Evolving Beyond Determinism

Kevin J. Mitchell's book Free Agents argues the long evolutionary path that created the human brain iteratively yet incidentally birthed free will.

Consciousness and free will are the heady topics that keep intellectually curious college kids up late for dorm-room debates. They’re also fodder for pop intellectual content that has taken off in types of long-form friendly media like podcasting and Substacking. However, these often hazy and genuinely unwieldy concepts have increasingly attracted real scholarly attention over recent decades. In fact, some readers may be surprised to learn that consciousness research was once (and in some places still is) considered an intellectual backwater. It was deemed the purview of mystics, gurus, charlatans and at best theologists and philosophers. Although many intellectually dubious figures still attach themselves to topics in the cognitive sciences, a bonafide cognitive neuroscience is emerging in fits and starts.¹ Outpacing but trending with the growth of this field is the public’s hunger for a naturalistic explanation of the human mind. Fewer and fewer are settling for the empty but uplifting woo of gurus or the hollow clichés mouthed by religions. Instead, there is a growing reliance on science to deliver understanding and a way forward.

One of trends that really speaks to the public’s hunger has been the blooming of books on cognitive neuroscience and related ideas. In 2023 alone, numerous titles, including many highly anticipated works from eminent scholars, have (or will) hit bookshelves: Psych by Paul Bloom, The World Behind the World by Erik Hoel, A Brief History of Intelligence by Max Bennett, The Four Realms of Existence by Joseph LeDoux, Determined by Robert Sapolsky, I've Been Thinking by Daniel C. Dennett, and Putting Ourselves Back in the Equation by George Musser. And it is not like there was a shortage of similar books prior to this year either.

Midjourney image: neurons

The public is inundated with these ideas. It’s a veritable flood. It’s clear publishers think there is an eager audience out there, and judging by the number of people piping Sam Harris into their ears, they’re right. At least, I know that I’m drawn to these ideas. I’ve read four of the 2023 titles I listed above, and these works have been a steady component of my reading diet for years now!

Given all this, a reader may ask, “Why, oh why did you read another?” And apart from confessing to an addiction of sorts, I would respond that the science in this space is evolving rapidly and perspectives diverge wildly. This makes for exciting discourse, harkening back to dorm-room days. Plus, no two books in this space are alike. So let’s focus on a particularly interesting one, Free Agents: How Evolution Gave Us Free Will by neuroscientist Kevin J. Mitchell. Free Agents presents a sophisticated, scientific response to determinism, a seductive idea to the scientifically minded who know all we have (or at least can know) is the material world, and it is whirlwind tour of the history of evolution, the human brain, and ideas about causation.

“Our thoughts come to us freely. Our action go from us willfully.”

~William James, “The Dilemma of Determinism” (1884)

Main Characters in the Ultimate Role-Playing Game

To start Free Agents, Mitchell draws on the now familiar distinction between playable or main characters and NPCs or non-playable characters in video games to sketch out the different positions on free will. An NPC is programmed to have a limited range of responses and actions that are activated by a particular set of inputs. This represents the determinist’s position. Alternatively, the playable character is not a mere stimulus-response machine. Although he is hemmed in by the parameters of the game and influenced by his prior experiences, his choices nevertheless direct the action of the game and are made for reasons known to the player. The main character, a self, is a causal agent. This represent the free will position. Jumping from this analogy, Mitchell previews the contents of Free Agents:

What follows is thus a full-throated defense of the idea of free will. Despite many claims to the contrary, the latest science - whether physics, genetics, neuroscience, or psychology - does not in fact imply that we have no choice or control over our actions. It’s true that we are learning more and more about the mechanisms underlying our cognition and behavior - from neural systems and circuits down to the level of cells and molecules or even atomic physics. But even though our cognitive systems have a physical instantiations, their workings cannot be reduced to this level. We are not a collection of mere mechanisms. As we will see, the nervous system runs on meaning.

To many, Mitchell is defending what our day-to-day experiences of decision-making feel like. It is is the common sense position that anyone’s brief introspection immediately suggests. To understand why Mitchell’s position is heterodox, it requires noting that some flavor of determinism reigns supreme in the minds of many, if not most, scientists.² This perspective can be summarized in short: 1) the world is made of matter; 2) the behavior of matter is governed by the laws of physics; 3) to the extent there are causal relationships, they follow exclusively from physical causes; 4) the brain, being obviously made of matter, is not free from the laws of physics.

It is also important to note that Mitchell’s book isn’t a rejection of this type of materialism. As his above quote highlights, he thinks the determinists simply have a mistaken reading of the science. They have concluded that ultimate causes of human behavior go all the way down to the fundamental subatomic particles and attendant forces. Alternatively, he thinks that conscious experience is the only level that deliberative human behavior can be explained at. In other words, the meaning of thought and the goals of the self are the true causes of human action. He doesn’t fault the determinists too much. It takes an entire book to point out the issues with their interpretations after-all.

“In a minute there is time / For decisions and revisions which a minute will reverse”

~T. S. Eliot, “The Love Song of J. Alfred Prufrock”

The Evolution of Free Will

Free Agents is ultimately "a naturalistic framework for thinking about agency and free will." To construct this framework, Mitchell has to take readers on a tour of science and the history of life. He gently guides readers through millennia of evolution, the anatomy and physiology of the brain, the vagaries of quantum physics, and myriad philosophical ideas about causation and agency. Throughout this tightly packed yet accessible journey, Mitchell successively, like the process of evolution, builds his case for how agency emerged in complex organisms. And though I fear his model is incomplete and imperfect, it is nonetheless persuasive. It is a robust challenge to determinism that at the very least complicates its boilerplate arguments.

Arguments about free will, even among many hard determinists, tend to veer into nebulous philosophical realms. This book is no exception, but Mitchell studiously keeps this to a minimum, foregrounding scientific findings and reserving philosophizing and abstraction for the later chapters. Mitchell spends the entire first half of his book recapitulating the science of how life emerged and how life became complex and capable of agency. A critical part of this effort includes demonstrating the sophisticated capacity of the brains of seemingly simpler organisms and expanding a typical reader’s understanding of what agency is. Mitchell’s definition includes almost any action that isn’t entirely determined by external stimuli and internal presets. In any instance where organisms are making decisions from among a range of options guided by a goal (even base biological ones like survival or reproduction), is an example of agency. Given this broader definition, Mitchell is laying the ground work for just having to show how evolution can scale this agency up to a level where mental states are the primary influence over human action.

Mitchell’s argument about agency has an important two-part biological premise: organisms "cannot be understood as static machines or instantaneous arrangements of matter,” rather “they are patterns of interlocking dynamical processes that actively persist through time." This biology is a product of the pressure to persist and reproduce (natural selection). Natural selection is the engine of evolution, driving the unfeeling, undirected process that increased the complexity of life over the last few billion years. This incidentally bequeathed complexity: the coding of stimuli into good and bad boxes (valence), the integration of many sensory stimuli in a control center (a central nervous system), the specialization of cells and cell states (multicellularity), the representation of the environment within organisms, simulation of possible events in a mentally mapped environment, and eventually an abstract and recursive map of mental experience itself. Mitchell argues this final step of reasoning about reasons (metacognition) along with the indefiniteness of the basic rules of matter allow a potent form of agency to be facilitated via physical mechanisms. It makes consciousness the special locus of control over deliberative behavior. We act for reasons, and these reasons emerge from the collection of our experiences, the goals we've set (via metacognition and planning), our innate proclivities, and the set of choices available to us in any situation.

Mitchell tentatively offers up the term “cognitive realism” to label his defense of free will. Cognitive realism, in contrast with dualism or reductive materialism, asserts that 1) causation behind behavior is thought; 2) thought cannot be discovered by quantifying neural activity; yet 3) thought is nonetheless facilitated by physical processes. Points 1 and 2 without 3 would be dualism, while the escape route that keeps point 3 from leading back to reductive materialism is the contention that mental states do not reliably follow from patterns of neural activity. How could this be? This is quite the needle to thread, but the idea is that the activity of the brain is noisy and imperfect and that consciousness has to sort this out. The longer an individual persists, experiencing new things, integrating that information, and engaging consciously in decision-making, the more degrees of freedom accrue. This argument is similar to Erik Hoel’s theory of causal emergence as an outgrowth of integrated information theory (IIT), but Mitchell appears to house these similar ideas under a different theory of consciousness, global neuronal workspace theory (GNWT).

Is Cognitive Realism a Viable Defense of Free Will?

Mitchell has done his homework and thought carefully about this topic. There's a great deal of scientific merit to Mitchell's claims. It is genuinely a materialist defense of free will without special pleading, a masterful feat especially with an audience that will be more scientifically educated and inclined (Free Agents is from an academic publisher Princeton University Press). The erudition and clarity with which Mitchell handles complex concepts is praiseworthy too. However, there is also a nagging measure of incompleteness or something that just doesn’t completely ring true. Plus, Mitchell’s claims seem particularly vulnerable to future advancements in cognitive science. In some ways this is admirable because it means his claims are at the very least testable, but it also makes them far from the final word.

Midjourney image: brain maze

Some immediately complicating questions jump to mind: What if the recursive mechanisms of consciousness are elucidated and thought and decision-making is shown to be entirely outside of conscious control? What if all or even a significant portion of complex human behavior becomes predictable or explainable using more fundamental inputs that mental states? What if human-level intelligence and consciousness is achieved artificially and those entities are clearly without agency? Now, I think definitive answers to the questions above are unlikely, but they can't entirely be ruled out either. The most pressing threat may be artificial general intelligence (AGI), but it may be the easiest argument to adjust too. Anticipating this, Mitchell’s epilogue does meditate on recent trends in AI. It seems he expects that AGI will require evolutionary processes to emerge and thus likely be agentic if achieved.

It is certainly possible that Mitchell's claims will survive new advancements in our understanding of the mechanisms of human cognition and behavior. In fact, one of the strengths of his arguments is that he is quite willing to delineate the ways in which agency is constrained by genes, development, other biological factors, culture, and the environment. His model of constraints on free will integrates well with his overall model of cognitive realism. The weak point is that his model doesn’t specify how many degrees of freedom are created by cognitive realism. Can we quantify this freedom? Moreover, his model appears to suggest that some people benefit from more free will than others. This raises thorny ethical issues that probably deserved greater consideration. However, I enjoyed that Mitchell was somewhat loathe to read too many normative conclusions into his theory of free will. Such tendencies run rampant among many determinists. For instance, Sapolsky often invokes anti-free will claims to argue for greater leniency in correctional practices or advocate anti-meritocratic sentiments.

In sum, Free Agents is a provocative and special contribution to the discourse on free will. It is a welcome challenge to determinism. I too feel that conscious mental states are important to guiding human behavior, but I remain wary about whether they’re truly irreducible and whether it is appropriate to label mental causation free will. And if it is free will, how much freedom is there? Regardless, these ideas are lovely to engage, and it is a balm of sorts to be reminded of the power and sophistication of our neural hardware.

1. If I was forced to put a date on when questions about consciousness and human agency jumped from the realm of the humanities to that of the sciences, I would probably (like Erik Hoel) point to the early-to-mid-1990s when two Nobel laureates, Francis Crick and Gerald Edelman, launched competing, parallel efforts to formally study consciousness. This was also around the same time that David Chalmers, a giant in the consciousness field, defined the “Hard Problem” of consciousness.

2. Part of the popularity of determinism is owed to a set of experiments by Benjamin Libet where EEG was used to predict the action of experimental subjects prior to action via detection of something called a readiness potential. In Free Agents, Mitchell points out that Libet’s finding has been dramatically over-extrapolated to all of human action and that the experimental design precludes any insight on free will because the experimental instruction compelled subjects to make unplanned actions. Additional empirical work suggest that deliberative decisions cannot be predicted in the way whimsical ones can be using the readiness potential.