Wednesday, June 04, 2008


If introspection is essentially a matter of perceiving one's own mind, as philosophers like John Locke and David Armstrong have suggested, then just as one might lose an organ of outer perception, rendering one blind to events in that modality, so also, presumably, could one lose an organ of inner perception, leaving one either totally introspectively blind or blind to some subclass of one's own mental states, such as one's beliefs or one's pains.

Is such self-blindness possible? There are, as far as I can tell, no clear clinical cases -- no cases of people who feel pain but consistently have no introspective awareness of those pains, no cases of people who can tell what they believe only by noticing how they behave. (I'm excluding cases where the being lacks the concept of pain or belief and so can't ascribe those states at all; and with apologies to Nichols and Stich on schizophrenia). Sydney Shoemaker suggests that the absence of such cases flows from a deep conceptual truth: There's a fundamental connection between believing and knowing what you believe, between feeling pain and knowing you're in pain, that there isn't between being facing a red thing and knowing you're facing a red thing -- and thus in this respect introspection differs importantly from sensory perception.

Since I'm generally a pessimist about the accuracy of introspective reports, my first inclination is to reject Shoemaker's view and allow the possibility of self-blindness. But then why do there seem to be no clear cases of self-blindness? I suspect that in this matter the cases of pain and belief are different.

Consider pain first. Possibility one: The imagined self-blind person has both the phenomenology of pain and typical pain behavior (such as avoidance of painful stimuli), maybe even saying "ow!" If so, on the basis of this behavior, she could determine (as well as anyone else could, from the outside) that she was sometimes in pain; but she would have no direct, introspective knowledge of that pain. Contra Shoemaker, this seems to me not inconceivable. However, it also seems very likely that a real, plastic neural system would detect regularities in the neural outputs generating pain behavior and respond by creating shortcuts to judgments of, or representations of, pain -- shortcuts not requiring sensory detection of actual outward behavior. For example, the neural system could notice the motor impulse to say "ow!" and base a pain-judgment on that impulse (perhaps even if the actual outward behavior is suppressed). This then, might start to look like (might even actually become?) "introspection" of the pain.

Alternatively, the self-blind person might show no pain behavior whatsoever. Then the person would behave identically to someone with total pain insensitivity (and cases of total pain insensitivity do exist). But now we're faced with the question: Do people normally classified as utterly incapable of feeling pain really feel no pain, or do they have painful phenomenology somewhere with no means to detect it and no way to act on it? The latter possibility seems extravagant to me but not conceptually impossible. And maybe even a fuller understanding of the neuropsychology of pain and pain insensitivity might help us decide whether there are some actual cases of the latter.

Regarding belief, I'm more in sympathy with Shoemaker. My own view is that to believe something is just a matter of being prone to act and react in ways appropriate to, or that we are apt to associate with, having the belief in question (taking other mental states and excusing conditions into account). Among the actions and reactions appropriate to belief is self-ascription of the belief in question, self-ascription that doesn't rely on the observation of one's own behavior. Someone who had the concept of belief but utterly lacked direct self-ascriptive capacity would be in some way defective not just as a perceiver of her beliefs but as a believer.

(Thanks to Amy Kind and Charles Siewert whose excellent articles criticizing Shoemaker on self-blindness prompted this post.)


Anibal said...

Eric, in line with your skeptical view of inner experience awareness i would like to say following Thomas Metzinger (Being No One, 2003) that the Wittgenstein/Shoemaker-principle of immunity to error through misidentification is somehow challenged in some neurological conditions
Alexithimia is a neurological condition in which the patient experience an emotion (i don´t know if pain have to be included) but he cannot tell us what it is.
Reverse intermethamorphosis syndrome is another in which the subject believes he is physically changing.
And many others delusional misidentification dyndromes.

Anonymous said...

Objects in the world have backs and insides. You can never see the entirety of a tangible object at once. Being perceptually acquainted with an object out there in the world involves positing that there are parts of it you don't see. (Holograms or projections trick you into thinking that they have parts that you don't see, when actually they do not have parts that you don't see).

But pains and beliefs do not have backs and insides. You can, and generally assume that you do, perceive the entirety of a pain or belief. So being perceptually acquainted with a pain or a belief involves positing that there are no parts of it that you don't perceive.

Another, Kantian way of saying this, is that our perception of objects is bound by the characteristics of both space and time, whereas our perception of pains or belief is bound by the characteristics of time, but not space.

These are just basic phenomenological differences, and don't speak to verdicality at all. Pains and beliefs are experienced by me as so intimate because I do not, in experiencing them, posit unexperienced parts.

Anonymous said...

Sorry, in the second paragraph I said, "So being perceptually acquainted with a pain or a belief involves positing that there are no parts of it that you don't perceive," when I should have said, "So being perceptually acquainted with a pain or a belief does not involve positing that there are parts of it that you don't perceive."

Badda Being said...

What would be an organ of inner perception?

Josh Weisberg said...


I've been trying to think of plausible cases of "self blindness." I take it the idea you have is someone who has conscious experiences but does not believe that they do, or something like that? (It does have a Moore's paradox feel, no doubt...)

One possibility is cases of illness denial. At some level patients know they can't move their limbs, but they claim to be healthy. They know at some level to confabulate. But one might worry that they are not conscious of their illness--that the confabulation is driven nonconsciously. In general, though, cases of denial and confabulation might be what you're looking for. See Hirstein's good book *Brain Fictions*.

Split brain patients also offer a possible example of "self blindness," though again, problems of interpretation lurk. The "left hemisphere interpreter" may deny consciousness of stimuli presented to the right hemisphere. However, lack of reportability is generally the best sign of lack of consciousness, so again, one might deny the break between consciousness and introspection.

I would also echo two of the previous comments: One, I'm not sold on immunity to error--is that supposed to a conceptual truth? Two, if one rejects a perceptual model of introspection, self blindness might be easier to accept. If introspection is theorizing about oneself, than why should we think the theory need always be accurate? There might be degrees of accuracy here: I might think I'm smarter than I am, I might think I'm better at pool or trivial pursuit, etc. Expectations might make me misinterpret a phenomenal state, etc. Pain is most likely the hardest case, but even that may occur. I think the theory model allows for a more nuanced approach here.



Anonymous said...

Does everyone find Moore's paradox-style sentences paradoxical?

It would sure take a lot of wind out of Shoemaker's sails if we could identify a class of human thinkers who do not see the need to make their assertions about the world and their assertions about their beliefs about the world jibe together.

Eric Schwitzgebel said...

Thanks for all the interesting comments, folks!

Anibal: I'm not sure what I think about the "immunity to error through misidentification" thesis. I suspect under some conditions it might be trivially true, similar to ordinary uses of "I": the judgment/sentence automatically refers to the subject. No identification is necessary, and no misidentification is possible. On the other hand, I am inclined to think that pretty much every sort of error is imaginable in some way or other. But the cases you describe don't seem to me to violate the thesis -- at least not without further explanation -- for example in alexithymia, people can't identify their emotions, but that seems to me not a matter of misidentifying oneself so much as misidentifying, or missing, the emotions that that self has.

Anon 1:32: Might there be parts of a pain I don't perceive? This is an issue I'd take slowly. If you think (as I do) that one can be mistaken about one's pains, then maybe the following kind of case is possible: You have a pain in one's forehead going down into your cheeks, but you judge mistakenly that the pain is only in your forehead. Is there then a part of the pain you didn't "perceive"? Well, what counts as perception, anyway, in this sort of case? Hmmm....

Nice points, Josh! I especially like your idea that denial cases and split-brain might possibly be interpretable as cases of self-blindness. Blindness denial might involve inability to introspect one's (lack of) visual experience; patients with split brains might be self-blind to some of their phenomenology. But both types of case are complicated. Blindness denial also involves denying straightforward third-person evidence, so the mechanism of denial can't merely be a failure of introspection. But then whatever is driving the denial might then be operating despite accurate introspection, as it operates despite accurate bodily perception? Split brain cases raise questions of the unity of consciousness and, as you mention, whether there really is the phenomenology that is not being reported. Great issues, though!

On the issue of introspection as theory: It seems to me that it has to be more than that. It has to involve something quasi-perceptual in the sense that it is attuned to events in the here-and-now. But surely (like perception!) it is theory-laden; and that indeed may be a major source of error.

Anon 8:37: I guess I do find Moore sentences, especially about belief, pretty paradoxical-sounding -- though I wouldn't adamantly insist that they could never be rationally uttered. Brie Gertler in a forthcoming essay has argued that there are cases where they can rationally be uttered....

Eric Schwitzgebel said...

Whoops, I missed Badda! I'm not sure I myself want to be committed to the idea that there is an organ of inner perception -- but if there were, it would have to be a relatively functionally isolable part of the brain or mind dedicated to the detection of currently ongoing mental states. Obviously, it couldn't be a readily identifiable extra-neural structure like an eye!

Anonymous said...

how about blindsight?

onclepsycho said...

"There are, as far as I can tell, no clear clinical cases -- no cases of people who feel pain but consistently have no introspective awareness of those pains"

It's called "pain asymbolia" (Schilder, P. and Stengel, E. (1931). Das Krankheitsbild der Schmerzasymbolie. Ztchr. Neurol.
Psych., 129, 250-79). See also the more recent work of F. Mauguière in Lyon, who actually recorded and induced this condition through intracranial electrodes right into the insula (he has some astonishing videos, you actually SEE these patients expressing excruciating pain, but they claim they don't feel it at all).

Anonymous said...

People with pain asymbloia feel what's happening to them (ie they're not completely anaesthetized) but don't feel it as "painful".

They can sense the kind of pain (dull, stabbing, etc) and the intensity, but the affective quality of pain is missing.

So I would argue that something more complicated is going on that simple lack of introspective awareness of pain.

kvond said...

splintered "Is such self-blindness possible? There are, as far as I can tell, no clear clinical cases -- no cases of people who feel pain but consistently have no introspective awareness of those pains, no cases of people who can tell what they believe only by noticing how they behave."

I think actually, instead of looking for a class of individuals where such is the case, one needs to look at instances where this is the case. And I think that this is a pretty regular occurrence. A child falls down but does not really start crying until it is rushed upon by a doting mother. A sinner does not feel guilty until retroactively he experiences his sinfulness. And most certainly in cases where beliefs are said to be unconscious (descriptions of nonetheless intentional behavior) one can take a third person view of one's own "beliefs". A therapist readily might be able to tell you what you believe, long before you know you believe it.

Aside from medical conditions, regularly we are estranged from our beliefs, and sometimes even from our supposed experiences (have you ever driven a car for some distance while deep in thought?)

Part of being social is having a degree of self-blindness, so that others can serve as normative corrections, and meaning-givers of our own experiences and thoughts.

Eric Schwitzgebel said...

Thanks for the continuing comments, folks!

On blindsight, the standard interpretation is that there is no phenomenology, so people are right about having no visual experience. Maybe, instead, there is visual experience but no reportability of that experience, but if so it's not clear that that's the best interpretation.

Onclepsycho/anon: Thanks, I'd never heard of that! I'll have to dig up the references on this phenomenon you mention. Based on anon's description, it sounds like it shares something with what's sometimes said about people on heroin (I think that's the drug) -- that they still feel the pain (and can report that they do), but it doesn't bother them.

Kvond: I agree that such cases are commonplace, but "blindness" (as Shoemaker and others conceive it) would be a different phenomenon -- not just not seeing (or introspecting) something in a particular case, but being persistently incapable of seeing (or introspecting). The challenge for me is to think about why although we often make, in my view, introspective mistakes, there is is no compelling evidence of this kind of systemic blindness. As you point out, that's a different question -- maybe less interesting, but at least a little interesting in its own right.

kvond said...

splintered: "but "blindness" (as Shoemaker and others conceive it) would be a different phenomenon -- not just not seeing (or introspecting) something in a particular case, but being persistently incapable of seeing (or introspecting). The challenge for me is to think about why although we often make, in my view, introspective mistakes, there is no compelling evidence of this kind of systemic blindness."

If one is to say that there is a "fundamental conceptual" connection between believing and knowing what you believe, you would have to also accept that there just such a fundamental disconnection as well. Part of having beliefs is being able to be wrong even about the beliefs that you think you have. The systemic cohesion requires a systemic incohesion.

I think that this "conceptual" fact is deeply related to the nature of beliefs (beliefs about the world, as well as about oneself). If you would allow me a Davidsonian position, while it is possible for any of my beliefs to be false, it is impossible for all of them to be false. As beliefs approach this asymptotic limit, they cease functioning as beliefs at all (which means they cease to explain intentional behavior).

The beliefs and knowing glue that Shoemaker (and Wittgenstein) may appreciate is necessary for the functional possibility of their disjunction, a requirement of revision and public correction.

Far from making error impossible, it is the grounds for making error continuous as a possibility. It is a ballast of rational explanation.

Looking for a class of persons who are completely in principle self-blind, would be like looking for a class of persons that have entirely false beliefs. They simply would not be "persons".

As a side note, because the attribution of pain is an explanatory force, I disagree that a self-blind person would behave like a pain-insensitive person. One in all likelihood would flinch at pin-prick and not know why, the other would simply not flinch.

onclepsycho said...


Ramachandran used pain asymbolia in his astute theory of the evolutionary origins of laughter. You can find this here (among other scattered papers from the same author, as well as the book Phantoms in the Brain):
"Consciousness and body image: lessons from phantom limbs, Capgras syndrome and pain asymbolia.
Philos Trans R Soc Lond B Biol Sci. 1998 Nov 29;353(1377):1851-9."

Otherwise, pain asymbolia was quite a big deal in German and French neurology in about the first half of the 20th century (notably P. Schilder and H. Hécaen), then for some reason interest decreased and now is not even mentionned in some textbooks. Unless you read German and/or French, my guess is that you will have a hard time finding anything worth on the topic. However, current thinking on the neurology of pain could go some way in clarifying some philosophical issues, maybe: think of the insula, the anterior cingulate and the primary somatosensory cortex as different levels relevant for the phenomenology of pain. Bud Craig, for instance, uses a thermal grill that induces an ILLUSION of pain. The objective degree of (non-noxious) coldness correlates with activity in the insula, while the subjective rating of (illusory!) pain by subjects correlates with activity in S1. A beautiful study, would love to hear a philosopher comment on this.

Josh Weisberg said...


Craig's work is very cool (so to speak), though it's a matter of debate whether this is an illusion of pain (or if there can be such a thing, which is the issue here, I take it). It may be that there is no pain present (there is no stimulus that is so cold as to potentially produce damage, as registered by the insula), while there is an illusory experience of pain due to ACC activation (or is it S1?). But some just define pain as the conscious feeling, regardless of the connection to stimuli. So there is ACC activation and an experience of pain. The subject is under no illusion about this. But the cause of the pain is not the usual damaging stimuli, due to the thermal grill. So it's not a illusion of pain, it's pain with an odd cause.

A further question is is there an error in introspection occurring here, even on the illusion interpretation? The conscious experience is one of (apparent?) burning cold pain. The subject reflects and reports that "That feels like burning cold. Ouch." It seems that the introspective reflection is in line with the conscious experience. But the conscious experience is not properly connected to registration of objective bodily distress. So it's not self blindness; rather, it's body blindness. Or so might a philosopher argue!

(It was a quick scan of the Craig article, so I may well have missed the point there. Another risk of asking philosophers about scientific results!)



Eric Schwitzgebel said...

Thanks for the follow-up, Kvond! I think I more or less agree with you in what you say about belief. On pain, I could see "self-blindness" going either way -- why not?

Eric Schwitzgebel said...

Onclepsycho and Josh: You've definitely convinced me to look into this! If I can dig something up from the hints and references you've given, it would be fun to work up a post on it. (My hunch is that I'll probably have a similar reaction to Josh's -- that multiple interpretations of the evidence are possible.)

kvond said...

eric: "On pain, I could see "self-blindness" going either way -- why not?"

In thinking about it, at first blush one might think so, that is, it is possible for a self-blind person to be interpretively completely the same as an insensate person. But I think that because pain ascription (the experience) in an explanation of behavior, like belief ascription (the propositional content of being able to say "I am in pain"), they would have some divergence in use. The occasions in which one would say, "He feels pain" would not co-extensive with those that would require "He thinks he is in pain". One need only look to pain ascriptions to animals and the problems of belief ascription to the same. Belief ascriptions hold much more firmly to linguistic beings, while experience ascriptions are much more wide-spread. I think that in terms of the behavior of the animal "human being" there is this same non-correspondence in terms of behavior and circumstance. There are animal level behaviors (automatic responses for instance) that simply would not require that full belief ascription force of explanation. This difference would lead to one being able to find distinctions between a supposed "self-blind" person and an insensate person. Lepers can be insensate to pain, but there is no sense in which we are tempted to say that they are self-blind to pain. Strict self-blindness would require a perceived disjunction, the behavioral registery of pain, without its conscious awareness (otherwise we would be able to say something like, blind people actually can see, they just are not aware of it).

If I am reading your "either way" comment correctly.

Eric Schwitzgebel said...

Thanks for clarifying, Kvond. If I understand your point, I agree with you: The insensate and the self-blind won't be interpretatively identical. But both self-blindness and insensitivity to pain are conceptually possible, and in many actual cases the evidence (e.g., "pain asymbolia") may leave it unclear which interpretation is the better. Yes?

kvond said...

Eric: "The insensate and the self-blind won't be interpretatively identical. But both self-blindness and insensitivity to pain are conceptually possible, and in many actual cases the evidence (e.g., "pain asymbolia") may leave it unclear which interpretation is the better. Yes?"

Yes. I think this is so. It seems that this very same categorical unparse-ability (now, that is a word!) is operating in our everyday ascriptions of people, and animals as well. I see no reason why the extraordinary examples would be any different.

Good thoughts.

Ian said...

It isn't by any means self-blindness, but what about phantom pains that people have in their amputated limbs? In self-blindness, I can have a pain in my finger and not know that I have a pain in my finger. In phantom limbs, I can have a pain in my finger and know that I don't have a pain in my finger.

Shoemaker's point in arguing the impossibility of self-blindness is to show that my second-order beliefs about my experience are not in principle separable from my first-order experiences. He even says somewhere that the second-order beliefs supervene on the first-order experiences plus rationality.

But in a phantom pain scenario, it is specifically my rationality that insures that I won't have second-order beliefs that correspond with my first-order experiences. I feel a pain in my finger, but, since I know I don't have a finger, I don't believe that I feel a pain in my finger.

I don't know if this works at all. It might be that if this is a problem for Shoemaker, then any scenario in which we'd doubt our own senses is, (i.e., any skeptical scenario) which would probably mean that I'm taking him to be asserting more than he is. What do you think?

kvond said...

Ian: "Shoemaker's point in arguing the impossibility of self-blindness is to show that my second-order beliefs about my experience are not in principle separable from my first-order experiences. He even says somewhere that the second-order beliefs supervene on the first-order experiences plus rationality."

This really points up the grand problem that I have with Wittgenstein-like assessments. There is the desire (in Wittgenstein it is a therapeutic desire) to categorically index statements to be of a kind or a specific relation. This is in tension, in my mind, with the general contingent meanings-as-use, language-as-tool approach which the Language Game analogy brings. The phantom limb example, and I think it is a good one, shows that while there is a definite ballast to the rationality of lst person statements, this is not a conceptually rigid designation. Experience and report are not necessarily consubstantial. In fact, they must be able to diverge in order for the system to have the required malleability to adopt to new and unique situations and social configurations. In the end, statements like, "Only I can feel my pain" or even "I know I am in pain" are grammatical in a broad and historically contingent sense. There very well may arise circumstances (a favorite Wittgenstein mode of analysis), where it is not illogical to deny such sentences.

Eric Schwitzgebel said...

I agree with kvond here, Ian, but I also think Shoemaker would probably want to deal with phantom limb cases by doing something like separating the privileged content (something like "I have a pain-in-my-toe-like feeling") from the non-privileged (something like "my toe has a pain in it"). Phantom limb cases don't undermine the accuracy of claims of the first sort, so I don't think the case against him can be made that simply.

kvond said...

Eric: "something like separating the privileged content (something like "I have a pain-in-my-toe-like feeling")"

I don't know how one assesses the content of a statement such as the above. It is assumed from the Wittgenstein field that we do not "know" that we are in pain, we just are. But we supposedly "know" how to use the word "pain", as some sort of continuation of the half-way-house of expression, called "pain behavior". "Ouch" becomes, "I have a pain-in-my-toe-like feeling". "Ouch" like a grimace has no real "content", but the latter is supposed to. Did I use the words "pain" (is it a tickle) and "toe" (is it floating above the toe) and "like" (how much is it unlike)correctly? If the use of words (properly or not) in describing experiences can be doubted and judged then I would say that so can the "content".

One can just say that there is no "content" there is just expression (and this is something of which Wittgenstein means), but following this example then there is no such thing as content in the public sense either. There is just getting it right and getting it wrong, by consequence. "Content" just becomes an index of further distinctions.

I think the phantom-limb example is a good one as it opens up the contingency of the truth of statements about oneself. And the fact that we regularly take third-person positions of observation on our own experiences and states. There is an irrational reconciliation of two observed facts: I have pain in my toe; I don't have a toe. The sentence "I have a pain-in-my-toe-like feeling" only makes sense in specific circumstances where assertion is required, for instance if someone does not believe you, or as a detail of a diagnositic (the accuracy is still contingent).

But in relating to oneself, for instance the ways in which one allows oneself to admit pain and comes to negotiate with it, the two distinctions which vie with each other, "I have a pain in my toe" and "I can't possibly have a pain in my toe" are important in their contradiction. Think for example, "I am so happy and successful" and "But why am I so depressed?" There is a kind of self-blindness that ensues, wherein the body (or parts of the persona) are carrying on the requisite experiences, yet one is not able to "know" them as coherently real. The very irrationality of their combination is what forces the person (or an attendant caretaker) to look for internal reasons or causes for their conflict. "I know that I am feeling x, but I do not seem to be feeling x" is an important diagnostic which has a conceptual relationship to self-blindness, I think.

For me, "I have a pain" is a more physically entrenched, but still in terms of content doubtable version of "I am happy".