This is my research paper written for my Consciousness course. For a PDF version instead, click here. Enjoy!
In a monist, materialist world, how is it that the mind can act on the body? Virtually no one today believes that the mind is of a separate substance from the body, immaterial and ethereal; most believe instead that it is a product of the physical brain. If this is the case—and it certainly appears it is—must we contend that the conscious mind is epiphenomenal? The standard physics-inspired argument goes something like this: since the mind arises from the brain, and since causation acts from foundations to outward effects, the mind must not be able to change its material foundations.
Our own intuitions about ourselves seem to oppose this chain-of-causation, epiphenomenalist picture of the mind. After all, it certainly feels like we have beliefs, feelings, goals, and desires, which influence the decisions made in our minds, and it feels like these decisions are in turn played out in our bodies. If we can be justified in this belief, though, we must give a philosophical-scientific account of how mental causation might occur in a physical brain. Such an account—and nothing short of such an account—would reconcile the monist, physicalist world view with our intuitive beliefs about the mind.
In answer to this challenge, some philosophers conceive of the mind as a strongly emergent system. Things that exhibit strong emergence have properties which are systematically determined by low-level facts, but which are not deducible from those facts alone (Chalmers 2006). The primary characteristic of strong emergence is downward causation, where a system as a whole affects the parts that compose it (for a comprehensive account of downward causation, see Bar-Yam 2004, Murphy 2006). A strongly emergent system, by definition, has high-level properties belonging only to the “whole,” which can act causally on the system’s constituent parts. Thus, if the mind is such a system, we have a non-mystical, empirical reason to believe it could act causally on the brain; if it is a strongly emergent system, the mind is a supervenient property acting on its material underpinnings.
Nancey Murphy has proposed a model of the mind in which it affects downward causation on the brain. This model is built on the assumption that subjects of an individual’s experience (objects in the world, actions performed on objects, etc.) correspond to neurons or groups of neurons in their brain. Based on what we know of the brain, this is quite reasonable; even if such knowledge is broadly distributed across the brain, as a large body of evidence indicates, the model is unaffected. The model assumes, too, that sensory input creates connections between neuron circuits because of some recognized relationship between the outer-world correlates of those neuron circuits—again, a very reasonable assumption in light of what we know of neurobiology. As a consequence, when this connection-building occurs, the brain as a whole acquires a new supervenient property—it holds a representation of the relationship between those outer-world correlates (Murphy 2006).
It is worth noting that in this sense, the expression of these relationships is supervenient on both neural connections within a brain and circumstances external to the body. A by-product of this notion of supervenience is that changes in a subject’s external environment may render a relationship-connection true or false apart from any physical change in the brain. This characteristic of supervenience relationships becomes important when discussing the changing of a belief (for instance, if a person’s belief becomes falsified apart from their knowledge), as we will later see.
Murphy’s model of the mind leads to an interesting formulation of the mind-brain interaction. The representational relationships that the brain possesses in its neural connections may rightly be called beliefs—beliefs about the connection in the outer world between the objects represented in the brain. For instance, when a person notices, even unconsciously, that event A is always followed by event B, his brain can reasonably be assumed to connect the neurons corresponding to A to those corresponding to B (note that we will refer to these as Anc and Bnc, respectively). This connection has the representative relational characteristic, supervenient on both the physical connection between Anc and Bnc and the relationship between A and B in the outside world, of a belief “when A, then B.” This connection may be illustrated like this:
A key feature of this model is that it allows for neural connections to be changed by acting at a representational, conceptual, belief level (where a belief is a type of concept, and a concept is a type of representation). The beliefs are the “handles” by which the environment and subjects themselves can act to modify the neural connections. For instance, if the environment is to modify the connection linking Anc to Bnc, it must do so by changing the relationship between A and B; when the subject notices (again, even unconsciously) that A is no longer followed by B, he will no longer believe “when A, then B,” and the neural connection will change.
To see why this is, recall that the belief is supervenient on the neural connection between Anc and Bnc and the outer-world relationship between the two. Certainly subjects have no awareness of their own neuronal configurations (though they may be educated enough regarding their own neurobiology to know that they exist). Certainly the environment, too, cannot directly access the brain, hidden as it is behind a protective wall of flesh and bone. However, by modifying the relationship “out there” between A and B in a way that the subject may observe and understand, the subject’s own mental processes will modify the connection between Anc and Bnc—something that is impossible by any means other than the “handle” of belief, as Murphy refers to it (Murphy 2006).
This belief-as-handle is an instance of downward causation because the conceptual relationship between A and B resulted in the creation or strengthening of the connection between Anc and Bnc; the concept in the mind, which in our model is supervenient on brain states plus the environment, altered the brain state that gave rise to it. Prior to the creation of the belief “when A, then B,” both groups of neurons “would have been multiply connected to regions throughout the brain” (Murphy 2006, 234); the fact that such a connection exists may carry only minor informational significance. However, as Murphy says, the pairing of A with B in the experience of the subject “resulted in the selection of this particular connection for reinforcement” (Murphy 2006, 234). Thus, the connection has representational significance in light of the experience which resulted in the belief.
At this point, we have a working model of downward causation in the brain: it is the choice by a higher organizational level (what we refer to as the mind) between already extant possibilities in the lower levels (the brain, and, in particular, neural groups). In this way, the causal closure of the lower levels has not been violated—no “magic” has entered into our explanation. At the same time, the representational mind has exerted influence on the physical brain.
Francisco Varela conceives of the relationship just described as interplay between two kinds of causation in the brain. There is upward causation (and weak emergence) resulting from many disparate brain systems (auditory, visual, linguistic, social, etc.) coming together into a complex whole. At the same time, there is downward causation (and thus strong emergence) resulting from global patterns, what Yaneer Bar-Yam calls a “global constraint,” acting on local neural activity (Varela 2002; Bar-Yam 2004). Here, Varela states that the global pattern, which in a strongly emergent system is the “causer” of downward causation, is observed empirically to be the massive synchronization of neural groups which electrically oscillate in the gamma range, commonly referred to as “40 Hz” oscillations (Varela 2002). These oscillations are readily observed when a subject concentrates on a difficult problem, suggesting an important role for intentionality in the creation of a global pattern. In broader terms, we may describe the global patterns as the neural correlates of a particular belief (for our purposes, we are considering only beliefs and not things like reasoning processes).
We have briefly considered how a belief might act as a strongly emergent “causer”; let us now consider the importance of a given belief in discussing the causal forces acting on a subject. Can a belief really be said to have causal efficacy, or can we describe the causal forces acting on a subject without that belief? We return to the simple example of the belief “when A then B.” In this case, when A occurs, the belief causes the subject to expect B; an outside observer would see the subject prepare for B as necessary. For instance, if A is the sound of a blaring car horn, we could observe a driver become tense in preparation for a collision, B, whether or not B actually occurs. A driver would have a belief that when car horns honk steadily, there is imminent danger. However, if the driver hears enough obnoxiously blaring car horns which are not accompanied by danger—that is, if the relationship that exists “out there” between car horns and collisions changes—the person’s belief is empirically falsified, and the neural connection on which the belief is supervenient will be changed to reflect this. At this point, the subject’s reaction to A will no longer reflect a belief that B will follow it.
This is straightforward conditioning and habituation, the behaviorist will say. Why should we think a mental evaluation of this person’s belief is causally active here? We must acknowledge a causal role for belief because the subject’s knowledge of the situation’s elements is supervenient on the neural configurations Anc and Bnc, and the truth of the proposition represented in the connection between them is supervenient on both the neural connection and the outer world. Likewise, evaluation of belief is the sort of mental function that an animal must have if it is to be successful in the world, and there can be no evaluation of belief without an output of the decision. If a belief is in fact “expressed” in the brain as described here, a subject has no way of relating A to B apart from beliefs connecting the two. In this case, it does make sense to call the horn-hearer’s response a conditioned one, but such a conditioned response cannot exist apart from the belief.
With this model of the mind established, let us consider explicitly its significance in an account of free will. For our purposes, we will define free will as a subject’s ability to make decisions regarding his or her actions. This, then, stands in contrast to a determinist concept of the will, which says that a subject’s decisions are fixed by fate in the form of the atoms and molecules making up their body. It is worth noting that a key feature of this definition is that a subject is actually deciding. It is not enough to say that, due to chance at a quantum level, a subject’s behavior cannot be perfectly predicted by their physical state. This may be a denial of determinism, but it is not free will by our definition.
As we have seen, when it comes to modifying the brain (and thus any behavior stemming from it), both subjects themselves and the environment have access only to beliefs—not the neurons and neuronal connections themselves. Thus, we can imagine a hierarchy of beliefs in the mind, where some beliefs carry greater influence than others. Beliefs with the most weight are things like “the universe exists” and “the universe is logically coherent,” while beliefs with less weight are things like “it will rain this afternoon” or “the football team will do well this season.” We might then imagine a belief-plex wherein new knowledge is compared against old beliefs to check for necessary revisions. If the new knowledge stands in contradiction to heavily weighted beliefs, the knowledge will be dismissed or rewritten. If, on the other hand, it stands in contradiction to a less weighty belief, it will modify the extant belief or replace it altogether. In all cases where new knowledge contradicts old beliefs, it will modify the least weighty belief possible in order to fit in; sometimes, that least weighty belief will be itself. While this is probably not an exact model of the way belief works in real brains, it is certainly plausible. At worst, it is a useful metaphor.
A hierarchy of this sort, and a corresponding brain function to evaluate beliefs, is sufficient for mental downward causation. To see this in action, let us return to the case of the driver who at first believes that honking car horns A are followed by car crashes B. This belief we will call P. Her new experience indicates that, thanks to the strange psychological experiment she’s participating in, A and B are no longer linked this way. This new belief, which we will call Q, stands in contradiction to P, so P and Q must be evaluated to determine which will be changed. When Q is judged to be of higher significance than P, the physical neural connection will be altered accordingly. At this point, the subject’s belief about A and B is R, which might be either Q or some combination of P and Q, and R will guide the subject’s future actions—at least until R itself comes under fire. In this case, if brain states determine action at a given instant, it appears that the combination of the belief and the evaluative process correspond roughly to the traditional “will.”
The decision-making involved in evaluating P and Q in the previous example is relatively straightforward, requiring little real weighing of options. What about cases of intense deliberation, when it is not only a small group of a person’s beliefs that must be accounted for by the evaluative process, but a large number? In this case, Daniel Dennett’s characterization of the process, adapted from the libertarian Robert Kane’s, is helpful. Dennett imagines parallel evaluative streams, each tallying points in favor of a particular belief that needs considering. Whichever belief has the most points in its favor at the end of deliberation will later to be acted upon. At any point, a subject might probe these streams and find that one belief has 30 points in its favor, one has 50, and one has 55 (the brain certainly has no such discrete quality, but the metaphor is still useful). If the subject stops deliberating now, the final belief is the one that will “win.” However, a subject who is appropriately educated regarding their own brain functions will also be aware that further deliberation could change the answer. He could recall some crucial implication of that first, previously undervalued belief which, when realized, rockets it to the top of his consideration. This process, though deterministic, does qualify as a free choice—it’s entirely up to the subject where to halt his consideration. All of this is a function of the evaluative process, which, having made its decision, acts on the brain the same manner as the experiential belief of the driver hearing the car horn—by the selection of neural connections for strengthening.
Thus, even if it is true that in a given situation, a subject’s actions may be accurately predicted by considering only the neuron-level features (a proposition which may one day be verified by neuroscience), we may still be said to have free will. In this strongly emergent model of mind, downward causation acting from beliefs to neurons can modify the brain, shaping future action. If, in one instance, the mind can change the brain, then it can change action. This action can in turn shape future beliefs, which can themselves modify the brain, ad infinitum. While there may be no homunculus in the head directing a subject’s every action, there may yet be a structure of belief, monolithic in nature, molding an individual’s disposition to action. This is an unconventional picture of free will in the brain, but it is certainly free will; it may remove the free choice one step from the actual action, but the fact remains that the person’s decisions regarding belief determine their actions. This is perhaps as close as we can hope for to a scientific equivalent of the traditional dualist picture of a soul qua self “shooting arrows of decision into your motor cortex” (Dennett 2003, 223).
Much of what we know of consciousness and the relationship between the mind and brain casts doubt on our intuitions about free will. However, if the mind is a strongly emergent system, wherein beliefs serve as global constraints supervenient on outer-world objects and their internal neural correlates, then it may in fact be causally efficacious; though the belief properties of the mind are contingent in part on brain states, they may still act causally on that brain by the methods of downward causation examined here. This position may be categorized as a compatibilist view on free will; even if determinism is true (or approximately true, per quantum mechanics), in the sense of the “lower levels” of a creature’s organization being causally closed, we can still have free will. At this point, it seems science has come full circle: thinkers dating back to the ancient Greeks have noted the mechanical nature of the universe and the seeming inescapability of the snowballing consequences of prior forces, casting doubt on our own agency; now, through the science of emergence, we are finding that we can still be agents.
Bar-Yam, Yaneer. “A Mathematical Theory of Strong Emergence Using Multiscale Variety.” Cambridge, MA: New England Complex Systems Institute, 2004.
Chalmers, David. “Strong and Weak Emergence.” In The Re-Emergence of Emergence, edited by Paul Davies and Phillip Clayton, 244-254. Oxford: Oxford University Press, 2006.
Dennett, Daniel. Freedom Evolves. New York: Viking, 2003.
Murphy, Nancey. “Emergence and Mental Causation.” In The Re-Emergence of Emergence, edited by Paul Davies and Phillip Clayton, 227-243. Oxford: Oxford University Press, 2006.
Varela, Francisco J. “Upwards and downwards causation in the brain: Case studies on the emergence and efficacy of consciousness.” In No Matter, Never Mind: Proceedings of Toward a Science of Consciousness: Fundamental Approaches (Tokyo ’99), edited by Kunio Yasue, Mari Jibu, Tarcisio Della Senta, 95-107. Philadelphia: John Benjamins Publishing Company, 2002.