Are There Psychophysical Correlations?
The orthodox view is that mental states or attributes are correlated with physical states or attributes. For every mental state M, there is a physical state P such that M is correlated with P (not so in the opposite direction). That is, every mental distinction has a corresponding physical distinction, down to the last detail. By “physical” we are to mean “couched in the vocabulary of brain science”, so neurons, dendrites, action potentials, neurotransmitters—cell anatomy, brain chemistry, electric charge, blood flow, metabolism, connectivity. The classic example is “pain is correlated with C-fiber firing”. This thesis is taken to be both true a priori and also established by empirical investigation—both sound metaphysics and solid science. It is the first dogma of neuroscience. Is it true? What explains it, if true? Two interpretations immediately come to mind, which I will call correlation by identity and correlation by causation. The former models itself on so-called theoretical identifications such as “heat is molecular motion” and “water is H2O” (apologies for the cliched examples). Pain is correlated with C-fiber firing in just the way heat is correlated with molecular motion—by identity. Everything is self-correlated, so we can speak of correlation as identity under different modes of presentation: it is in virtue of identity that water is correlated with H2O. In this sense, Hesperus is correlated with Phosphorous. Such statements are sometimes referred to as bridge laws. This is an awkward way to speak, to be sure, but anyway it isn’t the best way to understand psychophysical correlation, as it is commonly understood. For it is highly doubtful that such identities obtain in the psychophysical case; at any rate, we don’t want correlation to depend on the truth of an identity thesis. Not all correlation is by identity, as a matter of definition or necessity. So, we might move to correlation by causation: brain states cause the mental states with which they are correlated. The correlation is a matter of causal law. There are three possible types of causal relation here: from physical to mental, from mental to physical, and by common cause. We can dismiss the last alternative because there is nothing that could be the common cause—mental and physical exhaust the field. The most favored candidate is physical to mental: brain states cause the corresponding mental state. There are three problems with this idea: the problem of simultaneity, the problem of intelligibility, and the problem of asymmetry. Such causation would make the effect simultaneous with the cause, which is a no-no. Such causation would be unintelligible, which is a disadvantage. Such causation would be asymmetrical, which is arbitrary (why not say the mental state causes its physical correlate?). I won’t labor these points, as the causal model of correlation is not generally accepted, and for good reason (no one wants to say that C-fiber firing is the efficient cause of pain). I mention it to get it out of the way. What people are inclined to believe—and it is a prima facie attractive viewpoint—is that the putative correlation is brute—it is not by anything. It just so happens that pain and C-fiber firing are correlated, but not in virtue of anything. There is no explanation of it, but then again, all explanation comes to an end in primitive facts—basic laws of nature, or brute psychophysical laws in this case. The correlation is projectible in that it extends beyond the observed and even actual cases; it supports counterfactuals, in the lingo. It has the force of necessity, nomological if not metaphysical (that seems a bit strong). Pain is everywhere and always correlated with C-fiber firing, as a matter of natural law. The problem with this is that it is unprecedented: in all other cases of correlation, we have an explanation of the correlation—sometimes by identity, sometimes by causation (over time), sometimes by common cause. Thus, we observe that health is correlated with longevity, height with fertility (Karl Pearson’s example), intelligence with adequate nourishment, strength with size of muscle. In all these cases we can say why the correlation obtains. But not in the psychophysical case—and this is hardly because we are dealing with elementary facts of nature (electrons, protons, etc.). The assertion of bruteness seems ad hoc, arbitrary, and none too credible. Such an association cries out for explanation: there has to be a reason why it is so. It can’t be just a primitive inexplicable fact that pain is correlated with C-fiber firing in all nomologically possible worlds. What is it about C-fiber firing that makes it correlated with pain in particular, and with nothing except pain? Surely, we don’t want to say that it could equally have been the neural correlate of pleasure or deep thoughts of God. We want to avoid such a bizarre conclusion if we can. We might then think to weaken the correlation to avoid problems of explanation: why not say the correlation is completely accidental, utterly contingent, inherently arbitrary? Then there is no nomological regularity to explain, no dangling counterfactual. It’s just like the correlation between coins having the property of being in my pocket and being all dimes—pure chance. Here we might turn to the dictionary (that philosophical treasure house): the OED defines “correlation” as “a mutual relationship of interdependence between two or more things”, or again “mutual close or necessary relation of two or more things”. But mere contingency doesn’t measure up to that kind of definition: where is the dependency, the mutuality, the necessity? C-fiber firing has to be correlated with pain, not just any mental state type. Why, we don’t know, but somehow it has to be. So, the brute contingency position is not recommended: it gives us mere juxtaposition not genuine correlation (interdependence). Our options are rapidly running out. A desperate option would be to go eliminative: there are no mental states to be correlated! The whole thing is a myth, like the correlation between witches and old lady traits—there is no correlation of real things to explain. But let’s not go there yet (or anytime), because there is another possibility: there are no correlations of the kind commonly alleged. The mind exists all right, but it does not stand in correlation relations to states of the brain. Yes, I know that’s a big pill to swallow, but let’s give it a run for its money (we don’t have much else to fall back on). What kind of conception of the mind would lead to such a view? I can think of three possibilities: dualism, behaviorism, and externalism. Dualism locates the mind in a separate realm cut off from the brain; it doesn’t make the brain a necessary foundation for the existence of mental states. Anything could be going on in this immaterial mind quite independently of the brain, so no correlation is to be expected: pain occurring in the mind might be accompanied by X-fibers firing on weekdays and C-fibers firing on weekends, or by no firing at all. Behaviorism locates the mind in overt behavior, so that there is mind if and only if there is suitable behavior, irrespective of what might be going on inside the body (nothing, possibly). Such a position could take in both Skinner and Wittgenstein (see Wittgenstein on the seeds). There need be no correlation between mental states and internal brain states, just identities between mental states and episodes of behavior. Externalism (social or environmental) maintains that the world surrounding the mind (head) fixes what is in the mind, the brain not being in on the act: if thoughts are causally connected to water, they are about water, irrespective of the neural facts. It is the idea that thoughts are purely internal that leads to the brain correlation hypothesis, but once we go external that motivation lapses. So, those are three colorable conceptions of mind that might see fit to abandon the correlation hypothesis, indicating that it may not be compulsory. But we can also step back and notice that there is no real empirical evidence for the metaphysical thesis being advanced: no one has ever observed a precise neural correlate corresponding to a specific thought, for example. How do thoughts about London differ cerebrally from thoughts about Paris? Such empirical evidence as we have concerns gross cerebral localization not precise mapping from mental states to specific brain states. That there is such a granular mapping is a matter of metaphysical faith not established scientific fact. Where is the neural correlate of predication or conjunction or the concept of an even number? What brain properties specifically might stand in such correlation relations—chemical, electrical, anatomical, vascular, metabolic? Is it the shape of a neuron that is correlated with a specific mental attribute? All this is left hazy at best. Granted, the brain plays a vital role in mental functioning, but does it really map so neatly and systematically onto the elements of the mind? Is it true that neural classifications are point-by-point correlated with mental classifications? Why exactly should we believe this? Notice that we can preserve a good deal of materialism by sticking to a token identity theory without presupposing any correlation between mental and physical attributes (our token monism might be completely anomalous). If that were so, there would simply not be the psychophysical correlations that so perplex us—and so nothing that we are failing to explain. There might well be associations, accompaniments, alignments, but not quasi-lawlike projectible generalizations linking mental and physical types. True, this would put us in uncharted territory, sailing towards some sort of dramatic Dualism (that dark and dreamy continent); but we might be willing to explore it given the trouble we are having with the idea of psychophysical correlation. As Kripke once poignantly remarked, the mind-body problem is “wide open and extremely confusing”, so we should be prepared for fundamental upsets in our accustomed ways of thinking. At any rate, the question is worth pondering: can the anti-correlation thesis be defended? Is the brain an adjunct to the mind or a foundation, a partner or a progenitor, servant or master? Are we even thinking about it in the right way? Is anti-correlational mysterianism the way to go? The mind-body problem is still wide open and extremely confusing, despite valiant efforts.
 Zettel 608. The idea is that two seeds could produce different plants without any internal difference between them. “It is thus perfectly possible that certain psychological phenomena cannot be investigated physiologically, because physiologically nothing corresponds to them” (609).
 We could also keep supervenience, since it calls for no correlation between mental and physical attributes, only determination of the mental by the physical, which could be quite holistic. Abandoning psychophysical correlation is not the end of the world for materialism in some form. Then too, there might be higher-order properties of the brain that constitute mental states and hence correlate with them, despite the variations at the basic neural level. The absence of genuine correlations at the level of neurons-as-we-now-conceive-them would explain a lot about our modal intuitions, the possibility of multiple realization, and the force of the knowledge argument. There just isn’t the kind of close bond between mental types and brain types that the usual discussions assume. I wonder when people started to talk this way, and why—was it an offshoot of the development of statistics with its notion of the correlation coefficient? I doubt it originated in direct investigation of the brain. This is a derivative form of description based on unrelated paradigms. We certainly can’t see psychophysical correlations.