I've tackled this question on occasion before, but it keeps coming up -- largely due to AI models that create human-like outputs and can effectively pass the Turing Test. Despite their complexity, modern large-language models (LLMs) are wholly deterministic and can therefore be understood in terms of mechanistic cause and effect: Put in this exact input, put it through these steps, and you get that specific output, 100% of the time. For the last couple of years, people have been talking about "generative AI" as though it was non-deterministic, but researchers at Cornell University in 2025 proved that this was false. You can in fact work backwards from an AI output to the prompt that was used to generate it. The idea at one time was that generative AI was "creative" because it assigned probability-based weights to various outputs and selected the most likely one. But the new research shows that if you tightly control inputs, you do always get the same output, and that if you know the output and the steps that modified the model over time, you can in fact work backward to the input. As a side note, there also may be an upper limit on AIs' ability to learn and change over time -- after that, the probability weights stabilize and they just begin to confabulate outputs instead of continuing to improve.
The important point about recent developments in AI is just that they might lead us to question our own input-output system, the brain. If AI is completely deterministic, are we deterministic as well? Or, in other words, do we have free will? The AI example is only one of several challenges to free will that have emerged from neuroscience research, and it's probably not the strongest. The easy rejoinder to the AI challenge is that the way AI does language seems to be quite different from the way we do language. There is no reason to believe, for example, that AI models are conscious. So let's at least for now set aside the AI argument that raised these questions, and look at other compelling challenges to free will:
1. The argument from material determinism. To me this one seems like question-begging, but the basic argument is that everything we can observe in the world around us seems to have a physical cause, therefore mental activity also must have a physical cause. One might counter that mental activity is the one great example where we haven't been able to identify a physical cause! But when we look at things like the motion of the planets, the interactions of chemicals, or the activation of muscles by electrical stimuli, we do tend to see physical causes for things. A contrary position, idealism, suggests that mental activity is primary and actually causes physical phenomena, rather than vice-versa. This debate is probably unsolvable because both explanations seem compatible with physics as we know them. In fact, quantum physics suggests a strong effect of observation on physical phenomena, which requires either a causal arrow from mind to matter, or else a complex many-worlds hypothesis to explain it away. Biologist Kevin Mitchell argues that people evolved a mental system to generate alternative scenarios and evaluate potential actions -- what I call the Intuitive Mind -- which controls their behavior, and that this type of control meets every type of criterion we might impose for "free will" (e.g., choice, ability to take alternative actions in similar circumstances, internalized control of behavior). In addition, we have a Narrative Mind layered on top of that which can perform the same type of evaluation in a slower, more conscious, and language-based manner, if we put our mind to it. Even if you are a committed materialist, you have to admit that brains running simulations and selecting alternatives based on their expected results seems something like the traditional definition of free will.
2. The effect of unconscious processes on behavior. Another strong argument for many people is the fact that psychology has identified unconscious influences on behavior. In Two Minds Theory those are grouped together under the label of "Intuitive Mind," and include processes like social perceptions, learned habits, and heuristic responses. Many people find that when studies show they aren't consciously aware of all the factors that affect their behavior (especially when advertisers capitalize on those cognitive blind spots!), that undermines their faith in free will. Variations of this argument point to the genetic heritability of personality traits as evidence that my behavior is not really controlled by "me," as though my genes were not a part of me. My counter to this argument is that it relies on a philosophical dualism that has shaped Western culture, and that is probably false. I will be the first to say that my Narrative Mind does not control my behavior. But having my Intuitive Mind in the driver's seat doesn't undermine my faith in free will, because my Intuitive Mind is still me.
3. The Libet studies on timing of decisions. I have written about Benjamin Libet's intriguing studies that show activation in the brain's motor cortex, suggesting a nascent movement, before any corresponding activation in the prefrontal lobes, which are supposed to the be brain's executive decision-making area. In these studies, a "decision" to move one's hand seems to originate in the less-conscious areas of the brain, and only then to trigger a conscious decision to act. Libet and others have argued that the prefrontal area can still exercise a sort of veto power over the action before it comes to fruition, which may be a form of free will -- but it isn't the kind we imagine, where the conscious mind makes a deliberate choice and directs its execution by other parts of the brain. I agree that conscious control over behavior is illusionary, while nevertheless still arguing for free will. Again, the fact that my Narrative Mind doesn't direct the action (I have characterized it as "more like a sports commentator than a business executive") doesn't negate the fact that my actions still originate from within my own brain. Much of the argument over free will goes away if we stop demanding conscious control of behavior. Of course, I don't think that my conscious self is completely without influence: I have argued for the importance of a convincing narrative, or the timely exercise of attention, as two ways in which the Narrative Mind can affect the Intuitive Mind's effortless and ongoing control of behavior.
4. The argument from damaged or impaired brains. People with brain damage often experience changes in their behavior, with examples ranging from Phineas Gage to me. Alternatively, people sometimes point to their own experiences feeling out-of-control due to substance use or strong emotion. But the fact that people's free will seems "impaired" in these examples is the exception that proves the rule: If we are sometimes not in control of our behavior, doesn't that imply that in other circumstances we would have been? I'm not denying that people can experience a feeling of more or less control at different times and under different circumstances. But the out-of-control experience is the exception that proves the general rule.
5. The argument from social constraints. One more argument against free will says that we simply act in the way that other people expect us to act. Social perception is indeed a very strong component of the Intuitive Mind, because human society started in small groups where one's survival depended on maintaining good relationships with others. This is a form of the argument about unconscious processes, because social pressure only works to the extent that I perceive it via my Intuitive Mind. In fact, some people are more sensitive to that type of pressure than others. Another insight from Mitchell is that people are interested in having different social roles only within a limited range connected to their own human context. He writes (p. 228) that people often say "I wish I were more confident like Gary," but they rarely say "I wish I were more chilled-out like a walrus"!
Mitchell's answer to the question of free will refers us back to Aristotle: the structure of the brain is a material cause, and external forces that act on the brain are efficient causes, in Aristotle's 4-level typology of causation. Together these two types of causation can add up to a formal cause that explains just what is happening in the brain at a particular moment, and that brain state is presumed to produce behavior. But Aristotle identified one more type of causation, final cause, which addresses the goal or purpose of an action -- also called a teleological argument, an explanation by referring to the ends which some action is designed to accomplish. Scientific thought generally doesn't allow for teleology, because it seems like a kind of backward causation: "the purpose of a car is to drive places" or "the purpose of the eye is to see." Evolutionary biology, in particular, sought to eliminate final-cause language from its arguments, suggesting that if an eye happens to see that's a useful adaptation which may lead more creatures with eyes to survive and reproduce, but the purpose of an eye isn't anything -- it's just there. Mitchell points out that this logic breaks down once we talk about mental activity, because thinking does include purposes -- I go to the coffee shop because I want coffee, or for the purpose of getting coffee. The goal I'm trying to achieve is the cause of my behavior.
Once we have a world-simulation system, a hypothesis-generator-and-tester, online in the brain, the actions that we carry out can readily be explained with reference to those simulations and hypotheses. As Karl Popper said, we test out ideas so that "our hypotheses can die in our stead." Mitchell argues that "agents can act with causal power in the world because, biologically speaking, they have been paying attention" (p. 168). Goal-directedness of this type is what we generally mean when we talk about free will, not total freedom from constraints. A totally unconstrained system would just produce random behavior, which isn't what we want. And teleological models of behavior give us that type of goal-directedness, which doesn't have to originate from the Narrative Mind in order to count as free will.
Comments
Post a Comment