Epistemic Necessity and the Self

Consider the two statements, “This table necessarily exists” and “I necessarily exist”, where “necessarily” is construed epistemically (“I could not be wrong that”). The former is clearly false, the latter apparently true. Why is the former false? It is false because I could be hallucinating or dreaming or otherwise under a misleading impression. Putting it in the jargon of epistemic contingency, there could be an epistemic counterpart of my present experience as of a table that is not of an actual table. I could be a brain in a vat imagining a table not really seeing one. We cannot deduce the existence of a table from an impression of a table; the two things can exist independently. There might not be any tables, though there are many instances of seeming to see a table. In normal life, such hallucinations occur, and they undermine claims to certainty (as the skeptic argues). This is all very intuitive and virtually inarguable; it doesn’t violate common sense or invoke recherche possibilities. Beliefs about the external world are epistemically contingent, notoriously and obviously. You don’t need to be a trained analytical philosopher, adept in far-out thought experiments, to see the point; being a nightly dreamer suffices. But the self is another matter, or so it has been thought: this is the domain of the Cogito and its associated convictions (“I know that I exist”). However, it has not been easy to trace out the logic of this intuitive conviction—the classic Cogito has met with a good deal of resistance. In what follows I will attempt to articulate the underlying reasoning behind the belief that the existence of the self is an epistemic necessity—the thought that I necessarily exist. Put more cautiously, I will spell out the difference between material objects and selves in respect of epistemic necessity.

It will be helpful to have a stalking-horse: this is the contention that I may have epistemic counterparts that have no associated self—mental states that are indistinguishable from mine but correspond to no self. These are conceived as “free-floating”, not anchored to any entity that has them. In the case of my actual mental states, we can suppose for the sake of argument that a subject exists, but we can imagine an epistemic counterpart of these states that has no subject—so how do I know that this is not my situation? It would seem to me just as it does now, but there would be no existent subject. It would be subjectively the same but ontologically different: same appearance (phenomenology) but different reality (ontology)—self versus no-self. Couldn’t it be just like the table case? I think not. First, we must ask whether the epistemic counterpart conveys an impression of selfhood: does it seem to itself to have a corresponding self, or is it neutral on the question, or even negative? Well, does it seem to me as I now am that my experiences have a subject distinct from them? I think the answer is yes. It seems to me that my experiences have intentionality—they are as of various states of affairs—and it also seems to me that they have a subject, viz. me. That is, they have an as-ofcontent and an as-by content: as of a certain object and as by a certain subject. That is, an impression of selfhood is present in my experiences, as well as an impression of objecthood (at it might be, a table). Here a neologism will be useful: states of consciousness have both intentionality and “subjectality”. They are intrinsically subject-indicating (referring, representing); they are not subject-denying, or subject-neutral. Subjectality is part of phenomenology. Thus, any epistemic counterpart to my current state of consciousness will also have subjectality, by definition of epistemic counterpart. Put simply, it will seem to itself to have a subject—as my consciousness seems to have a subject. Consciousness is I-referential. But of course, it doesn’t follow from this that the subject referred to actually exists—maybe this is an error on the part of consciousness. Maybe consciousness hallucinates a self—as it can hallucinate a table. At this point things start to get hairy: for what kind of hallucination might that be? Are we familiar with such hallucinations, has anyone ever had one, how might they be produced? In the case of external objects these questions are readily answered: yes, yes, and easily. We have a model, a theory, of sensory hallucination: it happens a lot and is easily brought about. It isn’t just philosophical word-spinning. But the same isn’t true of supposed self-hallucination: have you ever heard of someone being under the illusion that they have a self? Are there patients in psychiatric wards suffering from hallucinations of selves? That is, we know they have no self (their states of consciousness have no subject) and yet they are under the impression that they have a self. The mind boggles: what could this even mean? We are being asked to accept that there could be, or are, cases in which a mind seems to itself to have a subject, an “I”, but doesn’t really. Surely, that is not possible; or if it is, such cases never actually occur and are impossible to comprehend. They are certainly not part of common sense and everyday life. Of course, there might, as a matter metaphysical possibility, be cases of people with a sense of self that have no body: you can hallucinate having a body. But there are no cases of people under the illusion that their mental states are had by someone (something). No one ever has the feeling that their consciousness is had but actually it is not had. We cannot make sense of subjectality without a subject. We can say the words, but we can’t provide any examples, or explain how the hallucination works, or suggest how it might be mended. So, the very thing that powers the intuition that this table’s existence is not known with certainty is absent in the case of the self. Hence, we quickly see the epistemic contingency in the table case, but not in the self case—here we are presented with just a jumble (or jungle) of words. The skeptic is limping at this point, but with tables he is off to the races. Maybe he can wheel in extra machinery (the skeptic is nothing if not resourceful), but he cannot rely on commonplace facts and powerful intuitions. He thus has a lot of work to do; he can’t just point to the existence of hallucinations and dreams (do you know of a case in which a person without a self had a dream in which it seemed to him that he had a self?). I am strongly of the opinion that there cannot be errors of selfhood—cases which subjectality is present but not a corresponding subject—but I have no direct proof of this. The point I am making is that the model of the table won’t work to derive skepticism about the existence of the self. Hallucinations of external objects are facts of nature; hallucinations of selves are figments of the philosophical imagination—would-be thought experiments not empirical facts. This is what lies behind our ready acceptance of the epistemic contingency of “This table exists” and our resistance to a like conclusion about “I exist”. The latter strikes us as a lot more necessary than the former (as that is more necessary than “Dark matter exists”). Epistemic necessity comes in degrees, and the self is at the high end of the spectrum (though perhaps slightly less high than “This pain exists”). What is interesting are the reasons for the difference, specifically the absence of demonstrable hallucinations of the self. The thought never occurs to us that our impression of our existence as a conscious self might be a lifelong delusion, possibly not shared by others, precisely because no such cases have ever been recorded. We might become convinced of it by a philosophical or scientific argument against the existence of the self (though I know of none such that really succeed), but we won’t be budged just by pointing to mistakes induced by hallucinations—because there are none. Our position ought to be, “Unless you can prove to me that the self doesn’t exist, I see no reason to abandon my strong (certain) belief that my self exists”. We would be right, however, to refrain from such a pronouncement regarding the table, given what we know about the human nervous system and the powers of certain drugs (you might have hallucinated a table only yesterday).

How do these points bear on the Cogito? Not very directly. That is a different argument altogether, proceeding from the existence of thoughts to the existence of a subject (a substance) that has them. It has been questioned on a number of grounds, persuasively enough. Descartes never argues that his premise about thought includes a thesis about self-indication (subjectality); nor does he invoke considerations about the constitutive conditions of hallucination. However, it may well be that the points I have adduced are subconsciously influencing our response to the Cogito, giving it an appearance of cogency it might not otherwise possess. Reasons for accepting a philosophical claim do not always coincide with the content of that claim (indeed, they often diverge).[1]

[1] Let me make clear, if it is not already, that I am working with a minimal view of the self or subject (as was Descartes). I don’t mean an animal with a certain kind of body (a human being), or a persisting self, or a type of substance, or a unified self, or even a knowable self; I just mean a thing that acts as a bearer (logical subject) of a mental state—something that has it. That could be ever so etiolated, so long as it doesn’t collapse into the mental state it is supposed to bear. The idea, then, is that states of consciousness make it seem as if they are had or possessed by something distinct from themselves (the conscious states don’t have themselves). We know with virtual certainty that this thing exists, however it may be with “thicker” things.

Again, Supervenience

There is a well-known problem with physicalism: the problem of defining it. Briefly: do we mean the mind is reducible to the brain as now understood, or do we mean the mind is reducible to the brain as it may be understood in the future? Do we mean current actual neuroscience or do we mean a future possible neuroscience? The former seems too restrictive and the latter too open-ended. I am not going to go over this problem now; my aim is to extend the argument to supervenience claims. The mind is held to supervene on physical properties of the body and brain, but what is meant by “physical” here? The dilemma is that this will either be too restrictive or too liberal, and hence either false or empty. Is there any version of the claim that can escape this dilemma? Suppose supervenience were propounded two hundred years ago when very little was known about the brain, specifically the nerve impulse; perhaps only the brain’s gross anatomy was known. Is the mind supervenient on that? Clearly not, because a completely insensate object could share that gross anatomy (e.g., a plastic model of a brain). The details of neurons and neural impulses matter. And it would be premature to suppose that nothing further will be discovered as to the brain’s neural properties. We should not limit the supervenience base to what we currently know of the brain. But then we are left with a lack of clarity about what the “physical” supervenience base consists in.

There are subsidiary questions. Is anything about the mind supervenient on what we now know of the brain? That does not seem out of the question: perhaps dispositional properties of mental states are so supervenient, or some structural and temporal properties. Not qualitative intrinsic properties but behavioral ones: maybe these depend wholly on neural impulses as we now understand them. The mind isn’t then wholly physically supervenient in this sense but it is partially so supervenient. Or perhaps not. The question is more empirical than a priori. Next: Is the mental supervenient on a subset of the brain’s properties not all of them? That is surely very plausible: some properties will be irrelevant to the mind, e.g., the color of the brain or its particular shape or whether it tastes good with spinach. It would be nice to know which of the known properties of the brain form the supervenience base. The bare assertion that the mind depends on the brain is unilluminating because much too coarse—we want to know which physical properties are crucial. We know that some parts of the brain have no mental correlate (e.g., the brain stem), so what is it that makes other parts of the brain capable of generating the mind? What is the physical difference? It can’t be the presence of the neural impulse, because that does not always lead to a mental correlate. Supervenience alone tells us almost nothing of interest; and it is either false or an empty truism. What we need is a theory that specifies precisely what aspects of the brain form the supervenience base. The muscles are supervenient on the body—yes, but what aspects of the body? If we knew the answer to that, we would be close to explaining the supervenience; but then, we wouldn’t be limited to a bare supervenience claim. Such a claim is next to useless unless it permits of conversion to an explanatory theory, but then it isn’t necessary. In short, brute supervenience is a pointless idea—an empty slogan. It should be banned.

Suppose we held that mental properties are physical properties of the brain, in some workable sense of “physical”. Then, trivially, mental properties would be supervenient on physical brain properties. That would be a doctrine of zero interest. No, we must mean by “physical” non-mental. So, the claim must be that the mental is supervenient on the non-mental. Do we really want to say that? We must rigorously exclude any hint of the mental from the putative non-mental supervenience base (so panpsychist properties must be excluded). This would mean the supervenience base would have to be compatible with the absence of the mental—but that contradicts supervenience! For example, if electrical properties were held to constitute the supervenience base, we would face the objection that electricity doesn’t entail mentality, which contradicts supervenience. But electricity is exactly what the brain traffics in, so it can’t be sufficient. How can the mental supervene on the non-mental, i.e., that which doesn’t require a mental correlate? If, however, the mental supervenes on some sort of proto-mental or quasi-mental base, then we don’t have supervenience on the non-mental. Either way psychophysical supervenience doesn’t work. Perhaps we can identify some properties of mental states that supervene on electro-chemical properties of the brain, such as dispositional-behavioral properties; but that is not to say that any perfectly general supervenience claim can be intelligibly formulated. Supervenience on what is known of the brain seems to be either relatively trivial or false. It is really a foggy magical idea captured in a fancy word. All we have is the idea that the mind might be semi-supervenient on the properties of the brain currently recognized (and as currently recognized). For there is really no content to the claim that the mind (as a whole) is supervenient on the physical properties of the brain (present or future). It is certainly not the case that mentality supervenes on electricity and chemistry, or else mentality would be everywhere. As the term “physical” is understood now, it is clearly false that the mind is supervenient on the physical; and no other sense can be plausibly stipulated.[1]

[1] The fact is that physical supervenience inherits all the disadvantages of physicalism in respect of definition but none of the advantages in respect of ontology. It doesn’t tell us the nature of the mental and how it exists in the brain, and it makes no progress with saying what physicalist doctrines amount to. So, why has it been such a popular idea? Because it papers over the difficulties. The analogous doctrine in ethics tells us nothing about the nature of the ethical and also faces the problem of saying what precisely the supervenience base is to include and exclude (what exactly is a “descriptive” property?).