Biological Agency and Free Will

Given my recent interest in the idea of naturalized biological agency (the topic of my last post), I was excited to read Kevin Mitchell’s book—Free Agents: How Evolution Gave Us Free Will—which develops this notion and connects it to the debate over human free will. Mitchell is a neurogeneticist at Trinity College (Dublin); he blogs at Wiring the Brain.

This is an accessible and engaging book that I recommend to anyone interested in this subject. Mitchell brings a wealth of information about biology and neuroscience, along with a dose of philosophy, to his exposition. His goal is to show that humans can fairly be said to have free will, given a careful look at the capabilities endowed by our biological inheritance.

It is a detailed and compelling case, and I found myself mostly nodding in agreement. My main worry is that Mitchell’s most convincing examples of the (genuine) freedom we exercise as biological agents actually do not conform to what we typically think of as “free will.”

The Roots of Agency

Like the enactive philosophers whose views I discussed before, Mitchell sees all living things has having “some degree of agency (19)”, and so in the early chapters he focuses on simpler organisms before considering the human case. He argues that the special persisting organization of these entities confers a kind of autonomy within the world. Their insulation from the rest of the world (highlighted by semi-permeable cellular membranes) creates a distinction between an inside and an outside that defines the organism as “an entity (33).” The entity strives to continue its existence, and its interactions with the environment are laden with value: they are good or bad in light of the goal of survival.  Given the role of genetics and inheritance, the persisting organism’s causal organization has a historical character that spans not just its lifetime, but reflects its lineage as well.

With regard to “goals”, Mitchell notes that he is not positing a cosmic purpose (nor does he think evolution is a goal-directed process), but “once life does exist, everything changes (42).” With the origin of life, goals enter the world and this brings along related notions, such as the idea that the organism’s sub-processes serve functions.

An important development comes when certain internal states embody signals of happenings that matter to the organism. This brings an informational component to bear on the organism’s causal patterns (discussed in Ch. 3). The organism’s subsequent survival-fostering reactions to these signals indicate that the signals have meaning. Organisms, then, can be said to be acting for reasons. Of course, multicellular organisms build on this beginning, with nervous systems greatly enhancing the capabilities of animals. This leads to sophisticated cognition whereby certain neural patterns carry meaning in the form of semantic representations. Mitchell focuses more on the human case in later chapters, but human capabilities are recognizably shared with those of other living things.

The Causal Backdrop

Throughout the book, Mitchell discusses topics in the philosophy of causation that are relevant to this story. In Chapter 7 (“The Future is Not Written”), he pauses to recap his views on these issues. There are two key points: 1) the world is indeterministic; and 2) causation occurs at multiple scales.

Mitchell starts with the fact that quantum mechanics shows physical processes are irreducibly probabilistic. He gives only a brief defense and discussion of alternative positions—those who adopt one of the deterministic interpretations of QM won’t find a detailed refutation of their views here. But since I agree with him, I will not discuss this further in this post either.[1] He also argues that there are no good reasons to think things become deterministic as you move to macroscopic or “classical” scales. A nice bonus here is that Mitchell mentions some work that I admire by Nicolas Gisin and colleagues who argue that the familiar determinism of classical physics rests on the implausible assumption that initial and boundary conditions can be specified with infinite precision.[2] This is an artifact of our mathematical models, not something physically plausible. The big picture is one of an open, undetermined future, and this leaves “causal slack” for agents to act freely.

On the second key point: Mitchell emphasizes from the start of the book that causation is not just a matter of the lowest-scale physics. Acting according to reasons, say, could not truly enter the picture (except perhaps as an explanatory convenience) if the micro-scale was all that mattered. Life “demands a wider concept of causation over longer time-frames and an understanding that the dynamic organization of a system, which encodes meaning, can constrain and direct the dynamics of its component parts” (21).[3]  The lowest level of physics is not enough to account for phenomena we observe, and indeterminism opens space for organizational patterns at higher scales to have causal influence alongside that of the smaller scale workings. As Mitchell notes, some who make similar arguments describe this as “downward causation.”  My own preference is to avoid this terminology, since there are no synchronic causal relations between entities at different scales or “levels” (and I believe Mitchell agrees).  I would call the view “multi-scale causation”, where repeated patterns of interacting small-scale processes form composite systems with their own causal profiles expressed over larger spatio-temporal scales.[4]

To conclude: indeterminism and multi-scale causation provide the backdrop for true biological agency.  But how this works in detail, and how it might relate to the notion of free will, is where things get really fascinating.

“Harnessing Indeterminacy”

This is the title of Chapter 8, and it describes how biological entities can act in way we can call “free.” Mitchell has already given some examples in the earlier part of the book.  A classic case is bacterial chemotaxis: E. coli will use its flagella to move up a gradient of increasing concentration of available food, but in the absence of such a gradient, it will switch to a “tumble” motion and then set off in a new random direction (55).[5] A paramecium, encountering an obstacle or noxious substance, will reverse and move in a new random direction, and in this case the relevant internal signals are electrical, using a “process very similar to that used by nerve cells in more complicated organisms (56).”  These creatures are acting based on meaningful signals, but their actions are not “determined,” since “there is substantial variability in the behavioral outcomes arising from any given stimulus (60).”

The hydra is a well-studied animal with a basic nervous system and a simple repertoire of movements. These are deployed actively when, say, searching for food:

It is important to emphasize that the organism is not just sitting passively, waiting for some stimulus. Hydras engage in spontaneous behaviors, and their nervous system shows endogenous patterns and oscillations of activity, even in the absence of any external stimulus. (87)

Spontaneity is a behavior that relies on indeterminacy as well as a looser connection between perception and action allowed by a more complicated body and nervous system.

In our human brains, Mitchell explains that “indeterminacy manifests as variability in the cellular processes and electrical activity of neurons (170)”. This “noise” is in some ways a challenge for efficient functioning, and Mitchell explains how this is overcome. But he argues it is also “essential to enable flexible behavior…the brain has evolved to take advantage of the noisiness of its components to allow the organism to make some decisions itself (170).” Given the constant surprises offered by the environment, variability in how we behave is an advantage, just as it is for simpler organisms. Indeterminacy can be harnessed in a number of ways:

It can be drawn upon to resolve an impasse in decision-making, to increase exploratory behavior, or to allow novel ideas to be considered when planning the next action… Organisms can sometimes choose to do something random. (175)

It certainly seems clear that being unpredictable can be advantageous. Mitchell discusses research on cockroaches that shows their movements have a definitive random component. Another example from the animal kingdom comes from work on leeches, where, in carefully controlled conditions, there is unpredictability in neuron population firing patterns that govern the choice of movement modality (between swimming and crawling). Neuron populations involved in many functions will accumulate noise and settle in one particular possible state in the absence of compelling inputs driving a direction.  An organism can use this process to make a spontaneous choice.  This can be utilized when a choice isn’t important or “where options are equally attractive, given the information available (180).”

This comes into play in an interesting discussion of the oft-debated experiments originally designed by Libet. Mitchell argues that the onset of detectible readiness potential prior to subjects deciding when to move their finger (a totally arbitrary decision) is a sign of ongoing spontaneous fluctuations reaching a threshold for initiating movement. In similar experiments involving important decisions that involve weighing reasons, no association between the timing of the decision and readiness potential is apparent (186).

Another possibility is that sometimes it is the set of options for action that arise spontaneously, even if the final decision is driven by reasons. This recalls the two-stage model of free will discussed by William James. This can be particularly useful in novel situations, and/or when facing time pressure.

Mitchell gives other examples and arguments, and I found the case that organisms, including us, harness indeterminism to foster our goals to be persuasive.

“Is This Free Will?”

I am going to give a compressed account of the later chapters, but this is not to say that they aren’t worthwhile.  In Chapters 9 and 10, Mitchell considers the human case in more detail, including: an examination of how neural patterns encode meaning; a discussion of the role of the mental and of subjective experience; and the interplay of agency with biological constraints.

In Chapter 11, he considers the important question of whether our metacognitive and introspective faculties mislead us about how we make decisions. He argues that we should assume these faculties are mostly reliable. Our evolutionary endowment includes the ability to model not just the world but also ourselves (254). We update these models as we learn, including those of our own reasoning process. We obviously don’t yet fully understand the role of consciousness in our cognitive processes, but when it comes to our actions, we can generally consciously control them when needed, and it seems implausible that this is somehow illusory.  This is supported by the fact that this conscious self-control does vary across individuals. This is a “source of evidence that what we think of as free will— the ability to consciously monitor and control our cognitive processes, to reason about our reasons, and thus regulate our own behavior— is not some abstract metaphysical postulate, but an evolved function, or suite of functions, with a very real biological basis.” (272).

In the concluding chapter, he considers the account he has given and asks the question: “Is this free will?” He has described how we are the inheritors of a naturalized biological agency that brings with it the concepts of purpose and meaning, and acting in light of reasons. Obviously, we are constrained by causal influences, including that of internal consistency with our own history as a persisting biological entity—this continuity is what defines a self:

Selfhood thus entails constraint. It is only constraint. The freedom to be you involves constraining the elements that make you up from becoming not you. (279).

Free will is clearly, then, not an all or nothing thing. But we have degrees of freedom to make decisions given our relative autonomy. This is in the context of a world that is indeterministic and therefore leaves room for free action. At the human level we have some insight into our own exercise of freedom in our decision-making process. As a result, Mitchell feels confident in concluding: “If free will is the capacity for conscious, rational control of our actions, then I am happy in saying we have it (281).” 

“Free Will” or “Strategic Spontaneity”?

Now I’ll offer a criticism of this conclusion. I begin by noting that the idea of harnessing indeterminacy for action and decision-making—a compelling feature of early chapters and the emphasis of Chapter 8—recedes from view in Mitchell’s concluding chapters. And yet they are the cases that are the most convincing examples of free acts. When we utilize noise to arbitrarily choose between options (and perhaps spontaneously generate those options to begin with), our actions are undetermined. On the other hand, when we have strong reasons for our choices, and we follow through on them, it is much less clear that these acts can be considered free. And I worry about cases where introspection may mislead us into thinking our acts are not determined: perhaps there are causal factors driving us that are opaque to our metacognitive lens.

So, while Mitchell helpfully makes the case for the existence of genuinely free, noise-harnessing acts, he is subsequently a bit too eager in his conclusion to convince us we can hold onto an ordinary intuitive notion of free will: that “we” are usually in control of our acts. I’m not convinced this is true.

Before finishing up this point, let me be clear what criticism I am not making.  In free will debates you will often hear folks say something like: “well if indeterminism is true of the world, that just means things happen at random and that’s not a basis for free will either.”  This is a shallow assertion that misses the mark.

Assume a scenario where, torn between two options with equivalently strong reasons for them, I harness internal noise to “just choose” one of them. Then, while the specific outcome is arbitrary and its consequences will involve luck or chance, it would be misleading to say the outcome just happened randomly.[6]  This would ignore the context: the act is a part of a larger, goal-directed pattern of activity. I agree with Mitchell that humans and other organisms are causally productive entities, operating alongside their parts at their respective time-scales, and as such they can be considered agents that are owners or authors of (at least) this kind of free act.  The label I might propose for this capability is “strategic spontaneity.”

So, the “good news” is that this kind of act is free.  It is authored by me, and it was possible for me to have done otherwise right up to the point where the choice was made. The “bad news” is that strategic spontaneity is not what we seem to have in mind when we think about free will. A revision to the ordinary idea would be needed to bring it into alignment.

Now, perhaps one might say that if an agent is capable of some free acts, then it is a bona fide “free agent”, and everything it does can appropriately be labeled as free. However, I’m not sure this move is justified.

In any case, Free Agents has advanced the debate, and has certainly given me a lot to think about!

Reference

Mitchell, Kevin J. (2023). Free Agents: How Evolution Gave Us Free Will. Princeton: Princeton University Press.

---

[1] My (surely correct!) view on interpreting QM is outlined in this post.

[2] Here is an interesting Gisin paper from 2019.

[3] I would note it is not only biological phenomena that demand a multi-scale view of causation. I would argue chemistry and condensed-matter physics do as well.

[4] For my own approach, see this post.

[5] Mitchell later explains that the tendency to initiate random tumbles is “itself subject to some randomness (174).” He notes that even sets of genetic clones will behave differently.

[6] A philosopher who has carefully explored the question of how indeterminism can ground freedom is Mark Balaguer. I discussed his 2010 book on free will on my old blog.