Davidson’s biggest enemy is himself, or so he thinks. He is an early and influential proponent of a causal theory of action (CTA) that necessitates a primary reason, which has earned him his fair share of criticism over the years from other philosophers; however, the counterargument that seems to scare him the most is the one laid out in his essay “Mental Events,” a counterargument crafted by himself. In this essay, he presents three principles that he believes to be unequivocally true that are also seemingly contradictory, at least prima facie. Principle one is that mental events can cause physical events, which is mandated by his CTA. The second is that “where there is strict causality, there must be a law” (208). The last is that “there are no strict deterministic laws on the basis of which mental events can be predicted and explained” (208). The contradiction that arises out of these principles led Davidson to adopt anomalous monism to bridge the explanatory gap between his CTA and the lawless nature of the mental realm. As a constitutive Russellian panpsychist (a broad physicalist perspective), I am tempted to write a theory of mind paper defending materialism, or nomological monism, against Davidson; however, in keeping with the theme of action and intention, I would like to explore some alternative methods of navigating this apparent paradox that do not force us to adopt anomalous monism. I will argue in this paper why I am unconvinced by Davidison’s arguments for principle three, and why I believe that psychophysical laws exist. Davidson lays out three main arguments for why he believes there are no psychophysical laws: the normativity of the mental argument, the holism of the mental argument, and the linguistic argument. I will use the rest of this paper to tackle each of these arguments, in order.
Davidson notes that the mental realm is subject to different standards than the physical realm: “physical change can be explained by laws that connect it with other changes and conditionals physically described,” whereas “mental phenomena must be responsible to the background of reasons, beliefs, and intentions of the individual” (222). Because of these “disparate commitments” between the mental and physical schemes, there cannot be any strict laws that relate physical commitments like the transitivity of length to mental commitments like the transitivity of preference (222). The issue with this argument is that the qualities like belief, rationality, and intention that are apparently exclusive to the mental realm don’t exist in spite of physical laws, they exist because of them.
All of these normative judgments that exist in the mental realm are due to complex interactions of atoms that have been selected through natural processes in order to aid in our survival and reproduction. The fact that we believe we ought to think rationally and align our intentions with our beliefs and desires is a mere coincidence by virtue of the specific mental realm that we exist in. We could quite easily conceive of a mental realm that is guided by different normative values, yet would still be able to hold certain beliefs, intentions, and desires. For example, if someone receives brain damage that results in them losing the ability to compute arithmetic, they can still intentionally make financial decisions (albeit irresponsible ones), but now the mental commitments that their intention listens to are different. Clearly, the ability to listen to logical reasoning for making financial decisions was encoded physically in the brain, meaning it is a property of the physical world and the mental world. The behavior of the mind is defined by the physical structure of the brain, or in other words, the mind has certain commitments because of the physical structure of the brain.
Consider as an analogy how a neural network like ChatGPT operates. It is clear that there are simple laws that govern how its computation is done on a bit-arithmetic level, much like how there are simple laws that govern how the transitivity of “longer than” operates in the physical realm. Furthermore, ChatGPT is able to craft language that follows syntactic and semantic rules that Davidson would have to claim cannot reduce to those simply computational laws. For example, ChatGPT knows how to be polite, which is a behavior that hardly exists on the bit level. Still, any computer scientist who knows about the structure of transformers would be able to tell you the generalizable and predictive laws that determine why ChatGPT is polite, and how it could be altered to start being rude. In fact, a mere backpropagation of the gradient will be able to tell you which weights do the heavy lifting for each predicted word. If we had a similar level of understanding of the human brain, we would be able to know which neurons and structures do the heavy lifting for each mental state. Davidson would like to think these laws would be heteronomic generalizations at best, but the mind is not random. The fact that we are able to replicate the brain’s normativity into a baby using just our DNA means that there exist reproducible and generic descriptions of our brains that can be used to make new ones. This genetic map of the brain, much like the map of ChatGPT’s architecture, is precisely one of the psychophysical laws that Davidson rejects the existence of.
Davidson also points to the holism of the mind as an argument against psychophysical laws. In essence, the argument is that mental states cannot be made sense of on their own; instead, they must be contextualized within an individual’s vast network of beliefs, desires, and other mental states. Since mental states are inherently tied to this vast web and considering them in a vacuum would be inaccurate, attempting to create a law that explains mental states without considering the entire mental ecosystem would be impossible.
If we are looking for a law with the flavor of something like Archimedes law to explain type-type identities of physical and mental states, then I don’t think we will ever find it. This is because electromagnetic and gravitational forces sum up nicely to construct the abstract forces that we draw coming from an object’s center of mass in a free-body diagram. But that’s only possible through ignoring the internal forces of the object since they cancel each other out to within any noticeable amount. We more often care whether a ship will float or sink than which exact atoms in the wood are under what internal forces. Davidson seems to expect some kind of abstract hand-wavy law like Archimedes to connect the mental and the physical, but a law that is this abstract would only be powerful enough for something simple like determining if someone is alive. In reality, unlike in the boat case where we don’t actually care about the state of individual atoms, the states of these individual atoms are what leads to the expansive richness of the mental.
This doesn’t mean we cannot have laws. I cannot hypothesize about the specifics of these kinds of laws because we simply do not know enough about the brain to craft them, but Davidson is right that they would need to incorporate the holism of the mental in some regard. Davidson thinks the issue is conceptual and could not be solved empirically; however, there are the same conceptual problems with ChatGPT, yet there are still laws that explain the identity between the individual parameters and the entire mechanism, namely the rules that define its architecture. There exists an analogous “holism of ChatGPT” where the existence of any individual layer (like an attention layer, feed-forward layer, or softmax layer) only makes sense in the context of all of the other layers in the network, but there still exist laws that define the behavior of each layer.
Davidson’s third argument against psychophysical laws is based on his other work in truth conditions. He writes that “there can be no psychophysical law in the form of a biconditional (x) (x is true-in-L if and only if x is φ) where φ is replaced by a physical predicate (a predicate of L)” (214). Davidson thinks this biconditional is impossible due to the inability to translate between truth using these two kinds of predicates despite being able to “pick out each mental event using the physical vocabulary alone” (215).
I believe this argument is inextricably linked to the previous arguments since it is simply a rephrasing of the problem into linguistic terms. Finding this physical predicate for mental states is precisely the same problem as finding the law that connects the mental and physical realms. The linguistic take is a useful analogy to use for the problem, but it can hardly be said to prove that psychophysical laws do not exist. If they did exist, finding this predicate would be trivial.
I believe that Davidson’s worries about psychophysical laws are important, but not definitive. His first reason to believe that they do not exist is the inherent normativity of the mind; however, these normative rules are engrained in the physical structure of the brain in such a way that they can emerge from simple physical interactions. If we want to find the laws, we have to understand the structure of the brain. His second reason is the holism of the mind, thus, the inability to have generalizable psychophysical laws. I agree that contextualization is essential for the brain, but there can still be laws that connect physical and mental types as long as they marginalize out this context. His third reason is more of a restatement of the problem in the language of linguistics in order to highlight the inherent impossibility of translating between physical and mental types deterministically. I believe this is a helpful analogy, but does not actually hold any new argumentative power. Therefore, I do not see Davidson’s arguments for principle 3 as convincing, and the dilemma that he presents is fallacious. Consequently, we are not forced to hold a positive of anomalous monism while still maintaining a causal theory of action.
Davidson, Donald. “Essays on Actions and Events”. Clarendon Press, 2013.