Life, agency, and the evolution of free will

How lifeless matter gained the power to choose

Traversing billions of years of evolution, Kevin Mitchell tells the remarkable story of how living beings capable of choice arose from lifeless matter. 


Across a billion years of Earth's history, life evolved the power to choose. Most things in the universe are caught up in the flow of physical causation around them but life doesn’t like to be pushed around. Life holds itself apart. First, living beings maintain a barrier with the outside world – this is what distinguishes them as entities. And second, they do work to keep themselves organised, and in doing so seem to resist the second law of thermodynamics.

Living organisms are thus causally insulated from the rest of the universe. They are buffered from the effects of all kinds of physical and chemical goings-on outside them, not pushed around by every passing potential cause. Instead, they become causes in their own right. Not merely as complex mechanisms, which you might say are pushed around by their own parts, but as holistic selves – as agents.

These causal capacities evolved with the earliest life-forms, but were greatly elaborated along certain evolutionary lineages, especially our own. By following this story, we can understand the emergence of sophisticated systems for cognition, decision-making, and behavioural control that ultimately manifest as what we call “free will” in humans.

related-video-image SUGGESTED VIEWING Free will is not an illusion With Denis Noble

This story begins with doing things for reasons. Non-living things don’t have reasons and they don’t do anything – they’re merely involved in happenings. But living things have a prime directive: to stay alive. They persist through time, in exactly the way that a static entity like a rock doesn’t – by being in constant flux. Living organisms are not just static patterns of stuff, they are patterns of processes.

Keeping those processes going requires energy, which for many organisms means taking in some kind of food. This simple fact presents a fundamental sort of goal-directedness. We can say that an organism – even a simple one like a bacterium – has a goal of getting food in order to support this over-arching purpose of persisting. It thus becomes valuable, relative to this goal, to develop systems that can detect food in the environment and that mediate directed motion towards it. The organism is now not just being pushed about by mechanical forces – either inside them or outside them. Instead, it is responding to information about things in the world and reacting appropriately. It is doing things for reasons.

Of course, bacteria are not aware of those reasons. They’re not apprehending the signals from the receptors on their surface and then deciding what to do. Instead, the appropriate control policies are pragmatically wired into their biochemistry and cellular physiology.


Sensation became decoupled from immediate action.



Our little bacterium is thus not pushed around by its parts. It is not just a locus of some happenings, driven deterministically by simple stimulus-response mechanisms. Rather, it is a constantly active – indeed, pro-active – organism, reading the cues from the environment and adjusting its behavior appropriately, in ways that favour its persistence. It is a holistic and integrative system that does things for the reasons of the whole organism – not the reasons of its parts.

Over evolution, that basic framework was elaborated on, in ways that granted greater and greater autonomy from the exigencies of the environment, and greater control over the farther future. Multicellular creatures evolved, which faced essentially the same kinds of ecological challenges, namely figuring out what is out in the world and what to do about it. Specialised cell types emerged to meet these challenges, including sensory cells for various kinds of information in the world, as well as muscles and neurons to coordinate motor responses.

Simple animals have pretty proximal couplings between sensors and motors, but over time, more complex nervous systems evolved with additional intervening layers of neurons. These extra layers of neurons made it possible to extract higher order information from the sensory data – parsing it to infer the presence and location and identity of objects, for example – exactly the information that organisms need to guide their behavior.  Sensation became decoupled from immediate action. Instead, these intervening layers represent information about what is out in the world, making it available to a central cognitive arena.

TN 9 SUGGESTED READING A new theory of matter may help explain life By Lee Cronin

Now the organism could think about things and, crucially, it could also learn about things. Nervous systems are adaptive control systems. Organisms learn about the properties of objects they’ve encountered and their causal relations. In particular, they learn about their “affordances” – what the organism can do with them, or what they can do to it. And they also learn about events, and sequences of events, about which things tend to follow each other. They are thus able to build up much more powerful models of the causal structure of the world.

This includes models of themselves as causal agents. They learn from experience which actions turned out well or turned out badly across various situations. The next time they encounter a similar situation, their behavior can be guided accordingly. In mammals, a sophisticated system evolved to predict the likely outcomes of various possible actions in any given scenario, to evaluate the predicted utility of each of these outcomes, and to use this information to inform the choice among the competing options. This internal simulation provides another layer of insulation from the immediacies of the environment, and thus a greater degree of autonomy. Organisms with such systems are really deciding what to do.

One final step sets human beings apart. Like these other organisms, we can decide what to do, for our own reasons. But we have enough levels of the neural hierarchy that the topmost ones can in a sense “look down” on the ones below and see the reasoning going on. We can reason about our own reasons. We’re not driven blindly by them – we are aware of them and can evaluate them. We can attach varying degrees of certainty to different beliefs, for example, to better guide behavior in an uncertain world.


We have the collective ability to shape the far future in ways that only our imagined gods could have done before.


Even more – we can talk about our reasons. We can communicate with each other. We don’t have to reason alone. The emergence of language and culture boosted us from merely smart apes to the dominant super-species on the planet, able to behave collectively in ways that have allowed us not just to respond to the world we encounter, but to actively shape it and bend it to our ends.

How this evolutionary story will end is, like all things in the future, not yet determined. We have a degree of cognitive control over our individual actions that warrants being called “free will”. And we have the collective ability to shape the far future in ways that only our imagined gods could have done before. And yet, whether we will develop the collective capacity to wield this power wisely remains to be seen.

Latest Releases
Join the conversation