Friday, December 21, 2007

Introspective Infallibility, Causation, and Containment

In his second Meditation, Descartes gives the impression that he thinks self-knowledge of current conscious experience is indubitably certain, immune to error, infallible. (Whether he consistently espouses this view throughout his corpus is another question.) Ever since, infallibilism about introspection has been a mainstream position in philosophy of mind -- sometimes dominant, sometimes (as now) out of favor but nonetheless with prominent proponents.

If we suppose that introspection is a causal process between two distinct events, it's hard to see how infallibilism could be plausible. What sort of event can't be brought about in strange ways? If we suppose, for example, that introspective judgment is a brain process, couldn't -- at least in principle, by dint of genius neuroscience, but probably much more easily than that -- that brain process be brought about non-standardly?

One way out of this is to deny that the introspective judgment and the introspected conscious experience are indeed distinct events ("distinct existences" in Shoemaker's sense). For example one might "contain" the other, as is sometimes suggested (e.g., Shoemaker, Burge). Consider as an analogy: "This sentence contains the word 'pixie'". The sentence is infallibly true wherever it appears because the conditions of its existence are a subset of the conditions of its truth. Could introspection work the same way?

Well, one fella's modus ponens is another's modus tollens: If containment implies infallibility, the case against infallibility is, I think, so compelling that we ought to deny containment. But let's consider containment independently of that. Does the judgment, "I'm visually experiencing redness" (for example) contain a visual experience of redness? Does it itself, somehow, contain the phenomenology of red -- not merely assert the existence of red phenomenology but actually include that phenomenology?

Let's suppose -- I don't quite buy this, but it's probably close enough for the purposes of this argument -- that the components of judgments are concepts. Concepts may be reshuffled and combined to make new judgments, right? Now the judgment "I'm visually experiencing something caused by Martians" cannot literally contain something caused by Martians because nothing is caused by Martians. And the judgment "Looking at Mars can cause people to visually experience redness" cannot contain an actual experience of redness because it can be uttered by a blind woman. But now we can recombine elements of the two to get "I'm visually experiencing redness". It's odd to suppose that this recombined product must contain actual red phenomenology if it's composed only of elements none of which contain that phenomenology and that can occur independently of it.

Or: The judgment "I visually experienced redness" does not contain red phenomenology (since I might now be experiencing no redness). Similarly for the judgment "I will visually experience redness". Is the present-tense version of this judgment so radically different in structure from the past and future tenses that it must contain redness -- a totally different kind of thing from what the others contain -- while the others don't?

The most plausible case for something like containment might be the following (bastardized and simplified from Chalmers 2003): "I have *this* phenomenology" -- where *this* is an act of "inner ostention", cognitively pointing toward one's own phenomenology. Such a case might be a case of self-fulfulling containment, but it is no more substantive or necessarily introspective than "I'm located here".

6 comments:

  1. Introspection, infallibility, inner ostention - this makes me think of Colbert's notion of truthiness: don't confuse me with the facts, I know what i know! Or, in his words:
    'I don't trust books. They're all fact, no heart.'

    ReplyDelete
  2. Eric:

    I've always wondered what could justify containment, especially in the face of many plausible cases of introspective error.

    But I wonder what your think of this sort of claim, often made by philosophers, and in the 2nd meditation as well:

    "You can be wrong about the world, but you can't be wrong about how things *seem* to you."

    Can you be wrong about how things seem to you, in your opinion? Is this just a restatement of the cases you've mentioned, or does the word "seem" add something?

    ReplyDelete
  3. Whoops, Josh Weisberg, sorry for the slow reply. Your comment got lost in the flood of comments on the dreaming posts!

    I'm inclined to think the word "seem" is toxically ambiguous, in such arguments, between a phenomenal sense and an epistemic sense. You *can* be wrong about how things phenomenally seem, but a sentence like "It seems to me that Hillary will win the nomination" is a different sort of matter (not a claim about phenomenology) that has some epistemic complexity to it. Under certain conditions it may be infallible or (differently) not functioning primarily as a statement evaluable in terms of truth and falsity.

    I have some more detailed remarks on this in section ix of my forthcoming essay The Unreliability of Naive Introspection, if you're interested.

    ReplyDelete
  4. Couldn't one prove, however, that a photon of a certain wavelength which people in the English language agree to call "red" is or is not able to be detected by the eye of a particular person, and that that photon hitting the photo receptor results in a cascade of neural activity to and through the brain? And suppose that the detection of said photon were able to create the same neural pattern of firing in the brain each time. And when we asked people what they experienced, if they said "red" each time, isn't that then proof of an accurate experience of redness?

    ReplyDelete
  5. That would make a pretty good case. I don't say we're never accurate in our introspective judgments -- just that we're less accurate, overall, in our judgments about our stream of experience than in our judgments about the stream of outward events.

    ReplyDelete