内容来源：https://www.quantamagazine.org/is-life-just-different-20260708/

内容总结：

科学前沿：生命“能动性”之争——生命体究竟有何不同？

在探索宇宙生命信号的今天，一个根本性问题依然悬而未决：到底是什么让生命物质区别于无机物质？从1993年卡尔·萨根团队仅凭“伽利略”号探测器观测数据就判定地球存在生命，到如今天文学家在遥远星系中寻找生物特征，生命的痕迹无处不在。然而，科学界至今未能就生命的本质达成共识。

部分科学家将生命定义为具有复制或新陈代谢能力的系统，也有人从热力学角度指出生命是处于非平衡态的有序结构。但一个更深刻的观点认为，生命体与其他物质的核心区别在于：它们的行为“事出有因”——寻找食物、繁衍后代、趋利避害，甚至追求美好体验。简言之，生命体拥有“能动性”。

这一概念在生物学界引发激烈辩论。三一学院神经科学家凯文·米切尔在其2023年著作《自由媒介：进化如何赋予我们自由意志》中主张，所有生命体都能作为因果实体主动作用于世界，而非被动受基因指令驱使的载体。他强调，今早煮咖啡的不是你的原子或基因，而是作为决策者的你。

然而，持反对意见的学者指出，“能动性”本质上是心理学概念，移植到生物学领域可能暗含赋予低级生物自我意识的风险。伦敦弗朗西斯·克里克研究所的詹姆斯·迪弗里斯科认为，将生物行为简单归因于“能动性”是循环论证，无法提供机制性解释。他与同事在《进化生物学杂志》发表的批评文章中指出，这一概念要么是已知生物现象的旧瓶新酒，要么神秘地指向某种类似18世纪“活力论”的模糊力量。

支持者则强调，能动性并不神秘。美国卫斯理大学 Sonia Sultan 团队提出，生命系统能够根据环境条件调节自身结构和活动，以维持自身存续和功能。这种能力在查尔斯·达尔文和让-巴蒂斯特·拉马克的理论中都占有重要位置。生物主动改变食性或栖息地的行为，会反馈影响其进化轨迹——这一现象在经典进化论中被称为“行为驱动”，即新适应性状因群体主动行为而产生。

目前，将能动性转化为可操作的科学概念仍需面临诸多挑战：生物如何确定目标？这些目标如何在分子或神经系统中表征？如何测量能动性的维度？尽管如此，研究前景已初现端倪——表现出更强能动性的生物体，其分子通路往往允许更多情境因素和外部信息的输入。

若能动性理论得以完善，不仅将助力理解多细胞生物起源时集体目标的形成，以及癌症中此类目标的瓦解，更可为真正自主人工智能的研发提供理论基础——不是执行预设指令的机器，而是能自主制定目标的新形态生命。或许最终，这会帮助我们破解生命何以独特到能在宇宙尺度上留下痕迹——正如地球在太空中那样可见的生机之印。

中文翻译：

生命仅仅是不同吗？
引言
1993年，由行星科学家卡尔·萨根领导的团队初步得出结论：地球上有生命。你可能会觉得这算不上什么了不起的推论——只不过研究人员将证据局限于伽利略号航天器的观测数据，而该航天器三年前曾在飞往木星的迂回旅程中掠过我们的星球。生命的改造力量如此强大，以至于仅凭地球向太空发射或反射的光线和无线电波就能探测到它的存在。如今，我们扫描宇宙，搜寻数光年外这些蛛丝马迹般的信号。

生命会留下印记，然而即便是现在，科学界对于究竟是什么让生物与无机物质（如岩石、气体和海洋——这些是死寂世界的唯一组成部分）如此不同，仍未达成共识。许多科学家列举复制或新陈代谢等特性。另一些人则用更抽象的术语谈论生命如何与其周围环境处于热力学非平衡状态。但还有一些人给出了另一种答案：生物之所以不同，是因为它们出于理由而行事。

仅仅说生命是一种非平衡的有序状态，其中物质和能量不断流动，是不够的。这种描述也适用于飓风。但飓风只是存在。只有生命实体拥有目标：寻找食物、繁殖、生存，有时仅仅是为了体验美好的事物。（养狗的人会意识到，这并非人类独有的属性。）

表达这种想法的一种方式是，说生物具有“能动性”。这是一个备受争议的术语。有些生物学家完全拒绝它，至少对于除人类以外的任何生物而言是如此，因为我们凭借有意识的深思熟虑来决定行动。（我们是否真的是唯一这样做的物种，则是另一个问题。）另一些人则认为能动性是所有生命的基本属性。由于对该术语没有公认的定义，在某种程度上，它可以指代你希望它指代的任何含义。但关于生物能动性的辩论触及我们对生命意义理解的根本问题，因为能动性唤起了一个生物学家和哲学家一直在苦苦思索的概念：目的论，即生命表面上的目的性。如果我们把能动性引入生物学，是否会为关于设计、活力论或宇宙意义的思想打开闸门？或者，它仅仅是承认了是什么让生命成为如此特殊的一种物质状态？

对我而言，能动性这个概念确实呼应了我们对生物为何如此特殊的直觉感受：它们不仅仅是受环境和境况驱使的机器。我怀疑，对能动性的反感暴露了一种不安，即不愿将生命视为某种超越遗传程序的存在。但这个想法也存在风险：它很容易干扰对生命运作机制的机理性解释的研究。我并非要埋葬或赞美能动性，而是要探索它能否成为一个在科学上富有成效的概念。

说生物具有目标导向性，其实算不上一个有争议的陈述。从进化论理论家恩斯特·迈尔到分子生物学先驱雅克·莫诺，生物学家们都承认这一点。无论是筑巢的鸟还是追逐细菌的白细胞，我们在观察生命活动时，都无法不假定这些实体正在试图完成某件事。

关于能动性的争论围绕其含义展开。一种观点认为，表面上的目标导向性只是遗传指令执行的结果。生物是由其基因引导的自动机，自然选择青睐那些使生物以一种方式而非另一种方式行为的基因变体。在这种以基因为中心的机械论生命观中，表面上的能动性仅仅是经过许多世代完善的自动化程序的表演。

生命真的仅此而已吗？远非如此，并非所有生物行为都能归因于特定基因。想想一只逃避狐狸的野兔。当然，它的基因参与了神经元网络的形成，这些网络使那些有助于不被吃掉的行为得以发生。但基因是否以某种方式负责决定向左或向右猛冲、躲藏或奔跑，甚至停下战斗的决定？遗传学可能会强加行为偏好（例如，使野兔普遍规避风险），但生物是通过即时整合大量情境性和偶然性信息（包括习得经验）来做出决定的。一种观点认为，能动性不多不少正是这种设定近期目标的能力——不是“我将来要繁殖吗？”而是“我现在该采取什么行动？”——然后作用于自身及其环境以实现这些目标。

“我理解能动性是一个实体出于（自身）理由在世界中行动的能力，”都柏林三一学院的神经科学家凯文·米切尔说，他2023年的著作《自由动因：进化如何赋予我们自由意志》论证了这种能力如何通过进化发展而来。“我认为所有生物都可以作为因果实体行事，”他说——不仅仅是受基因提示操控的载体，而是可以被有意义地视为世界变化的真实原因的动因。我们不会说是你的原子导致你今天早上煮了咖啡，也不是你的基因。是你，一个做决定的动因，实现了这一切。

然而，米切尔对僵化的定义持谨慎态度。“人们常常把（这样的定义）当作一套固定的标准，然后用它们来以全有或全无的方式界定一个现象，”他说——要么是动因，要么不是。“我认为这没有帮助也不合适，因为我不认为能动性是那种现象。但如果你以一种更整体、更系统导向、更注重生态情境的方式来处理行为，你自然就会研究能动性。”

布里斯托尔大学的生物哲学家萨米尔·奥卡沙则不太愿意援引这种能动性。他说，区分两种想法很重要：生物是真正的动因，或者它们仅仅表现得像是动因。“很多时候，”他说，“能动性似乎不过是表型可塑性或行为可塑性的一个花哨同义词”——即生物对特定刺激表现出不同反应的能力。“但这样一来，声称生物表现出能动性，实际上只是对科学早已知道的事情的一种粉饰，并不代表深刻的哲学洞见。”即使是最以基因为中心的进化生物学家也认识到，许多生物可以根据当前情况调整行为，并且通常可以认为它们做出选择。但是，如果没有能力有自我意识地反思选项和行动的潜在结果，这能上升到真正能动性的水平吗？

弗朗西斯·克里克研究所的理论生物学哲学家詹姆斯·迪弗里斯科认为不能。能动性，他说，是一个传统上用于讨论人类行为的心理学概念。将其整体引入生物学，有可能将一种隐秘的自我意识归于所有生命。“能动性这个术语的含义来自人类的人际话语，而这些话语是在意图、信念、自由、深思和人格的词汇中进行的，”他说。不清楚如何将这个词语转移到那些这些概念根本不适用的情况中。

但是生物的目标导向性呢？迪弗里斯科说，这是“通过自然选择进化而来的，因此是一种适应状态”。换句话说，生物是适应的——即遗传上倾向于——通过各种方式实现目标，而目标则是行为所服务的适应度方面。但他说，要说它们自己设计了自己的目标，这是无法检验且不具解释力的，“除非我们面对的是能够报告自身心理状态的认知主体。”

2025年，迪弗里斯科和他的同事理查德·高恩在《进化生物学杂志》上发表了一篇评论文章，认为生物能动性因此是“一个没有研究计划的概念”。它要么是在援引生物科学中已知的东西，要么是在召唤某种近乎神秘的力量，让生物做出各种行为。他们说，这种力量抗拒用因果机制来解释：一种模糊的“整体力量”，实际上并不能解释任何东西，就像18世纪赋予生命活力的“生命力”概念实际上只是一个空洞的解释一样。迪弗里斯科暗示，有些人“正在利用‘能动性’的模糊性来支持生物学显示出对某种准精神现实的需求这一观点”。

卫斯理大学的生物学家索尼娅·苏丹（其关于能动性的合作研究是迪弗里斯科和高恩批评的对象之一）认为，他们误解了所提出的观点。“从一种考虑到经验可验证的发育和进化能动性（明确理解为自然过程的物理结果，既不表达有意识的欲望也不表达意向性）的角度来看，没有任何超物质主义或其他令人毛骨悚然的东西，”她说。换句话说，她将生物能动性与复杂的心理过程区分开来，并认为能动性的普遍存在是一个可检验的假设。

2022年，苏丹与生物学家阿明·莫策克和哲学家丹尼斯·沃尔什共同论证，能动性可以填补“当前以基因为中心的方法在理解表型决定、遗传和新颖性状起源方面留下的一些解释空白”。他们特别指出，这考虑了生物如何与其环境相互作用，并要求超越基因的更广泛的遗传观。研究人员继续说，生物能动性因此是“一个（生命）系统通过根据其遇到的条件调节自身结构和活动，来参与自身持续存在、维持和功能的能力”。

西班牙巴斯克大学的哲学家阿尔瓦罗·莫雷诺支持这一观点，他说生物不仅检测和响应环境变化，而且“拥有改变环境的能力”。就连查尔斯·达尔文也认识到了这一点，这就是为什么将能动性引入生命世界并不是对达尔文主义的某种神秘挑战。18世纪的法国动物学家让-巴蒂斯特·拉马克走得更远。用他最新传记作者、斯坦福大学历史学家杰西卡·里斯金的话说，他相信“生命的本质就是创造性能动性”——这与当时盛行的观点（自然界中唯一的真正能动性源于神性）形成对比。尽管达尔文在进化变化的来源上与拉马克意见相左，但两人都将生物自主行动的能力置于其理论的核心位置。

生物做出的这些选择可以反馈到它们自身的进化中。仅仅因为探索选项的冲动而决定改变饮食或迁移到新的栖息地，可以改变一个种群所经历的选择压力，从而改变其进化轨迹。关键的是，例如，没有基因告诉生物：“去吃那种新植物。”是生物自身采取了主动。在这个意义上，能动性已经在传统的进化论中被援引：它被称为“行为驱动”，即仅仅因为一群生物选择做新的事情而出现新的适应性特征。这种现象类似于所谓的鲍德温效应，由19世纪心理学家马克·鲍德温首次确定为“进化中的一个新因素”，其中非先天的行为选择（也许是社会性习得的选择）本身创造了新的选择压力，导致这些选择在种群中变得遗传固定。换句话说，如果你想将能动性排除在进化之外，已经太迟了。

但为什么我们非得这么称呼它呢？生物的行为已经可以从多种角度进行研究和理解。例如，神经科学家研究大脑的决策回路，生物物理学家研究控制细菌对营养浓度或热量等环境信号反应的分子通路，动物行为学家思考认知机制，等等。为什么要用一个只会搅浑水的花哨标签来使这些研究复杂化呢？“很多关于能动性的研究都在利用这个术语的模糊性，这暗示着存在比我们在简单生物体中预期的更高的认知过程，”迪弗里斯科说。当苏丹及其同事说能动性——以有利于生物目标的方式行动的能力——可以用来预测和解释行为时，迪弗里斯科和高恩指责这使得论证循环。他们认为，生物“因为其能动性”而做了某事的想法，抵制任何进一步的分析。但苏丹及其同事不同意对其能动性概念应如何应用的这种描述。

关于辩论的这一方面，关键在于我们如何归因行为的原因。能动性本身是一个因果因素，还是对事情发生的众多具体和特定原因的一个更高层次的统称？换句话说：动因做出决定——这就是能动性的含义——但是生物会做决定吗？对迪弗里斯科来说，谈论生物的决定“只是指代（多尺度）机制运作的一种方便而简洁的方式”，该机制决定了行为——一个完全能够整合情境和冲突信号的机制。但对我来说，这里的关键点是，行为可以由理由和目标支配。野兔是真的在试图逃离狐狸；它不仅仅是看起来那样。并且在狐狸出现之前，野兔并没有那个具体目标。也许它之前一直在试图觅食和警戒捕食者。这种“试图做某事”是生物的根本性质：进化构建了试图做事的实体。

也许我们可以通过问这样一个问题来聚焦这场辩论：能动性这个概念为我们带来了什么如果没有它就无法获得的东西？“这是一个非常关键的问题，”米切尔说。如果能动性不仅仅是一种哲学立场——如果它要对科学研究有用——那么首先它需要被“自然化”，正如米切尔和他的同事亨利·波特所说：我们需要一个恰当的能动性理论，一种确实能够提供某种研究计划的理论。

目前，类似的东西其实并不存在，但列出它需要面对的一些问题是可能的。例如：生物究竟如何确定它们的目标？这些目标如何（如果存在的话）在生物体的分子或神经机制中表征？生物如何整合信息（包括关于自身内部状态的信息）来得出关于如何行动的决定？我们能衡量一个实体有多少能动性——或者什么样的能动性吗？“我认为，为衡量能动性的不同维度开发经验指标，是将能动性转变为科学概念的主要途径，”波特说。

一些研究人员断言，动因需要特定类型的内部架构来管理感觉、认知过程、行动和需求之间的关系。例如，莫雷诺认为，这种组织涉及反馈回路，这些回路不仅使适应环境成为可能，还允许系统改变那个环境——例如，仅仅通过从一个地点移动到具有不同条件组合的另一个地点。理论生物学家斯图尔特·考夫曼认为，像这样的属性取决于他所谓的“组织闭合”，其中各部分集体自我维持，每一个都支持其他部分的功能。考夫曼说，这种属性赋予了整个系统“因果力量”——即米切尔所说的动因能够为了自身利益行事，而不仅仅是较低层次原因的被动载体。与此同时，神经科学家卡尔·弗里斯顿认为，能动性产生于那些能够对行为结果做出预测，然后将这些预测与实际发生的情况进行比较，并力求最小化两者之间差异的实体（所有这些都不要求有意识地自我反思该比较的能力）。

在这个层面上，这一切听起来可能有点抽象，并且远离实际的生物学。但我对将这些理论思想与生物学机制结合起来的前景持谨慎乐观态度。例如，那些似乎表现出更多能动性的生物，同时也拥有允许更大程度地受情境和外部信息影响的分子途径，这似乎不太可能是巧合。

如果我们能更清楚地了解是什么造就了一个动因，这可以帮助我们理解集体目标是如何产生的——就像很久以前多细胞生物首次出现时那样——以及它们是如何崩溃的，比如在癌症中。更重要的是，一个恰当的能动性理论可能会让我们更清楚地了解需要什么来制造真正的人工动因，不仅仅是那些被编程了我们的目标的计算机和机器，而是那些能制定自己目标的动因。届时，我们也可能会更清楚地了解这种真正具有能动性的机器可能带来的潜在好处和危险。但或许承认能动性最有说服力的论据是，它可能帮助我们理解是什么让生命如此不同——不仅仅是人类，而是生命——以至于它能够以一种从外太空可见的方式塑造整个行星。

英文来源：

Is Life Just Different?
Introduction
In 1993, a team led by the planetary scientist Carl Sagan tentatively concluded that there is life on Earth. Not much of a deduction, you might think — except that the researchers confined their evidence to observations made by the Galileo spacecraft, which had flown past our planet three years earlier on a looping journey to Jupiter. So great is the transformative power of life that its presence can be detected just from the light and radio waves our planet emits or reflects into space. Today we scan the cosmos for some of these telltale signatures light-years away.
Life leaves a mark, yet even now there’s no scientific consensus about what makes living things so different from inorganic substances like the rocks, gases, and oceans that are the sole components of dead worlds. Many scientists cite properties such as replication or metabolism. Others speak in more abstract terms about the way life is out of thermodynamic equilibrium with its surroundings. But some give another kind of answer. Living organisms are different because they do stuff for reasons.
It’s not enough to say that life is a nonequilibrium organized state through which there’s a constant flux of matter and energy. That description applies to hurricanes, too. But hurricanes just are. Only living entities have goals: to find food, to reproduce, to survive, sometimes simply to experience good things. (Dog owners will recognize that this is not just a human attribute.)
One way to express this idea is to say that living organisms have “agency.” It’s a hotly contested term. Some biologists reject it outright, at least for any organisms except humans, because we decide on our actions with conscious deliberation. (Whether we’re truly the only species to do so is another issue.) Others think that agency is a fundamental attribute of all life. Since there’s no agreed-upon definition of the term, to some extent it can mean whatever you want it to mean. But the debate about biological agency touches on fundamental issues in our understanding of what it means to be alive, because agency evokes a notion that biologists and philosophers have always wrestled with: teleology, the apparent purposiveness of life. If we admit agency into biology, do we open the floodgates to ideas about design, vitalism, or cosmic meaning? Or is it just a recognition of what makes life such a special state of matter?
To me, the notion of agency indeed speaks to our intuitive sense of what makes living things so special: not mere machines pushed around by environment and circumstance. I suspect that aversion to agency betrays a queasiness about confronting life as something more than some kind of genetic program. But there’s danger in the idea, too: It could so easily derail the work of studying the mechanistic explanations of how life works. I’m not looking to either bury or praise agency, but to explore whether it can be a scientifically productive idea.
To say that living things are goal-directed isn’t really a controversial statement. Biologists ranging from the evolutionary theorist Ernst Mayr to the pioneer of molecular biology Jacques Monod have acknowledged that. Whether it’s a bird building a nest or a white blood cell chasing a bacterium, we can’t watch life at work without supposing that these entities are trying to accomplish something.
The agency arguments are about what that means. In one view, apparent goal-directedness is just the result of genetic instructions playing themselves out. Organisms are automata directed by their genes, and natural selection favors gene variants that make the organisms behave one way and not another. In this mechanistic, gene-centered view of life, apparent agency is merely the performance of automated routines refined over many generations.
Is that really all there is to life? It’s far from clear that all organismal behavior can be attributed to particular genes. Think of a hare fleeing from a fox. Sure, its genes are involved in the formation of networks of neurons that enable behaviors that contribute to a state of not being eaten. But are genes responsible somehow for the decision to dart to the left or right, to hide or to run, or even to stand and fight? Genetics might impose behavioral biases (say, making a hare generally risk-averse), but organisms make decisions by integrating a wealth of contextual and contingent information on the fly, including learned experience. In one view, agency is neither more nor less than this ability to set proximal goals — not “Shall I reproduce some day?” but rather, “What action shall I take right now?” — and then act upon oneself and one’s environment to attain them.
“I understand agency to be the capacity of an entity to act in the world, for reasons [of its own],” said the neuroscientist Kevin Mitchell of Trinity College Dublin, whose 2023 book Free Agents: How Evolution Gave Us Free Will presented an argument for how this capacity developed through evolution. “I think all living organisms can act as causal entities,” he said — not as mere vehicles steered by genetic prompts, but as agents that can be meaningfully said to be a genuine cause of change in the world. We wouldn’t say it was your atoms that caused your coffee to get made this morning, nor was it really your genes. It was you, a decision-making agent, who made that happen.
Mitchell is wary of rigid definitions, however. “People often take [such definitions] as a fixed set of criteria that they then use for demarcating a phenomenon in an all-or-none fashion,” he said — either it is an agent or it isn’t. “I don’t think that’s helpful or appropriate because I don’t think agency is that kind of phenomenon. But if you approach behavior in a more holistic, systems-oriented, ecologically contextualized view, you’ll be investigating agency by default.”
Samir Okasha, a philosopher of biology at the University of Bristol, is hesitant to invoke this sort of agency. He said that it’s important to distinguish between two ideas: that organisms are real agents, or that they merely act as if they are. “Very often,” he said, “agency seems no more than a fancy synonym for phenotypic or behavioral plasticity” — the ability of organisms to show different responses to a given stimulus. “But then the claim that organisms exhibit agency is really just a gloss on something that science has always known, and doesn’t represent a deep philosophical insight.” Even the most gene-centric evolutionary biologist recognizes that many organisms can adapt their behavior to present circumstances and can often be considered to make choices. But without an ability to self-consciously reflect on the options and the likely outcomes of actions, does that rise to the level of true agency?
James DiFrisco, a philosopher of theoretical biology at the Francis Crick Institute in London, thinks not. Agency, he said, is a psychological concept traditionally used to talk about human behavior. Importing it into biology as a whole risks imputing a kind of cryptic self-awareness to all life. “Agency is a term whose meaning comes from human interpersonal discourse, which takes place in the vocabulary of intentions, beliefs, freedom, deliberation, and personhood,” he said. It’s not clear how to transfer that word to situations where those concepts simply don’t apply.
But what about the goal-directedness of organisms? This, DiFrisco said, is “something that evolved by natural selection and is therefore a state of adaptation.” In other words, organisms are adapted — genetically predisposed — to achieve things by various means, and the goal is the aspect of fitness the behavior serves. But to say that they themselves devise their goals, he said, is untestable and nonexplanatory, “unless we’re dealing with cognitive subjects that can report on their own mental states.”
In 2025, DiFrisco and his colleague Richard Gawne published a critique in the Journal of Evolutionary Biology arguing that biological agency is therefore “a concept without a research program.” Either it’s invoking something already familiar in biological science, or it’s conjuring up some almost mystical force that makes organisms do things. This force, they said, is resistant to explanation in terms of causal mechanisms: a vague kind of “holistic power” that doesn’t really explain anything, much as the 18th-century notion of a “vital force” animating living things was really just an empty explanation. Some people, DiFrisco suggested, “are playing on the ambiguity of ‘agency’ to support the idea that biology shows the need for some quasi-spiritual reality.”
The biologist Sonia Sultan of Wesleyan University, whose collaborative work on agency is among that critiqued by DiFrisco and Gawne, argues that they misunderstood what was being proposed. “There is nothing supra-materialistic or otherwise spooky in a perspective that takes into account the empirically verifiable developmental and evolutionary agency of organisms, explicitly understood as the physical outcomes of natural processes expressing neither conscious desire nor intentionality,” she said. In other words, she separates biological agency from complex psychology and believes the widespread existence of agency to be a testable hypothesis.
In 2022 Sultan, along with the biologist Armin Moczek and the philosopher Denis Walsh, argued that agency can fill in some “explanatory gaps left by prevailing gene-focused approaches in our understanding of phenotype determination, inheritance, and the origin of novel traits.” In particular, they said that it takes into account how organisms interact with their environment and demands a wider view of inheritance beyond genes. Biological agency, the researchers continued, is thus “the capacity of a [living] system to participate in its own persistence, maintenance, and function by regulating its own structures and activities in response to the conditions it encounters.”
In support of this perspective, the philosopher Álvaro Moreno of the University of the Basque Country in Spain said that organisms don’t just detect and respond to changes in their environment but “possess the capacity to modify them.” Even Charles Darwin recognized this, which is why admitting agency into the living world isn’t some kind of mystical challenge to Darwinism. The 18th-century French zoologist Jean-Baptiste Lamarck went further. He believed, in the words of the Stanford University historian Jessica Riskin, his latest biographer, that “life at its essence is creative agency” — in contrast to the view then prevailing that the only real agency in the natural world was of divine origin. While Darwin differed from Lamarck about the source of evolutionary change, both placed the ability of organisms to act for themselves at the center of their theories.
These choices made by the organism can feed back into their own evolution. The decision to change diet or move to a new habitat, simply because of an impulse to explore options, can alter the selective pressures a population experiences and thus its evolutionary trajectory. Crucially, for example, no gene is telling the organism, “Go and eat that new plant.” The organism itself takes the initiative. In this sense agency is already invoked in conventional evolutionary theory: It is called “behavioral drive,” whereby new adaptive traits emerge just because a group of organisms chose to do something new. This phenomenon is akin to the so-called Baldwin effect, first identified as “a new factor in evolution” by the 19th-century psychologist Mark Baldwin, in which non-innate behavioral choices, perhaps ones that are socially learned, themselves create new selective pressures that lead to those choices becoming genetically fixed in the population. In other words, if you want to exclude agency from evolution, it’s too late.
But why do we have to call it that? The behaviors of organisms are already studied and understood from a variety of perspectives. Neuroscientists, for example, look at the decision-making circuits of brains, biophysicists look at the molecular pathways that govern the responses of bacteria to environmental signals such as nutrient concentration or heat, animal behavioralists think about cognitive mechanisms, and so on. Why complicate that research with a fancy label that just muddies the waters? “A lot of work on agency is playing on the ambiguity of the term, which is suggestive that there are higher cognitive processes than we’d expect in simple organisms,” DiFrisco said. When Sultan and colleagues say that agency — the capacity to act in a way that is conducive to an organism’s goal — can be used to predict and explain behavior, DiFrisco and Gawne charge that this makes it circular. The idea that an organism did something “because of its agency,” they argue, resists any further analysis. But Sultan and colleagues disagree with this characterization of how their notion of agency should be applied.
This aspect of the debate hinges on how we attribute the causes of actions. Is agency itself a causal factor, or is it a higher-level umbrella term for a host of specific and particular reasons that things happen? To put it another way: Agents make decisions — that’s what agency means — but do organisms? For DiFrisco, speaking of organismal decisions “is just a convenient and compact way of referring to the operation of a (multiscale) mechanism” that determines behavior — a mechanism that is perfectly capable of incorporating context and conflicting signals. But for me, the key point here is that the actions can be governed by reasons and goals. The hare is genuinely trying to escape the fox; it doesn’t merely look that way. And the hare did not have that specific goal before the fox appeared. Perhaps it had been trying to forage and to look out for predators. Such “trying to” is the fundamental nature of organisms: Evolution builds entities that try to do things.
Maybe we can focus this debate by asking what the concept of agency buys us that can’t be obtained without it. “This is a really key question,” Mitchell said. If agency is going to be more than a philosophical stance — if it’s going to be useful for doing science — then first it needs to be “naturalized,” as Mitchell and his colleague Henry Potter have put it: We need a proper theory of agency, of a sort that can indeed furnish some sort of research program.
At present, nothing like that really exists, but it’s possible to set out some of the issues it would need to confront. For example: How exactly do organisms determine their goals? How are those aims represented (if at all) in the molecular or neural mechanics of the organism? How do organisms integrate information (including information about their own internal state) to come up with a decision about how to act? Can we measure how much agency an entity has — or what kind of agency? “I think developing empirical indicators for measuring different dimensions of agency would be the primary way to go for turning agency into a scientific concept,” Potter said.
Some researchers have asserted that agents need particular kinds of internal architectures to govern the relationships between senses, cognitive processes, actions, and needs. Moreno, for example, has argued that this organization involves feedback loops that don’t just enable adaptation to the environment but let the system alter that environment — for example, simply by moving from one location to another with a different set of conditions. The theoretical biologist Stuart Kauffman has argued that properties like this depend on what he calls “organizational closure,” in which the parts are collectively self-maintaining, each of them supporting the functions of the others. This property, Kauffman says, gives the whole system “causal power” — Mitchell’s ability for an agent to act on its own behalf rather than just being a passive vehicle for lower-level causes. Meanwhile, the neuroscientist Karl Friston has argued that agency arises in entities that can make predictions about the outcomes of behaviors and then compare those with what actually transpires, and seek to minimize the disparity between the two (none of which demands an ability to reflect on that comparison with conscious self-awareness).
At this level it can all sound a bit abstract and remote from actual biology. But I’m cautiously optimistic about the prospects of uniting such theoretical ideas with biological mechanisms. It seems unlikely to be coincidental, for example, that organisms that seem to show more agency also have molecular pathways that permit more openness to the influence of context and external information.
If we can get a clearer idea of what makes an agent, this could help us to understand how collective goals arise — as they did when multicellular organisms first arose long ago — and how they can break down, as in cancer. What’s more, a proper theory of agency might give us a clearer idea of what’s needed to make genuine artificial agents, not just computers and machines programmed with our own goals, but ones that can formulate their own. We might then also get a clearer idea of the potential benefits and dangers such truly agential machines might bring. But perhaps the most compelling argument for recognizing agency is that it might help us understand what makes life so different — not just humans, but life — that it is able to shape an entire planet in a manner visible from outer space.