Peacocke gives the outlines of his view of self knowledge in ¤II of the essay. But it will be useful to begin by sketching some points he makes in ¤I.
Central to ¤I are two distinctions between classes of mental states. First, Peacocke introduces the distinction between those mental states which do, and those which do not, “contribute to what, subjectively, it is like for the person who enjoys them” (64). The paradigm case of such a state is a sensation or perceptual experience. But, Peacocke notes, such states include also at least some propositional attitudes - such as judging that so-and-so or having a thought go through one’s mind -- and acts of trying to perform certain actions. His interest in this essay is mainly with conscious propositional attitudes.
Second, Peacocke distinguishes between occurrent and non-occurrent mental states. This is a distinction that we have made before; it is hard to spell out clearly, but easy to get some grip on. Having a thought go through one’s head is an occurrent mental state, whereas
Peacocke in this essay is interested in mental states which are occurrent conscious propositional attitudes. He wants to claim that these mental states can help explain self-knowledge. To understand this claim, though, we need to say a bit about the role of attention in conscious propositional attitudes.
One of the difficult parts about this essay is to understand how the discussion of attention fits into the theory of self-knowledge, and why Peacocke begins with a discussion of it.
One interpretation is as follows: Peacocke is out to show that knowledge of our own beliefs is partly explained by awareness of conscious states. But there is a kind of view in the philosophy of mind on which this seems to involve a kind of category mistake. On this view, there are ‘phenomenal states’ and ‘non-phenomenal states’ - perceptions, pains, and other bodily sensations go into the first category, and all non-sensory states go into the second. Peacocke does not completely reject the idea that there is some such distinction to be had; but he does reject the idea that types of mental states can be so easily categorized as phenomenal or non-phenomenal. Thus he thinks, for example, that episodes of thinking or even of believing can have a distinctive phenomenology.
Two strategic points about this: (i) If Peacocke is right, then this lends credence to his view of knowledge of belief as based on kinds of awareness of conscious episodes. (ii) If he is right, then this also seems to count against a Shoemaker-style account of our knowledge of our own beliefs (for the same reason that such an account seems ill-suited to the case of knowledge of our own sensations and perceptual experiences).
One point of the discussion of attention is to support Peacocke’s claim that not only sensations, but also thoughts and other propositional attitudes (as well as episodes of trying to perform some action) have a distinctive phenomenology: they affect, as he would say, what it is subjectively like to be the subject in question. (They are apart of one’s ‘stream of consciousness.’) This has two parts.
1. He points out that attention to one’s environment can be interrupted by a thought occurring to one.
2. He points out that thoughts can be actions, and trying to do something has a distinctive phenomenology which seems not to be reducible to sensory awareness of any objects.
These both support the idea that some thoughts can be conscious, in Peacocke’s sense.
But there is also a more direct role played by attention, which Peacocke discusses beginning in ¤II.
Peacocke notes that, in general, conscious states can provide a subject with reasons for making a certain judgement or pursuing a certain action. A pain, for example, can give me a reason to pull my hand back; a visual experience as of a car rushing towards me can give me a reason to move out of the way; a visual experience as of a red table can give me reason to believe that there is a red table in front of me.
There are two different ways of understanding how such conscious states provide reasons for action and belief. (i) Such states cause you to form the belief/judgement that you are in that state, which in turn provides you with a reason for action or further belief/judgement. (ii) Such states themselves provide you with a reason for action or belief.
It is crucial to Peacocke’s account of self-knowledge that (ii), rather than (i), be the case. He gives the following argument for (ii) on p. 72: Experiencing pain can provide a subject with reason to judge that he is in pain. Moreover, we cannot gloss this in terms of the experience causing a belief (or other propositional attitude) which gives the reason for judgement; for what could the belief be? As Peacocke puts it: “Am I supposed to rationally reach the conclusion that I am in pain from the premiss that I am in pain?” So we must conclude that cases which fit model (ii) are at least possible, and so that the following general claim is true:
A conscious mental state can provide a reason for judgement without causing another mental state from which the judgement is inferred.
The next step in developing Peacocke’s model is to note that, if conscious sensations of being in pain can themselves provide reason for an agent to judge that he is in pain, the same is at least prima facie possible in the case of conscious propositional attitudes. Peacocke introduces the phrase ‘consciously based self-ascription’ (CBSA) as a name for any ascription by a subject of a proposition attitude to himself on the basis of (for the reason that) his having an occurrent conscious attitude.
He gives an example of a CBSA on p. 71. Someone is asked: ‘Whom do you believe was Prime Minister in 1956?’ The project is then to explain how that subject comes to have knowledge of his own belief on this score. Peacocke’s explanation comes in three stages: (1) The recollection that Dubcek was Prime Minister then. (2) The judgement, made on the basis of this memory, that Dubcek was Prime Minister then. (3) Since this judgement is a conscious attitude, it, like a pain, can itself provide a reason for judgement. In this case, it provides a reason for the judgement (self-ascription) that he believes that Dubcek was Prime Minister then.
This raises the question: is Peacocke giving a perceptual model of self-knowledge? The answer is: no. The key question here is: what provides agents with justification for their beliefs about themselves?
A perceptual theorist will give the same kind of answer to this question as he would give to a question about what provides agents with their beliefs about their environment. Just as perceptions of one’s environment provides one with reasons for beliefs about that environment, so perceptions of one’s mental states provides one with reasons for beliefs about one’s mental states.
It is crucial to note here that Peacocke never makes use of ‘perceptions of one’s mental states.’ It is not perceptions of one’s conscious mental states which provide reasons for judgements about one’s mental states, but the mental states themselves.
This can make the view seem puzzling. We can grant that a perceptual experience with content that p can justify a belief that p; but how can a judgement that p justify, not just the belief that p, but also the judgement that one believes that p? How does one, so to speak, get access not only to the content of the original judgement, but also to the fact that one has judged it? Needn’t one have a mental state in the content of which not only that p, but also one’s judging that p, figures?
This kind of puzzlement is the point of the discussion on p. 73. It is important to see why Peacocke must have a reply to this kind of puzzlement. The worry in the background seems to be this: one might think that the only kind of justification is inferential: the things which provide justifications are premises used in inference. (A further step would be to identify the premises available for inference with the contents of one’s beliefs.) The very idea that conscious states, as well as premises for reasoning, can provide reasons/justifications for beliefs/judgements is mysterious. We have already considered Boghossian’s argument that the view that basic self-knowledge is inferential leads to a vicious regress; since Peacocke is trying to give an account of at least one kind of basic self-knowledge, his account had better not involve the subject in question forming a belief on the basis of inference. In particular, it had better not portray the subject in the above example as inferring the judgement that he believes p from the belief that he judges p (for this would presuppose self-knowledge rather than explain it).
Peacocke’s reply is to defend the coherence of CBSA’s by analogy. Consider the case of perceptions giving rise to beliefs about the external world. It is plausible that the following are all true: (i) perceptions provide reasons for belief about one’s environment; (ii) in such cases the thinker must be sensitive to the distinction between having a perceptual experience that so-and-so and imagining that so-and-so; (iii) and yet the thinker does not take the belief that he has perceived so-and-so as a premise in his reasoning. A perceptual experience can provide a reason for belief without going via a belief about the perceptual experience. (Another argument is the case of pain, discussed above.)
This is supposed to help because it is supposed to allay the puzzlement about how a conscious judgement that p can provide reason for the judgement that one believes p: even though the fact that the thinker in question judges the relevant content is not a part of any of the thinker’s relevant mental states, the thinker can still be sensitive to that fact in the formation of further mental states. This is just what happens in the case above, or when one infers one beliefs from another: one has to be sensitive to the fact that the original state was a belief rather than a conjecture; but one can do this without forming a new mental state with the fact that it was a belief as part of its content.
Still, one feels like saying: there must be at least some minimal sense, even it does not amount to a belief, in which the thinker in question is aware of the fact that he is judging that Dubcek was the Prime Minister then. How else could he come to make the relevant self-ascription? And how is this awareness supposed to differ from a kind of ‘inner sense’?
We’ll return to this question when we discuss the dilemma Martin raises for Peacocke.
Peacocke conceives of his view as a middle way between two opposed views of the nature of self-knowledge. At one pole is the view that self-knowledge has an epistemology similar to perceptual knowledge: introspection is a kind of inner sense which allows us quasi-perceptual access to mental objects about which we then form beliefs. At the other pole is the view that knowledge of our own mental states is not based on any real kind of reasons, or justification. Peacocke thinks that each view is flawed, and tries to position his own view between them. It will therefore to be useful to sketch Peacocke’s view of these two polar opposites and his reasons for rejecting each.
Peacocke characterizes the internalist introspectionist as someone who holds that conscious attitudes are subjective states, that their contents are internalist, and that we know about them via introspection.
This is a bit puzzling at first. It seems that introspection and internalism are separable views; why does Peacocke saddle the introspectionist with internalism?
The reason is an argument from Boghossian, which Peacocke treats in kind of a peculiar way. The example of the coin, and whether we can perceive its extrinsic properties. As Boghossian notes, the property of being a coin is extrinsic, and yet seems in at least one sense to be perceivable. Nonetheless, Peacocke concludes that an introspectionist will for this reason have to be an internalist as well.
His arguments against internalist introspectionism would not impress a determined internalist; he relies on the claim that “it is a datum that we do know the full, ordinary, externally individuated content of our own thoughts ...without reliance on inferences from . . .something weaker” (78-9). While this seems plausible to me, a thoroughgoing internalist will try to explain this datum away.
Shoemaker’s arguments against introspectionism.
The view that our beliefs about our own beliefs qualify as knowledge in virtue their being reliably caused. The basic idea: one need not form a belief on the basis of reasons to gain knowledge; it can just be a matter of the causal process by which the belief is formed.
There are really two versions of this claim: one on which beliefs and beliefs about those beliefs are ‘distinct existences’, and one more like Shoemaker’s positive functionalist picture.
Peacocke’s main argument against this view is a general argument against reliabilist accounts of knowledge: if reliabilism is generally false, it would be odd if it could be invoked just as an account of the nature of one kind of self-knowledge.
There is a second kind of no reasons view which Peacocke occasionally mentions: the view that beliefs about oneself needn’t be based on reasons to count as knowledge because the facts about oneself known are not independent of the process of forming beliefs about them. This is a view often associated with Crispin Wright; we’ll take it up later in the semester.
One way to get a grip on the view is to recall Burge’s claims about self-verifying judgements like ‘I think: I am thinking about horses’ which are such that the act of thinking them makes them true. The extreme version of the Wright view holds that all mental states are a bit like that: thinking that one is in such a state can go some distance toward making one be in that state.
Peacocke’s central idea in this section is that both internalist introspectionism and the no reasons view gain an illegitimate appearance of plausibility from the ‘spurious trilemma’ that self-knowledge must be based on either introspection, inference, or on no reasons at all. If one has this in mind, and grants that inferentialist views of self-knowledge are a non-starter, then the failure of introspectionism might seem to count in favor of the no reasons view (and vice versa). But Peacocke thinks that this is an illusion: his view of self-knowledge provides a way out of the trilemma.
Peacocke claims that his own view is ‘an intermediate stance’ between these two views. In order to understand his view, then, we need to know what the axis of comparison between the introspectionist view and the no reasons view is supposed to be. By measure of what property is Peacocke’s view halfway between these two?
It seems to me that the idea is this. We are trying to explain how we can know that we are, e.g., thinking about philosophy. The no reasons theorist says that one has this knowledge even though one has no justifying reasons for the belief that one is thinking about philosophy. The introspectionist thinks that we have a reason in the fullest sense: one’s reason is that one is in a mental state which has as its content the proposition known (the proposition that I am thinking about philosophy). Peacocke’s intermediate position is that one has a reason for the belief about one’s own thoughts, but that it is not a reason in this full-blooded sense.
This suggests that Peacocke may be open to the criticism that Shoemaker brought against the broad perceptual model of self-knowledge. If we construe self-knowledge along the model of some kind of introspection or awareness, then we seem to have no explanation for the necessary truth that any subject with beliefs and sufficient conceptual capacities also has immediate access to her own beliefs.
Shoemaker and other no reasons theorists can respond to this demand for explanation by claiming that first- and second-order beliefs are not distinct existences. That is: they explain the necessary truth in terms of the nature of belief.
In our discussion of Shoemaker I suggested that it was at least possible that we could explain the necessary truth, not in terms of the nature of belief, but rather in terms of the conceptual capacities which figure in the definition of self-blindness. This is the path that Peacocke follows.
The key move that he makes is to note that the conceptual capacities involve possession of the concept of belief; and to note further that anyone who possesses the concept of belief must also have a kind of immediate access to facts about their own beliefs. To understand this, we’ll have to say a bit about how Peacocke conceives the task of giving a theory of concept possession.
To give a theory of concept possession is to give the conditions an agent must satisfy in order to have thoughts involving a given concept. The simplest kind of example of this is the concept of conjunction. Roughly, for C to be the concept of conjunction for an agent, that agent must be disposed to find the transitions from p and q to pCq and from pCq to p and q primitively compelling (in virtue of their form - but don’t worry about that part). For at least the concept of conjunction, it is very plausible that such possession conditions can be given - we simply would not count someone as having thoughts involving the concept of conjunction if they failed to satisfy these conditions.
The important point for our purposes is the sort of answer that possession conditions seem to give of certain kinds of epistemic facts. Note, for example, that it seems to be a necessary truth that anyone who grasps a conjunctive proposition p&q will find the transition from it to p primitively compelling. How are we to explain this necessary truth? Do we have some kind of privileged access to logical relations of this type?
Peacocke’s view of possession conditions seems to offer a deflationary kind of answer to this question. This is a necessary truth because anyone who lacked this knowledge would not grasp the concept of conjunction, and so would be unable to entertain the original proposition in the first place.
This is relevant to Shoemaker’s challenge to explain the impossibility of self-blindness. There we are asked to explain the necessary truth that anyone who has beliefs and grasps the concept of belief also has a kind of privileged access to her own beliefs. Shoemaker answers this challenge with a view of the nature of belief; Peacocke suggests that we can answer it via a theory of the nature of possession of the concept of belief. Anyone who lacked this sort of immediate access to her own beliefs would fail to have the concept of belief, and so would not even qualify as a candidate for self-blindness.
What, then, are the possession conditions for belief? It’s not so obvious that there are such conditions as it was in the case of the concept of conjunction. In his 1992 A Study of Concepts, Peacocke gave possession conditions for the concept of belief, a simplified form of which is as follows:
R is the concept of belief for an agent only if
(i) when the agent consciously judges p, she finds the thought that she stands in R to p primitively compelling (and does so because she consciously judges p), and
(ii) when the agent judges that some other agent a stands in R to p, she commits herself to the claim that a is in a state which plays the same psychological-explanatory role for a as would a state she would be in if she bore R to p
This first half of this possession condition is the important part for our purposes: it says that it is a requirement on possession of the concept of belief that one be disposed to make certain transitions from consciously judging p to endorsing the claim that one believes p. As Martin nicely puts it:
“On this condition there is no simple necessary link between two beliefs, but rather a necessary connection between two beliefs and the concept exploited in one of them: one would not be in possession of the relevant concept, were one not to form the one belief in the presence of the other.” (109)
And this seems enough to explain why it is a necessary truth that anyone who grasps the concept of belief has the kind of privileged access which consists in making immediate transitions from conscious judgings to judgements about belief in the contents judged. So Peacocke promises an explanation of the impossibility of self-blindness without appeal to the ‘no reasons’ view.
In terms introduced by Ned Block, we can distinguish two ‘kinds’ of consciousness. Very roughly, a state is phenomenal-conscious if it contributes to what, subjectively, it is like to be a given thinker, and is access-conscious if it is available for use in reasoning, action, and speech (intuitively, if we have immediate access to it).
It seems as though some states can be phenomenal-conscious but not access-conscious. (Cases of perceptual experiences to which you are paying no attention; the feeling that yu had been hearing that ringing noise for a while before you noticed it.) Peacocke suggests that attention to the phenomenon of first-person ascriptions of propositional attitudes shows that there can also be states which are access-conscious but not phenomenal-conscious.
But the cases on which this claim is based also prove problematic for Peacocke’s view. According to Peacocke, in the case of CBSA’s the fact that the judgement on which the self-ascription is based was conscious is important, and is part of the justification for the self-ascription. But now he points out the possibility of apparently similar cases in which there is no conscious judgement on which the self-ascription could be based. (E.g., the case in which I say ‘I know that my name is Jeff Speaks’ without first consciously judging ‘My name is Jeff Speaks.’)
In these cases, in which there is no intermediate conscious state (NICS’s), how are we to explain our justification for the second-order belief? The problem is that Peacocke’s explanation of CBSA’s is not available here; but it may well seem as though the two kinds of cases should be given a unified treatment. And this seems to count against Peacocke’s view.
Peacocke’s account of the justification of NICS’s:a self-ascription of a belief that p is justified (in the absence of a conscious judgement that p) iff the agent in question is also willing to judge that p. This is what Peacocke calls the ‘requirement of first-order ratifiability.’
On this view, as Peacocke notes, the kind of justification we have in the case of CBSA’s is more fundamental than in the case of NICS’s: NICS’s are justified to the extent that an agent is in a state which is similar in certain respects to the state that she is in in CBSA cases.
Martin presents an important challenge to Peacocke beginning at the top of p. 110:
“How, then, are we to model the kind of evidence of grounds the subject has for a higher-order state about that first-order state? In consciously thinking a thought, the subject’s attention is directed at its subject matter. So could the evidence that the subject has for consciously believing that also be evidence for her higher-order thought that she has that belief?”
The point is that, since Peacocke rejects the no reasons view, he must think that the self-ascription is based on reasons. But, Martin asks, what could these reasons be? In the ordinary cases of CBSA’s under consideration, the agent’s attention is focused wholly on the subject matter of the thought in question, rather than on herself. This suggests that the evidence in question must be evidence for the thought. But this, as Martin points out, cannot be right, for two reasons: (i) often we have very good justification for the second-order thought without such justificaton for the first-order state, and (ii) it is just mysterious how evidence about the world could provide any justification for a claim about the beliefs of an agent.
We have already seen Peacocke’s response to this kind of point: we have reasons for the second-order judgement, but not evidence in any normal sense. This is his point about the sensitivity of inferences to the fact that the premise of the inference is believed, even though the fact that the agent believes it is not a part of the content of the premise.
But Martin thinks that the challenge can be given a more general form which does not rely on any contentious link between reasons and evidence. Martins puts it as follows:
“We can restate the puzzle which motivated the ‘no reasons’ view thus: in relation to cases where the subject is self-ascribing beliefs which form part of how the world is for her ...the subject is aware of aspects of the world, and her attention is drawn out into the world. Yet, by directing her eyes outward, so to speak, she gains knowledge of her own mind. Why should this be so?”
Martin seems to be thinking that the no reasons theorist has an answer to this question. What it is isn’t so clear; but perhaps the idea is that the no reasons theorist thinks the the justification for the self-ascription along reliabilist lines, and so owes no account of how attention to the world can provide reasons for judgements about one’s own mental states.
Martin suggests another answer which seems to be open to someone who rejects the no reasons view: the view that “consciousness presuppose[s] self-consciousness.” (119) On this view there is a kind of self-awareness which goes with all conscious experience, and which provides the basis for self-ascriptions.
Problems with this view: (i) It seems that consciousness is possible without self-consciousness. (ii) It seems likely that this kind of view will land is with commitment to an irreducible and inexpliccable kind of self-awareness. (iii) It does not seem that far distant from perceptual models of self-knowledge. (iv) It’s not obvious that this provides any explanation of privileged access, above any beyond a restatement of what privileged access is.
Moreover, this is not Peacocke’s view. So how can Peacocke answer the question at the end of Martin’s quote? Or are his remarks about sensitivity to the attitude in which one stands to a content enough to defuse the puzzzle?