While growing up, the question of whether I had free will used to keep me up at night. When that question got me stuck in the nihilist quicksand, the only solace I could find was that maybe it didn’t matter either way. It’s a fundamental part of being human to question ourselves and our nature. Intuitively, at least in most western ideologies that hold individuality as supreme, free will seems to be an essential quality of being human. Without some sense of agency and autonomy, what meaning could my life, actions, and responsibilities possibly hold in the universe? Dennett put it nicely when he said, “when we consider whether free will is an illusion or reality, we are looking into an abyss. What seems to confront us is a plunge into nihilism and despair” (Dennett). If we are all just cogs in god’s great machine, what purpose do I have to be good? What reason do I have to not be bad? We intuitively feel that there is a link between free will and moral responsibility. While some philosophers, like Spinoza, might argue that moral responsibility does not necessitate free will, it is hard to see how someone can be blamed for an evil action or praised for saintliness if they did not do that action on their own accord. Even our own justice system believes this premise and acquits some defendants based on grounds of insanity because some sort of mental disorder may have caused them to act unfreely.
What role, then, if anything, does consciousness play in this debate? In this paper, I will first examine three definitions of free will that commonly appear in philosophy and how consciousness fits into these three definitions. Then I will explain how the debate over free will is merely a verbal one and assert a certain compatibilist definition of free will that necessitates consciousness: a view that I think is most useful and intuitive. I will finally show the consequences of this definition for a few theories of consciousness and what that means for us as humans and the rest of this universe. First, here are three common views of free will from traditional western philosophy.
This first view was the dominant view for most of recorded human history: the cartesian perspective of free will. Under this dualist interpretation, free will requires a kind of soul or immaterial self that compels you to do things. Actions and decisions caused by this spirit would be free actions, but anything that has only physical causes would not be free. Dennett ironically refers to this definition as the “real magic” form of free will due to its implication that there is some superphysical realm that tinkers with our material reality (Dennett). Consciousness is an integral part of this definition because of the inherent tie between phenomenality and the immaterial self. Depending on which dualist you ask, consciousness is equivalent to the immaterial self, part of the immaterial self, or a property of it. So the cartesian view necessitates consciousness for free will. These days, however, this is not a particularly helpful or popular view of free will. All that this definition accomplishes, according to Dennett, is that it hides “the problem behind an impenetrable shield of mysterious stuff” and “postpones the problem indefinitely” (Dennett). If the immaterial self that contains this executive function is not in any way caused by the physical world, then it has nothing to base its decisions on besides maybe randomness or “god”. On the other hand, if the immaterial self is in some bilateral communication with the physical world, this mind merely becomes an abstraction of the real world. In either case, this definition is unhelpful because it merely pushes the problem off to another realm instead of addressing it.
Another popular definition is the libertarian view of free will. This view posits that free will requires the ability to have done otherwise. It’s also sometimes referred to as the principle of alternative possibilities (PAP) because it declares that an action is only free when there exist possible alternative outcomes. This is the intuitive definition that most people give because, prima facie, it makes perfect sense: how could an action possibly be free if that was the only available action? However, something known as Frankfurt cases (named after Harry Frankfurt) shows why this doesn’t always hold up to our intuition. If I unknowingly had a brain chip implanted that would force me to eat a cookie even if I didn’t want to, but I ate one anyways without the chip having to activate, we would probably say that that action was freely chosen. I wanted to eat a cookie and acted upon that desire, even though there was no other alternative. If I didn’t want to eat the cookie, the brain chip would have changed my mind, and I would have eaten it anyway. This view also does not provide a firm stance on the role of consciousness. A robot doesn’t need consciousness in order to be able to have done otherwise. If you believe we do have this version of free will, then presumably your zombie self would also have this version of free will, and if you don’t believe we have this version of free will, then presumably your zombie self would also not have this version of free will. In either case, consciousness appears to be an unrelated property to this definition, although libertarians probably would want to include some kind of consciousness into the theory to avoid the case where either everything (including quarks and stars) or nothing has free will.
This leads us to our final definition of free will: compatibilism. Compatibilism is the view that free will is simply our ability to act upon our beliefs and desires without external coercion. It’s aptly named after its compatibility with determinism since our actions could be predetermined (or even random), and we would still be able to classify some actions as being caused by internal beliefs and desires and others as not. Dennett would argue that we don’t have the cartesian or libertarian version of free will, but that we do have the compatibilist definition of free will, which in his eyes is “all the varieties of free will worth wanting” (Dennett). If given an identical mental state and collection of sensory inputs, I would want myself to make the same decision every time; otherwise, I would be guaranteed to be making decisions I don’t think are best at least some of the time, and I would presumably be basing which decision to make on nothing. The way that consciousness plays into this definition is the way in which we define these beliefs and desires: what does it mean to have a genuine preference, and does this require phenomenality?
I’ll attempt to answer this question in a moment, but first, I think it’s important to note that this debate is mostly a verbal debate. Very few philosophers take the cartesian view seriously and this view is not only unsupported by scientific inquiry but also faces major issues like the causation problem (Chalmers). If we ignore the cartesian view, the only difference between libertarians who don’t think we have free will and compatibilists who do think we have free will is that they disagree on what to call free will. This is because free will is inherently a social construct. Unlike other phenomenal properties and mental properties, free will is not an objective quality that something can have, it is merely an abstraction that we can use to say something about ourselves. It is a story we tell ourselves to make the world go around. In many ways, the debate about free will is not a debate of objective truths, but of what definition we think is most useful and intuitive. The compatibilist definition is most intuitive because it aligns with what we think free will requires, like acting according to personal preferences and beliefs while also not being coerced by external forces. It is also most useful because it fits neatly into our current understanding of physics. How, then, should consciousness play into this theory?
As previously stated, it depends on what it means to have a preference or belief. Does a machine learning algorithm “want” to decrease its loss function? “We talk loosely of a ball wanting to roll downhill, but surely the ball does not have any real preferences. So where do we draw the line” (O’Rourke). O’Rourke speaks of this in the context of moral status: that in order to be worthy of moral respect, you must have genuine preferences. Although this scenario is slightly different, since being a moral agent is insufficient for having free will, the question of having genuine preferences is related. This is hard to elaborate on because O’Rourke leaves his paper on a bit of a cliffhanger on what defines a genuine preference; however, he would like to believe that having genuine preferences does not require consciousness, whereas I think that the beliefs and desires necessary for free will do require consciousness.
Why would I define genuine beliefs and desires this way when I could have just as easily defined free will as not requiring consciousness? The reason that free will should require consciousness is that it aligns most closely with our intuitions about what free will ought to mean. We tend to think that we have agency in a certain decision when we feel internally “compelled” to make that decision, and there is something that it is like to be compelled. When you are zoned out while driving, you are not consciously deciding to turn and follow the road or slow down when you see brake lights in front of you, but you do it regardless. Compare that to the first time you’ve driven down a certain route, and you are consciously aware of all of the turns and stops you are making. In both cases, your brain is making decisions about how to act, and you are doing the same actions (assuming you get to your destination safely); however, it is clear that you do not feel consciously compelled to make those decisions in the unconscious case whereas you do in the conscious case. We are making decisions in either case, so the reason we call the conscious ones free must be because they are conscious.
A counterargument to this could be that the free will we are referring to in this example is actually due to the conscious brain’s role in executive function and flexibility of decision-making in the brain and not the consciousness itself. Maybe the reason we are free in the conscious example is that we are making more complex decisions that require more computing power in the brain than simply repeating a habit. However, I think this is simply a coincidence because the conscious part of our brain just so happens to be the executive control part. Consider you were driving down this route for the first time ever and had to devote a lot of executive control to following the route and not crashing into obstacles. But in the scenario, also imagine that this could all be done unconsciously while daydreaming about what you will have for dinner. When you arrive at your destination, it will feel like the previous zoned-out driving example, even though your brain was making many complicated decisions and reacting flexibly to the environment. Since it still won’t feel like you freely decided to react to turns and obstacles correctly, consciousness must be required for free will. However, since free will is inherently about complex decision-making, this flexibility and executive function are necessary for free will, just not sufficient.
This is currently what I believe is the most helpful and intuitive definition of free will: acting consciously in accordance with your beliefs and desires. So even though a zombie equivalent of me will act identically to me and make all the same decisions, he does not have free will, whereas I have some nonzero amount.
I say vaguely “nonzero amount” because the amount of free will you have depends entirely on the scenario and the degree to which unconscious and external influences affect your decisions. Mudrik enumerates many examples of how unconscious influences undermine our decision-making practices in his paper “Free Will Without Consciousness” (Mudrik). For example, people’s decisions about whether to be an organ donor are significantly influenced by whether it is opt-in or opt-out; judges are “more likely to award parole at the beginning of the day or after a lunch break”; and “the race and gender of defendants, victims, and jurors…influence the given sentence for different crimes to different degrees” (Mudrik). It’s clear that these unconscious states and biases have significant consequences on our decision-making, even our conscious decision-making, so how does this affect the free will debate?
Mudrik agrees that consciousness is necessary for free will, but not sufficient. In other words, “an agent must be conscious while making a free choice,” but we still haven’t answered the question of what they must be “conscious of, when making a decision?” (Mudrik). Mudrik lays out two candidates for what is required of consciousness for free will: control and reasons-responsiveness. Control refers to “being conscious of our motivations, reasons, decisions, and actions,” which gives us a degree of executive control that we would lack otherwise, and reasons-responsiveness means we are reactive to reasons for/against acting in a certain way (which are distinct from explanations of why you did) (Mudrik). It’s unreasonable to think that any decision or action could be fully free because there is so much unconscious processing that occurs before thoughts arise in our conscious mind; however, the more we are aware of these unconscious influences and the reasons for making a decision, the more free the choice is. It could be argued that the conscious knowledge of reasons will always be inconsequential compared to the immense noise of subconscious neural activity, so how could any decision ever be remotely free? However, this is making a crucial mistake of looking at the wrong level of abstraction to draw conclusions. Free will is a psychological concept, so the psychological level is the right level to look for free will. Some unconscious thoughts and biases affect conscious decision-making, but any submental representation is too low-level to have any implications on free will.
So we’ve defined free will as the intersection of phenomenality and executive control, now, we must consider the implications of defining free will in this compatibilist consciousness-requiring way. Let’s look at three different theories of consciousness–global workspace theory, higher-order thought (HOT) theory, and panpsychism–to see how they might respond to this definition. I’ve chosen these three as an overview of the various (broad) physicalist theories which could complement compatibilism.
First, what does global workspace theory have to say about free will? According to this theory, consciousness results from subconscious brain modules “broadcasting” information to the rest of the brain. Since only thoughts in this global workspace are conscious, the only decisions that would be free are decisions made with broadcasted information. This would mean that your actions are free but limited by the constraint of what the various modules choose to broadcast to the global workspace. This nicely parallels our intuitions about free will and the subconscious influences that Mudrik was concerned with. We can’t directly choose which thoughts arise in our brains, and it’s not always clear what unconscious biases or influences cause those thoughts to arise in the first place. Therefore, the intuitions about global workspace theory align with our intuitions about free will in everyday decision-making: we can’t necessarily choose our beliefs, desires, and biases, but we can consciously react to thoughts that arise because of them and make decisions based on them.
What about higher-order thought theory? HOT theory says that conscious mental states are certain mental thoughts that have a higher-order thought about them. Like global workspace theory, the consequences of this theory align well with our natural intuitions about free will. We still can’t explicitly choose which thoughts we have and which become conscious, but once they are conscious, we can actively make decisions based on our current beliefs and desires. Once again, under this theory, we have free will, yet it is limited by the extent to which we can and can’t control which thoughts arise in our minds.
Finally, what would a panpsychist say about this compatibilist definition of free will? There are many different kinds of panpsychism which would all have different views on the matter, but for the sake of this paper, I will focus on Constitutive Russellian Monism, a form of broad physicalism (sometimes) held by Chalmers which could be sympathetic with this version of free will. The “Russellian” part means that there are some microphysical roles (like mass or charge) that are played by microphenomenal properties (Chalmers). In essence, there is something it is like to be a quark. The “Constitutive” part means that these microphenomenal properties sum up (with the help of structural properties) to create macrophenomenal properties like human consciousness. It may seem like this view, which posits that even quarks are conscious, would be a problem for a compatibilist consciousness-requiring definition of free will; however, once again, we must look at the right layer of abstraction to understand how free will fits into this picture of the universe. Yes, quarks may be conscious, but the kind of consciousness we care about for free will is macroconsciousness, not microconsciousness. We don’t know what it would be like to be a quark, but presumably, it would be a much simpler version of consciousness, one that is too simple to account for things like preferences and desires. So, panpsychism seems to fit well with free will so long as we consider Constitutive Russellian Panpsychism which also incorporates structural properties in the explanation for macrophenomenal properties. Russellian forms of panpsychism also uphold an even stronger role for phenomenality in free will than the other theories since it claims that phenomenal properties are the foundation of physical interactions. In other words, experience is fundamentally consequential in the physical world, so it would not even make sense to separate consciousness from free will.
This paper has taken a meandering course, one which parallels the meandering nature of the debate over free will. I began by explaining why compatibilism is a more intuitive and useful definition of free will than the cartesian and libertarian views. I then argued why consciousness should be included in our definition of free will in order to support our intuitions about it. Then I explored some of the implications and limitations of this view of free will, including subconscious biases and what is sufficient and necessary for free will. Finally, I gave what I think three different physicalist theories of consciousness would say about this definition of free will, showing that this definition holds up to our intuitions even when tested by various schools of thought. Am I just creating a definition of free will that happens to fit into my current understanding of the world and eases my fears that my conscious experience bears no weight in this world? Maybe. Would my zombie self read this paper and be angry that I excluded him from the most basic theory of human autonomy? Maybe, although he presumably wouldn’t know that he is unconscious, so maybe he wouldn’t care. Maybe I am unconscious, and you just read a paper trying to convince you I don’t have free will. But the world doesn’t care how we define free will, so we might as well give it a definition that makes sense and makes us happy.
Dennett, Daniel C. "Some Observations on the Psychology of Thinking About Free Will." Incompatibilism and "Epistemic Push": Proceedings of the 2011 Sociedad Filosófica Ibero-Americana Conference, edited by Ignacio Angelelli and Eduardo Rivera-López, Sociedad Filosófica Ibero-Americana, 2012, pp. 29-48.
Mudrik, Liad, et al. "Free Will Without Consciousness?" Philosophical Psychology, vol. 33, no. 6, 2020, pp. 872-895. DOI: 10.1080/09515089.2020.1784723.
O'Rourke, Joshua. "Consciousness, Coherent Causation, and the Moral Status of Zombies." March 2023.
Reed, Matthew. "Who is Will, and is he Free?" December 2022.
Rosenthal, David. "A Theory of Consciousness." Journal of Philosophy, vol. 83, no. 7, 1986, pp. 403-429.