Finding the Freedom in Free Will
New theoretical work suggests that human agency and physics are compatible.
Posted February 10, 2022 | Reviewed by Vanessa Lancaster
- Some physicists believe there is no place for free will in a scientific worldview.
- The probabilistic nature of reality and top-down causation explains how living systems can affect physical reality through intentional action.
Whether or not we have free will is arguably the most important philosophical and practical question of all time. As reported in this New York Times article, studies have shown that convincing people they do not have free will can lead to immoral behavior. In one well-known experiment, participants presented with a passage by Francis Crick—explaining how a deterministic view of reality appears to leave no room for free choice or agency—were more likely to cheat on subsequent tests because they no longer felt personally responsible for their actions.
Other studies have shown that people who believe that free will is an illusion are more likely to experience depression and anxiety. That means that if we do have the freedom to choose our future path out of the realm of the possible, then scientists and philosophers have a responsibility to broadcast this information to the public for matters related to mental health and morality.
For this reason, it is troubling that some famous (and brilliant) physicists, particularly those clearly in the reductionist camp, have gone out of their way to ensure that the public believes there is no place for free will in a scientific worldview.
For example, in a blog post from 2016 titled Free Will is Dead, Let’s Bury It, theoretical physicist and popular YouTuber Sabine Hossenfelder wrote, “I wish people would stop insisting they have free will. It’s terribly annoying. Insisting that free will exists is bad science, like insisting that horoscopes tell you something about the future—it’s not compatible with our knowledge about nature.”
Another prominent free will-denier is Brian Greene, as discussed in this article by science journalist Philip Ball, who tried to bring some clarity to the great causation debate by explaining that fundamental physics does not have much, if anything, to say about agent causation. As expected, the article was misunderstood or intentionally straw-manned by “philosophically naïve anti-free will enthusiasts like Sam Harris and Jerry Coyne,” to use the words of philosopher and evolutionary theorist Massimo Pigliucci. In this recent blog post, Pigliucci argued that when we deliberate over a problem and then make an informed decision, that decision is entirely real. However, he is adamant about not calling the causal power associated with conscious thought “free will,” though we will see that this is splitting hairs.
Other famous scientists—physicists, neuroscientists, and complexity scientists among them—have a position virtually opposite of Greene’s and Hossenfelder’s. In this recent interview, quantum computation pioneer David Deutsch told science journalist John Horgan he believes humans have free will. The philosopher Dan Dennett also believes free will is real, as revealed in new detail in this 2021 interview. The neuroscientist Giulio Tononi—creator of Integrated Information Theory—has described a mathematically-precise mechanism for free will in this recent lecture, in agreement with the view of his colleague Christof Koch, as expressed in a 2018 interview with Lex Friedman. Do anti-agency reductionists believe that the theories of these scientists are not worth serious consideration?
Origins-of-life researcher Sara Walker, who is also a physicist, explains why mainstream physicists in the reductionist camp often take what most people would consider a nonsensical position: “It is often argued the idea of information with causal power is in conflict with our current understanding of physical reality, which is described in terms of fixed laws of motions and initial conditions.”
Where did this assumption come from?
The 18th-century polymath Pierre-Simon Laplace pictured the Newtonian cosmos as a giant deterministic machine. According to this picture, every event that occurs is a strict consequence of the preceding event, which is itself a consequence of a prior event, and so on, leaving no room in the causal picture for human decision making. The unfolding of reality is merely particles of matter obeying fixed laws of motion, such that their trajectories through space and time are entirely decided at the moment the universe came into existence.
According to Laplace’s model, all causation in the universe is “bottom-up,” meaning the particles that all objects and organisms are made of at the bottom are the physical entities that do all the pushing around. Despite the feeling of agency, you don’t lift your arm—a chain of physical events at the level of atoms is what underlies the arm’s movement.
According to classical determinism, it is not even genes or brains that are doing the causal work because if everything is determined by microphysics, organisms aren’t objectively real either. Everything that looks to us as if it is moving with purpose is an illusion.
Fortunately, advances in science and philosophy have all but killed determinism of the Laplacian variety, known as hard determinism. When quantum mechanics emerged about a century ago, the experimental evidence suggested that reality is fundamentally probabilistic, reviving the idea of indeterminism, which says the future is not determined in the strict sense once imagined. If reality is probabilistic, then it would mean that nature is inherently noisy, unlike the clockwork universe popularized by Laplace.
But quantum indeterminism alone does not help the notion of free will much since a reality with some randomness is not the same as one where an agent has control over that randomness. While indeterminacy does not provide a mechanism for free will, it provides the “wiggle room” not afforded by Laplace’s model of reality; the necessary “causal slack” in the chain of cause-and-effect that could allow room for agency, as physicist George Ellis describes it. The combination of the probabilistic nature of reality and the mechanism known as top-down causation explains how living systems can affect physical reality through intentional action.
Top-down causation used to be a vague concept of philosophy, bordering on the mystical, but now it is beginning to be described in the language of mathematics and quantified by measures derived from Claude Shannon’s Information Theory. Top-down causation is the opposite of bottom-up causation, and it refers to the ability of a high-level controller (like an organism or a brain) to influence the trajectories of its low-level components (the molecules that make up the organism). This means that biological agents are not slaves to the laws of physics and fundamental forces the way non-living systems are. As adaptive or cybernetic systems with control and causal power, we create our futures in accordance with physical law but not subservient to it.
With the transition to life, macroscopic physical systems became freed from the fixed trajectories that we see with the movement of inanimate systems, which are predictable from simple laws of motion. Animate systems can climb uphill, going against the force of gravity, because the system is internally powered by the process of metabolism, and infused with agency, which is a result of information processing and integration, as well as controlled feedback loops.
The emergence of top-down causation is an example of what philosophers call a “strong emergence” because there is a fundamental change in the kind of causality we see in the world. When life arose in the universe, wherever it first occurred, it was the moment when information gained causal power. According to physicists Paul Davies and Sara Walker, agency emerges when the information encoded in the patterned interactions of molecules begins telling the molecules what to do. At this point, the chemical system, which can now be called a biological system, can begin to push matter around.
They refer to this ability as informational control, and it is this property that can explain the unique behavior of life, such as its purposeful, goal-oriented movement. Informational control, which emerges with an event they describe as algorithmic takeover, gives life the “ability to harness chemical reactions to enact a preprogrammed agenda, rather than being a slave to those reactions.” Once algorithmic takeover occurs, the whole is truly more than the sum of its parts, because the adaptive system can now control the physical behavior of the matter in and around it.
The evolutionary mechanisms that lead to such an event are beyond the scope of this article, though they are explained in this 2019 Quartz article I wrote about Jeremy England’s theory of dissipative adaptation, along with Eric Smith’s phase transition theory of life, which together provide a series of mechanisms that do a pretty good job explaining how life got started.
Once informational control is enabled, causation in nature is no longer strictly bottom-up. The mind is not a “nonphysical substance” like Descartes imagined, but a monitoring-and-control unit that allows the agent to stay out of thermodynamic equilibrium, a state of death and disorder. Top-down causation means control. The bottom-up flow of causation is never disrupted—it is just harnessed and directed toward a goal. We still inhabit a cause-and-effect cosmos, but now the picture is more nuanced, with high-level causes being every bit as real as the low-level ones.
Erik Hoel, a neuroscientist trained under Tononi, claims that a higher-level analysis of a system’s dynamical evolution—its lawful transition from one state to another—may reveal that the causal work is predominantly being done by the agent rather than the molecules that make up that agent. Tononi, Hoel, and another former doctoral student, Larissa Albantankis, have taken up the task of mathematizing the emergence of causal power, or causal emergence, using formalisms adapted from Judea Pearl’s causal calculus.
Hoel explains the primary claim of his theory of causal emergence in his award-winning essay, Agent Above, Atom Below: How Agents Causally Emerge from Their Underlying Microphysics:
"Recent research has argued exactly this by demonstrating the possibility of causal emergence: when a macroscale contains more information and does more causal work than its underlying microscale.”
According to Hoel, the causal power is located not in the microphysics but at the macro-level because the information is encoded in a pattern of organization, which is multiply realizable. Because an informational state can correspond to many different specific molecular configurations, many microstates can realize an equivalent pattern or macrostate. We can imagine the microphysics of a system as noisy and chaotic, with the macrostate playing the role of noise reduction and error correction.
Kevin Mitchell, a neurogeneticist who is currently writing a book about agency and free will, explained that when it comes to causation, the macrostate is what matters:
The macroscopic state as a whole does depend on some particular microstate, of course, but there may be a set of such microstates that corresponds to the same macrostate…Suppose the system's evolution depends on those coarse-grained macrostates (rather than on the precise details at the lower level). In that case, this raises something truly interesting—the idea that information can have causal power in a hierarchical system, and, more generally, in the universe.
Let us now ask, "How is it possible for mind to push around matter?" but let’s replace the word “mind” with “information” and see if the mind-body paradox remains. How does information push particles around? That is a question we are beginning to answer with mathematical and mechanistic precision.
We have explained agent causation as information in action. Still, it doesn’t seem to get us quite to “free will” because there’s not much autonomy involved in automatic or reflexive behavior. The bacterium performing chemotaxis will predictably swim toward chemical food and away from toxins—and it will do so without any hesitation, so there are no behavioral indicators of reflective thought or deliberation.
Similarly, if you’re only able to do what is programmed by the older and more primitive networks in the brain, that would not seem very free because you’d be like the bacterium obligatorily performing chemotaxis. In next week’s post, Free Will is a Higher Level of Control, we will see that an agent becomes “free” when they can override preprogrammed behavior with cognitive control.
This article was adapted from chapter 11 in my forthcoming book The Romance of Reality: How the Universe Organizes Itself to Create Life, Consciousness, and Cosmic Complexity, which is available for pre-order.