A Review of Determined, by Robert Sapolsky
In Determined, Dr. Robert Sapolsky shows his Stanford chops and builds a powerful argument that free will is a terrible basis for accountability. Our ability to take account of the various determinant factors that lead to behaviors that one can get into trouble for has increased along with the development of modern communications technology and as a result, it is increasingly clear that we need to spend the time and energy it takes to figure out how to do a better job of making our social systems help people. Modern blame culture doesn’t take circumstances into account enough, and the reason Sapolsky ventures is simple: everyone assumes people have free will, that all choices are equal. In Determined, we find out that it is most decidedly not so - our past plays a major role in our decisionmaking and the sooner our social systems begin to take this into account, the better.
Along the way, Sapolsky faces down several philosophical chimeras to argue that much if not all of our experience of living is determined. He identifies this stance as Hard Determinism, from the philosophical literature, and everything about the work is cutting edge. Unfortunately, Sapolsky does not simply discard the dialectic of Free Will and Determinism (the FWD dialectic). Sapolsky is mainly interested in using philosophical research as a point of evidence, as one would do with scientific research. Scientific papers are atomic sets of facts about the world; philosophy papers are explorations of possible arguments for possible conclusions.
By engaging with Sapolsky and his strong scientific background, I hope to make a very strong argument in favor of updating the philosophical literature with some insights from the nascent school of cognitive science to better suit his purpose. What I take Sapolsky’s purpose to be, to make things plain, is a goal consisting of two parts: 1: undercut Free Will as a widely-held point of view and argue in favor of Hard Determinism instead to create a strong foundation for a more mechanistic view of the mind, in support of 2: update the modern judicial/moral apparatus to take into account all of the myriad nuggets of wisdom we’re developing around what it actually means to be a person with a brain in a body in an environment. While both of these goals are great, choosing determinism as a philosophical position is not as cut and dried as Sapolsky makes it out to be.
This review will constitute an analysis of Sapolsky’s view and a deep philosophical response to it: instead of concluding that we are deterministic machines (which is flawed for all the reasons Sapolsky himself lists), we should scrap the Free Will/Determinism dialectic altogether and replace it with cognitivism. Cognitivism is a cutting edge scientific framework that allows us to explain what is physical and what is virtual in the mental apparatus and in this bizarre, wonderful subjective experience of life each of us is fortunate enough to have.
Responding to Sapolsky
To start us off right, it’s necessary to offer a salute to Dr. Sapolsky just for getting through that dung heap of philosophical literature about Free Will. From Libet (who Sapolsky describes, correctly, as irrelevant) to LaPlace, Sapolsky has a thorough grasp of not only what great philosophers of the past have said about these things, but also the modern literature of academic philosophy’s responses to such thinkers (which I grew quite weary of back in philosophy grad school).
However, it is also important to get things right. Hard Determinism is the belief that the LaPlace daemon (a thought experiment involving a magical creature with perfect information) would in fact be able to calculate any and all future states of all systems in the universe, on the basis of information, which is possible because the universe moves from state to state in ways that are reducible to simple rules that merely need to be applied to find out what happens next from any particular point in the past. The LaPlace daemon is assumed to have some magical codex containing perfect information as well as a perfect algorithm, but this assumption is very very bad because it’s actually impossible for it to be right, knowing what we know about language: the information would presumably need to consist of linguistic representations of the world, and to make a long story from my book, Formal Dialectics, very short, language isn’t good enough to model the world to this extent. Borges’ perfect map of the empire that is scaled the same as the empire itself is a better-known description of this general concept. (more information about Borges' piece can be found here: https://en.wikipedia.org/wiki/On_Exactitude_in_Science )
The “less free” position Dr. Sapolsky advocates would perhaps better be termed Soft Determinism, but aside from the apparent misnaming of it, Sapolsky has done a great job articulating the philosophical concept and its power. We may not be able to say that the universe is deterministic at the quantum level, but at the level of experience almost everything that can be observed can be said to be deterministic. In the philosophical sense, determinism is something a lot like modelability; i.e., with all the information plugged into the right equations, a deterministic system is predictable if it is linear and “chaotic” if it is non-linear. This paper will push back against Sapolsky to present an alternative idea, the cognitive idea that the mind is an enactive complex that evolved as a relationship between organism and environment, to serve the purpose of making decisions about what to do to promote the health of the organism whose mind it is (and the community it is a part of).
Now, some systems that seem like they should be predictable aren’t, and despite a very nice introduction to emergence and chaos theory, Sapolsky doesn’t want us to get lost in the weeds here. My own philosophical approach is almost the exact opposite; I love to jump right into the chaos theory weeds and I feel I’ve brought some very interesting facts back from such ventures. That said, we can all see why it’s difficult to argue that a brain is deterministic: we have nothing like the sophisticated techniques we’d need to deeply model the brain’s trillion synapses at scale, much less to leap beyond this point to the point where our consciousness resides, in the virtual space created by the actions of these neurons in time. In fact, this may require anthropological techniques like asking people questions, and that comes with the built-in uncertainty of whether they’re able and willing to tell us the truth.
Sapolsky doesn’t quite reach this conclusion, and instead settles for calling consciousness a deterministic (if emergent) phenomenon outright. One wonders whether Dr. Sapolsky decided to think of his own consciousness at all in the writing of the book. There are a few things we need to be sure we don’t miss in the text, for the critique this article advances of Dr. Sapolsky’s position to hit home. For the next few sections of this review, we’ll focus on the chapters toward the end of the book, where the more advanced concepts can be found. The earlier chapters do a good job of explaining the status of the present philosophical literature around free will and determinism, as well as chaos theory and emergent phenomena. These are recommended reading, but we’ll focus on the juicy stuff for this review. Chapters 12, 13, 14, and 15 are independently interesting, but they also form a wildly useful whole, to which the next four sections are generally oriented.
The twelfth chapter of the book is an absolute masterpiece, providing evidence that was ultimately absent until the end of the twentieth century and the onset of the cognitive neuroscience revolution. Despite this laudable achievement, however, the chapter leaves a few open questions including this one: Where does a person’s sensation of deliberately considering options and then deciding upon a particular choice figure in, and why does Sapolsky seem to completely miss this core component of conscious thinking?
The virtual component of consciousness is not easy to spot in the data because subjective experience is too complex to fully describe in words, just as a post-hoc analysis of a modern microchip would be unlikely to provide you with any information about what sorts of programs it was used for at a high level of specificity. This disconnect between the real and the virtual is good for us because it is the reason our brains can think so many different thoughts and we can consciously experience so many different things. It also doesn’t hurt our ability to adapt to new situations.
Whether my Apple M1 microprocessor is used to run Google Docs in the browser or MS Word locally does not translate into easily accessible circuit-level understanding because the programs are virtual and the processor is a real-world artifact which can run arbitrary programs because they are identical for its purposes. We could say that the user interface is richer, more rewarding than the binary code the processor executes. And we could argue that this happens many removes of complexity above the processor, in a far more abstract space.
Imagine what your dog, who is likely unable to comprehend the images on the screen, thinks of you staring into your phone or laptop all the time. The pet perceives events as you do, but only up to a certain level of complexity, beyond which you leave it in the dust - it has no idea why you’re behaving as you are. For that matter, it would scarcely be obvious to any human from an arbitrary point in our species’ past up to about twenty or thirty years ago.
In the next section I’ll provide another layer of gloss on the issue, but the problem comes from the difficulty inherent in knowing what something is for, if you only know what it’s made out of.
Culture is one example of a virtual phenomenon that emerges from the activity of enough neurons at scale that Sapolsky should know well enough to take it into account here, but the utter complexity of the problem is perhaps grounds for us to forgive him for not doing so. Nonetheless, think of any interaction you’ve had with anyone and the impact that it has had on you - this is what we mean by culture, and by 2-way causality. If you impact someone else, you have an impact upon your culture, and anything anyone does that you notice or that affects something about your life is your culture impacting you. Writing a book about determinism in an attempt to sway the minds of your fellow citizens to do a better job with the legal system is bottom up causation; that change being implemented and making things better for everyone is top-down causation where culture is the higher level of abstraction and an individual person is the lower level causal agent.
Deriving an account of cultural phenomena from the activity of any subset of available neurons in particular human brains at scale for anything even remotely complicated would be a ludicrously challenging task, were it even possible. Assuming technology eventually connects everyone to the internet sufficiently for a model of individual cognitive states to become feasible, the differences between individuals’ experiences would still be a very daunting obstacle to effective modeling of the virtual contents of people’s experiences - so culture would still be more or less inscrutable due to linguistic bandwidth restrictions and the need of useful symbols to be generally useful enough to be known by others.
We also have to question why Sapolsky left out the concept of agency, which occasionally makes its way into FWD debates. In the philosophical vein of cognitivism, agency is the ability of an organism to see the world change in accordance with that organism’s worldview. So we each have at least two parts: how we see the world, and how we act in it (of which deciding what to do is a nontrivial part!). Perhaps this is a small, virtual component of consciousness that attempts to defy explication in the traditional scientific vein where Sapolsky is most comfortable, but nonetheless it is highly important to follow all available leads when working such a difficult case! The argument that consciousness is best described under the philosophical Hard Determinism model may be the best available in the philosophical literature today, but it still falls far short of wrestling with the totality of human knowledge of cognitive phenomena and their origins.
Sapolsky discusses epilepsy a good deal in Chapter 13. His arguments’ implicit premise: due to the nature of the damage epilepsy can do to conscious bodies and minds, we understand that the self governing the actions of these bodies and minds is a deterministic one. As he loves to say, “it’s turtles all the way down.” The implication is that life starts with a determined event and ends with one, and everything that happens in between is also determined. No matter how far back in time you went, it would all have been determined. But we have no good reason to believe in this infinite regress, despite our charitable attitude toward Sapolsky and his generally-sound reasoning in this work!
How do we keep the good but help Sapolsky update his way out of the woods if 2-way causation between a virtual layer of conscious experience and the neural circuits he describes is the inception point of the phenomenological sensation of making a choice, and Hard Determinism ends up being just as false as the indeterminist’s view of an unfathomable world where things just happen? In the cognitive account of consciousness, we construct an additional dimension on the same biological structure and it adds information about the biological interactions we are observing, which are identical in either case.
Would things generally be the same if we went back and restarted our universe from a given point in its history? It seems like a pretty grandiose claim to make. It is more likely that our instinct for creating order for ourselves is making claims in a space that is too abstract to have much relevance to our lives, that Hard Determinism is a myopic and deeply flawed view of the cognitive apparatus.
Though I don’t know that he actually gets all the way to this conclusion in the book, Sapolsky’s brushing-away of Libet and the line of study he engendered is related to the fallacy inherent in the Libet study’s core premise, which is that if someone freely decides to push the button, activation should occur differently than it does when we measure it. This is rather ridiculous, because it assumes that the activity of making the choice should occur somewhere else and at some other time than it does. What if choices simply take time because the frame of mind one must be in to participate in the experiment involves endogenous processes that have to circulate to the prefrontal cortex (pfc), then back to the relevant cortical/motor regions, then back to the pfc again, and the process just simply involves the deliberated-upon physical action at a slightly different point than Libet expected? It could be a physical basis for an abstract process that freely chose things to do, and Libet’s experiment could either be proof of how the higher order process chose things to do, or it could be the workings of a mechanistic process that subconsciously made choices arbitrarily, allowing conscious thinking to later come back and justify them to itself. There are many other critiques of this line of experiments, but we needn’t delve further into them because the information they gather is organized in such a way that it gives us no reason to believe in Free Will or Determinism.
Pleasure and pain drive action, we can say. But they don’t explain behavior at a granular level in complex organisms. If consciousness is a virtual process, involving arbitrary real-world mechanistic chemical interactions and they are organized, instead of mere biology, around compressed data, then determinism is false high up the causal chain because the physical substrates of the conscious states of mind we experience are arbitrary relative to the actual physical system’s particular states - we can neither model the activity we see predictively nor create a thorough enough linguistic account of what might be happening there from our vantage, looking as we are into the physical substrate. This substrate holds some but not all of the information in the network, leaving us without the key insight we came here for about how to understand the virtual layer it supports.
So we can observe the brain waves, the patterns of electrical impulses, and the metabolic fluctuations that make consciousness possible, but we still cannot take an experience out of someone else’s mind and simulate it in our own. Experiences are not reducible to language; which means they can only be real and never themselves virtual, and hence the best we can do is make imperfect speech acts, by definition, will never convey the entirety of our point of view to someone else. Language is always virtual; it models arbitrary subjects and compresses them into symbols like the Grand Canyon. Imagine thinking that the phrase ‘the Grand Canyon’ conveyed the full experience of the place the Grand Canyon! The closest we can get is the experience someone at the Grand Canyon might have if they read the phrase while there, and experiences are non-transferrable, which is why it’s such a challenge to convey exactly why we loved our recent camping trip to our friends and family.
To take Dr. Sapolsky’s turtle analogy all the way to the bank and cash it out, we can simply say “Dr. James replied, ‘consciousness is not actually turtles all the way down, Dr. Sapolsky. Never was. The bottom turtle is standing on metabolism.’” In a sense, this is the argument that the bottom turtle is the metabolism’s recurring need to reduce entropy, which, in the body, essentially boils down to the need for every cell to receive waste management and nutrition. All of the turtles are standing on the presupposition that a turtle stack is the best way to combat entropy, which disrupts metabolism and is a core fact of the environment in which people’s minds are found.
Perhaps what happens is this: consciousness supports a self-process that reverses the deterministic causal chain Sapolsky argues for, alongside the exact process he describes, yielding a situation in which the neurons fire predictably but a user interface (like a conversation, perhaps about the specific attributes of the Grand Canyon) is needed to truly understand the activity of the system on its own terms.
The modern computer is a perfect example of this sort of system, so everyone reading this should be familiar with the possibility of a piece of hardware running an arbitrary set of computations. If entropy is relevant, it opposes determinism in such a way that cognitive control and other operations are carried out to provide determinism by reducing entropy. All of the biological effort to minify entropy serves metabolic needs over the course of time. All animal consciousness emerges from a cellular-level drive to ensure the endurance of life, ultimately creating a conscious self-process that decides, acts, and moves on, leaving automated subprocesses in charge of stabilizing the entropy-reducing strategies it produces.
Sapolsky sells it with public policy around crime. There’s just no other way to put it. This book has been written for a purpose, and regardless of the leading edge philosophical nuance we can pick out, the purpose is a good one. The book is great because it effectively argues for its purpose with the available evidence. The point of Determined is that free will is bad language, punishment tactics make punished people better or worse insofar as they are good or bad punishment tactics, and we can evaluate social methods of dealing with misbehavior according to the success they have in improving the lives and health of the human beings they involve. Punishment as retribution has no place in modern society, whose purpose is to make everyone - even wrongdoers - better. Determined gets this exactly right.
Still, we need to keep our philosophical grounding to evaluate the role that Determined has in the contemporary philosophical canon. Free will is false. Hard Determinism is false. Indeterminism is generally still true insofar as we know little about the character of general high level virtual processes aside from our phenomenological experience of them, which is unfortunately famous for being unreliable, but generic determinism generally is the rule of the day in lived experience, just as determinism is the outcome achieved by a successful computer programmer - note, in this case, breaking the motherboard of the computer or the decay of a core component thereof will still break the determinism of the programs a frustrated user might try to run. Still, in the vein of criticism, it is important to understand that it is possible the virtual layer of experience shapes deterministic biological processes in individual people, just as a computer processor receives input from its user and runs programs on the basis of the inputs received.
Basic enactivism is a more efficient model of the human mind derived from contemporary science than Hard Determinism, creating a better way of modeling the actions of an individual away from “free will” and toward a part of a sociocultural whole that everyone is part of and that we all should be working to improve.
What we can be conscious of is a measure of our agency, which relates to our intelligence and our ability to care, but the virtual component of lived experience may indefinitely escape arbitrary precision in physiological description. Care relates to our ability to change things outside ourselves but it also determines which changes happen within us; where our attention goes, etc. Care provably relates to intelligence in modern cognitive science. (see: https://www.mdpi.com/1099-4300/24/5/710 )
Suppose that the level of determinism of an environment relative to a person is inversely related to entropy in that organism’s impression of that environment, in the Claude Shannon, information theory sense of the term entropy. Perhaps entropy within the organism, even in its virtual representation of itself and its surroundings, increases determinism in that organism’s personality by reducing the agency of its mind. The body reduces physical entropy for every cell by gathering information from its environment and stacking the deck in favor of metabolic success at the level of the whole, a process driven by consciousness. We could say that indeterminism in our environment is a major source of anxiety and feelings of powerlessness because we use deterministic conditions at multiple levels to sustain our health. Our environment is always changing, and it impacts our ability to sustain our health just as our history is a chain of events we can never be free from. And yet there we are, still alive in the moment, seeking to exert our will to continue to exist - we cannot simply will our way to more willpower, but we can train ourselves to be the way we need to be by investing energy over time. We have the ability to change, particularly when we are patient and focused, and the better we understand our environment (read: the more determinism we can conjure up to make our actions effective), the better we are able to act in the world to reduce our own metabolic entropy.
Sapolsky is right that it is tremendously unfair that it takes so little to damage our brains and forever alter our minds for the worse, but we human beings are resilient and mostly we cope. Also, no one knows for a fact what the right answers are to lots of questions - what to eat, where to live, what to do with our time… the only way for the human species to solve these conundrums is to try a lot of things, like an ant colony choosing paths but with many orders of magnitude more complexity. Determined or not, human ingenuity is found in the continuous unfolding of our success in our struggles against adversity in many more instances than it gets credit for in Sapolsky’s book here.
History has led to this moment, which unfolds as it will around us, but we exist and our influence is real. To the extent that our consciousness is prosocial, our world improves incrementally because we are good to one another. A cognitivist view provides insight into the mysterious self-shaping core processes at the heart of the human body’s neurological circuits, providing far greater resolution when compared to the low resolution descriptions provided by Free Will or Hard Determinism, both of which are modeled on the basis of the old dualist view that mind and body are separate. Is this bundle of neurological processes completely determined? Sapolsky says yes, but as reader we have strong reason to disagree - the highest resolution scientific picture of the mind is still incapable of reducing the complex inner world of consciousness to a simple reductive description.
Once we have more information about the virtual processes that take place in our consciousness, we’ll perhaps be able to revisit the Free Will/Determinism dialectic and make more definitive claims. It’s unlikely that either view will survive the cognitive revolution. Unfortunately, the complete scientific description of consciousness continues to retreat just beyond the horizon, even as our science progresses and our understanding grows. It is good that we have, in cognitive neuroscience, begun the process of observing many many minds under different circumstances - gathering data like this is the best way to further our knowledge.
Perhaps one might say that, although the free will/determinism dialectic must cede its ground to cognitivism as an explanatory framework to describe lived experience and our perception of our own mental action, Sapolsky may also be required to dispatch the free will component of this dialectic on its own terms because we cannot simply wave a magic wand and change the social concepts of individualism directly. After all, what is empathy if not understanding? Is it not likely that the accounts of determined behavior Sapolsky ventures throughout the text of his work are capable of facilitating empathy for the real people he describes?
In the end, the scope of a person’s behavior is the result of external and internal causal drivers, and in this we agree with Sapolsky. However, we don’t have to believe that it’s actually turtles all the way down. The reason for this is that we see the bottom, and the bottom is the metabolism. So each of the turtles is a layer that has evolved for a purpose; i.e., the optimization of the metabolism, such that the top layer is evolved to manipulate all of the lower layers to optimize them for survival in any and all circumstances. We should show sympathy to people who have bad turtles, and help them make their turtles better if we can.
This layer, which Sapolsky does not wrestle with, is the layer at which self-consciousness takes place. Self-image is one part of this, as is the sum of determinants from the environment that interact with us on the physiological level, including Maslow's hierarchy of needs, among other variables that need to be in position for the highest level of consciousness to emerge - i.e., an intact frontal lobe, sleep, exercise, purpose, comfort from a sense of meaning, relationships, and food.
Nonetheless, once established, the virtual image of self that each self-conscious human being possesses becomes its own causal force in that person’s life, consciously and unconsciously. This can be disabled by modifying neural circuitry (a la learning, reading, injury, disease, getting older, etc), but this reviewer would once again like to suggest to Dr. Sapolsky that, with all due respect, when all of the parts are working, they enable a powerful virtual process to “run” on our neurons just as software enables us to write processes that “run” on silicon which we could never deduce from a mere analysis of the parts of the machine. We know this process exists because each of us experiences it.
Our understanding is that the arbitrary code that enables binary transistors to solve problems for us is thus an irreducible property of the modern microprocessor and not a deterministic output that emerges from the hardware in an observably reductive way, just as our experience of consciousness is a strange phenomenon that emerges from our neural hardware. When the hard drive on a computer gets old, it sometimes loses files, or tries to open them and lacks the right program or finds them corrupted - just as hypometabolic, aged human brains often have difficulty remembering. The self-process Antonio Damasio describes can be related to the neurobiological machinery Dr. Sapolsky has described in Determined just as a user of a modern computer involves a virtual interface enabling deterministic read/write operations to disk in a modern microprocessor.
In a robust model of consciousness, we might term this virtual layer as agency and plot it as a spectrum to show the vast range of characteristics it can have. When things go as predicted, people feel happy. We also gain the ability to make better decisions over time with repeated successes, and better decisions lead to more complex metabolic frameworks that eventually lead to new levels of homeostatic stability from which to then support another layer of risky experimentation that can ultimately lead to even more complexity being stably integrated. In general, stable complexity is good because it makes us more flexible and adaptable in our endless pursuit of ever-more-effective metabolic agency.
Thus, Sapolsky’s claim that it’s “not possible for (subatomic weirdness) to percolate all the way up to actual behavior” is both true and utterly irrelevant to the issue at stake here - consciousness isn’t quantum (or magical!), it’s virtual. Consciousness does not need to emerge in random places or orchestrate uncaused causal influence upon individual neurons; rather, a back-and-forth 2-way causation model enables us to see that the virtual layer of consciousness can impact the physical components of the brain system, just as the physical interactions of the cells and particles involved can deeply change the virtual layer’s state.
Yes, bit of code x must be run by the processor on someone’s laptop to display an email, and we can model that and see that the cause and the effect are always related, but even if we do, we don’t understand that Pat wants to open an email on the laptop because their phone just dinged. To understand that, we have to quit looking inside Pat’s laptop and just simply ask them what they’re doing with it. It isn’t that anything undetermined is possible (in fact, that would destroy Pat’s laptop and, as Sapolsky well knows, the program wouldn’t run!), it’s that the causal model of the mind/brain complex is 2-way between physiology and personal virtual reality. Note: both are equally real, emergent dimensions of the same structure’s reality. So perhaps the only valid conclusion we can reach is that there will always be a causal component of the action of the neurological system that will elude attempts to explain it in a mechanistic way.
Human uniqueness is extremely multifaceted and manages to evade the reductionism at the heart of Hard Determinism. The choices we make are heavily influenced by our environment - the mind generates a contextual set of options based upon the input of the senses, after all (a man with no arms will not be an arm wrestling champion!). Nonetheless, the experience of conscious steering through life by making choices is anything but illusory. Choices ranging from workout routines to which friendships ought to be developed provide evidence not only of the integration of preference into the enaction of a person’s cognition, but also of an irreducibly unique kernel of that-person-ness, a pattern we call personality, which cannot be abstracted away from its circumstances.
Should the legal system do a better job of taking into account people’s histories when it acts in their lives so that it does a better job of not indiscriminately making things worse for them all? Absolutely. But is human life determined in advance? The answer from philosophy and from science is that there’s no way we could prove that our consciousness is determined. Too many improbable things happen all the time. Determined was a thought-provoking read that the author did an extremely good job of writing, and it will no doubt be a jumping-off point for many philosophical conversations featuring philosophers and others with a curiosity about the nature of living, but analytic philosophy needs to catch up to cognitive science - and we should scrap the Free Will/Determinism dialectic in favor of a cognitive humanist view that takes enactivism as its primary assumption about the relationship between brain, behavior, and world. Insofar as this impetus to kick the philosophical canon from a lower state to a higher one was the effect Sapolsky’s work had, we could say that it was perfect. Insofar as the minutiae of the philosophical literature are concerned, we have a few suggestions for him.
*Thanks for reading this piece! It is part of a broader investigation into the literature around consciousness and artificial intelligence. Find the index of these works at https://worldviewethics.cent.co/ .*