Last spring, the mereological sum of D.M. Armstrong and me was the Whichard Distinguished Visiting Professor at ECU, giving a seminar for the Philosophy Department faculty.  (Long story and you do not want to hear it.)  One week I gave a presentation on HOP, and we considered objections.  Armstrong replied to the objections and then made his three new arguments in support of Inner Sense.  Here I will repeat the objections and report DMA's replies, and then reproduce his new arguments.

    The appearance-reality objection (rediscovered by Chuck):  Some philosophers have complained that representational theories leave introspective beliefs too fallible and underrate the privileged access we have to our own mental states. An internal monitor is a mechanism, and every mechanism is fallible and works only contingently.  But the objectors contend that our awareness of our own mental states is either infallible or, if not flatly infallible, constrained against unreliability.  Sydney Shoemaker, for example, grants that pain can “occasionally” escape awareness, but he insists that that could not happen “as a matter of course; it may be true in Lake Wobegon that all of the children are above average, but it can’t be true everywhere.”  Reply:  The Wobegon analogy fails; nothing about introspection affords any arithmetical calculation such as that of an average.  Rejoinder (Wittgenstein):  But there are concepts which impose rules of the form, “X can and occasionally does happen without Y, but this couldn’t be common or normal or without some special explanation”; to grasp the very concept is to know that any X without Y is a necessarily rare aberration.  To have the very concept of pain is in part to know that although in special circumstances there can be pain-behavior without pain, pain-behavior is overwhelmingly good evidence of actual pain.  So too with introspection.  It can and occasionally does go wrong, but that has to be an aberration and it requires a special explanation; it couldn’t happen as a matter of course.  But (again) the “inner sense” model predicts that if you have a broken introspector, introspection could go wrong all the time, even systematically.  If our introspectors were all to break, introspection would be completely unreliable—which idea makes no sense.  Variation:  Jason ingeniously proposed an analogue to J.L. Austin's antiskeptical argument about sense-perception, to the effect that misperception conceptually must be the abnormal case; see Austin's Sense and Sensibilia.

    The false positives objection (Neander):  Like any mechanism that is a monitoring-and-reporting device, the internal scanner can be expected to issue the occasional false positive.  Even though you are not in any pain at all, your introspector could start yelling, “Yow, terrible pain in your left elbow!!!”  But what would that be like??  (a) Neander's claim: It would be exactly like feeling pain; yet supposedly there is no pain at all; but that’s as near as matters to a contradiction..  (b) It would be like hearing a voice saying “Yow, terrible pain…” even though you felt no pain.  –But that wouldn’t really be introspection, but only having the experience of hearing a voice.  (c) You would be in a weird dissociative state.  As I said, (c) is my choice, because the basic functional effects of the pain would be missing and that would register phenomenologically.  And (a) is required to make the objection go through.

    Inner Sense meets methodological solipsism:  Propositional attitudes themselves are in the head, but if Fodor-Putnam-Stich externalism is true, their contents are not.  An internal monitor can scan what is in the brain, but it cannot scan causal-historical-or-whatever processes extending outside the head and connecting one’s brain states to water.  Yet you know introspectively that what you desire is some water.  Introspection tells you effortlessly and reliably that that belief is (indeed) about water—water, not XYZ.  But an internal monitor can’t know whether you’re on Earth and in contact with H2O rather than on Twin Earth and facing XYZ.  Thus, introspection cannot be, or cannot simply be, the operation of an internal monitor.  (Of course, we already know about the problem of externalism and self-knowledge; the present point is that HOP makes that problem worse.)

    Mark and Rey on laptops:  Internal monitoring per se is cheap.  Every laptop does it.  Any halfway competent computer has proof-checkers and fail-safes of various kinds that monitor its own first-order computational states.  So every laptop is conscious? Reply:  It’s only mental states that are made conscious by being monitored.  So every laptop has mental states?  Rejoinder:  All right.  But if we just assume that some creature or device does have first-order mental states but no conscious ones, could we make those states conscious just by adding a simple monitor and turning it on?  As before, monitoring is cheap; buy a little one at the hardware store and you’re in business.

    Here are Armstrong’s replies.

    Appearance-reality:   There is mounting empirical evidence of the fallibility of introspection (hypnotism, etc.).
    Rejoinder:  Shoemaker et al. never denied that introspection is fallible; their complaint is that, as a conceptual matter, introspective error must necessarily be rare, exceptional, an aberration, in need of special explanation, etc.  And the “inner sense” theory does not secure that.  It leaves it a contingent matter that introspection is as reliable as it is, even if there are good contingent reasons why introspection is very, very reliable.
    Of course we have not really seen an argument for Shoemaker’s thesis.  At this point, the Inner Sense theorist can just insist that the reliability of introspection is a contingent matter.  (But we would eventually have to look at Shoemaker’s argument to the contrary, if we can find one.)

    False positives:  We already know empirically that introspection can give false positives.  So what’s the problem?
    Rejoinder:  In an extreme case—the introspector reporting terrible pain when there’s no pain at all—the experience cannot be satisfactorily described (see above).  That problem remains unsolved.  And now, Greenville strikes a nasty blow (I did not confess this in class):  George Bailey alertly pointed out that we can just have the introspector give false positives of all the functional mental effects too, making it seem to me that I have the relevant urges, distractions etc. even though I don't.  Gaaahh; I hadn't thought of that and have no very good reply to it, except to note that there would still be a bit of dissociation because there would still be no actual pain behavior even if the introspector is reporting an urge to pain-behave, and that discrepancy would probably register phenomenologically.

    Methodological solipsism:  (1) This is more an objection to Fodor-Putnam-Stich externalism itself than it is to Inner Sense.  (In fact, it already is a standard objection to externalism.)  (2) Since there aren’t, in reality, any Twin Earth, XYZ etc., some internal features of your desires and beliefs about water are reliable indicators of the external fact that water is what the desires and beliefs are about.  That’s how you know.  (This assumes an epistemology according to which reliable indication suffices for knowledge.  Armstrong has defended such an epistemology at length, in his book Belief, Truth and Knowledge.)

    Laptops:  Right, just monitoring per se won’t guarantee consciousness.  (The CNN example.)  Point taken; further conditions must be added.

    And here are the three new arguments in favor of Inner Sense.

    1.  What’s characteristic of ordinary external perception is that it gives a continuous flow of information about the environment.  Introspection is the same in that regard; we have a continuous acquisition of information about our own mental states.  William James called it the “stream of consciousness.”  This is another way in which introspection is perception-like.

    2.  Assume “empiricism” about concepts, according to which we acquire concepts by abstracting them from things we perceive (“table” from tables, “bully” from bullies, etc.; Jesse defends an extreme version of this view).  Then where do we get our mental concepts?  Presumably from (quasi-)perceiving our own mental states.
    (But this empiricism is very contentious and controversial as a general doctrine.  Also, such an account of mental concepts is also in considerable tension with Armstrong’s own analysis of mental concepts as third-person causal concepts.)

    3.  Think evolutionarily: What is the adaptive value of consciousness in the present sense?  We humans are the only species that can solve practical problems in our heads.  (Though in a famous ethological experiment, banana-seeking chimps made small achievements of this kind.)  Problem-solving in the real world requires constant perceptual feedback; indeed, just walking from one room to another requires constant perceptual feedback.  So, probably, problem-solving in our heads, walking through the steps, requires feedback.  But the feedback doesn’t come from the real world, since we’re only in our heads.  It must come, instead, from the mental states we are running through.  But that means there is an information channel monitoring those mental states.  Hence, some version of Inner Sense is correct.