

contrails is worth any extra fuel burn from shifting flight paths, Paul Hodgson at Google UK said during the briefing. Re-routed flights may have to burn an additional 100 kilograms of fuel per aircraft while crossing the airspace, resulting in an additional few hundred kilograms of $CO_{2}$ emitted. By comparison, “the contrail impact on those flights, especially at those times of day, is measured in the tonnes or even the tens of tonnes”, he said.

AI could help design more efficient plane wings

Avoiding contrails is just one of the ways AI is being used to reduce the climate impact of air travel (see main story). AI agents have also devised a way to reduce the friction of an airplane wing model.

Steven Brunton at the University of Washington and his colleagues created a platform, HydroGym, in which many AI agents could tweak how a fluid flowed over virtual objects to decrease the drag they experienced. The AI agents were first trained using relatively simple computer simulations, before tackling the task by using a trial-and-error approach known as reinforcement learning.

The team discovered that the AI agents could apply lessons learned from more simple examples in a computer simulation to work out more complex scenarios – even without access to a computer simulation of those complex scenarios (Nature, doi.org/rkjs).

For instance, after learning how to control flow of a turbulent fluid in a flat channel, the agents worked out controlling fluid surrounding a curved, three-dimensional airplane wing model, managing to decrease the energetically wasteful friction between the wing and the fluid by 38 per cent.

"The financial and the ecological impact is profound for the tiniest improvements," says Brunton. Karmela Padavic-Callaghan

The project follows previous trials of Google's AI contrail forecasts, which have been used to inform airlines' flight plans before take-off and routes set by air traffic dispatchers. This work suggests re-routing a plane can cut contrail formation by 50 to 60 per cent.

But this is the first time contrail forecasts will be used across an entire airspace, and airlines will not be able to opt out. "This is what we call an airspace-scale trial," said Hodgson.

Physics

A 30-year mystery over lost neutrinos may have been solved

SOME missing neutrinos in experiments have been puzzling physicists for more than three decades, but a new theoretical study suggests there may be no mystery after all. Rather than offering a hint of a new kind of particle, the unexpected result may come down to a calculation oversight, which would reaffirm the laws of particle physics instead of challenging them.

Neutrinos are ghostly, slippery particles that are hard to detect and can spontaneously oscillate between their three types. Since the 1990s, several experiments with gallium-based detectors have found about 20 per cent fewer neutrinos than expected, kick-starting ideas about some neutrinos having turned into a new type of “sterile” neutrino that interacts with matter less and so eludes detection. If this were the case, the standard model of particle physics, which tabulates all existing particles and forces, would have to be amended.

Matteo Cadeddu at the University of Cagliari in Italy and his colleagues now propose that such a drastic intervention may

"Several tests with gallium-based detectors have found 20 per cent fewer neutrinos than expected"

not be necessary. Their calculations show that the problem might lie in the mathematical modelling of the experiments.

In the experiments, when an atom of gallium is hit by a neutrino, it is transmuted into an atom of germanium and an electron. To count the neutrinos in the gallium-based detector, researchers measure the interaction between the germanium and the electron, then work backwards.

Until now, they have routinely assumed that the quantum wave functions of the electron and the neutrino, which mathematically encode their properties and behaviour, don't vary across the nucleus of the transmuting atom. This was the detail that Cadeddu and his colleagues focused on.

To their surprise, they discovered that dropping this assumption could account for the reported 20 per cent neutrino deficit, given certain properties of the atom's nucleus (Physical Review Letters, doi.org/rkgq). "We do not need to add a new particle or a new interaction," says Cadeddu.

"It is an interesting lead that deserves further study," says Joachim Kopp at the Johannes Gutenberg University of Mainz in Germany. In his view, attempts to extend the standard model to fit the experiments require "really bizarre and fine-tuned

theoretical models", so the new analysis is more promising. But details of the nuclear structure of gallium and germanium would have to be independently measured or calculated to fully resolve the neutrino mystery, he says.

Ante Ravlić at Michigan State University says the new study is a step in the right direction. "We have to wait for more sophisticated calculations based on microscopic nuclear models. However, the work does demonstrate a clear impact of treating the electron and neutrino wave function more rigorously in the possible resolution of the anomaly."

Ravlić expects to see quite a few follow-up studies from nuclear physicists, especially as neutrinos don't just test the standard model, but are also important for astrophysical processes such as explosions of massive stars. ■ KPC

Ageing

Older women reap major benefits from lifting heavy weights

WOMEN in their 50s to 80s who learned to deadlift 50 kilograms became stronger, leaner and more mobile.

In the study, led by Juezhen Zhang at the University of Taipei in Taiwan, 32 postmenopausal women aged 52 to 84 were randomly allocated to participate in a strength-training programme or to maintain their usual lifestyles.

Those in the strength-training group were shown how to use a hexagonal barbell, a six-sided frame that goes around the body and has weights on the side. For

24 weeks, the women did two supervised sessions per week involving multiple sets of deadlifts.

Over the course of the study, the women in the strength-training group lost fat and gained lean muscle mass around their middles, while simultaneously losing fat from their thighs and buttocks. Their strength also improved: the maximum amount they could deadlift five times in a row increased from an average of 45 to 58 kg.

By the end, the women also wobbled less while standing with their eyes closed and were able to walk further in 6 minutes, indicating better balance and mobility. In contrast, those asked to maintain their usual lifestyles didn't see any improvements (Experimental

CHUIKWANHO/AFP VIA GETTY IMAGES

If you learn to deadlift, you too could one day enter a competition for older lifters

Gerontology, doi.org/rkgp).

Most strength-training programmes for older adults involve the use of machine weights. But Zhang's team believes that lifting free weights, like barbells and dumbbells, is probably more beneficial because it requires activation and coordination of more muscles, including stabiliser muscles in the core that are important for real-life tasks.

Elissa Burton at Curtin University in Perth, Australia, says older women who are interested in doing strength training for the first time should start with a qualified trainer to make sure they perform the exercises correctly and don't injure themselves. Alice Klein

Comment Neurodiversity

Suggesting that ADHD is a myth is pseudoscientific and dangerous ADHD is a real condition. An ill-informed documentary that suggests it is a modern, social construct could cause harm, says Caroline Williams

IF I were to accept the premise of Channel 4's documentary The Great ADHD Myth?, then I have spent years labouring under a false belief. The diagnosis I received for my so-called ADHD is apparently an arbitrary concept, invented by popular consensus and promoted by publications like New Scientist.

National Health Service doctor Max Pemberton presented this idea in a documentary that aired in the UK on 18 August, and in the days that followed, many articles and columnists commented on why it's so offensive. While I have my own thoughts on that – and trust me, I'm offended – the most urgent issue to address is the show's suggestion that ADHD may not be a real condition.

One claim is that ADHD is a new condition, dreamed up to force square pegs into the round holes of modern life. This is simply not true. An early documented medical description of what we now call ADHD dates from 1798, when Alexander Crichton from the Royal College of Physicians described people who had problems paying attention for long enough to complete tasks, saying they had “the fidgets”.

But the results of an analysis published in The Lancet this month

Since then, the condition has been medically described and renamed (who can forget “abnormal defect of moral control in children” in the early 1900s?) many times. While Pemberton is right to note that rates of diagnosis have soared in recent years, diagnoses in the UK are still below estimates of population prevalence from around the world. Studies that screen large groups of people for symptoms consistently put rates of ADHD in the general population at around 5 per cent in children and 2 to 3 per cent in adults.

MINNOW PRODUCTIONS

found that just 1 per cent of the population is diagnosed overall in England. Even among boys and young men, the group most likely to be diagnosed, the recorded rates are 2.9 per cent and 3.6 per cent, respectively. If anything, ADHD is underdiagnosed on these shores.

Then there is the claim that ADHD is a social construct, not a neurodevelopmental condition. This is something Pemberton ponders after talking to neuroscientist Katya Rubia at King's College London. Rubia says that while there is evidence of differences between the brains of people with ADHD and neurotypical people at the group level, it isn't possible to diagnose ADHD from a brain scan.

She is right, of course. You can't diagnose ADHD from brain scans. The same can be said for most, if not all, psychiatric and neurodevelopmental conditions. Most of these, like ADHD, are diagnosed by assessing symptoms and the effect they have on a person's everyday ability to function. "My interview was hours long and they cherry-picked the bits that suited their push for presenting ADHD as a myth," Rubia told New Scientist. "Biased

"We have evidence that treatments, drug-based or otherwise, improve quality of life and self-esteem"

Max Pemberton, a National Health Service doctor, fronted the show

programmes like this one that push their view of ADHD as a mythical entity are irresponsible and will have detrimental effects."

Pemberton proceeds to seek out a diagnosis for ADHD via an online assessment, despite thinking he probably doesn't have it. In reality, ADHD is diagnosed via a lengthy process that includes multiple questionnaires and an interview. To get a diagnosis, ADHD traits have to be causing significant problems in two or more settings, such as at work or with managing bills. If it doesn't affect your life, mental health or relationships, you don't qualify for the second D of ADHD: disorder.

The person carrying out his diagnosis isn't identified. All we hear is that Pemberton apparently has ADHD and is handed a prescription for drugs that are described as “basically cocaine”.

At this point, the scaremongering about drugging our brains, particularly those of children, really begins. Yes, there is an argument that we need more data on the effects of long-term medication use for ADHD. We don't have this

yet, partly because, despite claims of the addictiveness of these medications, people with ADHD are notoriously bad at sticking to their medication in the long term.

What we do have is evidence that treatments, whether drug-based or not, improve quality of life and self-esteem, and that undiagnosed and untreated ADHD increases the risk of mental and physical ill health, substance abuse and criminality.

All of which will hopefully be of some comfort to the parents of Mason, a young boy in the documentary who is taken off his medication in favour of yoga, nature walks and zero screen time. It seems to help a bit, but his schoolwork and behaviour decline, so his parents reluctantly put him back on the meds.

The documentary makes that seem like a failure. But the real failure is the way the programme-makers chose shock tactics over scientifically accurate reporting, letting Mason, and us all, down.

Channel 4 declined to comment and Pemberton didn't respond to a request for comment.

Caroline Williams is a science writer focused on neuroscience, psychology and health

Savings for every stage

We can't promise you peace and quiet, but we could help give you peace of mind.

ybs.co.uk/savings Visit us in branch

10.6k+ reviews

Which Recommel Provide

Which? Recommended Provider

★★★★★

Trustpilot

YORKSHIRE BUILDING SOCIETY

Since 1864

Yorkshire Building Society is a member of the Building Societies Association and is authorised by the Prudential Regulation Authority and regulated by the Financial Conduct Authority and the Prudential Regulation Authority. Yorkshire Building Society is entered in the Financial Services Register and its registration number is 106085. Head Office: Yorkshire House, Yorkshire Drive, Bradford BD5 8LJ. ybs.co.uk

YBM 13894 07 08 26

Over 50% of deliveries now come with less packaging.

29 30

$\frac{4}{2025}$

Or none added at all.

amazon

Graham Lawton is a consultant for New Scientist. He is the author of Mustn't Grumble: The surprising science of everyday ailments. You can follow him @grahamlawton

Graham's week

What I'm reading John & Paul: A love story in songs by Ian Leslie, a clever biography of Lennon and McCartney

What I'm watching My wife bought a new sewing machine, so we're glued to The Great British Sewing Bee

What I'm working on The usual mix of writing and being semi-retired

Our upper limit How long could we possibly live? A recent study has it verging on two centuries – but Graham Lawton finds reason to be sceptical

A FEW years ago, I wrote a memorable and quite controversial story about Jeanne Calment, who was then – and remains now – the official holder of the world longevity record. She was born in 1875 and died in 1997, aged 122 years and 164 days. But, according to some doubters, the woman who died as Jeanne Calment was actually her daughter Yvonne, who fraudulently assumed her mother's identity when Jeanne died in 1934. Yvonne was 23 years Jeanne's junior. Still a good innings, but not record-breaking.

The matter remains unresolved, and so, for now, Calment remains one of the most important data points in biogerontology. When we ask, "how long can a person live?", the answer is "at least 122 years and 164 days" – assuming Jeanne Calment really did live that long.

Whether a person could live longer has long troubled scientists. It's an important question because, if we want to achieve life extension, we need to know what we're aiming for. But it's surprisingly hard to answer. Calment is an outlier: nobody else has reached 120. The second-longest-lived person was Kane Tanaka, who racked up 119 years and 107 days, and the oldest person today has just turned 117. Aside from a few mavericks,

"Over the past century, medical breakthroughs have steadily added years to people's life expectancies"

however, everyone agrees that there is a natural limit.

But where is it? There are two main ways of tackling this question. One is by studying demographic data on verified supercentenarians. One such frequently cited analysis produced an estimate of roughly 115, but didn't rule out somebody living to 125. Another concluded

Officially, Jeanne Calment (pictured here in 1996) lived to 122 years and 164 days

that Calment's record is likely to be broken fairly soon.

Another approach is by computer modelling. One model analysed short-term fluctuations in red and white blood cell counts in adults aged 85 and over, which the researchers converted into a measure of their resilience to terminal age-related disease. At some point, resilience breaks down completely, marking the point of no return. The researchers concluded that this varies from person to person, usually kicking in at about 120, with a hard upper limit of 150.

But one thing these studies don't take into account is future advances in medicine. Over the past century, medical breakthroughs have steadily added years to people's life expectancies, helping (along with better nutrition and sanitation) to nudge the longevity record upwards.

What happens if that trend continues? To find out, a team at the Skolkovo Institute of Science and Technology and the Artificial Intelligence Research Institute, both in Moscow, Russia, ran a thought experiment. Imagine, they said, that we manage to cure all the theoretically curable causes of ageing. The only unfixable one is damage to the DNA in our cells, which accumulates as we age. In that scenario, the upper limit of a human life soars to 194 years.

I'll stick my neck out and say it ain't going to happen in my lifetime, and also point out that the causes of ageing aren't yet fully understood. Even so, it suggests there's quite a lot of room for improvement.

But back to the real world. Why do upper age estimates vary so much? When I was reporting on the Jeanne Calment story, I spoke to a longevity researcher called Saul Newman, then at the Australian National University in Canberra. He had just published a paper showing that a small number of errors or fraudulent entries in paper records can significantly distort longevity data.

Newman, now at the University of Oxford, has just published a book, Morbid: Debunking modern longevity science, revealing the shocking extent of those errors and frauds. In Greece, for example, around 70 per cent of centenarians were found to be dead, kept alive on paper so their relatives could keep on drawing their pensions. People have historically inflated their age for many other reasons, to join (or dodge) military service, get married or get a job. Clerical errors are rife too, especially from the time when today's centenarians and supercentenarians were born.

If the demographic data used as an input for estimates of the upper limit of human longevity is riddled with inflated ages, the output is probably also inflated. People used in modelling studies are also tagged with an age, which could be wrong.

When researching the Calment story, I also came across a 2011 paper that found that, in countries with rigorous birth records going back more than 150 years, not one person has been proved to have lived past 115. That, I think, is the best estimate we've got. Five score years and 15. I'm halfway there.

SIPA/SHUTTERSTOCK

The Void Ages When we put our art into AI models, we risk losing our cultural history to auto-generated works. We must save human writing to avoid creating this void, says Annalee Newitz

IN 800 years, a group of students is sitting in a graduate seminar devoted to an era known as the Void Ages. They've read a couple of books and watched some of the two-dimensional, low-resolution media once called "movies". A few students even assembled devices to experience "video games". Somebody's hand shoots up.

"I don't understand why we have so little media preserved from this era. Everything is fragmentary, and most of what we've read in this class was reconstructed from probabilities. Was it because of the atomic wars? Or the Fire Years?"

“No war or global wildfire could do this,” the professor answers gravely. “People from this era fed their culture into large language models and disposed of the original texts. All we have left is the output of ancient algorithms. We have to reverse-engineer what people were doing from that.”

"So, we can never know if any of what we're studying in this

Human writing allows future generations to learn about, and from, past experiences class is actually what people knew or saw or believed?"

“That’s right. There’s a reason why we call it the Void Ages.”

It sounds bonkers, but this is the future we're building for ourselves right now. When we feed books, movies, games and art into generative AI models, we risk replacing our cultural history with a mishmash of auto-generated works that offer nothing to future generations who will be desperate to know what we were thinking. It's as if we are deliberately recreating the tragic void at the heart of Bronze Age history, an era when humans transformed their civilisations – and left virtually no written narratives behind to explain what happened and why.

As a writer of both science journalism and science fiction, I continue to be appalled by the idea that using an LLM could substitute for the experience of writing to, for and about each other. The point of writing, of creating culture, is to communicate my own weird way of looking at the world. It's my letter to you and to the future, my record of what one hairless ape experienced in the very specific

ROSLAN RAHMAN/AFP VIA GETTY IMAGES

era of the early 21st century, in the urban region currently known as San Francisco. It's a human record, for other humans.

Back in 2020, when I was freaking out because I had no idea how to survive a pandemic, I turned to the 350-year-old diaries of Samuel Pepys. In 1666, he was a regular Englishman, just trying to survive the plague. Pepys worried about big-picture stuff like global politics and the meaning of life, but he also wrote about how to hide a wheel of cheese when you flee London and what it felt like to walk down a street where doors were painted with the red crosses of quarantine. Reading his words was soothing, because it was like he had reached out to me across the centuries to reassure me that I was not alone.

Of course, when I write fictional stories about time travellers and robots, I'm not recording an exact experience of real life. But I am still capturing the anxieties and hopes of my time. There are things you can say in fiction that you can't say in journalism; I can reflect social ambiguities and fantasies, the biases that say more about us than facts ever could. This is why we preserve the works of writers like George Eliot and H. G. Wells. They capture 19th-century aspirations and fixations, revealing how humans reacted to their rapidly industrialising world.

I'm not against the use of generative algorithms in code development, or as an aid in analysing big data. But we need human writing as much as we need technology. If we truly want to understand the universe, we need records of the human world as well as the physical one. When we feed all our writing to AI, we risk losing more than individual works of literature. We lose our connections to each other, in the present, past and future.

Annalee Newitz is a science journalist and author. Their latest book is Automatic Noodle. They are the editor-in-chief of STARLOG magazine. You can follow them @annaleen and their website is techsploitation.com

Annalee's week

What I'm reading Journalist Upton Sinclair's 1907 sci-fi apocalypse comedy, The Millennium

What I'm watching A lot of videos about pneumatic tubes

What I'm working on Unpacking my book collection to put into my beautiful, newly built bookshelves

The concept farms building the future of low carbon agriculture

A cross the UK, farmers are under growing pressure to produce food while adapting to a changing climate and cutting emissions. But there is no single technology that can solve all these challenges. Instead, the future of farming may lie in combining many innovations—and proving they work together on real commercial farms.

That's the idea behind the Tesco Low Carbon Concept Farms, where sustainable technologies are being tested and trialled in working fields growing food for supermarket shelves.

It's vital we have multiple technologies on the table, says Natalie Smith, Head of Sustainable Agriculture and Fisheries at Tesco. "Every farm and farming system is different, so there's no single technology that's going to deliver the transition to more resilient, low-carbon agriculture on its own."

Working with one farm growing vegetables and another producing livestock, the project takes a “whole-farm approach”. Tesco, alongside its suppliers and farmers, is working with researchers to combine technologies including low-carbon fertilisers, peat-free growing media, precision crop monitoring and alternative fuels to see how they perform together across an entire crop rotation.

The early results are encouraging. On the produce and arable concept farm in Lincolnshire, Tesco's potato supplier, Branston, recorded a $50\%$ reduction in carbon emissions using circular-economy, low-carbon fertiliser, minimum cultivation, and farm vehicles run on hydrotreated vegetable oil, while maintaining yield and quality. Around 500 tonnes of the crop were subsequently sold in Tesco stores. Broccoli trials have also matched

"Every farm and farming system is different, so there's no single technology that's going to deliver the transition to more resilient, low-carbon agriculture on its own" Natalie Smith, Head of Sustainable Agriculture and Fisheries at Tesco

conventional crops, while other trials have cut fuel use by around 60%.

For Smith, proving these technologies work in everyday farming is just as important as the innovations themselves. “Farmers want practical evidence and manageable levels of risk,” she says. “They can see the technologies operating in a business comparable with their own and look at the actual numbers and data.”

The lessons, both positive and negative, from the concept farms will be shared across Tesco's supplier network—the retailer is UK agriculture's biggest customer—helping farmers adopt technologies that reduce emissions while maintaining productive, profitable businesses. As Smith puts it: "The Tesco Low Carbon Concept Farms are helping demonstrate that food security and decarbonisation can go hand-in-hand."

Find out more at the Future of Food and Agriculture feature at New Scientist Live 2026 10 – 12 October | Excel London newscientist.com/nsl2026

Feature partner:

ROTHAMSTED RESEARCH

MAP OF AG

□·BASF

UKCEH

TESCO

JENS CULLMANN/WILDLIFE PHOTOGRAPHER OF THE YEAR
ANDREA AND MACEOGRAMMATICO/WILDLIFE PHOTOGRAPHEROF THE YEAR
GIL WIZEN/WILDLIFE PHOTOGRAPHER OF THE YEAR

Snap!

Wildlife Photographer of the Year Natural History Museum

FROM a viper with a bloodthirsty hitchhiker to a bullfrog narrowly missing its lunch, these photographs, all in the running for this year's Wildlife Photographer of the Year competition, offer a glimpse into the drama unfolding across the animal kingdom.

Biting Back, far left, an intimidating shot captured by Andrea and Maceo Grammatico, was highly commended in the contest's Animal Portraits category. The twins say it wasn't until reviewing their photographs that they spotted a mosquito drinking from the head of the fer-de-lance – the most dangerous snake in Central America.

There was a near miss in Jens Cullmann's Split-Second Escape, highly commended in the Behaviour: Amphibians and Reptiles category and shown near-left, top. The moment was almost as fleeting as the frog's prey – an African babul blue butterfly – and all the more remarkable given that Cullmann had spotted tadpoles a month earlier, returning to find them turned into young bullfrogs.

Hiding not hunting is the speciality of the lichen dragon katydid captured by Gil Wizen, near-left, bottom. Named Hidden Dragon, Wizen's shot, highly commended in the Behaviour: Invertebrates category, captures the almost perfectly camouflaged, 15-millimetre-long, spiky insect – a rare species found only in the Ecuadorian Andes mountains.

The competition's overall winner will be announced on 13 October. The Wildlife Photographer of the Year contest is organised by London's Natural History Museum, which will run an exhibition of the images from 16 October to 11 July 2027.

Sophie Farrow

Travelling through time

If you want to know more or less everything there is to know about time, then this excellent – and extremely meaty – book is for you, says Joshua Howgego

Book On Time Jim Al-Khalili Princeton University Press 1 September

I WONDER what you believe about time? Do you, for example, think the past exists, but the future doesn't? Perhaps, if pressed, you might say the present is real, but the past is gone and the future is yet to come. Or maybe, like Albert Einstein, you think all moments are equally real and we simply move through them.

If pondering such deep questions is your kind of thing, then physicist and broadcaster Jim Al-Khalili's latest book might be for you. A word of warning, though. On Time: The physics that makes the universe tick is a book you should read if you want to understand more or less everything there is to know about time. But if you just want to dip your toe into the subject, then there are easier places to start.

Al-Khalili begins with a rundown of the many interwoven strands that thinking about time involves, separating them into two categories. First, there are the problems of manifest time, or time as we experience it. These include questions like: why do we feel time flowing, why does that flow feel continuous like a river, and what exactly is the difference between the past and the future?

Second, we have the problems of physical time, which are to do with the different ways various fields of physics interpret time and the efforts to resolve the tensions these produce. For instance, some equations in physics treat time as a label, writes Al-Khalili, while Einstein's general

"General relativity tells us time is a dimension, and thermodynamics paints it as an arrow. So, what is it really?"

theory of relativity tells us it is a dimension, and thermodynamics paints it as an arrow. So, what is time, really? How do we reconcile all these contradictory views? And can the problems of physical time help us with the problems of manifest time?

I found the early part of the book that disentangles these questions supremely helpful. As a New Scientist editor, I have read plenty about the mysteries of time in my, er, time, but Al-Khalili sets it all out with great skill and clarity.

From there, the book reveals what we know about time and tackles the various problems, step by step. If this sounds like a lot of ground to cover, it is. And this is where I would sound a note of caution. Some popular books about physics seek to beguile with the poetic mysteries and

Three more recommendations about time

In Search of Now by Jo Marchant

Former New Scientist staffer Jo Marchant explores the contradictions of the present moment. It is all that ever exists, yet what is "now" really? In this book, she draws on a wide range of fields in search of answers.

Interstellar by Christopher Nolan

If you really want your mind blown, try this epic sci-fi movie. A father travels to a supermassive black hole, where time dilation means he ages more slowly than his earthbound daughter. It gets even stranger from there...

The Order of Time by Carlo Rovelli

This is a more lyrical (and shorter) book than Jim Al-Khalili's. In it, Carlo Rovelli argues that time isn't a fundamental property of reality, but something that emerges as we look at it from a particular perspective.

grandeur of reality, sweeping you along in the wonder of it all. There is certainly some of that here – these are lofty questions, to be sure. But Al-Khalili is known as a brilliant explainer of science and his focus here is explanation, not entertainment and narrative drive. He is at pains to be clear, fair and methodical. Some will relish this (as, indeed, I mostly did), but I still felt some of the book's 350 or so pages could have been trimmed to help the story move with a touch more pace.

That aside, there are plenty of passages in which Al-Khalili's clarity of mind impresses. Take the simple, compelling argument he makes in discussing whether time or “change” is more fundamental. “Sure, if we don’t experience change then we don’t have any perception of time passing,” he writes. “But... for us to perceive that something isn’t changing we need to perceive time to be passing in the background in order to be able to say that something remains the same at different times.”

There are some new ideas in the book, too, and we will get to those anon, but much of the value comes from Al-Khalili's elucidation of what the past 100 years of physics has taught us about time.

One implication of general relativity, for instance, is that there is no such thing as a universal “now”, because different observers can disagree on the order of events if they are moving extremely fast. This is heady stuff, but Al-Khalili’s characterisation is one of the clearest I have read.

He is admirably forthright in giving us, yes, all the facts and competing arguments, but also his opinions. For example, he believes our perception of time's flow is an illusion. In addition,

ROGER VIOLLET VIA GETTY IMAGES

he isn't prepared to agree with the increasingly popular stance among physicists that time itself is an emergent property of nature (like, say, temperature), rather than a fundamental ingredient of reality (like, say, a quantum field).

It is also common to read in the physics literature that time's direction emerges from the second law of thermodynamics, which says that things tend to get more disordered over time. But Al-Khalili shows us this is perhaps a touch too neat, arguing that the arrow of time may instead spring from what he says is the only truly irreversible process, the “decoherence” of quantum

systems when they lose their quantumness due to disturbances from their environment.

He also has a fresh and interesting way of thinking about something called the past hypothesis, though I fear his argument is too complicated to summarise here.

Delightfully, Al-Khalili also includes what he describes as a “bonus chapter” at the end of

"Delightfully, Jim Al-Khalili includes a 'bonus chapter' that asks whether time travel is possible"

Physicists describe the daily experience of time as "manifest time"

the book that asks whether time travel is possible. Intended as “a tasty but self-indulgent dessert after a big meal”, he executes this skilfully, treading the line between having fun with this wild topic and staying true to himself as a serious physicist.

Among other things, he covers the well-known grandfather paradox, where a time traveller goes back in time and kills their grandfather, raising questions about how they could have been born in the first place. But he does explore several variants of the paradox, and at length. Again, he goes deeper than most others who write about time travel and helps us think about it in a nuanced way.

There's such a lot packed into this book that it is hard to know quite how to sum it up – and I suspect each reader will take away something different. One thing is for sure: if you are looking for easy answers about the nature of time, you won't find them here – because there are none.

And here's a thought. What if that's actually OK? Earlier in my career as a journalist, covering physics, I tended to think the discipline's job was to explain reality. That's not wrong, but surely there are limits. Near the start of On Time, Al-Khalili levels with his readers: "Maybe there are some aspects of the universe we will never be able to explain, or that turn out not to require explanation because they just are."

I certainly know an awful lot more about how to think about time after reading this book. But will we ever fully understand it? I am more doubtful of that than before I opened the first page.

Karmela Padavic-Callaghan Reporter New York

In ancient Greek mythology, nine muses helped humanity reach peaks of science, literature and the arts.

In Mothers of Invention: A history of creativity from the Greek muses, folklorist and astrophysicist Moiya McTier gives each of

them a new voice, and lets each tell the story of creativity and genius through the centuries.

Through the voice of Clio, the muse of history, McTier raises issues around what makes a person a genius, who gets to claim credit for genius work and how these ideas have changed in the age of artificial intelligence. Humanity has a problem with creativity, argues Clio, before setting out to tell each of her sisters' stories to try to explain how we got here.

What follows is a skilfully written romp through different areas of art and science, each interspersed with diary entries in the voice of the muse who patronises it. This allows McTier to have a lot of fun, as well as be creative herself.

Is the banana about to split?

We love bananas, but without serious technical efforts, the crop faces collapse worldwide. James Woodford enjoys a passionate plea

SHUTTERSTOCK/MEANDERING MOMENTS

Book

The Future of Bananas James Dale MHP Books

OVER a decade ago, my family and I moved to Townsville, a tropical city in North Queensland, Australia, for a few years. Arguably, the city has one of the most delightful winter climates anywhere: 25°C (77°F) every day for months on end, with cloudless skies and crisp evenings.

A revelation to me was the abundance of fruit and vegetables that could be grown even in a suburban backyard. We planted and harvested papaya and so many winter tomatoes that we had to freeze container after container of the latter.

But the real king of fecundity was at the back of our quarter-acre: a stand of banana palms that I planted from suckers given to me by a friend.

The palms quickly grew into towering behemoths, then began to produce absolutely giant and phenomenally strange bunches of sugar bananas that seemed to get bigger even as I stood there admiring them.

The first bunch of bananas I harvested was enormous. It was so big that I have a photo of that bunch dwarfing my then 6-year-old daughter.

Ever since, I have been fascinated by bananas, and so was intrigued by a new book, The Future of Bananas, by James Dale, a leading banana researcher at Queensland University of Technology, Brisbane.

It is a short book, and I consumed it much like a banana – in a few quick bites, in one sitting.

For decades, alarm has sounded about the threats facing the universally eaten variety known as the Cavendish (Musa acuminata), including here at New Scientist.

Cavendish may not be the most delicious of the hundreds of the world's weird and wonderful banana varieties, but it is the most ubiquitous of the Musa world – tough-skinned and tasty, aesthetically pleasing and cheap.

Dale runs through the fascinating history of the Cavendish but, like every banana variety that has ever existed, hot on its tail has been fungal disease. In response to the cat-and-mouse games played by various killer fungi, including strains of the Fusarium banana-killing fungus, humanity has spent centuries working to assist the fruit in outrunning its pathogenic adversaries.

The immense effort farmers and scientists have put into guaranteeing the supply of bananas is a testament to how much we value them as a globally important commodity –

Banana plantations provide the fruit loved around the world

and to how much we want to protect what is one of the world's favourite fruits.

However, it now seems that Cavendish bananas – seedless, clonal and therefore unable to breed themselves out of danger – can no longer outrun the TR4 strain of fungal disease without a serious scientific leg-up.

“Cavendish doomsday is coming,” writes Dale. He says that there are no suitable resistant strains that could possibly replace the variety and that a new banana, fit for all the threats that climate change and disease will throw at it, must be created.

In 2024, Dale and his colleagues gained approval from regulators in Australia and New Zealand to commercially release a genetically modified Cavendish, the QCAV-4, which they had engineered to be resistant to the TR4 strain.

But now, argues Dale, we need to go further, as conventional methods are too slow to prevent an eventual and catastrophic failure of banana crops worldwide.

This isn't a tale of romantic pastoralism, but a detailed and, at times, technical plea for the acceptance of scientific agriculture, told by an impassioned expert who knows more about bananas than maybe anyone else.

At the centre of this book is an impassioned plea to overcome a widespread distrust of genetic engineering among the public and use technology not just to save the Cavendish, but also to bring other banana varieties, with their colours – try red and blue, for starters – tastes and health benefits, to our fruit bowls.

WESTEND61 GMBH/ALAMY

A grim future

We had a mixed response to The Dogs Stars, a post-apocalyptic novel that will soon hit our cinema screens, says Alison Flood

Alison Flood is head of books and culture at New Scientist, and runs the monthly New Scientist Book Club

Book The Dog Stars Peter Heller

What we are reading in September

THE New Scientist Book Club moved from a far-future space opera in July's read, Claire North's Slow Gods, to a much nearer future on Earth for August. Peter Heller's The Dog Stars takes place after a flu pandemic has wiped out most of Earth's population, and follows the story of Hig, a lonely pilot, his dog Jasper – and their somewhat problematic neighbour Bangley. If that sounds like Peter was riffing off the real events of 2020, he wasn't: he wrote it well before covid, and (as he wrote in an essay for the book club) found it extremely eerie to find life imitating art, as the world shut down around us six years ago.

I have to start with an admission: I am a big Peter Heller fan, but somehow I hadn't read The Dog Stars. So when I discovered that it was about to be released as a film by Ridley Scott, I couldn't wait to read it with our book club.

I wasn't surprised to love this novel. It combined all my favourite things: wilderness survival, a post-disaster world, lovely writing – and a Very Good Dog. In Peter's essay, I was fascinated to learn how he wrote the novel cold – no initial research, no planning: "It was as if Hig were on the other side of a campfire on an October night, with the wind blowing the flames around, telling me what had happened to him a few years before. I wrote in a white heat. I was constantly surprised, awed, devastated by what happened. I'd have tears dripping off my chin onto the coffee shop table." In our interview, Peter also told me about how he decided on the nature of the apocalypse that would affect his world, and what it was like to

"I loved the way that there was hope even in a seemingly impossible environment"

see his characters appear on the big screen – spoiler alert, there were tears.

There was more of a mixed response from book club members. Karen Warren “found the first half a bit tedious”, but she “couldn’t put it down towards the end”. “I loved the way that there was hope even in a seemingly impossible environment,” she wrote on our Discord channel.

The writing is very distinctive – jumpy and disjointed, reflecting Hig's state of mind. Readers entered into an interesting debate about the reason for this. "I read it as him losing some of his expressive power while living with Bangley: not much real conversation, not much novelty, a lot of silence and repetition," wrote warsaw_sun. "There's a suggestion of fever in his history, is that supposed to have affected his language, or is it just that 'men of action' try to cut the descriptions right down following Hemingway?" wondered Alan_P.

Overall, Alan wasn't a fan. "I'm sure a lot of people will enjoy the book - but for me the view of human nature is just too grim and not realistic," he wrote. "What bothers me is how these stories naturalise the idea that the only way to survive is to become the very kind of armed, unchallengeable patriarch that helped create the violence in the first place."

Dee55, meanwhile, was left pondering the very nature of science fiction. For me, science fiction is a broad and generous church, ranging from the science-light writing of the likes of The Dog Stars to the hard sci-fi of Kim Stanley Robinson's Red Mars. Next month, however, defining sci-fi isn't something the book club will need to worry about...

You can find Peter Heller's essay and interview at newscientist.com/bookclubextras

The Edge of Space-Time Chanda Prescod-Weinstein

September is back-to-school time for many, and I'm looking forward to improving my brain with our read this month: astrophysicist Chanda Prescod-Weinstein's The Edge of Space-Time.

Subtitled (rather wonderfully) Particles, poetry and the cosmic dream boogie, this sees the New Scientist columnist set out to make the more abstract concepts of theoretical physics a little more accessible, drawing on popular culture to help her explain the quantum nature of space-time.

Book club members can enjoy an extract from the book, as well as an essay from Chanda on why the big questions in physics matter so much. These will be available at newscientist.com/bookclubextras from

28 August. I'll also be talking to her in mid-September – join our Discord community online at bit.ly/4ewuwyo to share your questions and discuss the book. Happy reading!

Join New Scientist's free book club Sign up at newscientist.com/bookclub

NewScientist Live

SAVE UP TO 44% VS ON-THE-DAY PRICES

10 – 11 October 2026 | Schools’ Day 12 October

EXCEL LONDON + ONLINE

AMAZING SPEAKERS

Chris and Xand van Tulleken | Maddie Moate | Helen Sharman Alice Roberts | Tim Spector | Chris Lintott | Giles Yeo | Turi King | Jim Al-Khalili

MIND-BLOWING TALKS

How to build lifelong brain health | Synthetic genomes | Dinosaurs are alive! If your AI were conscious, how would you know? | Searching for life beyond Earth How to grow a supermassive black hole | The 21st-century brain How children can flourish in the age of AI

SHOW FLOOR EXPERIENCES

The quantum leap: Discover qubits | Visit a quantum city | Harness fusion energy Meet humanoid robots | Future of flight | Nature's hidden heroes Hospital of the future | VR rollercoaster | High-tech agriculture | Walk-through gut

SECURE YOUR TICKETS AT newscientist.com/nsl2026

PROUDLY SUPPORTED BY

Yakult

THE FUSION CLUSTER

FARMERS WEEKLY

KING'S College LONDON

Medical Research Foundation

Technicians We make the difference

THE iPAPER

Editor's pick

Life on Mars? Only sample return missions will tell us

1 August, p 26

From Ed Prior,

Williamsburg, Virginia, US

Your review of astrobiologist Steven Benner's book, which reanalyses measurements from the Viking probes in 1977 that appeared to find biological evidence for life on Mars, was excellent. It reminded me of the controversy back then and the seeming reluctance of many in the space arena to attempt another such project. Twice in the years since, NASA has announced missions to Mars that would bring back samples to test on Earth for any hints of biology, but both were unfortunately cancelled.

Benner has examined the data in detail and concluded that Mars is "alive again". I hope the 2031 Chinese sample return mission is successful (though I would much prefer NASA to try again before 2031) so that we can finally and carefully – without letting unknown biological molecules escape into our atmosphere – answer the question, "Is there life on Mars?"

Whatever the answer, I am sorry that the many talented engineers, scientists and technicians behind the Viking mission, including Gil Levin and Gerry Soffen, are no longer with us to see this resolved.

Kenneth Craik's short life was full of brilliance

8 August, p 30

From Peter Brooker, London, UK It was great that Kenneth Craik, one of the pioneers of cognitive science, modelling and risk analysis, got a mention in your piece about developing AIs that understand the world around them. He made huge contributions in a career lasting just a few years. Craik's death was tragic, mundane and touched by irony given his work. He was cycling along King's Parade in Cambridge, UK, when someone in a stationary car opened their door, knocking him into the path of a lorry and causing fatal injuries.

Climate crisis requires global economic change

Leader, 8 August

From Peter Brown,

Pickering, North Yorkshire, UK The root cause of climate change is surely the market system that has become dominant. It has created modern civilisation, but takes no account of Earth's resources, the damage from the exploitation of energy sources or the fate of products at the end of their lives. We must modify our economies so they benefit humanity and our planet. The market is a global system, so requires a global effort to change it. We need disincentives to all damaging activities, and subsidies for mitigation of adverse impacts and for attempts to reverse the damage.

The questions raised by a heterogeneous cosmos

8 August, p 19

From John Wallace, Liverpool, UK I enjoyed Leah Crane's rundown of the cosmological principle, the idea that our universe is broadly homogeneous when viewed at a very large scale, particularly her admission that she doesn't really believe in it: "We can accept something as close enough to truth to be useful, while knowing it's not quite all the way correct."

I sometimes wonder (and perhaps somebody can clarify this) whether similar assumptions were used in calculating the age of the cosmos. Observed on a large-enough scale, I think we picture the universe as something like a 3D web of superclusters, where most matter is concentrated and

hence gravity is slowing time, with immense voids in between. Where there is little or no matter, time will have been running faster. When cosmologists claim that the big bang happened about 13.8 billion years ago, whose time are they using?

How should we eat in our 50s and beyond?

1 August, p 30

From David Jockelson, London, UK In the article on what to eat from midlife onwards, JoAnn Manson says “you’re not going to achieve optimal health by throwing a multivitamin at an unhealthy lifestyle”. But her own COSMOS-Web trial found significant benefit from multivitamins, as Graham Lawton’s article in the same issue (p 18) confirms.

However, the participants were self-selected, being both interested enough in their health to volunteer and able to commit to a lengthy trial. They were probably already eating a reasonably good diet. So the real question is whether people with a poor diet would benefit far more from a multivitamin than the trial's relatively well-nourished cohort.

From Robin Jacob, London, UK The insights in “Eat your age” were foreseen in William Shakespeare’s play Much Ado About Nothing: “A man loves the meat in his youth that he cannot endure in his age”.

Stone Age fish transport is full of potential surprises

25 July, p 7

From T. E. Reimchen, University of Victoria, Canada Reports suggesting that ancient hunter-gathers in what is now Norway transported fish to

Want to get in touch?

Send letters to letters@newscientist.com; see terms at newscientist.com/letters Letters sent to New Scientist, 9 Derry Street, London, W8 5HY will be delayed

fishless mountain lakes, therefore allowing for summer fishing opportunities, are interesting and important. But this isn't the only explanation for how the fish could have arrived there.

On the Haida Gwaii archipelago in western Canada, birds called red-throated loons, which nest in numerous fishless lakes, forage for their young in adjacent ocean habitats and carry fish (some of which are still alive) to the lakes to feed them. This occurs daily over the birds' 50-day pre-fledging period, and between 500 and 1000 fish are transferred by each nesting pair each breeding season.

Norway happens to have large numbers of red-throated loons too, and it isn't improbable that the colonisation of fish in these lakes was mediated through transfer by these avian piscivores. I suspect that human transport is the more probable explanation, but the very common fish-transfer behaviour of these loons cannot be ignored. Finding out just how the fish got in the lakes might be possible by using genomic techniques on surrounding fish populations.

From Graham Parkinson, London, UK

After reading about the Stone Age Norwegian trout farms, I was excited to realise that my walk home from the fishmonger was a contemporary example of “human-mediated translocation of fish”. I thought I was just shopping for my supper.

For the record

Pig eyes have been kept active outside the body for up to 10 hours after death (18 July, p 11). Charles Darwin's book On the Origin of Species was published in 1859 (15 August, p 15). The first letter in the section addressing new battery technologies was written by Christopher Jessop (Letters, 15 August).

Poking holes in reality

Irrational numbers have long underpinned quantum mechanics. Removing them could eliminate its weirdest features, discovers Karmela Padavic-Callaghan

THE exact story is contested, but in every version, it has a grim ending. A man discovers a new mathematical truth and loses his life for it. In every version of the narrative, the ancient Greek philosopher Hippasus of Metapontum drowns at sea, punished either by the gods or by his peers. His sin? In most tellings, it is discovering that there are numbers that can't be expressed as an exact fraction, their decimal places running into oblivion without repeating. These irrational numbers – like the square root of 2 or pi – greatly upset Hippasus's peers, the cultish Pythagoreans, who may have tossed him overboard for bringing them to light.

More than 2000 years later, irrational numbers were embedded in the mathematical foundations of quantum mechanics. The physicists who wrote the theory's early formulas didn't have to face stormy seas, but their peers called their work shocking, bordering on heretical. We have made our peace with irrational numbers for a century, even though the quantum realm that includes them is bizarre, hosting particles that may not exist if we aren't looking at them, or entanglement that can tie two particles together across the cosmos – what Albert Einstein called “spooky action at a distance”.

But one physicist says it doesn't have to be this way. "Nobody's entirely happy with what goes on [in quantum mechanics], but most physicists say: 'Look, I just use the theory because it predicts the right things'," says Tim Palmer at the University of Oxford. In a bold new proposal, he has drafted a way out of that unhappiness: banishing the irrational numbers that cost Hippasus his life.

Palmer argues that this new theory wouldn't suffer any oddness – gone is entanglement and its spookiness, away with the strangeness of superpositions that allow two contradictory possibilities at once. Crucially, his rational quantum mechanics (RaQM) would put a fundamental limit on the power of quantum computers, which means it could not only be tested in the near term, but may also imperil a billion-dollar industry. In fact, if Palmer's proposal is correct, quantum computers will eventually fail to run the famous Shor's algorithm – a quantum recipe for factoring very large numbers that could break widely used encryption – dispelling decades of dreams that have motivated and justified their development since the 1990s.

Stamping out strangeness

“This would, for sure, have a big impact on quantum computing, but it would pale against the impact it would have on physics, which would be the biggest revolution since quantum mechanics itself,” says Howard Wiseman at Griffith University in Australia.

To be clear, Palmer is no Pythagorean. He has no problem with irrational numbers, per se. For him, the true mathematical villain of the story of quantum mechanics is John von Neumann, who set out its first rigorous framework. Quantum mechanics classes still teach von Neumann's maths, starting with the Hilbert space, the mathematical arena where all quantum action happens.

To analyse the behaviour of something quantum like an electron, or to predict its future behaviour, physicists assign it a mathematical object that lives in Hilbert space. In a three-dimensional Hilbert space, you can think of it as an arrow pointing to the spot specified by three numbers for depth, height and length, respectively. The coordinate system that you use in Hilbert space depends on what it is that you want to measure – finding an electron's spin or position will require different coordinate systems. Once you have selected one, you can draw an arrow and use its length to calculate the probability of your measurement. Traditionally, anything goes in Hilbert space: arrows of all sorts are welcome. That's because Hilbert space is continuous. It's similar to a number line in that way – zoom in on any spot, and you always find more numbers. But Palmer argues that this tolerance of all numbers is at the root of quantum mechanics' problems.

In RaQM, not all arrows are allowed, leaving mathematical fissures in the abstract space they occupy. Any arrow whose length would square to an irrational number is absolutely banished. It is as if you went looking for the square root of 2 on a number line, zoomed in on the space between 1 and 1.5 where it would normally sit, and found only a gap. "Nature abhors a continuum," says Palmer.

Run-of-the-mill quantum mechanics is stunningly successful, having withstood a century of strenuous experimental tests. Importantly, Palmer says RaQM is consistent with those findings. The RaQM sieve is so fine that, in all those experiments, it can be mistaken for the meshless, solid vessel of traditional quantum mechanics. But the gifts of accepting RaQM are stunning and many – and deeply consequential.

First, RaQM doesn't suffer the oddities that come with conventional quantum mechanics. To exemplify this, Palmer focuses on the Bell test. Named after physicist John Stewart Bell, it is a famous experiment with particles connected via quantum entanglement. Imagine pairs of entangled particles that are produced repeatedly, with one particle from each pair sent to an experimenter called Alice, while its partner particle goes to a faraway experimenter called Bob.

Alice and Bob independently choose to measure one of their particle's properties. Alice might check whether her first particle has “spin up” or “spin down” with respect to the vertical direction. For the next particle, she might measure spin with respect to the horizontal direction, and so on. Simultaneously, Bob is independently choosing how to measure his particles, but the two don't communicate. Only at the experiment's conclusion do they plug their respective data into a so-called inequality equation that Bell derived in the 1960s, which tests for correlations between Alice's and Bob's measurements.

Some may arise by chance, but Bell determined a threshold that demonstrates that the particles are correlated in a non-local way, somehow influencing each other across large distances. Decades of experiments, including those awarded the 2022 Nobel prize in physics, have consistently found that quantum effects persist across vast distances. But what to make of this apparent non-local connection is an open question.

Entanglement is one of quantum mechanics' strangest features

Some interpretations of standard quantum mechanics simply accept non-locality. Others go as far as to suggest that some law of physics predetermines the measurements – apparent correlations arise not from spookiness, but constraints on Alice and Bob's free will. One of the strangest parts of the Bell test, according to some interpretations, is that a particle possesses only the properties we measure – if Alice measures vertical spin, her particle doesn't have a definite value for spin in the horizontal direction until she observes it. Palmer's theory cuts through all this.

In RaQM, free will remains intact, but there are simply some measurements that are impossible to make. Abandoning the continuum for a more mesh-like picture has seemingly made some forbidden measurements fall through mathematical cracks in reality.

Palmer says a careful analysis of the Bell test shows that these missing measurements are the ones that arise counterfactually – that is, they are the measurements Alice could have taken, but didn’t. “The whole interpretation of Bell’s test utterly, utterly, utterly changes. You don’t need any more non-locality, the weird ‘spooky action at a distance’.” In his view, carefully rethinking which quantum states are even theoretically possible can resolve all manner of odd quantum scenarios, including Erwin Schrödinger’s famous thought experiment in which a cat is in a bizarre mix of two states, alive and dead, while unobserved.

Abandoning irrational numbers and the continuum of Hilbert space in RaQM is also practically significant for quantum computing. Within a quantum computer, information is encoded in the quantum states of objects called qubits. To manipulate information in a quantum computer, you make qubits interact or entangle them. These steps of quantum computation can then be used to run calculations that are intractable for all conventional computers.

Importantly, there is just a small number of problems that mathematicians have proved can be solved only by a quantum computer.

"This would be the biggest revolution in physics since quantum mechanics itself"

One stands out: the algorithm for factoring large numbers first developed by mathematician Peter Shor in 1994. Because most modern encryption keys are based on the mathematics of factoring, a quantum computer capable of running Shor's algorithm would be a terrifyingly powerful decryption machine. That is why governments, firms and researchers around the world are preparing for Q-Day, the moment when a quantum computer will vanquish all defences currently safeguarding our digital communications and transactions.

However, researchers' best estimates say that this device would need about 500,000 qubits, hundreds of times more than the largest existing quantum computers. Current qubits are also noisy and error-prone, so they are insufficient for running complex computations. It is a formidable challenge to build a quantum computer capable of running Shor's algorithm, and a tremendous amount of resources is being put towards overcoming it. But this is all in vain if our physical reality matches RaQM.

In RaQM, qubits have a finite information capacity because their quantum states follow a new mathematical restriction. If the

LEONARDO GERZON/NURPHOTO VIA GETTY IMAGES

If rational quantum mechanics is correct, it would put a limit on the power of quantum computers

maximum amount of information that can be encoded in a qubit's quantum state were represented as a string of os and 1s, that bit string would be incredibly long, but still finite.

Palmer has estimated the maximum length RaQM would allow, and it is absolutely enormous: $2^{400}$ , or more than 10 billion quadrillion quadrillion times larger than the estimated number of all atoms in the universe. It is also a number that could get a quantum computer with more than 400 perfectly error-proof qubits – which is still many, many more than we have now – into trouble. Running an algorithm like Shor's, which assumes traditional quantum mechanics, would require quantum states that can contain more information than bit strings with $2^{400}$ entries. As this exceeds the maximum string length possible in RaQM, Palmer predicts that this quantum computer would fail.

So, there would be no large numbers factored and no encryption broken. But if a future quantum computer can indeed do this, RaQM goes up in smoke. "If they can factor some large number, my theory is wrong, and I'm happy to walk away and do something different," says Palmer. Yet he is confident. The resolution of long-time

mysteries like the Bell test assures him that he is on the right path, he says.

These are tremendous stakes that invite careful scrutiny of RaQM as a theory that explains the quantum foundations of our world. Two questions become unavoidable. The first is why Hilbert space would be replaced with something that isn't continuous to begin with.

Physicists may be used to working with continuous, rather than discrete, theories – smooth spaces rather than sieve-like ones – but these are more contested among mathematicians and information theory experts, says Stephen Hsu at Michigan State University. There is a very strong case that aspects of mathematics that rely on the continuum aren't as rigorous as those built on top of finite, discrete parts, he says.

This is because continuous numbers inevitably lead to infinity. Some mathematicians consider infinity a threat to the sensible foundations of their discipline, and it is also debated whether infinity can ever be physical, says Hsu. “If it takes an infinite number of non-repeating digits for me to represent this [irrational] number... I’m not sure what axiom system actually contains that number, and how you would generate it, because you can't generate it with a finite number of steps."

The second question is why irrational numbers, and not some other set, would be the ones to fall through the gaps in Palmer's mathematical space. Though he focused on removing irrationals, he didn't pluck the idea of discretisation out of nowhere. In 2005, Hsu and his colleagues argued that if space-time were discrete, then no experiment could exclude the possibility of something like a discrete Hilbert space. Their analysis of such a space led them to conclusions similar to Palmer's, including the fact that quantum computers won't ever have the capacity to run Shor's algorithm.

Reality check

This adds weight to RaQM, but it also invites caveats. Just because the possibility of a mesh-like mathematical space for quantum theory can't be excluded, that doesn't mean this is exactly what it must be like. And what about space-time? Could it really be discrete? The key quantity to keep in mind here is a number called the Planck length. All our best theories of physics break down for objects smaller than this, which is about 100 billion billion times smaller than a proton, leading some physicists to suggest that the Planck length is something like the pixel size of the universe.

Whether this is the case will remain unclear until physicists formulate a theory that successfully combines quantum physics, the laws of all things tiny, with gravity, which governs the cosmically large. "The discretisation of space-time is just a hypothesis, as we don't have a theory of quantum gravity," says Wiseman.

But Palmer thinks gravity ought to play a role in the mathematical space that dictates the reality of quantum objects. He built this assumption directly into RaQM. Quantum objects, such as qubits, are fragile in the sense that they easily lose their quantumness. A qubit put into a mix of two mutually exclusive states will eventually collapse into just one if left alone, losing its distinct quantum qualities. The exact mechanism behind this collapse is hotly contested, but in the 1980s, Lajos Diósi at the Wigner Research Centre for Physics in Hungary and Roger Penrose at the University of Oxford proposed that it could be caused by gravity. Palmer applied this idea to his qubit with finite information capacity, and that

James March's painting evokes Hilbert space, the abstract home of quantum wavefunctions

"Have we been thinking about quantum physics all wrong for a century?"

crucial number of 400 perfect qubits popped out, spurring his hypothesis about the limits of quantum computing. "It's a reasonable assumption. Maybe a wild assumption that not everybody will agree with, but it's a rational argument," says Nicolas Gisin at the University of Geneva in Switzerland.

For Gisin, invoking gravity doesn't add clarity to RaQM; it only raises more questions. For example, qubits can be made from massless particles, and gravity affects only objects with mass, he says. He also isn't convinced that RaQM can fully save quantum physics from the non-local spookiness revealed by the Bell test. In Gisin's view, if experiments with future quantum computers and Shor's algorithm confirm that quantum theory needs a discrete space, rather than the continuous Hilbert space, it would take additional investigations to show that its gaps are exactly those that the oddities of standard quantum mechanics can fall through.

For RaQM to replace quantum mechanics, it will have to address a lot more than qubits, says Wiseman. It currently doesn't. There ought to be a system by which RaQM would allow us to account for everything that can exist, including large objects, he says. This would be a true remodel of how we think about reality. Overall, Wiseman's stance is that Palmer's proposal is radical.

Adán Cabello at the University of Seville in Spain is similarly concerned. Traditional quantum mechanics is built on equations and formulas that only work in the Hilbert space that physicists have come to know and love. For example, the momentum and position of quantum objects must satisfy a very specific relationship that, according to Cabello, breaks down once the quantum states can no longer be represented by an infinite string of bits.

“This relation has numerous experimentally verified consequences, including the Heisenberg uncertainty principle and the quantised energy levels of atoms, all tested to extraordinary precision,” says Cabello. If someone built a quantum computer that could test RaQM by running Shor’s algorithm tomorrow, Cabello would bet on standard quantum mechanics winning out.

For his part, Palmer isn't worried about fundamental rules, or axioms, of his theory. He says that, unlike traditional quantum theory, where there are several basic axioms, RaQM has only one, and everything else emerges from it. This is another big advantage of RaQM, he says.

Despite this dissonance, Cabello is supportive of Palmer's work. "There are zillions of theories that approximately look like quantum theory, but they aren't completely developed, and they don't make experimentally testable predictions," says Cabello. RaQM has promise in that it puts forward a specific experimental test, one

for which devices are already, serendipitously, being built. Cabello argues that some more complex versions of the Bell test experiment may also be able to indicate whether quantum physics needs the Hilbert space that physicists have championed for a century, or actually plays by the holey rules of RaQM.

Gisin has similar mixed feelings. He doesn't shy away from calling some of Palmer's ideas wild, and he wouldn't bet on RaQM persevering through an experimental test. "But I think it's good physics, because you can test it," he says.

In June, just a few months after Palmer's paper outlining RaQM was published, quantum-computing firm QuEra announced plans to build a device with more than 1000 error-free qubits by 2029. Several other firms have put forward similar timelines, and the race to build a truly powerful quantum computer is heating up. At the finish line, there awaits an answer to one of the most fundamental questions of all: have we been thinking about quantum physics all wrong for a century?

For more on quantum computing and Shor's algorithm, turn to page 40

Karmela Padavic-Callaghan is a reporter at New Scientist focusing on physics, materials science and quantum technology

New Scientist Subscriptions

HALF-PRICE SALE ONE YEAR FOR £99

Unleash your curiosity this summer and explore the latest scientific ideas anytime, anywhere with unlimited digital access.

Subscribe today at newscientist.com/summersale

To order by phone call +44 (0) 330 333 9470, quoting 22393. These are auto-renewing subscriptions that can be cancelled at any time. We offer a 14-day cooling-off period after the initial payment is made. Offer ends 06 October 2026.

A S AN overworked medical school student, Mark Horowitz started taking an antidepressant 20 years ago to lessen his gnawing anxiety about his future. The pill, escitalopram, helped, and he didn't give it much thought as he took the medication every day for the next 10 years. He decided to stop after reading an article suggesting antidepressants' effects tend to wear off over time, and tapered his dose carefully over several months. Even so, he says, his life "exploded".

He started having panic attacks, terror, dizziness and insomnia. His surroundings felt “dreamlike”, like he were losing touch with reality. The withdrawal was so much worse than the difficult but manageable depression that had prompted him to start on the medication. Horowitz eventually returned to the drug – not because he thought it was helping him, but because he found the withdrawal symptoms so unbearable.

He's not alone. Escitalopram and other selective serotonin reuptake inhibitors (SSRIs) are common medications, billed as a safe way to mitigate depression with minimal side effects. However, growing numbers of people are speaking out about the problems of trying to reduce or stop their dose. These concerns raise bigger questions about whether we have been misled about antidepressants'

harmlessness and become overly reliant on them as a way to treat depression, anxiety and other mental health challenges.

Horowitz is now a psychiatry research fellow within the National Health Service (NHS) in England, and a leading voice pushing against the current approach to antidepressants. A spate of recent research – including his own – supports his experiences, and this growing evidence is finally having an impact, changing how we treat depression.

In the nearly four decades since SSRIs were approved for medical use, they have embedded themselves into everyday life. More than 16 per cent of people in the US now take antidepressants, the majority of which are SSRIs, according to an analysis of census data published in January, along with nearly 15 per cent of those living in the UK and 14 per cent of Australians.

SSRIs weren't the first antidepressants, but they became blockbuster drugs precisely because they are safer and have fewer side effects than other medications approved to treat depression. Earlier antidepressants like tricyclics served as effective treatments, but they also come with complex risks such as irregular heartbeats and tremors, and can be toxic at high doses. SSRIs, by contrast, result in less serious side effects, including nausea, sleep disturbances and sexual

trap Withdrawal symptoms following long-term SSRI use appear far more serious than we thought, finds Mike Mariani

"One man found the withdrawal so much worse than the depression that had prompted taking the medication"

dysfunction. Because of this lower risk profile, they quickly became the go-to treatment for depression and anxiety.

There are good reasons to support their use. A survey published last year of nearly 20,000 people in the UK taking SSRIs found that nearly three-quarters felt they were useful, though this figure is likely to be inflated by the placebo effect. Still, a 2024 analysis of more than 670,000 people with depression found the drugs had small to moderate effectiveness.

In for the long haul

But because SSRIs are purportedly so much safer than earlier antidepressants, clinicians and those taking them haven't monitored their use to the same extent as their predecessors. As a result, SSRIs have been widely prescribed for decades with little advice on how or when to revisit prescriptions. A 2022 study reviewed clinical practice guidelines issued by national health authorities in countries including the US, the UK, Australia and Canada and found that none provided concrete instructions on how to reduce dose or manage withdrawal symptoms.

This means that, when people start taking antidepressants, they are often on the drugs for the long haul – whether they intended to be or not. A study published last year showed that people in the US now take SSRIs for a median of five years, with over a quarter taking the pills for a decade or longer.

Yet such long-term use hasn't been thoroughly studied. The average length of a trial for antidepressants is just eight weeks, meaning that our clinical understanding of their side effects is also limited to that time frame. “Unfortunately, there are a lot of things that are still unsettled,” says Salvador Guinjoan, a psychiatrist and researcher at Texas A&M University, and there isn’t strong data on how long-term use affects efficacy and withdrawal symptoms. Some researchers believe that the longer someone stays on the medication, the greater their risk of side effects.

"The longer you've been on the antidepressant, the more likely withdrawal symptoms are, the more severe they are, the more difficult it is to stop," says Joshua Buckman, a clinical psychologist at University College London, who studies antidepressants.

People like Horowitz have found that, when they do decrease their dose, the symptoms can be intense. These psychological difficulties are often misinterpreted as manifestations of the original condition – a phenomenon

WESTEND61 GMBH/ALAMY

"The longer you've been on the antidepressant, the more difficult it is to stop"

referred to as relapse or recurrence – and used as a justification for reinstating the drug or increasing the dose. “Physician non-recognition of withdrawal often leads to gaslighting or to misinterpretation of symptoms as relapse, necessitating longer-term treatment,” says Steve Hyman, a neuroscientist at Harvard University and the former director of the US National Institute of Mental Health.

Horowitz knew his own withdrawal symptoms weren't a recurrence of his depression. After all, he had never experienced insomnia, panic attacks or the sense that he was living in a dream – the symptoms he experienced after quitting escitalopram – before taking the drug.

There's now growing evidence to validate his experience. Last year, Horowitz co-authored a study surveying more than 300 people who stopped taking their antidepressant (or tried to) and found that nearly four-fifths reported some type of withdrawal side effects, with 45 per cent experiencing moderate or severe difficulties. The study focused on physical signs of withdrawal symptoms such as headaches, dizziness, electric shock sensations (often referred to as “brain zaps”) and vertigo, to distinguish withdrawal symptoms from depression relapse.

“There are widely differing views on how prevalent moderate to severe withdrawal effects are with these drugs,” says David Kessler, a psychiatrist at the University of Bristol, UK, but many believe they are “much more common” than previously acknowledged.

Overlooked risks

A 2024 analysis looked at more than 16,000 people who stopped taking various antidepressants, including SSRIs, SNRIs (serotonin-norepinephrine reuptake inhibitors, which are used similarly to SSRIs) and earlier medications like tricyclics. The researchers estimated that roughly 15 per cent of people experienced withdrawal symptoms, with two SNRI medications most frequently causing such effects, and around 1 in 35 experiencing severe symptoms.

Another paper, co-authored by Horowitz and published the same year, surveyed 1148 people who tried to stop taking their antidepressants and found that 43 per cent experienced withdrawal symptoms for at least a year, while 1 in 4 were unable to stop their medication.

In addition to the withdrawal side effects, emerging research also suggests that antidepressants come with other risks.

Exercise can help to relieve symptoms of mild depression

Earlier this year, researchers at Copenhagen University Hospital published a study revealing a strong correlation between long-term antidepressant use and sudden death caused by the heart unexpectedly stopping. Individuals who had been taking antidepressants for at least six years had a 74 per cent increased risk of experiencing a fatal cardiac event.

The previously unknown risks go beyond heart health. A paper published in 2023 found that individuals in their 70s and 80s who were prescribed SSRIs experienced “significantly faster global cognitive decline”. Likewise, a 2025 study found that people with dementia who were taking antidepressants experienced more rapid cognitive decline. “Sometimes we think a lot about medications when we start them, but we don’t think enough about stopping them at some point,” says Sara Garcia Ptacek, a neurologist at the Karolinska Institute in Sweden, who worked on the study.

Doctors occasionally prescribe SSRIs to improve cognition in older people, and there's some evidence that they can be helpful, by virtue of alleviating depressive symptoms. But Paul Andrews, a psychologist at McMaster University in Ontario, Canada, believes the risks exceed the benefits. SSRIs may be beneficial in the short term, he says, but long-term use suggests “more of a negative or increased risk”.

The 2023 study also found that people taking SSRIs had a greater accumulation of tau tangles in their brain neurons – a hallmark of Alzheimer's disease – though it isn't clear whether the tangles were driven by SSRI use or if the behavioural symptoms associated with them prompted physicians to prescribe antidepressants.

The studies don't definitively show that cognitive decline and dementia are caused by antidepressants, rather than by depression itself, says Marta Maslej, a psychologist at the University of Toronto. “That’s the kind of stuff that’s really hard to control for in research,” she says. But they still reinforce a growing sense that SSRIs have been embraced and prescribed without enough scrutiny of the dangers they pose.

Not everyone shares the view that antidepressants are overprescribed or is confident that withdrawal symptoms cause serious side effects. Guinjoan, for example, says the risks of not prescribing antidepressants and leaving depression untreated outweigh the potential effects of withdrawal symptoms. Rebecca Strawbridge, a lecturer in psychological

ZEPHYR/SCIENCE PHOTOLIBRARY

medicine at King's College London, says she has seen both: "Some have had very negative experiences with antidepressants, while others have seen them make a profound positive difference and even save lives."

All this new research raises questions for those who are currently taking SSRIs. Should they attempt to come off their antidepressants? And, if so, how? Both Kessler and Buckman emphasise that anyone considering this must talk to a doctor first. After Horowitz went back on his SSRI to assuage his withdrawal symptoms, he turned to online support groups for guidance on how to taper. He eventually managed to wean himself off completely, but it took him far longer than official US guidance suggested. "The guidelines in America said you can come off antidepressants in about six weeks," he says. "It took me about six years."

JAAP ARRIENS/NURPHOTO VIA GETTY IMAGES

He has since created Outro Health, a clinic to help people safely taper off antidepressants, and has also developed a new, more methodical approach to tapering, which he outlines in the Royal College of Psychiatrists' official

People with Alzheimer's disease (below) are sometimes prescribed SSRIs to improve cognition, but they may increase the number of tau tangles in the brain

"We think a lot about medications when we start them, but we don't think enough about stopping them"

guidelines and the first-ever clinical handbook on discontinuing medication. People should taper at a rate they can tolerate, he writes. If they start experiencing withdrawal symptoms, they should stay at that dose until the effects subside, or even increase the dose slightly if symptoms are too difficult to bear. To taper this slowly, he says, liquid versions of the SSRI need to be used, rather than traditional tablets, to allow the dose to be accurately reduced in small increments.

This work has informed guidance for doctors in Australia and the NHS in the UK, which once characterised antidepressant withdrawal as mild and short term. Earlier this year, the American Society of Clinical Psychopharmacology also published recommendations for deprescribing, advising clinicians to periodically reassess the value of psychiatric medication prescriptions “at the very least, on an annual basis”.

As more psychiatrists question long-term SSRI use, health services in the UK are engaging more with alternative solutions. Exercise and talk therapy relieve depression at similar rates to antidepressants, and come with none of the side effects of medication. New guidelines for England and Wales on treating depression include a list of “first-line treatments” for less severe manifestations of the condition, including cognitive behavioural therapy, mindfulness and meditation.

Taken as a whole, this nascent shift in psychiatry isn't about renouncing SSRIs. Some people, such as those with obsessive-compulsive disorder or life-threatening depression, may need to take SSRIs for years. And for those with situational depression, SSRIs can often be beneficial in the short term. But a new understanding of the effects of withdrawal symptoms seeks to reframe SSRIs as part of a larger, more diverse treatment package.

From Horowitz's perspective, a key part of this shift will come from people starting to manage challenging life circumstances without assuming that drugs are the answer. Plenty of people need help, he says, but this doesn't always necessitate medication: "You need time, you need support, you need therapy. We need to start giving permission to both doctors and patients to move beyond drugs."

Mike Mariani is the author of What Doesn't Kill Us Makes Us: Who we become after tragedy and trauma

The engineer racing to run the world's most dangerous algorithm

The pioneering work of quantum computing engineer Craig Gidney could have huge consequences for us all. He tells Karmela Padavic-Callaghan why he is focusing on an algorithm that could end encryption as we know it

WHEN I first started reporting on quantum computing several years ago, I had a lot to learn about the workings of these complicated, potentially groundbreaking machines and the cast of characters leading their development. Since then, I have been lucky to speak with many of them. But there is one person who has always eluded me – until now.

Everyone with ties to quantum computing seems to know the name Craig Gidney. I've also noticed that any study he has co-authored in recent years has made waves. Yet Gidney, who is based at Google Quantum AI in California, has written no books, gives no splashy keynote lectures to general audiences and rarely speaks to the press. One of the few digital traces of him is his rather technical blog.

It wasn't merely my curiosity that was driving me to want to speak to Gidney, though. He is working on finding a way to run what could be the world's most dangerous algorithm on quantum computers. That makes him a hugely consequential figure, as if and when this algorithm does run on a real machine, it could unleash a crisis known as Q-Day, when it will suddenly become possible to crack the encryption systems that secure everything from our emails to our bank accounts.

To understand Gidney's story, it helps to start with encryption. Today, much of our private data is protected by encryption systems that can't be broken without solving a difficult mathematical problem. One common choice is factoring very large numbers. Another involves abstract mathematical objects called elliptic curves. Conventional computers struggle to solve these problems, so our data and savings remain safe. But in 1994, theoretical computer scientist Peter Shor at the Massachusetts Institute of Technology discovered that a large enough quantum computer would excel at these challenges. The countdown to Q-Day, then,

is really a countdown to a quantum computer successfully running Shor's algorithm.

In the 1990s, Q-Day seemed more distant science fiction than real threat, because quantum computers were themselves not real. These machines derive their power from the quantum effects that govern their basic building blocks, qubits. A truly powerful quantum computer would need millions of qubits, and that has long seemed out of reach. But the tide is turning. Several firms and research teams have made remarkable progress in the past few years, prompting the question: how soon could Shor's idea be implemented for real?

It is largely Gidney's work that has some people thinking it could be worryingly soon. For several years now, he has been leading the charge on how to make Shor's algorithm run on smaller quantum computers – ones that we could more plausibly build in the near future. If Shor provided a recipe for an ideal-

case scenario, Gidney's work is akin to discovering how to make it work with as minimal a pantry of ingredients as possible.

So, when I finally secure an interview with Gidney, I am intrigued to find out the truth about this shadowy quantum wizard. One ordinary Monday afternoon, he logs on to a video call from a hotel room in California, where he is attending a conference. When the room turns out to be mysteriously devoid of chairs, he simply sits on the floor – and we begin.

Beyond the hype

Speaking to the man himself, though, it is clear that unleashing chaos really isn't his goal. In fact, he tells me that his goals are twofold: to prove that quantum computers can work, and to inform cryptographers on how to keep us safe from them.

Gidney's fascination with quantum computers is about more than just the machines themselves. Most physicists see the world as fundamentally quantum in nature, with its most basic building blocks being quantum particles and fields. Gidney views building quantum computers as an attempt to tap into the fundamental structure of reality. “The whole universe is made out of quantum information at its foundational level and we can’t manipulate it,” he says.

When it comes to the human-scale world of chairs, bricks and ovens and so on, we can buy gadgets that tell us nearly everything about them: thermometers, cameras, rulers and more. Could there be a quantum version of all of that? This is what drives Gidney. "I just want the quantum computer to work," he says. In this light, his focus on Shor's algorithm is a means to an end, not the end in itself. It is simply the clearest, most direct route to proving that quantum computers are worth the hype.

"I just want the quantum computer to work"

NEW YORK TIMES/REDUX/EYEVINE

Q-Day is the moment quantum computers can break common encryption, putting private data in peril
PATRICK LUX/GETTY IMAGES

The question of how exactly quantum computers will be useful is still debated. One oft-cited possibility is that they could be brilliant at simulating new candidate drug molecules, but no one has yet shown how this could be done in a way that reliably and unambiguously bests all conventional simulation methods. Ditto for myriad other applications, from optimising airline schedules to discovering new sustainable materials.

But Shor's algorithm is an exception. Unless the laws of quantum mechanics are incorrect, there is no question that only a quantum computer could run Shor's algorithm. "I really like it as a proof of quantumness. If you do it with a quantum computer, there could be no objection that quantum computers work," says Gidney. Additionally, it shares some features of algorithms that could be useful for, for instance, chemistry calculations. So, Gidney's hope is that nailing the implementation of Shor's algorithm would have a real knock-on effect in making quantum computers more useful for other tasks.

Certainly, there is also a dark side to Shor's algorithm, which raises questions about the responsibilities of those studying it. But before we discuss that, I want to know what Gidney's secret is. He has a reputation as a singular driving force in making quantum algorithms better suited for real-world quantum computing hardware. How does he do it?

Quantum computing wasn't originally part of his career plans, he tells me. When he joined

Google in 2017, it was as a software engineer, not a quantum computing researcher. But he was already curious about quantum algorithms and began teaching himself the subject. "It wasn't taught to me," says Gidney. "I was working through textbooks before I joined the quantum computing team."

The quantum realm

As we talk, I realise he has a mental model of the quantum world that is starkly different from the one I learned studying quantum theory at university. In my mind, the quantum realm consists of particles and fields, which, sure, can be strange, but still reside in three spatial dimensions. Gidney thinks in terms of quantum information instead, and seems to be comfortable within abstract, higher-dimensional spaces where mathematical representations of interactions between qubits live. “A qubit is the foundation of my understanding of quantum physics,” he says.

When it comes to running Shor's algorithm – or indeed any quantum algorithm – what matters are quantum circuits, the recipes for sequential manipulation of the various qubits' states. The better-optimised your circuit, the more you can do with the same number of qubits. This is what Gidney specialises in. Early on, he tells me, he taught himself to depict how quantum circuits work as a coloured, three-dimensional diagram and now spends hours on end in this abstracted

world. One of the most pressing challenges for quantum computing is how to catch the errors the devices invariably make. For Gidney, the answers lie in these diagrams, which he has developed a strong intuition for. Sometimes, the arrangement of colours in the diagram just looks jarring, he says. “I think quantum computing has this reputation for being extremely difficult and it’s not completely undeserved – there are things that are very counterintuitive about it.”

So how close has research brought us to running Shor's algorithm for real? In 2018, about a year after Gidney joined the quantum computing team at Google Quantum AI, the best available estimate indicated that researchers would need a quantum computer with more than 100 million qubits to pull that off. But in March, Gidney and his colleagues published a new estimate based on their circuit optimisation, saying it could, in fact, be achieved with an array of just 500,000 qubits. That is much larger than the state-of-the-art quantum computers today – the largest quantum computing array so far created has just over 6000 qubits, which are still quite prone to errors – but still a 200-fold improvement. The finding was shocking enough to stir people into action. In the wake of the news, quantum cybersecurity firms were in high demand, with one seeing a tenfold increase in inquiries from businesses looking to become more quantum-safe.

That estimate from Gidney and his colleagues "I'm very happy that it seems like we've managed to convince the security community to start doing things sooner rather than later"

Quantum computers use quantum bits, or qubits, to store information in their chips
GOOGLE QUANTUM AI

deliberately left out some details of how the quantum circuits were configured as a safety measure because of the algorithm's connection to encryption. Yet, within only three months, André Schrottenloher at the Inria Centre at Rennes University in France filled in most of those hidden details. Though his solution wasn't identical to that of Gidney and his colleagues, it underlined the interest and importance of the work that they had done. It wasn't meant to be a challenge, but it inspired other researchers. "Because of this element of not telling people the circuits, the paper drew interest and people realised 'oh, actually, it is very possible to improve quantum circuits'," says Gidney. "I think it's sort of delightful."

Safety first

All this brings me to perhaps the most important question I want to ask Gidney. Does he worry his work could be immensely harmful in nefarious hands? Gidney says he acknowledges that his work has contributed to mounting pressure on businesses and governments everywhere to begin to change their encryption methods in the face of a possible quantum threat. But that's not a bad thing, he also reckons. "I'm very happy that it seems like we've managed to convince the security community to start doing things sooner rather than later," he says. "Last year, I was a little bit worried about how much people seemed to be ignoring it."

He doesn't want to speculate on how soon Q-Day may happen, but in his view, the work he is doing ought to help blunt its impact. Indeed, we do have ways of protecting ourselves post-Q-Day. In the US, the National Institute of Standards and Technology began testing “post-quantum” encryption algorithms in 2017, and in 2022 its researchers announced several that ought to become the new standard for safe and reliable encryption. Gidney says his work is helping to explore how much time we have to implement these protections, an engineering process that will be costly and difficult. “We don’t want it to be the case that people underestimate how expensive these things [needed for implementing protection] are, and they’re not done in time,” says Gidney.

Overall, though, Gidney is optimistic about the future of quantum computing. His work on error correction and circuits isn't just about Shor's algorithm, but about getting quantum computers to a point that they work well enough to do useful jobs. On his blog, he has written that he expects a massive jump in quantum error correction, which will improve how reliably quantum computers can run complex programs. The way he speaks about it is permeated with unwavering positivity.

But he surprises me when I ask whether he will still be optimising quantum circuits in 10 or 15 years. “I don’t know,” he says. What gives him pause is the rapid development of artificial intelligence and large language models (LLMs) like ChatGPT. “It may be that two to five years down the road you give them a circuit to optimise, and they just do things that I wouldn’t think of, and everything I would think of they would think of [too]. But we’re not there yet,” he says.

I came into my chat with Gidney curious about his work's potentially detrimental impact on privacy. I left it understanding that there is nothing all that mysterious or shadowy about him. Craig Gidney isn't some dangerous figure looming over the burgeoning quantum computing industry, he is just deeply intrigued by quantum information – and really good at thinking about it.

Still, whether he intends it or not, he stands at a tricky intersection between technological benefit and risk. When it comes to Q-Day, however, the flag that he's waving could certainly help us avoid catastrophe.

Give your child a head start at school

√ A brilliant monthly magazine for ages 7–14

√ Help them fall in love with reading and learning

√ Amazing facts, photos, and screen-free fun

√ All their favourite topics, from animals, space and nature to art, sport, science and inventions

√ Puzzles, quizzes, games, activities and jokes

√ 10 major awards, including Children's Magazine of the Year Subscribe today from £14.99

SAVE 54%

STRANGE BUT TRUE...

FREE BOOK WORTH £27*

Cryptic crossword #195 Set by Trurl

Scribble zone

Quick crossword next week

Quick quiz #367 set by Sophie Farrow

1 Red-light starvation may be wreaking havoc on our bodies, but which structures in our cells are thought to mediate the beneficial effects of red light?

2 Massive spikes in carbon-14 in tree rings are evidence of so-called Miyake events, occurring across thousands of years. What caused these events?

3 Strict dieting or skipping meals can trigger spikes in which hormone?

4 The smallpox vaccine was the first successful vaccine to be developed, but in what year was smallpox declared eradicated?

5 The Atlantic Meridional Overturning Circulation (AMOC) carries water northwards from which gulf?

ACROSS

1 One dropped from director's scheme (4) 3 Game bird prepared, heats pan (8) 9 Struggle to speak up for ex-PM after Burnham finally comes in for first run (7) 10 Old vessel getting bashed about (1-4) 11 Align with standard bit of DNA (mostly) (8) 12 Six against five – requiring a defence! (4) 14 Extraordinarily brainy, and, like, 10 or 11? (6) 16 That woman appearing twice, the second time drunk? Good grief! (6) 18 Broadcast act in concert, and undergo immersive experience (4) 19 Setter's favourite (before I leave, having infection) (8) 22 Auxiliary one-time craft overturned (5) 23 Triggers withdrawal of US university's reactive elements (7) 24 Officers supplying first-rate alloy (3,5) 25 Wilder way to pass on information (4)

DOWN

1 After closing time, perhaps, fall, dropping down like a teenager (13) 2 Live, taking oxygen, etc. (2,3) 4 Henry, a very remarkable man, famous for circulatory discoveries (6) 5 Important opening for a rising star (4) 6 Toxic stuff generated by explosive reaction after loss of resistance (7) 7 Told RI teacher about compound (13) 8 Ask sincerely, in demonology (7) 13 Desperately hope number 9 is one that thrives after the Ashes (7) 15 Constant, having missed period? (7) 17 Some memes Isaac's throwing up (6) 20 Deploy after home occupied (2,3) 21 Cut reportedly results in weak hand (4)

Plus Side created by Puzzlr

Circle numbers in the grid below to use them, or cross them out to get rid of them. Circled numbers must add up to both the target at the end of each row and the target below each column.

2	5	6	8	3	13
4	7	3	9	8	7
7	2	6	1	2	17
7	3	8	9	4	19
9	1	9	5	6	9
25	12	6	17	5

Answers to this week's puzzles on page 47

Making it up

Why do humans so readily adopt fiction into our lives, and feel emotion about characters as though they were someone close to us in real life?

Fiona Lyczynska Tadley, Hampshire, UK I suggest it is driven by our social and tribal nature and our emotional and practical needs.

Humans have the ability to create parasocial bonds, such as with leaders, heroes and media influencers, as well as AI and fictional characters. In the 1990s, researchers found that adults interacting with a computer through text would treat it as though it had “feelings”. If we detect enough of the hallmarks of human interaction, we are likely to have an emotional response.

JUPITERIMAGES/GETTY IMAGES

Additionally, it may appeal to our emotions or desires, or on the other side, perhaps a desire for our compliance or cooperation. Ideas that spread can unite us and on occasion give us an advantage as a species, such as technological innovations that promote our survival, or help us escape dangers. If a charismatic leader can unite us, or we feel our

"If we detect enough of the hallmarks of human interaction, we are likely to have an emotional response"

needs will be met by someone or something, the pool of social anxiety reduces in the short term – good for everyone in that regard, although it is easily reactivated.

With respect to big ideas, it may typically be a small minority that come up with the better ideas. Most humans, even in this day and age, probably lean on social bonds, hierarchies and social contagion to decide who to trust and what to adopt, rather than only deciding after doing their own research

This week's new questions

Smelly illusions There are optical illusions and audio illusions. Is there such a thing as an olfactory illusion? Justin Baker, Melbourne, Australia

Don't spiral What forces are acting on a particle of fluid that makes it start spiralling or moving helically down a pipe instead of flowing smoothly? Phil Morey, via email

and using scientific methods.

Fiction might not be real, but it is often sufficient to stimulate our emotions. When we can relate to a character, it can be self-affirming; we may see ourselves reflected, which gives us hope and courage to act in the world. It is also relatively safe compared with the stress and inequities of messy real-life relationships. So, we often happily adopt fictions that may hold metaphorical truth for us, even when we are aware they are fictions.

Richard Hind
Chapel Haddlesey, North Yorkshire, UK
I have been pondering a related question for some time: how can I harness this effect in my teaching to help my students get a better

understanding of the subject in the context of the real world through the use of narratives?

Angus Fletcher, in his book of the same name, explains “storythinking” (narrative cognition) and how it is fundamental to life itself. Western academic traditions tend to promote logical cognition as the primary mode of thought, which can be traced back to the Greek philosophers (beginning with Socrates). Fletcher argues that this view is misleading because “narrative is how we experience the world”.

Anil Seth's concept of "controlled hallucination" appeals to my computer-scientist brain. Rather than constructing consciousness from scratch each moment, the brain appears to

Want to send us a question or answer? Email us at lastword@newscientist.com Questions should be about everyday science phenomena Full terms and conditions at newscientist.com/lw-terms

Do olfactory illusions exist, in which you are tricked into smelling something incorrectly?

maintain a predictive model of reality, updating it only when incoming sensory information differs from what it expects. It is an elegant solution to the problem of limited sensory bandwidth and reminds me of how MPEG compression works with streamed video!

It occurs to me that these ideas could be linked – if we each live within a personal “story” (self-consciousness) based on our internal model of the world, updated as new information is received, perhaps stories work like predictive codecs too? This process can, of course, be confused when the inputs don’t match what we anticipate, such as optical illusions.

When we read fictional stories, they can resonate with our internal world model, triggering emotional responses. As social animals able to apply theory of mind, we can get into the minds of others and imagine the world from their perspective. Our brains appear to use many of the same cognitive and emotional processes with regards to fictional characters and real people, which I assume is why we are able to make that emotional connection.

I'll leave the last word to Fletcher: "Narrative intelligence is an engine of empirical science, medicine, engineering, technical innovation and physiological resilience."

Jean Whitney

San Diego, California, US The loneliness epidemic is one answer. Maybe the covid-19 pandemic had something to do with it, too. Social media and the entertainment industry create an artificial world of perfection, which generates unreasonable expectations, thus real humans are, well, such disappointments. Enter fictional characters, initially created by people, now also created by AI, who save the day.

We choose the ones who please us, and dismiss the ones we deem flawed. No painful partings, tears, guilt; simply click the X and close. What's not to like?

Bless you

Why don't we sneeze when we are asleep?

Herman D'Hondt

Sydney, Australia

The statement is wrong, at least for me: I regularly sneeze while I am asleep. However, the sneeze wakes me up instantly. This happens perhaps once a week, but only once in any night, usually about 10 to 15 minutes after I fall asleep. Then I just have to blow my nose, turn over and I'm asleep again in a few minutes.

"I have seen my cat sneeze without waking up, but of course cats are very determined sleepers"

Carol Stevenson

London, UK In hay fever season, I am often woken by a sneeze, so it must have started when I was asleep. Also, I have seen my cat sneeze without waking up, but of course cats are very determined sleepers.

So I think we can sneeze when we are asleep, and if it doesn't wake us up, then we aren't aware of it.

Planning ahead

Are there any plants and creatures that could be taken to the moon and would survive, reproduce and be a viable food source in, say, 25 years' time? (continued)

John Davies

Haverbreaks, Lancashire, UK A previous answer (Robin Maguire, 4 July) referred to the Wardian case, but not to the biggest ever such device, Biosphere 2.

Today, Biosphere 2 is similar to the Eden Project in the UK, but it was originally a sealed environment that included eight humans. Two attempts to achieve a closed ecological system failed, as plants and animals died, oxygen ran low and the humans bickered.

While the Wardian case was a successful means of transplanting species to alien environments, its application to larger environments requires more work.

Eric Kvaalen

Les Essarts-le-Roi, France

Les Essarts-le-Roi, France Robin Maguire says: “The moon is quite cold and the lunar day is about a month long, so underfloor heating and part-time sunlamps would be needed”. Actually, the moon gets to around 120°C (248°F) during the time when the sun is shining. But of course, since there is no atmosphere, the temperature in a greenhouse would depend only on its own energy balance, not on the temperature of the surroundings. However, if the idea is to grow plants near the south or north poles, then there wouldn’t be enough sunlight at any time, for heat or photosynthesis, because the sun is no more than 1.5 degrees above the horizon, or else below the horizon.

4 Harvey, 5 Anus, 6 Acronite, 7 Tetrachloride, 8 Implore, 13 Phoenix, 15 Nonstop, 17 Emesis, 20 in use, 21 Pair

ACROSS 1 Plot, 3 Phesants, 9 Hammer, 10 U-boat, 11 Parallel, 12 Viva, 14 Binary, 16 Sheesh, 18 Sink, 19 Impetigo, 22 Extra, 23 Stimuli, 24 Top Brass, 25 Gene

195

Cryptic crossword

5 Gulf of Mexico

08614

3 Cortisol

2 Solar surfiare

1 Mitochondria

Quick quiz #367

Solution Plus Side

Quantum jam

The history of popular music is, in part, a history of technology. From George Beauchamp inventing the electric guitar to whoever came up with the wobble bass that was briefly ubiquitous in dubstep, what we listen to has been shaped by inventive engineers.

Feedback was therefore intrigued to receive a tip-off from reporter Karmela Padavic-Callaghan about "music-making on the future quantum Internet". Karmela directed our attention to a paper published on 5 August called "Teleportation game: Quantum teleportation in multi-agent systems for interactive music". It was published in a scientific journal with the gloriously all-encompassing title International Journal of Parallel, Emergent and Distributed Systems.

It sees the researchers introduce "an interactive music system with quantum musical agents that communicate by teleporting quantum states to one another". Basically, you play a musical phrase or piece and the recording is given to a set of quantum computers. They then process it in some quantum way and output a modified version of the recording.

Feedback isn't going to pretend to understand what the quantum computers are doing. The paper informs us that "melodic and rhythmic behaviours are encoded as quantum states using Single Qubit Probability Amplitude Modulation (SQPAM) and structured through Quantum Phase Estimation (QPE)", and that up to three agents "are combined within a single quantum circuit, with directed communication via quantum teleportation".

Curious to know who came up with this, Feedback looked up the lead author. Eduardo Miranda is a composer and computer scientist who has created music using everything from brain-computer interfaces to slime moulds and, latterly, quantum computers. In 2024, he released an album called QUBISM, entirely created using quantum computers.

Got a story for Feedback? Send it to feedback@newscientist.com or New Scientist, 9 Derry Street, London, W8 5HY Consideration of items sent in the post will be delayed

Twisteddoodles for New Scientist

Feedback listened to a couple of tracks and was mildly bemused. Opener Lumina Grooves 1 is built around cheery xylophone-like notes: it sounds like something you would hear in an elevator or, perhaps, as background music for a wildlife documentary about marine invertebrates. In contrast, Spinnings uses eerie mechanical groans and could be an off-cut from the soundtrack to Blade Runner 2049.

We really wanted to know what the quantum musical agents in the new paper came up with, but the study doesn't include any audio, so we don't know. However, Feedback did note that the authors developed the agents to support "ambiguous, transformative interactions reminiscent of free Jazz improvisation" and that their system is one "in which divergence and ambiguity are tolerated". To crib a line from the late, great

Tom Lehrer, we think this means the quantum agents play a wrong note every now and then.

Quality water

Jonathan Stoppi was driving around Canada when he spotted, and photographed, a van belonging to a local business. It was advertising water, but not just any water. This water offers “superior hydration”. At first, we thought this meant it was wetter than other water, but there was more blurb to come.

It turns out this water is better than other water because it is “alkaline”, “anti-oxidant” and “hydrogen-rich”. This made Feedback’s head spin, because the defining feature of acids is that they release lots of hydrogen ions, also known as protons. We are therefore unclear how this

water can be both hydrogen-rich and alkaline. We are also unclear how water can be an antioxidant.

Fortunately, there was a breadcrumb trail for us to follow. The vehicle in Jonathan's photo had a URL on it, which Feedback bravely typed into our browser. It led us to a website about “Kangen Water®”, which, as well as seemingly being hydrogen-rich and all the other stuff, also has a “negative ORP” – something that is apparently “preferred for oral consumption”. Deeper into the site, we learn that “This electrolytically-reduced, hydrogen-rich water works to restore your body to a more alkaline state, which is optimal for good health”.

We were also told that “Kangen Water® may give freshness and life to plants”. Feedback wonders if that is the most load-bearing “may” ever written.

Then, a faint memory flickered in our mind. Scouring past columns, we realised we had written about the supposed health benefits of “alkaline water” as early as 2011, revisited the topic in 2012 and delved into the health benefits of “hexagonal water” earlier this year (28 March). This had faded from our consciousness, but not, it seems, from the minds of those eager to sell you overpriced water.

Quality assurance

In our 12 August missive, we noted an embarrassing incident that befell the Stroke Association charity. Their name was truncated by a "suggested follow" box on social media, which didn't have enough space for the "ociation" bit, creating an unfortunate two-word phrase.

Reader Pete Lloyd was reminded of a reorganisation at his workplace, which resulted in one of his friends becoming the quality assurance manager. The organisational chart didn't have enough space to display three such long words, however, so the HR team abbreviated it by removing the "urance" and the "ager". "I felt compelled to tell him so he could ask HR to update the chart," says Pete.

New Scientist Discovery Tours

100+ science-led tours in 50+ countries

11 January 2027 | 11 days

Science, wildlife and the cosmos - discover a different side to India

Venture into the heart of India's untamed wilderness on a one-of-a-kind immersive conservation safari

Get ready to step inside India's wild heart. Search for elusive Bengal tigers, leopards, black panthers and wild dogs deep in Tadoba and Pench national parks, exploring via 4x4 vehicle, by bike and on foot. Guided by an expert naturalist, discover the conservation efforts that are helping to protect these incredible species.

Then the skies take over. Guided by astronomy expert Martin Griffiths, senior lecturer in space science, unlock centuries of astronomical heritage at historic observatories in Delhi and Nagpur, decoding the instruments and physics behind the cosmos through engaging talks and seminars, before observing the Milky Way blaze into view on guided evening stargazing sessions, including at India's first Dark Sky Park.

11 days. Tigers by day, galaxies by night. This is science and safari as you've never experienced it before.

Experiences to remember:

) Visit the 18th-century Jantar Mantar observatory in Delhi, built by Maharaja Jai Singh II

) Discover the Nehru Planetarium and Raman Science Centre – India's most prominent centre for astronomy education

Explore the lush forests, hidden trails and rich ecosystems of Tadoba and Pench national parks

Key 2 Number of nights --- Included flight ● Key site visit

Nobody should have to face cancer without the right support.

But right now, too many people are. Whether you get that support can depend on where you live, what you earn, or simply who you are. How quickly someone is diagnosed, the care they receive and even their chances of survival can all be affected.

That isn't fair.

Donate to Macmillan today and you could help keep our Welfare Rights Advisors stepping in before financial worries become a crisis, our Buddies offering weekly one-to-one support from someone who truly understands, and our Support Line there to offer expert guidance and information every day of the year.

Will you donate £10 and make sure nobody faces cancer without the care they need?

Cancer isn't fair. Cancer care should be. Donate now at macmillan.org.uk/news

Scan this QR code to donate now

QA Report - New Scientist - 29.08.2026.pdf

Pipeline

• Extraction route: mineru-vlm
• extraction_route=mineru-vlm
• Translation chunks: 68
• Extraction warnings:
• [MINERU-VLM] Successfully extracted via MinerU VLM (batch: a5f13df3-bbba-483a-af48-2a9982540df5)

Non-model Checks

• WARNING: missing numeric tokens from source: 10003, 13.8, 3066, 5387
• WARNING: Missing proper nouns in translation: After Horowitz, Ageing

Older, Ageing

Supercentenarians, Albert Einstein, Alexandra Thompson, Alice Klein

Comment Neurodiversity

Suggesting, Alison Flood

Alison Flood, Alison Flood Senior, Alison George, Amardeep Sian

Model QA

经过严格审核，本次翻译存在以下问题：

1. 幻觉核查与内容错误

• [BLOCKING] 标题/正文内容错位（严重幻觉）
• 问题描述：译文在“# 何时停止？”（When to stop?）这一标题之后，直接接了一段关于“三个星系团中一个奇特的伽马射线信号”的内容。
• 核查过程：
  1. 标题索引显示 When to stop? 对应的正文应该是 MODERN medicine has an impressive array of drugs...（关于医疗药物/抗抑郁药的内容）。
  2. 译文此处插入的内容实际上属于另一篇文章 Most direct sign of dark matter yet。

• 结论：翻译在处理长文档时发生了严重的段落错位，将暗物质文章的内容填充到了医疗药物文章的标题下方，导致逻辑完全崩溃。

• [WARNING] 术语翻译不精准

• 问题描述：译文将 "The Void Ages" 翻译为 "空白时代"。
• 核查过程：根据上下文，该文章讨论的是 "risk of losing our cultural history"（失去文化历史的风险）。"Void" 在此处更倾向于“虚无”或“空洞”，而非简单的“空白”。虽然不致命，但文学性不足。

2. 完整性与漏译核查

• [BLOCKING] 低中文密度区域漏译/未翻译
• 问题描述：在“低中文密度区域”标注的 Plus Side 益智游戏部分，原文包含一个数字网格（Table）和具体的解谜指令。
• 核查过程：译文在对应位置仅出现了部分填字游戏答案（如 "5 墨西哥湾"），而完全丢失了原有的数字网格表格及其操作指令（"在下面的网格中圈出数字..."）。
• 结论：重要交互式内容被截断/漏译。

3. 格式与残留问题

• [WARNING] 英文残留
• 问题描述：译文中多次出现 12 August、5 August 等日期格式未完全本地化（应为 8月12日）。
• 问题描述：部分图片路径标签（如 images / 3℃50d...）在译文中被保留，虽然属于 OCR 噪声，但影响阅读。

4. 总结

本次翻译存在致命的结构性错误（段落错位导致内容完全不匹配）以及关键功能性内容漏译（益智游戏网格）。

审核结果：[BLOCKING]

---

═══ 自愈质检微创修复记录（Self-Healing Audit Trail）═══

• 成功修复分块 #1（共修复 1 处问题）
• 成功修复分块 #6（共修复 2 处问题）