There It Is

Presenter: Benj Hellie, University of Toronto

Commentator 1: Jacob Berger, The Graduate Center, CUNY

Commentator 2: Jeff Speaks, University of Notre Dame

  • Get Jeff’s paper in .pdf format
  • Commentator 3: Susanna Schellenberg, Rutgers University

    Commentator 4: Heather Logue, University of Leeds



    1. I want to start off by welcoming everyone to this special session. As I said in my commentary, many thanks go out to Richard Brown for making this conference happen–and in advance to everyone participating in this session!

      Before the conference began, Benj had an opportunity to write initial responses to our commentaries, so I’m going to post those for him. I hope this is just the beginning of a really rich discussion!

    2. Here are Benj’s initial replies to Jeff’s commentary:

      Most heartfelt thanks are due to Jeff for his sympathetic and accurate exposition of the paper, and for his typically sharp and incisive criticisms: in light of the paper’s epic length, I’m nominating Jeff for this year’s Distinguished Philosophical Service Cross.

      Jeff asks what demarcates, from the class of all properties, those few that are predicates of my Lagadonian language of perception. The quick answer is: any property the instancing of which is the directing of sense-perceptual attention so as to select some fragment of the totality of what is perceived or sensed. How do such properties manage, when instanced, to ascribe their own instantiation of their instantiators? I don’t know, and am inclined to think that no reductive answer can be given. I plead good company in that any nonreductive representationalist must appeal to some irreducible bit of metasemantics. (Perhaps my theory has some advantages of economy in that vehicle and content are one and the same, and the ‘expresses’ relation is just identity.)

      Jeff wonders what it is to regard one of my odd sentences as equivalent to another, or to some more familiar sentences. Jeff rightly surmises that regard as equivalent does not involve any explicit decision in which the things so regarded are objects of attitudes. What is it then? The regard is more ‘implicit’ as it were. Suppose that Sam reasons (non-ampliatively, and ex nihilo) as follows: Q, so P. Here Sam’s attitudes are focused on the content of ‘Q’ and ‘P’, rather than the sentences themselves. In reasoning with the sentences, Sam reasons about their contents: in this little case, to the doctrine that it is metaphysically necessary that Q and P just if Q. Since she used the sentences ‘Q’ and ‘P’ in this reasoning, she regards those sentences as equivalent, in my sense. None of this requires explicit bestowal of meaning on those sentences, or even any real awareness that they are there. What is important is not having attitudes “toward” the sentences, as Jeff suggests, but rather having attitudes toward their contents, with the sentences. Those who think (a) that entertaining contents requires the tokening of vehicles of contents, and (b) that some notion of logical or conceptual equivalence or implication makes sense, should, I think, find my notion of regard as equivalent to be rather anodyne.
      (Jeff’s fn1 suggests that I am in two moods about my story: in one, my story is advanced as the literal truth about intentional psychological explanation as it applies to us (if in a somewhat more restricted set of circumstances than many philosophers like to acknowledge); in another, it applies only to the ‘ideally rational’, and can be used to shed light on us (perhaps) by exploitation of certain (unspecified) analogical features we bear to those creatures. Not so: the second mood doesn’t exist. Jeff reads my remarks on p62 as an expression of the second mood. But those remarks are not intended as my fictional story that I regard as a useful analogue to the truth, but rather as a true story about a fiction that is part of our actual psychological-explanatory practice. I am raising the accusation of instrumentalism against everyone else, rather than engaging in it myself.)

      Jeff rightly notes that my view of content is a ‘coarse grained’ or ‘intensional’ one, and reminds us of some familiar spectres haunting that approach. The paper hastily sketches some of my proprietary attempts at exorcizing these spectres: 34–6, 41–3. (By way of a swift broadside against ‘fine-grained’ approaches: I understand what it is to assess Sam’s credal state for rationality if by this is meant that her ‘world-as-pictured’ is intelligible. But a ‘world-as-pictured’ is like a highly blurry and somewhat inaccurate world — a set of possible worlds, we might say. By contrast, I have no idea what the rational significance could be of Sam’s bearing the B-relation to some abstract entity representable by an upper semi lattice of objects, properties, and sui generis entities whose names are homophones with the sentential connectives. Rationalizing such a story as providing the ‘metaphysics of propositions’ is more of a shibboleth than an explanation, in my view. Conversely, I don’t see why every quirk of psychology needs to be explained by intentional psychology. Sometimes we just screw up, failing to picture the world in any intelligible way: in that case, we need to see explanations elsewhere. Fine-grained theorists accuse coarse-grained theorists of being unable to solve various problems; but in my view this inability is a feature rather than a bug.)

      The last half-and-a-bit of Jeff’s comment presents an accurate and mercifully concise rundown of the paper’s case that the phenomenon of being taken in by hallucination can’t be fully explained by intentional psychology. Jeff’s central question here is why, in my view, a hallucination has representational content: this position has the consequence — in Jeff’s view, the cost — that the subject taken in by hallucination is inconsistent. (I hope I’m reading Jeff right here, and the complaint is not that on some other theory of the content of hallucination, things would look better.)

      Note first that if the aim is to avoid my conclusion that ‘in such circumstances, all bets are off from the point of view of intentional psychology’, eliminating all content from the hallucination (perhaps the approach Jeff recommends to the truly thoroughgoing direct realist?) would not do the job: no content, no intentional psychology. Nor would this position enable us to answer Jeff’s question about what hallucinating Sam has over hallucinating Sam*: neither Sam nor Sam* have any content to respond to; explanatorily, in both cases, we are back in the brute neurocomputational soup.

      Note next that calling someone inconsistent is not the worst thing in the world: as Jeff notes, it doesn’t mean they are blithering about with a hatchet on some berzerker rampage, but rather merely that there is a local collapse of intentional explanation. If Sam’s perceptual beliefs aren’t rationally explicable, her inferences from those beliefs or actions based on those beliefs might yet be so.
      Note next that simulation is a big part of the story about what justification is: if it strikes us as us-like upon projection into their shoes, that is a sort of justification (thus the long prelude in section 1 on the nature of that phenomenon). Indeed, we might even use the expression ‘rational’ in its ordinary somewhat loose sense and thereby say exactly what Jeff does about Sam and Sam* (‘Surely there is a straightforward sense in which Sam’s response to his dream is more rational than Sam*’s’). My story merely observes a degree of ‘grain’ in this ordinary usage — a mixture of intentional psychology and plain old familiar habit — so I’m not sure there is a bullet to be bit here. If I’m right that on my story, the problems of the philosophy of perception go away, it’s hardly a crippling defect that the story brings into relief a distinction in ordinary usage we sometimes overlook.

      So why /do/ I think hallucinations are contentful?

      (1) Continuity with the four or five other sorts of cases discussed in the final section.
      (2) We can imagine people — Churchlands, say — who habituate to judging how their brains are in various hallucinatory (or veridical, for that matter) cases: no false presuppositions there need crash the rationality of their judgements (this is the ‘fourth’ case on p53: I address the ‘third’ case of seeing but thinking one is dreaming — see Jeff’s fn 2 — on 53 as a central motive for moving from unconditional to conditional evidential policies).

      (3) Ultimately, though, Jeff is right to wonder about this — I notice that some crucial discussion motivating the approach got lost on the cutting room floor (I’ll have to put it back in for the final version!) — so thanks to Jeff for astutely noticing that the crucial argumentation had gone missing. Part of what is striking about perceptual states (whether veridical or otherwise) is that it is obvious that we are in them: Chalmers’s menu of replies to the dualist does not include the obvious error theory about qualia to the effect that we seem to have them but don’t. This suggests a view on which any perceptual state provides indefeasibly trustworthy justification for something or other, though it may require some interpretation to determine exactly what that is. This ‘something or other’, though, can always be (or perhaps always is) captured in a demonstrative judgement in ‘thought’ that is the same in content and perhaps even the same in vehicle as the perceptual state itself. In this sense, whether I am seeing or hallucinating, I can still be certain that /there it is/.

    3. And here are Benj’s initial replies to Susanna’s commentary:

      I want to begin by expressing my extensive appreciation to Susanna for this most magnificently learned set of comments, which elegantly put the finger on two of a very small set of the most fundamental issues with which the paper attempts to engage: in crisply zeroing in on these issues, Susanna’s comments display her characteristically impressive directness in sweeping aside considerations that others find perplexing in order to boldly advance on the points she finds philosophically most central. So thanks a great deal to Susanna.

      1. I begin at the end of the comment, by attempting to make clear a point on which Susanna’s final paragraph professes fogginess: this will provide the context for a painless treatment of the discussion of the propositional attitudes which comprises the remainder of the commentary.

      What Susanna finds less than entirely ‘clear [is] why things that the hallucinating subject does not have access to should matter for whether the subject is justified’. This question absolutely cuts to the heart of the matter, and it can be answered as efficiently as it was asked: they don’t; the subject /is/ justified. Alternatively: they don’t; the subject /has/ access to the things in question.

      According to a slightly less efficient answer, the subject /isn’t/ justified, but that is the downside of our ‘bargain’ with perception; the upside is that when we approach perception with presuppositions congenial to its nature, we receive justification for beliefs about the external world.

      Less efficient still: Susanna’s presupposition that the ‘things’ that ‘matter for whether the subject is justified’ are ones ‘the hallucinating subject does not have access to’ is no assumption of the paper. According to the paper, we do have ‘access’ to our perceptual states. When we receive them with presuppositions that are mistaken, we wind up in a rationally defective condition in which justification of the finest variety is unavailable, but a lesser sort is still to be had. When we do not, our condition permits justification of both varieties.

      The need for multiple conflicting answers will no doubt be a sign to the reader that Susanna and I are not entirely on the same page in our conceptions of ‘justification’ and ‘access’, so I will summarize the paper’s position on these phenomena. Each of Jeff and Jake in their comments raise, inter alia, similar points, so I will briefly repeat what I said there at perhaps greater length.

      2. On ‘justified’: my chief pedagogical aim in this paper is to evangelize for a methodological nostrum I have come to regard as of the highest significance in the philosophy of mind, namely the need for a ‘divide and conquer’ approach to elucidating the nature of psychological explanation. For too long philosophers have burdened the notions of rationality and content with weights these notions are unequipped to bear. We have asked these notions to overextend themselves, to perform tasks above their pay grades. Someone makes an error of logic, thereby endorsing over the long haul a picture of the world of which coherent sense cannot be made: perhaps even ‘nonculpably’ so. Why? Here my nostrum says: Don’t expect the answer to come from intentional psychology: the reach of an explanatory practice is limited to its peculiar conception of the ‘normal’ case; and for intentional psychology, the normal case is that in which synchronically one’s picture of the world is coherent and diachronically one’s picture of the world equilibriates toward coherence.

      This is why I give central place to a distinction stressed by Susanna’s advisor, John McDowell: that between ‘exculpation’ and ‘rationalization’ (my terminology: McDowell’s, ‘justification’). The philosophical notion of ‘justification’ — so it strikes me, as in use in the commentary — seems to blur these two notions. We find rationalization when some phenomenon yields to explanation by intentional psychology; exculpation when it does not but yet still to explanation in terms of arational habits we find in ourselves.

      My position on the subject taken in by hallucination then is this. First, the content of the perceptual state and the content of belief cannot be mutually satisfied: very roughly, the former says ‘I am hallucinating in a certain way’ while the latter says ‘I am perceiving veridically’. Thus, the subject is in a pair of states which, collectively, fail to encode a coherent picture of the world. So in attempting to understand how they update from this point forward, we cannot assume that these responses are apt to their picture of the world: which is to say that we cannot use intentional psychology in coming to such an understanding. Since comprehensibility through intentional psychology is the only sense I can give to the notion of ‘rationalization’, I find myself compelled to conclude that their belief updates cannot be rationalized.

      Second: despite this, we do find some updates more ‘natural’ than others as responses to the sort of situation they are in (Jeff’s contrast between Sam and Sam* brought this out nicely, I thought). My apparatus addresses this, if rather thinly, through the notion of ‘habitual like reaction’ (introduced on p59). The thought is that I find that were I in the subject’s position, I would update as they do. Not because doing so is an apt reaction to my picture of the world (although of course I would be under the impression that it is) but rather /merely/ because /that is how I roll/. So in that sense a contrast can be drawn between the ‘plausible’ and ‘implausible’ updates of the subject taken in by hallucination. Here, although intentional psychology grounds out, something more like pure sympathy is still available as an explanation. Sympathy provides understanding ‘from the inside’, and with it ‘justification’ of a sort: the sort I label ‘exculpation’.

      3. Now on the notion of ‘access’. My position is, to recall, that the vehicles of one’s perceptual state are contributors to one’s picture of the world: one ‘affirms’ their contents, in my proprietary jargon. This is so regardless of the veridicality status of this state. Since the content of the perceptual state consists roughly in a self-ascription of the vehicle, one always has access to the state: indeed, perhaps the best possible access one could have to anything.

      Now, when one has false opinions about something lying in the background, one will doubtless arrive at further falsehoods or pieces of ignorance in the foreground: this is a cat before me, but in the background I mistakenly believe myself to be in the land of dogs, and so in the foreground mistake purring for growling. What goes for cats and dogs goes for perceptual states and of course any externalia they may include.

      I don’t understand why it would be an attractive requirement on a notion of ‘access’ that it should yield more than I already build in: why anything ‘accessable’ must be so self-luminously the way it is that it lasers through all possible mistaken presupposition so that its essence presents fully to the intellection, shining its truth unmistakably forth. Scrub away the rhetorical excess and this blandishment may, I fear, be what Susanna is hoping for.

      4. Of course, one might endorse all this and still wonder why it is compulsory to say that when taken in by hallucination, our position is incoherent. Here is where the metaphor earlier of a ‘bargain’ with perception comes in.

      Direct realists say that in the good case, perception is a relation to externalia. I say that this is both vehicle and content. The justification works in part by a different vehicle — the vehicle of the justified belief — having the same content. But on my rather Carnapian picture, this is a ‘voluntary’ matter: one decides to regard the two vehicles as equivalent. This is something like an ‘analytic’ or ‘logical’ equivalence, rather than an ’empirical’ equivalence, which allows us to evade the regress skeptic asking after the empirical justification for basing belief on perception. So then if one bases a good case belief on a bad case perception, one is making a mistake in effect of logic: thus incoherent.

      It is perhaps slightly tragic that we put ourselves at risk of such incoherence, but the payoff is something like analytic truth for external-world beliefs in the good case. (Since the flipside of analyticity is contradiction, we should be unsurprised at the prediction that those taken in by hallucination are incoherent.)

      Now why would we want /that/? Well, perhaps if we knew of some other theory of perceptual justification that does not fall apart under the impact of a hard glance, the view I characterize might not seem so attractive. Still, we don’t, and so for that reason I imagine some might find the view worth taking seriously.

      5. Finally to Susanna’s discussion of the propositional attitudes. My paper employs a certain implementation and interpretation of that old warhorse, intensional semantics: in large part cribbed from Tractatus, Carnap, Stalnaker, Lewis, and Rayo; in small parts (inter alia, in the phenomenological interpretation discussed, and in the link I draw to ‘simulationism’) my own invention.

      The result that intensionally (more specifically, logically) equivalent sentences share semantic values is immediate, was built into the theory deliberately (Peter Sullivan’s papers contain exquisite discussions of the Tractatus’s philosophical motivations for doing so), and has been rather widely discussed: for example, in my paper at 34ff and 41ff. According to some — including Susanna, apparently — this is a bug. According to others, it is a feature: including me, for reasons the paper discusses at 30ff. Primarily because psychology involves calculation in the service of representation, where the service often does not go so smoothly: when not, representation ceases but calculation runs on. The rubric of my discussion above files explanations in such cases as ‘merely exculpatory’ as distinguished from ‘fully rationalizing’. The Tractatus tradition affirms this point, and so, quite sensibly, avoids overgeneration by neglecting to provide a semantical story to distinguish among such cases of failed representation. A couple more quick moves reach the doctrine that cutting intensionally is cutting exactly finely enough (where what exactly it is to cut intensionally remains of course open to dispute).

    4. And Benj’s initial replies to Heather’s commentary:

      Thanks a metric tonne (I think that’s how we spell it up here in Canada) to Heather for the wonderfully clear and incisive comments. Heather’s comments reflect the position in the paper with a perfect balance of detail, accuracy, and crispness: for such a gifted exegete, expositor, and critic to present one’s views in a way that squares exactly with how one sees them oneself is a most gratifying experience indeed. Heather’s central concern — a point discussed also by Jeff and Susanna — is sharply stated and to the point, and she makes an admirable go at chasing down its source in search of a friendly repair. In light of the paper’s length and intricacy, the effort Heather put in is to be thoroughly appreciated indeed. Also, before beginning this final reply to my commentators, one last word of appreciation to Jake for putting the event together: thanks Jake.

      1. This is a good context in which to take the view from 50,000 feet, and thereby highlight the structure of the view in a way doubtless obscured amidst the paper’s extensive detail, apparatus, motivation, polemic, and stage-setting.

      Let’s suppose we have a fondness for the Tractatus-Carnap-Stalnaker-Lewis tradition in rational explanation. Then we think that that which rationalizes or is rationalized does or is so thanks to its *possible worlds truth condition*. Rationalization is a matter of /inferential validity/ or perhaps validity /granting of what is presupposed/: if S supports T (let these be tokens), then ||T|| (the set of worlds associated with T) is no more specific than ||S|| — perhaps somehow de jure and therefore independently of what is presupposed, perhaps /merely/ in light of what is presupposed. This fondness for validity manifests the rather Cliffordian bent of the tradition: whatever rational support may be, it had better not be able to lead us astray.

      This Cliffordian bent shows up also in the Tractatus tradition’s treatment of that which can be introduced into a train of thought ex nihilo: of ‘axioms’, so to speak. Since axioms are basic they’d better not be able to ‘direct inject’ falsehood; so any axiom had better be infallible.

      At one extreme lies the line on logic and metaphysics. Those doctrines in these departments an individual accepts are regarded by the Tractatus tradition as /meaning postulates/: doctrines stipulated by the individual to serve the purely calculational needs of the system through which they draw up a picture of the world, and therefore without representational content. Why? We need them, so eliminativism is unacceptable; infallibility through triviality reflects the ‘maximally general’ role these particular sorts of axioms play in theory; and we are unconstrained in our choice of them, so they’d better be subject-relative.

      At the other extreme lies the Tractatus tradition’s line on perception. Perception had better direct-inject substantive representational content (after all something does), so the infallibility had better be due to some sort of special relationship to its subject-matter.

      2. Now, if perception is infallible, why isn’t perceptual belief? We can imagine two styles of answer here — one traditional, another discussed in my paper — where the difference concerns what it is that perception is infallible /about/.

      The Tractatus tradition is (often) /internalist/: perception is infallible about one’s inner condition. Perceptual belief therefore /is/ infallible: but it is not very interesting. What is fallible is the interesting stuff: beliefs about externalia. These are fallible because we make ampliative inferences from boring perceptual beliefs about our internal conditions to interesting perceptually-grounded beliefs about their external causes. This approach faces familiar worries about psychological reality: we are unaware of the inner states; we have no idea about their causes; the necessary theory does not seem to be in place.

      My answer is /externalist/. I say that perception is infallible about the externalia that we are interested in, and rarely if ever exclusively concerns our external condition. If perceptual belief is not to be infallible as well, it had better be a different kind of animal than perception. Well different how? Difference in content seems like a bad idea: what are these magical facts that perception can but belief cannot represent? So instead I say the difference is one in vehicle. This is a Tractatus tradition friendly manoeuvre because of the tradition’s segregation of representation and computation (a point I discuss further in my reply to Susanna). When perceptual belief is false, this is due to a computational failure — akin to the running of a fallacious derivation.

      3. OK, to Heather’s central question: so what about this incoherence stuff? When someone’s perceptual beliefs are not merely /false/ but indeed /incompatible/ with the content of their perceptual state, this makes them incoherent. When does this happen? It depends on what those contents are of course, but let’s suppose they include the colors of seen objects. If so, then one can become incoherent when taken in by an illusion: perceptual content ascribes whiteness (perhaps along with redness of illumination); perceptual belief ascribes redness. Let’s suppose nothing could be white and red; if not, the person is incoherent.

      Incoherence emerges more broadly thanks to the way in which perception relates to perceptual belief. My story includes something maybe a bit like ‘recognitional concepts’: belief sentences with contents linked de jure to certain perceptual sentences by meaning postulates (‘evidential policies’, I call them); for a neutral example, suppose ‘*R* iff lo, a red thing!’ is analytic for Sam. Now suppose Sam is taken in by hallucination: she responds to a perceptual state in a certain case by accepting ‘lo, a red thing!’, despite tokening not ‘*R*’ but ‘*RD*’ (assuming that hallucination involves a similar sort of Lagadonian language: more on this in my reply to Jeff). Now, since Sam tokens ‘lo, a red thing!’, which, for her, is analytically equivalent to ‘*R*’, fulfilling her full budget of semantic commitments would seem to require tokening ‘*R*’. She doesn’t; and so in that sense falls short of a fully coherent world-view (one part of her mind accepts that /lo, a red thing!/; another part fails to accept it; these attitudes are not compatible toward the same content). Just a bit of metaphysics takes us further. Grant that if a state is of a determinable kind, it is of some determinate of that determinable. In that case, it can’t be that states of *RD* are a little bit vague, while states of *R* are just more determinate; rather, *RD* and *R* are incompatible kinds: it is impossible to be in both. So then Sam’s token of ‘*RD*’ has a content which is inconsistent with that of her token of ‘lo, a red thing!’, and so Sam’s picture of the world is incoherent in the most straightforward way.

      4. Put as bluntly as possible: in the good case, perceptual beliefs have a status we could think of as ‘contextually analytic’: Lagadonian sentences are semantically guaranteed true when uttered, and perceptual beliefs are their analytic equivalents. In the domain of the absolutely analytic, falsehood suffices for absolute incoherence; so we should not be surprised to find a quick path from falsehood to contextual incoherence in the domain of the contextually analytic.

      The internalist tradition in perception draws back from these extremes. If there is a benefit, it is that falsehood is declared less bad than (contextual) incoherence; but the cost is that truth is less good than (contextual) analyticity. That is what it means to say that perceptual belief acquires support from perception by ampliative inference. Those with Cliffordist leanings will regard such support as no support at all.

    5. Benj’s initial replies to my commentary:

      Many thanks to Jake for his probing and wide-ranging commentary. Jake raises many crucial questions pertaining to how the view developed in the paper relates to swathes of the existing literature: both in the philosophy of mind, especially on the more ’empirical’ side, and in epistemology. Having already expressed my extensive gratitude for Jeff Speaks’s reactions, I should highlight that I find Jake’s remarks especially valuable for the notable contrast in perspective from which they approach the paper. The opportunity to receive a constellation of responses is the distinctively valuable feature of a ‘symposium’-type session for this sort, so I should also express my thanks to Jake for putting the event together.

      I. Belief and consciousness

      Jake begins by asking: ‘I’m specifically interested in whether Benj thinks there can be justifications or rationalizing explanations holding between mental states that occur in the absence of consciousness’. Jake has just quoted my remark very early in the paper that ‘the most basic interpretation of the claim that A justifies B means something close to: from the first-person perspective, B was required to maintain coherence of the stream of consciousness in light of A’ (p. 2), so the reader will not suffer the disappointment of a failure of nerve at this very early stage: in a word, my answer is No.

      Let us consider an example of the sort the Freud allusion suggests Jake may have in mind. Chris, let us suppose, is looking for love in all the wrong places. As it happens, Chris’s appropriate-gender parent withdrew the lavish affection to which Chris had grown accustomed just at some crucial juncture in Chris’s maturational process. A therapist looking at Chris might fill out the picture by postulating a belief that the search underway will eventually culminate in a reinstatement of the lost security of envelopment by a loving authority. An ‘unconscious’ belief, mind you: Chris is neither especially articulate nor especially informed about prevailing motivational patterns in our species, so Chris would neither spontaneously affirm the alleged credo guiding the social activities at issue nor perhaps even understand it (though certain neighboring caricatures of the credo would likely be rejected with explosive disgust).

      Here I want to highlight a doctrine of the utmost significance for the philosophy of mind: *belief is spoken in many ways*. As I consider it, my silent affirmation of my sentence ‘the earth moves’ is a reflection of my belief that the earth moves; Galileo believed the same thing but never heard the English sentence (to my knowledge). Alternatively, my belief about where the nearby grocery store is located lie — under most circumstances — in the dispositional: in my habits in shopping. However, when actually on the way for milk, the belief rests in the default pathways I find in my environment: crosing Ossington is unnatural and repellent; as is turning back south on the eastern sidewalk; but heading up the hill to Dundas, crossing the street, taking a right, and walking a certain distance before entering the store is a course of actions (and a stream of consciousness) that ‘flows’ coherently. To believe that P, no distinctive sort of articulation of the belief is required: merely a ‘picture of the world’ according to which, somehow, P. My picture of belief is Stalnakean rather than Fodorean (the talk of ‘sentences of belief’ in the paper is perhaps somewhat misleading along these lines: I think of the ‘sentences’ at issue as just whatever are counted as the vehicles of the content in the relevant case). For the most part, beliefs will be anchored in patterns of affect or dispositions thereto rather than in any sort of explicit linguistic articulation.

      So we could say that Chris’s social credo is as it were ‘spoken’ in what seem to Chris to be natural reactions to various situations: the prospect of spending an evening at home reading with the cat carries an affect chilly with isolation; the affect of hitting the bar, by contrast, is rich with the warmth of social congress (thanks to the prospect of restoring what was lost in childhood); and this is why Chris hits the bar. In saying ‘Chris believes that P’, we are attempting to articulate linguistically a picture of the world that corresponds as closely as possible to the picture guiding Chris; the linguistic vehicle of this articulation need not in any way mirror in its intrinsic character the dispositional-affective vehicle of Chris’s belief.

      What about priming? My right ear hears ‘palm’; my left ear hears ‘finger’ (‘oak’); focusing attention to the right, I freely associate ‘hand’ (‘tree’). Why this pattern?

      Here we come to another doctrine I regard as of rather significant importance in the philosophy of mind: *explanatory burdens should be shared*. We try to get rationalizing explanation to do everything; but we shouldn’t. (Tamar Gendler’s notion of ‘A-lief’ is reflects a similar insight, though A-lief is a Thermidor: my approach is more radically separatist.) Sometimes reactions are entirely explicable along rationalizing lines; sometimes entirely inexplicable; sometimes a mix. I think the present case is a mixed case. I say something semantically similar to something in my lexicon phonologically realized as ‘palm’. That is the *associative* part of the task. That is rationalizable. What in particular I say goes this way or that way because of what ‘pops into my head’. That is the *free* part of the task. That is not rationalizable. This division of the aspects of my reaction squares with the phenomenology of the case. The metaphor of something’s having ‘popped into my head’ suggests something somewhat alien: somewhat incongruent with the remainder of the stream of consciousness. Why this? I dunno! The answer is inaccessible from the first person, tho perhaps the cog sci crowd will have something enlightening to say. So

      Nutshell: in the Freud cases the beliefs rationalize and are ‘conscious’ but we have to be careful what we mean by ‘conscious’; in the priming cases we see a partial rational explanation by the conscious together with a promissory note for a more detailed ‘cognitive’ or nonrational explanation by the nonconscious.

      II. Blindsight and rationality

      Now to blindsight. To begin with, I’m not sure the explanation isn’t just that the person is echolocating, something we can all do though most of us are not very good at it — if some are able to rollerblade gracefully about their culs-de-sac. (What is it like to be a bat? You already know.) Alternatively, either Chris Mole or Sebastian Watzl once informed me that blindsight may involve a visual perceptual state with a vastly lower signal-to-noise ratio than ordinary: too low for phenomenologically unsophisticated subjects to articulately explain what is going on, but high enough to enable crude discriminations of Xs from Os and the like. In either case, I don’t think anything especially essential in my story would require alteration.

      To see a more interesting possibility, consider an analogy. Some who can do can teach but not all can; those who can differ in the measure of various ingredients in their pedagogy: explicit articulation, metaphor, leading by example. For example, while many superb musicians can’t explain what they are doing at all, many others can cite chapter and verse of Schenker’s /Harmony/, and some say things like ‘people like to see a saxophonist’s fingers move’. This is of course independent of whether these or those aspects of the sensorimotor stream are present within consciousness. Engaged action is one thing; explicit articulation is another; both can be elements of the stream of consciousness. So maybe the perceptual beliefs of the patient in Jake’s example continue unaltered in guiding his engaged travels through the environment, despite the tragic loss of his capacity to make these beliefs articulate.

      Finally, I could plead that this is a case in which we should parcel out explanatory burdens. This is of course something of a Hail Mary and the explanatory significance of my views diminishes with increased appeals to the line, but it is perhaps not out of line in this case. Rational explanation is built for the normal case and this is not the normal case; to the extent that (as I and many others think) rational explanation involves a substantial ‘simulational’ component, our grave difficultes in understanding what things are like for the blindsighted may suggest that we have exited the range of application of rational explanation. A nonrational explanation perhaps compatible with aspects of my views not under discussion in this paper might be that the sensorimotor complex can go about certain articles of its business without much intervention from consciousness (why, going up the stairs, does your back foot never catch on the stair past which it is being raised and on which your front foot rests?; how, driving my standard transmission VW Cabriolet through the rolling hills of the Finger Lakes region, did I manage to hit all those shifts while at the foreground of my attention were such engaging philosophical conversations?). For TN, consciousness perhaps issues a self-command ‘walk over there’ and then the sensorimotor complex goes and carries the command out. The sort of careful or deliberate intervention of arbitrary levels of precision familiar to students of music would perhaps be impossible for TN in performing this task, but it could be accomplished crudely. We could flesh this out by making an appeal to a notion of ‘attention’ as a sort of selection from among the sensorimotor stream and a stage-level notion of the ‘attentiveness’ of an action over a period in terms of its constitution by the selected part of the stream. Our thought would be that TN traverses the obstacle course but does so as inattentively as I shifted gears on those rolling hills and as a result contributed no rational input to the fine structure of the means by which the traversal was carried out.

      Nutshell: either (A) TN’s actions are based in the entirely normal way on perceptual states distinct from the usual ones; or (B) they are based on the usual ones though other activities we ordinarily perform based on those usual ones (thinking and talking about the perceived) are no longer possibilities; or (C) what we are seeing is a mixture of (nonspecific) action and something more like the (indefinitely finely-grained) carrying out of a program, where the former is explicable by the rational but the latter is not.

      III. Consciousness and the mental

      Jake asks whether I think ‘mental states’ can occur without being ‘conscious’? Yes, all of them do. Or I don’t know what is being asked. (Notice that my paper does not use ‘conscious’ as an adjective.) I think that the experiences are the actions: a bit more precisely, the ‘actions-in-the-world’, actions laden with pictures of the world (‘beliefs’) and constituted by the attentive bit of the sensorimotor stream. Actions are processes; the sensorimotor stream is a sequence of states; a belief is of course a state. If you want to know what it is like for one at t, ask which variety of ‘action-in-the-world’ one is undergoing at t.

      Are there mental occurrences of other sorts? Sure, why not. I don’t want to butt heads with Chomsky. But they aren’t ‘intentional’ in any literal sense: the notion of an intentional state is the notion of a state in which one pictures or regards the world as being a certain way, and is therefore a phenomenological notion. Taking the intentional stance can be useful in understanding, say, language processing, but let us remember that it is a /stance/.

      IV. Blindsight and the ‘third person’

      Jake suggests:

      it would seem that TN’s actions are perfectly rational from the third person. TN is walking along, his eyes focus on a barrier, and he moves around it. Why did he move around it? Because there was a barrier. And it’s clear that the nonconscious beliefs formed on the basis of the nonconscious perception must interact in ways with his standing beliefs. Presumably TN believes that one cannot walk through a barrier, which, coupled with his newly formed belief that there is a barrier, causes him to alter his course.

      I’m inclined to be more hesitant about whether these actions are rational ‘from the third person’. In my view, this would require that I find them rational ‘from the first-person’ upon satisfactory sympathetic projection into TN’s position. Since (as above) I don’t know how to project myself into TN’s position, I decline to pronounce on whether I find his course of behavior a course of entirely rational action. If I am more cautious on this point than Jake, it is because while it is perhaps ‘clear that the […] beliefs formed on the basis of the […] perception must interact in ways with his standing beliefs’, the unclear status of those ‘standing beliefs’ renders it unclear whether the alteration in course is the commencement of a new (rational) ‘action’ or a new (arational) ‘motor program’. This goes with a correlative unclarity over whether the adjective I have elided applies in any reasonable sense to the occurrences in TN under discussion. Just to be explicit, options (A) and (B) reject Jake’s assessment of this status and declare the alteration of course to be a (rationalizable) ‘action’; option (C) accepts Jake’s assessment but declares the alteration of course to be an (arational) ‘program’.

      V. Consciousness and the first-person

      Jake considers an approach somewhat (perhaps not exactly) like (C), and makes the intriguing suggestion that ‘It’s not clear, however, that states outside of consciousness do not count as being from the first-person perspective’, meaning (as I understand him) to agree with me in tieing rationality to an expansive ‘first-person perspective’ but to disagree in severing it from a relatively meager ‘stream of consciousness’: if so, what option (C) labels TN’s arational ‘program’ would instead count as a rational action. Option (C) therefore would not serve its purpose of rebutting the assault on my consciousness-rationality link by pleading arationality.

      What does Jake mean by distinguishing ‘consciousness’ and ‘the first-person? In context, Jake has just quoted my remark that ‘the notion of ‘from the first-person perspective’ is not exactly the same as ‘according to the subject”. If am not entirely clear what Jake has in mind here, perhaps this is because he was not entirely clear what I had in mind /there/: perhaps blamelessly so. The picture I intended is something like this (and is the result of the extraordinary contortions into which fully accommodating the extent to which consciousness is transparent requires the theorist to bend). By ‘according to the subject’ I have in mind a notion related to the subject’s best attempt at an articulate account of things. Such an account would be prey to disruption by undischarged incoherence: according to one aspect of Sam’s picture of the world, P; according to another, Q; but P&Q |– the false. Sam is imperfect and so fails to notice this; as a result, Sam builds on it, inevitably reaching false conclusions; thanks to transparency, this falsehood reaches inward, resulting in self-ignorance — perhaps about P and Q themselves: Sam’s explicit, articulate sense is that P and ~Q but rather R, say. ‘From the first-person’, by contrast, is a sort of theorist’s ‘cleaned up’ version of ‘according to the subject’: how are things for various accurate and coherent so therefore partial *simulations* of Sam? Our ‘confabulated’ proposition R makes it into the story only later, and we are more liberal than Sam herself in ascriptions of arationality. We would say something like ‘from the first-person, P; and from the first-person, Q; and [blah blah blah about the weird structure resulting from the incoherence including R]; and, from the first-person Sam is fragmented, self-deceived, etc.’ (This is in essence the story I tell about hallucination.)

      Now, does this story allow for occurrences ‘outside of consciousness’ but nevertheless ‘from the first-person perspective’? Well perhaps so: perhaps the acceptance that Q — and, indeed, the fragmentation — are candidates for this status.

      Within my framework, Jake’s suggestion would be read as meaning that TN’s visual perceptions should also be treated in this way. And, within option (C), this would mean that my shifting of gears, moving my back foot just so to avoid tripping on the stairs, and so forth, should be treated in this way. I find this unattractive, since I don’t think there is any connection between rational incoherence and inattentive perception: inattentive perception, to reiterate, works its magic /arationally/.

      This can seem a bit like a shaggy dog story in that other attractive treatments (options (A) and (B), recall) of the case under discussion are available (though ultimately it is surely an empirical matter which story is correct). But at root is a deep and significant issue raised by Jake’s request for an explanation why ‘a better gloss of ‘A justifies B’ isn’t simply that B was required to maintain coherence /for one/ in light of A’ (better, recall, than my gloss in terms of coherence /in the stream of consciousness/). Well, the answer is that there isn’t any (comprehensible) difference here. Two lines of support are available here:

      1. What might /coherence/ for one be? I have suggested above (and I argue in the paper) that among the things assessable for coherence are the /hypotheses/. But a hypothesis is roughly ‘the world according to a certain picture’, and the notion of a picture of the world is at root phenomenological: we come at it via sympathy with or projection into the mind of the other, where the target of projection is the stream of consciousness.

      2. What might coherence /for one/ be? We can give synonyms: coherence as it were ‘by one’s lights’, where (to operationalize a bit) ‘one’ is understood as a candidate referent of ‘I’ or ‘you’ — a person or an agent: perhaps most generally, a subject. What kinds of things might these subjects be, though? To wax Cartesian, what kind of thing am I? Short answer: a subject is a substance ‘hosting’ occurrences with subjectivity, where subjectivity is the most important notion people mean by ‘consciousness’.

      There is of course a more lot to be said on these points.

      VI. Attention and consciousness

      Jake wants to know what I think about the prospect of ‘attention in the absence of consciousness’. Short answer: both notions here are pretty obscure. Longer answer: the notion of ‘consciousness’ can be teased apart into (a) subjectivity (b) attentiveness (c) attention (d) sensorimotor quality. (c): attention requires itself. (d) attention is a selection from sensorimotor quality and so requires it. (b) while attentiveness of an action over an interval requires constitution of that action by the stream of attention, what is important here is the converse: I’m somewhat inclined to think I can conceive of that via a sort of intricate quasi-zombie case. (a) subjectivity is composed out of candidates for attentiveness so what goes for them goes for it.

      There’s also a lot more to be said on these points.

      VII. Rationalization and bad cases

      Jake asks in closing how widespread rationalizability is. He worries that since there is a lot of change blindness there are a lot of bad cases, hence a lot of failures of rationalizing explanation. Here I reiterate a point Jeff made (and that I affirmed in my reply): such failure is usually local, partial, temporary, and easily contained. Change-blindness is especially easily contained in that the changes in question are usually entirely orthogonal to what we are about and hence stand in no connection to the explanandum. We’ve all been frustrated by staring at the slide for long minutes before discovering the shifting rail, but this sort of encounter is not exactly the stuff of daily life.

      About the blind spot, I say good thing they didn’t connect the optic nerve at the fovea: then we’d really have trouble!

      The Matrix is an interesting case. You Chalmers fans out there may have detected the possibility of rolling out ‘The Matrix as metaphysics’ in the context of the 1D, Millian intentional semantics I sketch in section 2 (see p42 for some especially relevant stuff).
      Chalmers advances his position as sort of Kantian: the Matrix shows that the phenomena, given in the primary intension, obscure the noumena, given in the secondary intension. My view one-ups the Kantianism: Chalmers’s PI does not capture ’empirical realism’ because it represents Hesperus as ‘that thing which manifests phenomenally as yada yada’; and the ‘transcendental idealist’ wouldn’t want the SI in his repertoire of psychologically explanatory devices.

      A bit less obscurely, my position is that the Matrix inhabitants don’t regard themselves as in the bad case: rather, they regard me as in the bad case, so in calling them arational I’m getting as good as I’m giving. And of course nothing prevents me in understanding them from just bracketing my belief that they are in the bad case, and thus being able to give what are formally perfectly fine rational explanations.

      VIII. Bad cases and uncertainty

      Jake concludes by observing that we are always partially in a state of uncertainty about being in the bad case. As a psychological claim, that is false, I think: we usually assume that our current preparations are good enough for our ongoing actions to succeed (so that there are no significant delusions) and push forward as if we are in the good case. As a normative claim, the claim that we should do otherwise seems dubious: acting out of certainty has tremendous pragmatic advantages, and I’m a Jamesian voluntarist about epistemic norms.

    6. I want to thank Benj for his terrific and extensive reply to my commentary. In it, Benj clarifies what he counts as being part of a subject’s stream of consciousness–and I thought I’d kick things off here by asking a question about that.

      I’m going to adapt Benj’s example of Chris. Chris looks for love in all the wrong places. His therapist hypothesizes that he does this because he believes he’s inadequate. Chris denies this, though–he says that he does not believe that he’s inadequate. And I agree with Benj that to “believe that P, no distinctive sort of articulation of the belief is required,” so I don’t think that Chris must say that he’s inadequate in order for Chris to believe this–his behavior, as it were, says it for him. But Chris says he doesn’t believe it. What to make of this? It seems to me that Chris is wrong: Chris believes he’s inadequate but doesn’t believe that he believes it.

      It’s clear that Chris’ (false) meta-belief is in his stream of consciousness–he expresses it in speech. But a natural thing to say is that Chris’ belief that he’s inadequate isn’t.

      That’s why I’m a bit confused by Benj’s claim that “in the Freud cases the beliefs rationalize and are ‘conscious’ but we have to be careful what we mean by ‘conscious’.” Does Benj want to claim that both of Chris’s beliefs are in his consciousness, but only the meta-belief is explicitly articulated? That seems quixotic–especially in the face of Chris’ fervent denial that he believes he’s inadequate. Or is Chris’ belief that he’s inadequate not in the stream of consciousness, but just not capable of entering into rationalizing explanations of his looking for love in the wrong places?

      It seems to me that both believes can be rationally characterized, even though it’s plausible that only the meta-belief is in Chris’ consciousness.

    7. Hi Jake, in answer to the question you pose, my view is closer to the former than the latter. But there are some presuppositions we don’t share. The notion of ‘belief’ in the paper is the ‘update’ or ‘test’ notion on which a belief-avowal and an affirmation of its content are first-person equivalent (and so in particular that BP and BBP are first-person equivalent): I talk about this in 1.1 subpoint 8, and further in 1.2. This view would say that Chris self-ascribes inadequacy and self-ascribes adequacy, but in different ways and in different fragments, and avows each belief at every level. The ascriptions are differently coded and exist in different ‘fragments’ so the incoherence doesn’t wind itself up immediately.

    8. Questioning the whole idea

      There’s something wrong with the way that perception is being looked at. In particular, I question the notion of veridical perception.

      To say that perception is veridical is to say that it conforms to some standard of truth. But what could be the source of such a standard? One might suppose that such a standard is part of a person’s genetic makeup. But this seems doubtful. It is hard to see how the DNA could have sufficient capacity to be the carrier an adequate truth standard. Moreover, the vagaries of mutation, development variabities, plasticity of the brain, etc, only serve to make it even more implausible that there could be such a standard transmitted through the genes.

      A creationist might assert that the standard of truth comes from a creator. But that doesn’t really solve the problem, for there would have to be a means available for a cognitive agent to consult this divine oracle, and there is no evidence to support such an idea.

      Yet another possibility is that the truth standard comes from the culture. But it is hard to see how that could work. For a child needs to already have a reliable perceptual system before it can make contact with the culture. So if perception needs to be veridical, then it must be veridical before the child can absorb ideas from the culture.

      My own view is that no truth standard is needed for perception. I see truth as tied to language. The child’s use of language needs to approach being veridical if communication is to be possible. However, there is plenty of time for the child to pick up a standard of truth from the culture for language use. However, I don’t see that “veridical perception” even makes sense.

    9. Neil, I hope you don’t think you’re objecting to /me/ here. I don’t have any use for (or real comprehension of) the notion of veridicality either: that is part of why I am a direct realist.

      Notice that ‘veridical’ has a rate of occurrence in my paper of once every 8,000 words; there as in its four occurrences in my comments here ‘perceives veridically’ is a rough and ready equivalent to ‘sees in the normal way’ and contrast to ‘is dreaming’.

      More grist for your mill: how exactly is positing veridicality supposed to help explain perceptual justification? Friends of this style of approach have never really explained it to my satisfaction. Is the story just one of good old ampliative inference? Then why does it matter whether perception can be veridical? I can make ampliative inferences from opinions about qualia, sense-data, or what have you.

    10. My comment was general. Several of the commentators used “veridical” as a way of discussing justification.

      I’m not sure that justification makes any sense either. If I report seeing something, and I am asked to justify that, I would probably tell the questioner to go look for himself.

      Sure, I might sometimes wonder about what I think I am seeing. But my normal reaction would be to take a closer look, rather than to try to decide whether the initial perception had been justified.

    11. Neil, I suspect we are in agreement here as well, though perhaps less strongly so. The notion of ‘justification’ is obscure in the extreme; the main reason the paper is so long is my attempt to formulate a notion that makes sense and has some significant theoretical role.

      In a nutshell, the view is that if there is a role for considerations of rationality, coherence, and the like in philosophy of mind it is in a certain maybe default (though not hegemonic) style of explanation of updates of one’s way of ‘being in the world’ — one’s take on how the world is, the actions one performs in light of this take. Roughly, updates preserve coherence ‘from the first person perspective’ in a certain somewhat refined sense treated in the exchange with Jake.

      So for example, I judge ‘this is red’. Where did this judgement come from? From perception of course. According to the framework there should be a first-person logical connection of some sort between the perception and the judgement. Then the question is ‘what is this connection?’. That is what the paper is about.

    12. Hi Benj. Thanks for the follow up about beliefs and the stream of consciousness. I agree that there are some presuppositions we don’t share, but your response really clarified things. I suppose one thing I’m worried about, then, is that it’s not clear what could demarcate mental activity that’s conscious from that which isn’t. But perhaps isn’t so much of a concern because, on your view, anything that’s really intentional (and not just regarded as intentional in a ‘stance’ way) is conscious. I’m not sure I agree we should think intentionality is a phenomenological notion–and it seems rather ingrained in, say, folk psychology that there are nonconscious intentional states–but that’s a big issue that I’m happy to set aside for now.

      But I’ve been wondering about something else too. In your exchange with Jeff about the Lagadonian language of perception, Jeff asked what demarcated the properties of a person that are predicates of the language. You replied that “[t]he quick answer is: any property the instancing of which is the directing of sense-perceptual attention so as to select some fragment of the totality of what is perceived or sensed.”

      You then said that you didn’t have a story for why those properties thereby represent themselves, but you said that wasn’t a problem because you think that it’s fine for such an account to include an “irreducible bit of metasemantics.”

      So I’m not sure I totally understand the problem, but it seems to me that your account doesn’t have to regard this representation as nonreductive. Why don’t those properties’ functional roles (i.e., in selecting fragments of what is perceived or sensed) determine the properties’ contents, as a conceptual/functional-role semantics would hold? Is the worry that it’s not clear why having that role means the property represents itself, as opposed to, e.g., the fragments of what is perceived or sensed?

    13. Hi Jake, on the metasemantics stuff: maybe one thought is that to the extent that doing the metasemantics (foundational semantics? the terminology is used in both ways) turns out to be hard we can always plead irreducibility thanks to the involvement of consciousness. Another thought is that we have additional resources to go with thanks to the involvement of consciousness. I’m sort of of the opinion that talk about inferential role is often a sort of euphemism for consciousness, or maybe a way of trying to get at what consciousness in fact does when hampered by an impoverished explicit theory of what consciousness could do. Think of Peacocke on ‘primitive compulsion’. That sounds pretty phenomenological. According to me for an inference to be primitively compelling is for it to be required for the maintenance of coherence in the stream of consciousness. This story has the advantage of (1) helping to operationalize Peacocke’s notion (2) making his meaning explicit (if that is what he intended) or advancing something similar that is comprehensible (if it isn’t) and (3) bundling the notion in with a story about the practical domain.

      Quickly on the stuff about folk psychology: I agree with the letter if not the spirit of your view. The intentional stance is part of folk psychology, as is significant unclarity in what is meant by ‘consciousness’ (by which I don’t mean that folk psychology has no clear concepts that ‘consciousness’ is used equivocally to express or clear words for expressing these concepts — in fact I think it does).

    14. Hi Benj. Perhaps you’re right that metasemantics is fundamentally intertwined with consciousness–and if consciousness is irreducible, then perhaps metasemantics is as well. I’m of the mind that it’s a good thing to keep as little as possible unreduced in our theories (and I think that’s independent of the fact that I’m optimistic about a reduction of consciousness). But maybe that’s not possible in this case.

      That said, I was thinking of conceptual/functional/inferential role semantics as not being distinctively connected to consciousness. CRS, and I’m thinking of it here as Sellars (or Harman) has it, characterizes contents not only in terms of what inferences a content will enter into (i.e., if I’m disposed to believe Q on the basis of believing P, then Q is part of the conceptual role of P), but also what cause and are caused by that content (if X’s typically cause me to believe P and believing P causes me to do Y, then X’s and doing Y’s partially characterizes what P is).

      (I think all that stuff can take place outside of consciousness, but I note that, on your view, it seems all thinking and intentional acting is part of consciousness, so maybe CRS is inextricably linked to consciousness).

      But maybe this kind of story won’t do for the contents of the sentences of the Lagadonian language. It’s not clear if there’s a way to characterize the conceptual role that a Lagadonian perceptual sentence plays (i.e., what causes it, what it causes, what can be inferred to and from it).

      Moreover, and this was something I was worrying about before, it’s not clear that, if there were such a way to characterize that role, it would follow that the contents of the sentence would be the objects and properties that compose the sentence–as opposed to something else.

    15. Thanks for your reply, Benj. I now see why taking the Lagadonian perceptual sentences and the belief sentence to be about the *objects* of experiences rather than the experiences themselves won’t allow us to avoid the conclusion that Sam is incoherent in the mismatch cases. This comes out especially clearly in the case of illusion, where the content of the perceptual sentence would be something like: there is a white widget bathed in red light before me, while the content of the belief sentence would be something like: there is a red widget before me. And I see now that in the hallucination case, accepting ‘there is a red widget before me’ without accepting a Lagadonian perceptual sentence with the same content amounts to incoherence: to affirming and failing to affirm the same content. (Actually, now that I’m writing this out, I’m not so sure– could the hallucinating subject *accept* the relevant Lagadonian sentence without *uttering* it? Perhaps she accepts it in the sense that she would utter it if she could, but she can’t…)

      My impression from your comments is that in giving an account of perceptual justification, we face two broad options: either it consists in infallibility about externalia, or it consists in ampliative inferences from internalia (claims about one’s experiences). Then, the thought seems to be that the latter option is hopeless (I seem to recall McDowell considering something like this option, and concluding that “the dreary history of epistemology” suggests that its prospects are dim), so we should go with the former option. And then the thought seems to be that this commits us to characterizing subjects in bad cases as incoherent. It’s this last bit I’m not so sure about– I think there’s a way of developing this option that doesn’t require us to characterize Sam as incoherent in mismatch cases. In my comments, I was casting about for a way of modeling this option in terms of your apparatus; I failed at this, so I’m just going to characterize in terms I’m more familiar with!

      Let the good case be one in which Sam sees the widget and sees that it is red, and bad cases be ones in which (i) Sam can’t tell her experience apart by introspection alone from a veridical perception of a red widget, and (ii) she either doesn’t see the widget (as in a widget hallucination) or doesn’t see that it’s red (as in an illusion). You might think that Sam has *some* perceptual evidence for the proposition that there’s a red widget before her in both cases (perhaps the proposition that it looks to her as if there’s a red widget before her). (This is to give up on the Cliffordian picture you favor– more on this below.) However, you might think that Sam has *additional* perceptual evidence in the good case (perhaps the proposition she perceptually represents, something like: that there is a red widget before her). Why doesn’t she have this evidence in the bad cases? You could say that only knowledge is evidence, and that only in the good case does she know that there is a red widget before her.

      On this picture, in a bad case in which Sam believes that she’s in the good case, she’s not incoherent– she perceptually represents the proposition that there’s a red widget before her, believes that there’s a red widget before her, but doesn’t have sufficient justification for her belief to count as knowledge– all she’s got is the proposition that it looks to her that there’s a red widget before her, and that’s not enough for her to count as knowing that there is. In the good case, she’s got additional perceptual evidence for the proposition that there’s a red widget before her: *that very proposition*. So this seems to be a version of the infallibility option: in the good case, her perceptual evidence for the claim that p simply is the claim that p itself– she can’t go wrong there!

      Now suppose we modify the good case to make it a mismatch one: Sam sees the widget and sees that it is red, but she falsely believes that she’s dreaming. In this case, you might say that Sam’s false belief about her experience functions as a defeater for her justification for believing that there’s a red widget before her– if it weren’t for that false belief about her experience, she would have infallible evidence for the proposition about her environment. So in this case too, Sam’s not incoherent: she perceptually represents the proposition that there is a red widget before her, but her total evidence isn’t sufficient to justify belief in this proposition (because one bit of it undermines another bit that would be sufficient if it weren’t for that other bit). (She might or might not believe that there is a red widget before her– this belief wouldn’t be justified, but either way she wouldn’t be *incoherent*).

      So it seems that we’ve got two versions of the infallibility about externalia option on the table: one that says that Sam is incoherent in the mismatch cases, and one that doesn’t. The latter involves giving up the Cliffordian idea that one’s perceptual evidence had better not be able to lead you astray…but is that so bad? Here’s how I see it: take the mismatch case I outlined in the comments, in which Sam is seeing but believes that she’s dreaming because of trustworthy testimony. It seems to me that this is a paradigm case of rational belief formation– ex hypothesi, Sam would be crazy to ignore this testimony and form the belief that she’s seeing anyway. This is not at all like a case in which one forms a belief as the result of being clonked on the head. I agree that if considerations of rationality apply in this case, then Sam’s overall mental situation is coherent. So I think we ought to conclude that her overall mental situation is coherent, which suggests that we should go for the non-Cliffordian version of the infallibility option.

      I take your general point that people sometimes think rational explanation is in order when it isn’t. It’s just not clear to me that the mismatch cases fall into that category.

    16. Thanks for the reply, Benj. A quick follow-up question on the “regard as equivalent” business as it applies to sentences of the Lagadonian language.

      One of the points you make in your reply is that to regard two vehicles of content S and T as equivalent, one needn’t take any attitude toward S and T themselves; one only need to take an attitude toward their contents by tokening those vehicles.

      Point taken. But I’m not totally sure I see how this helps in this case, since, as you point out, in the case of the Lagadonian language vehicle and content are identical.

    17. Hi Jeff, hm interesting point. Let’s see, one instantiates this property just if one affirms that one instantiates this property. That’s the sense in which vehicle and content are identical. My thought is that in affirming that one instantiates the property, one isn’t also affirming that one uses the property as a vehicle; or at least not in a way that readily leads to expression in English that one is doing so. My views might predict that in failing to affirm the expression in English of one’s situation, one falls short of ‘full coherence’. This seems like a species of the logical omniscience stuff.

      That’s a bit speculative; would need to think this through a bit more.

    18. Hi Heather, thanks for the response. I feel as though I don’t fully understand the point at which your picture and mine diverge. In order to chase this down, I’ve set up the argument for incoherence in the bad and mixed cases in the most explicit way possible, trying to chase everything back to the basic phenomenological and programmatic commitments behind the view.

      good case:
      perception content: there’s a red widget
      belief content: there’s a red widget

      bad case:
      perception content: something other than /there’s a red widget/
      belief content: there’s a red widget

      mixed case:

      perception content: there’s a red widget
      belief content: I’m hallucinating in a widget-free environment


      Why the perception content in the good case?

      1. Perceptual justification requires entailment
      2. The belief is justified

      Why the perception content in the bad case?

      3. Perception content is infallible

      Why the perception content in the mixed case?

      4. Perception content doesn’t depend on belief content at least not in the way that would save from accusation of incoherence

      Why incoherence in the bad case?

      5. Belief sentences are equivalent to the perception sentences that justify them
      6. Perceptual affirmation and belief affirmation are equally constituents of one’s perspective

      Why incoherence in the mixed case?

      6. As above

      Why 1?

      7. (a) Phenomenological: seeing a pig settles the question (b) Programmatic: using ‘justification’ to mean ‘rationalization’, our only theory of rationalization is Cliffordian

      Why 2?

      8. (a) Phenomenological: my operational sense for rationality comes from reflection on the sort of necessity involved in this kind of update (b) Programmatic: we should be antiskeptics

      Why 3?

      9. (a) Phenomenological: whatever it may be, /there it is/: no one responds to the Chalmers case for dualism with the obvious error theory ‘this is all just an illusion’ (b) Programmatic: ‘axioms’ in the Cliffordian system have to be analytic or contextually analytic

      Why 4?

      10. No real reason, seems plausible; in any event trying to take a stand against incoherence at this point wouldn’t solve the general problem

      Why 5?

      11. Entailment requires equivalence or something weaker; getting belief to be weaker wouldn’t save from incoherence

      Why 6?

      12. (a) Phenomenological: (i) there it is; in having a perceptual state there is that ‘phenomenological force’, where what this consists in is the acceptance of something; (ii) hard to see what could motivate updating one aspect of one’s perspective if not the world being a certain way according to another aspect; (iii) transparency (b) Programmatic: if derivation is to be the model we need to make sure the premisses are entered into the same argument as the conclusion and they need to be used rather than mentioned.

    19. Hi Benj,

      Many thanks for your helpful response. You address the main issue that I raise in the 5th part of your response and I’ll follow up on what you say there. You write: “The result that intensionally (more specifically, logically) equivalent sentences share semantic values is immediate, was built into the theory deliberately”. This raises the question whether (1) intensionally equivalent sentences can reasonably be considered as logically equivalent sentences and whether (2) logically equivalent sentences will carve contents finely enough for your purposes.
      Regarding (1): Logically equivalent sentences are not intersubstitutable salve veritate in contexts of belief ascriptions. By contrast intensionally equivalent sentences are arguably intersubstitutable salve veritate in contexts of belief ascriptions. For this reason, it seems problematic to understand intensionally equivalent sentences in terms of logically equivalent sentences. Now you may deny the principle that intensionally equivalent sentences are intersubstitutable salve veritate in contexts of belief ascriptions. If you do deny this principle, then it would be good to know why. I provide you with a few alternative conditions for intensionally equivalent sentences (I use the more neutral phrase “sameness of content”), which in conjunction with your conditions (1-4) provide a criterion for when sentences are intensionally equivalent. I don’t know if any of these suggestions are useful for your purposes. If you do want to operate with the idea that intensionally equivalent sentences just are logically equivalent sentences, then it would be good to know how you deal with the problem that logically equivalent won’t be intersubstitutable salve veritate in contexts of belief ascriptions.

      Thanks again. Great paper.


    20. Hi Susanna,

      My presupposition in that quote is not that if Nec P iff Q, P =||= Q, but rather its converse: if P =||= Q, Nec P iff Q. By ‘more specifically’ I indicate not that I am attempting to /define/ the notion of intensional equivalence (the conventional indicator for definition being some phrase like ‘by which I mean’) but rather my intention to /specify/ a variety of intensional equivalence (namely, the one under discussion in the arithmetic example that gets your dialectic underway — logicism strikes me as correct, perhaps somewhat eccentrically).

      You have evidently detected my agreement with Lewis’s view that ‘the connection between doxastic alternatives and the truth of belief sentences is far from uniform or straightforward’ (/Plurality/, 32). You want to know what the connection then is, and you provide several hypotheses. This is a very helpful approach, and I am touched by your gracious efforts on this front.

      So I hope you will forgive me for giving these short shrift. My excuse for this is that I have a few thoughts of my own on these matters, and I was trying to avoid muddling myself. I am sure that any of your reactions to these thoughts would be of considerable value, so if your busy schedule permits, I very much hope you will post them. You may find my discussion of these matters in the paper, at pages 2–44.

      Thanks again. I’m really glad we are getting a chance to interact in this forum! And I look forward to returning the favor at CMW.


    21. Hi Jake, sorry for the somewhat belated response. I think we’re in agreement about what the space of possible views looks like if not necessarily about what the most promising approaches might be (which anyway is a largely biographical matter of limited philosophical significance).

      On CRS: nice thing on Gleick by Dyson in this NYRB, which had a thought-provoking remark on Claude Shannon. Allegedly, his ‘central dogma’ of information theory is ‘meaning is irrelevant’.

      To me this indicates that functional reductions of intentionality have failure wired into their DNA. Information theory is a theory of translation, understood to /contrast/ with a theory of meaning. So functional reductions of intentionality say that meaning is ultimately to be understood in terms of a theory intended to have nothing to do with meaning.

      In this context it’s noteworthy that Harman’s approach has facets that are ordinarily overlooked. He is under no vulgar illusions about what the Shannon approach can or cannot do. His more broad-scope papers (‘Immanent and transcendent approaches in meaning and mind’) give a crucial and irreducible role to the first-person take as the locus of meaning. (This is also Carnap’s position in ‘Psychology in physical language’ — the key sections are excerpted out in the Chalmers collection.)

    22. Hi Benj,

      Thanks for those clarifications. Now I understand how you are thinking of the relationship between logical and intensional equivalence and we’re in agreement. I do hope we get the chance to discuss this more later. It’ll be great to see you at CMW and looking forward to hearing your comments then.

      Thanks also to Jake for putting together this special session and to Richard for organizing the whole event. Online conferences are the future; you’ve paved the way for others to follow!


    23. Many thanks for your thoughts on CRS, Benj. I was just offering it as a possible way to help in dealing with Jeff’s worry about the Lagadonian sentences’ contents–but I see your doubts about why my version may not be applicable in this case.

      Shifting gears, I wanted to pick up on another issue that you addressed in your back and forth with Susanna. I’m not sure if you addressed it or something like it already. If so, forgive me, there’s been a lot going on in this session–which, by the way, is awesome.

      In your reply to Susanna, you wrote about updates in bad cases that “despite this, we do find some updates more ‘natural’ than others as responses to the sort of situation they are in (Jeff’s contrast between Sam and Sam* brought this out nicely, I thought). My apparatus addresses this, if rather thinly, through the notion of ‘habitual like reaction’ (introduced on p59). The thought is that I find that were I in the subject’s position, I would update as they do. Not because doing so is an apt reaction to my picture of the world (although of course I would be under the impression that it is) but rather /merely/ because /that is how I roll/. So in that sense a contrast can be drawn between the ‘plausible’ and ‘implausible’ updates of the subject taken in by hallucination. Here, although intentional psychology grounds out, something more like pure sympathy is still available as an explanation. Sympathy provides understanding ‘from the inside’, and with it ‘justification’ of a sort: the sort I label ‘exculpation’.”

      So you’re willing to draw a distinction between ‘plausible’ and ‘implausible’ updates in bad cases–the former being ones that we can explain in terms of habits and the other not. But even if both updates are, strictly speaking, arational and beyond the purview of intentional psychology, it seems that only the plausible one deserve exculpation. Am I confused about that?

      Also, even if the plausible updates can’t be characterized by intentional psychology, it does seem (at least to me, unless I’m missing something) that we could develop some other explanatory scheme capturing those (strictly arational) habits, one that is perhaps law-governed and that might look a lot like intentional psychology. Or do you think no such alternative to psychology could be developed?

      In other words, are we justified in sympathizing with (exculpating) people taken in by bad cases but updating plausibly, but not with those updating implausibly? And if so, is there some system of explanation of plausible update that would explain our being so justified?

    24. Hi Benj,

      I think the main locus of divergence between your picture and mine is:

      “3. Perception content is infallible”

      On the view I was sketching, this isn’t true of perceptual content in general. The idea was that *in good cases* one possesses infallible perceptual evidence for claims about one’s environment (e.g., Sam has the claim that there is a red widget before her as perceptual evidence for the claim that there is a red widget before her). But one also has *additional* evidence concerning one’s experience (e.g., Sam also has the claim that it looks to her as if there’s a red widget before her as perceptual evidence). This is evidence that Sam also has in the bad cases, although it is on its own insufficient to justify beliefs about her environment– this is why I said the view I sketched abandoned the Cliffordian account of evidence you favor. It seemed to me that you thought a Cliffordian account was the only alternative to an account of perceptual justification in terms of ampliative inferences from claims about one’s experience. I was trying to sketch a third way.

      I’m afraid I didn’t quite follow what you said in support of 3:

      “Why 3?

      9. (a) Phenomenological: whatever it may be, /there it is/: no one responds to the Chalmers case for dualism with the obvious error theory ‘this is all just an illusion’ (b) Programmatic: ‘axioms’ in the Cliffordian system have to be analytic or contextually analytic”

      Here’s why I’m inclined to *deny* 3: it leads to the conclusion that subjects in mismatch cases are incoherent, and I find that difficult to swallow. It would be helpful to hear your thoughts about the case I outlined in my comments and in my last post– the case in which Sam is looking but believes that she’s dreaming because of trustworthy testimony. Your picture seems to predict that she’s incoherent and so that standards of rationality don’t apply; but it seems to me that this is a paradigm case of rational belief formation and hence that she must be coherent after all.

    25. Hi Heather, what didn’t you follow about the case for (3)? And why is the prediction of the conclusion that bothers you a reason to worry about (3) rather than about the other premisses in the case for that conclusion? And if (3) is rejected, how do you accommodate the phenomenological data on its behalf? And in virtue of what is it OK to regard perceptual sentences as axioms but not OK to regard any old sentences I might dream up as axioms?

    26. Hi Jake, the answers to your questions are: (1) no, you are not confused; (2) that’s above my pay grade; (3) not exactly: the notion of exculpation applies not to transitions we are /justified/ in sympathizing with but to transitions we /can/ sympathize with; (4) I think this question involves as a presupposition the doctrine just distinguished from my view, and so I think may need to be recast.

    27. Hi Benj,

      “what didn’t you follow about the case for (3)?”

      The “phenomenological datum” you offer in favor of (3) is as follows:

      “whatever it may be, /there it is/: no one responds to the Chalmers case for dualism with the obvious error theory ‘this is all just an illusion’”

      I’m afraid I don’t understand what you mean. If you could elaborate, I’d be grateful!

      “And why is the prediction of the conclusion that bothers you a reason to worry about (3) rather than about the other premisses in the case for that conclusion?”

      I am worried about some of the others: on the account I was suggesting, (1) and (5) are true only in the good case. (5) seems to presuppose (1), and (1) and (3) stem from your Cliffordian picture of justification. I was just trying to offer an alternative to that picture, one that isn’t a retreat to the old “ampliative inferences from claims about mental states” one.

      “And if (3) is rejected, how do you accommodate the phenomenological data on its behalf?”

      See above– I’m not clear on what the phenomenological data are supposed to be.

      “And in virtue of what is it OK to regard perceptual sentences as axioms but not OK to regard any old sentences I might dream up as axioms?”

      Because you just find yourself saddled with perceptual sentences– they’re forced upon you, so to speak. Not so for any old sentence I might conjure up. (I don’t see why the alternative picture I was suggesting would entail that it’s not OK to regard perceptual sentences as axioms, or that it is OK to regard any old sentence I’d conjure up as an axiom; I was just suggesting that the perceptual sentences in the bad cases have different contents than your view says they do.)

      “and what is wrong with my strategy for explaining away the sort of unease you are expressing by way of my distinction between justification and exculpation?”

      I’m worried that it doesn’t adequately address the case I mentioned in my comments: the one in which Sam is looking but believes that she is dreaming because of trustworthy testimony. This strikes me as a paradigm case of rational belief formation– which suggests that justification, not exculpation, is what’s wanted.

    28. Hi Heather, thanks for the elaborations. Let’s see if I can shed some clarity on my view of things (thanks for pressing these points by the way — this exchange is extremely valuable in getting me to see what exactly my view is).

      Yours between ***s, responses between ~~~s.

      *** “what didn’t you follow about the case for (3)?”
      The “phenomenological datum” you offer in favor of (3) is as follows:
      “whatever it may be, /there it is/: no one responds to the Chalmers case for dualism with the obvious error theory ‘this is all just an illusion’”
      I’m afraid I don’t understand what you mean. If you could elaborate, I’d be grateful! ***

      ~~~ Fine question. This is a crucial part of the paper that (rather stupidly) got left on the cutting room floor when I punched the draft out.

      Here’s the idea. I was combing through a bunch of lit in search of widespread undefended presuppositions — to me when this happens it is often evidence of a ‘phenomenological’ variety that the presupposed claim is someow manifest in experience: if we don’t see it as even in need of defense, we implicitly think of it as completely obvious, not in need of defense — ‘given’ or ‘manifest’ as it were.

      The claim then is that (3) is a widespread undefended presupposition of this sort.

      Of course to make this convincing, I need to produce the texts in question and justify my claim that (3) is an undefended presupposition in those texts. Obviously also we would need to rule out confounding factors — fashion, herd mentality, influence. So we want stuff like /it’s all over a contentious literature/ or /it’s all over in time/ or /geography/ or /it’s in phil mind *and* epist *and* phil pcn/ etc.

      I think you do find this.

      * Lewis’s ‘Why conditionalize?’. There he presupposes that evidence is infallible and evidence characterizes ‘experience’.

      * Price on ‘what cannot be doubted’: there’s gotta be /somethin/! Hard to say what it is exactly (and Price comes up with a cuckoo specification of what it is) but /some proposition/ given in perception is indubitable.

      * Chalmers on the incorrigibility of pure phenomenal judgements: although coming at a quale with a mistaken presupposition might lead us into an erroneous belief about how we are, if we don’t lard things up with bad presuppositions our bare ‘this’ judgement can’t go wrong. in my view there is no difference between the bare ‘this’ judgement and the perceptual state; so the perceptual state can’t go wrong.

      * The anti-Chalmers literature on dualism: in order to establish that if there /were/ any qualia, they would be primitive, the conceptual argument (that we can imagine qualia separately from the physical and that this imagining involves no confusions about identity) needs no further supplementation by empirical premisses. In order to establish that dualism is true of the /actual/ world, we need that there /are/ qualia. No one in the lit ever questions this. In my view, this is because they regard the contents of perception as infallible: infallible self-ascriptions of qualia, for the purposes of the lit, but that’s irrelevant to the general point.

      * Pigs and pig food.

      * David Lewis on evidence in ‘Elusive knowledge’.

      * Fallibilist intentionalists: people who say ‘the content of perception is fallible’ seem generally to regard something closely related as infallible, namely the proposition that /according to perception, P/ (where P itself might not be true). I say that according to such intentionalists, the /real/ content of the perceptual state is the infallible higher-order proposition, and perceptual justification involves an ampliative inference from this along the lines of ‘endorse whatever he endorses unless you have reasons for mistrust’ (compare DKL on the contents of perception in ‘EK’). (This is, I take it, not your view: for you the mistrustful ‘he’ is still ‘me’, as on my view — more below.)


      ***“And why is the prediction of the conclusion that bothers you a reason to worry about (3) rather than about the other premisses in the case for that conclusion?”
      I am worried about some of the others: on the account I was suggesting, (1) and (5) are true only in the good case. (5) seems to presuppose (1), and (1) and (3) stem from your Cliffordian picture of justification. I was just trying to offer an alternative to that picture, one that isn’t a retreat to the old “ampliative inferences from claims about mental states” one.***


      OK, good, thanks. The logical order here is still shaking itself out and running through the dependencies here is very useful in moving this along.


      ***“And if (3) is rejected, how do you accommodate the phenomenological data on its behalf?”
      See above– I’m not clear on what the phenomenological data are supposed to be.***


      Hope remarks above were clarifying


      ***“And in virtue of what is it OK to regard perceptual sentences as axioms but not OK to regard any old sentences I might dream up as axioms?”
      Because you just find yourself saddled with perceptual sentences– they’re forced upon you, so to speak. Not so for any old sentence I might conjure up. (I don’t see why the alternative picture I was suggesting would entail that it’s not OK to regard perceptual sentences as axioms, or that it is OK to regard any old sentence I’d conjure up as an axiom; I was just suggesting that the perceptual sentences in the bad cases have different contents than your view says they do.)***


      OK, thanks. I’m going to think out loud for a bit.

      So what about the clairvoyants. They are in a shallow sense (deeper sense discussed 2 paragraphs hence) ‘saddled’ with B-sentences with the same contents as the P-sentences they would be having if they had P-states (but which are not contextually analytic, because that status involves linguistic properties beyond the semantic). Their B-sentences are not justified but our B-sentences are.

      My thought is that perception makes the difference. The central concern of the paper is to explain /how/ perception makes the difference.

      The answer is this (section 3): (G) P-sentences don’t admit of justification (because they are not consequents of ‘laws of rationality’ — so we are in a deeper sense ‘saddled’ with them), so we don’t have an option to revise them — it’s either revise the belief or live with the incoherence. (T) P-sentences are indefeasibly trustworthy (because contextually analytic), so revising B-sentences to match is always a better policy than living with the incoherence.

      By contrast, the clairvoyant B-sentences /admit/ of justification, so we can ask whether they are justified; and they are neither analytic nor contextually analytic (indeed, contextual analyticity seems to be incompatible with admittance of a justification status), so they do not count as apt responses to the world-as-manifest: they are more like epistemic free choices.

      This is why our beliefs are better off than those of the clairvoyants.

      OK, so I’m imagining that your view preserves (G), so that we are ‘saddled’ in the deeper sense, but not (T), so that in the case of hallucination the P-sentence is false.

      An initial observation is that in ‘defeat’ cases the G+T– (‘g-plus-t-minus’) view predicts incoherence despite ‘reasonability’ in an intuitive sense. P-sentence says /there’s a red thing/, (false) presupposition is that I’ve gone inverted, B-sentence says /there’s a green thing/: crash, but (perhaps) still ‘reasonable’. So G+T– doesn’t fully respond to the worry motivating the view. (Maybe this means that the view you are pushing is distinct from G+T–.

      Then here’s the programmatic worry for your view. Without (T), we need some story of why we are better off than the clairvoyants (when we are?). Is it /always/ a better policy to revise the B-sentences? If so, why, given that sometimes they are false? Is it /sometimes/ a better policy to revise the B-sentences? If so, when; and how are we supposed to be in a position to /follow/ the policy, whatever it may be?

      The Cliffordist approach has the advantage of bone-simplicity on its side: as a piece of theory, it is very easy to say what it is. Anti-Cliffordist approaches cannot be so easy to state; and to my knowledge they have not been stated. This may suggest that they are at odds with our implicit understanding of the phenomena.


      ***“and what is wrong with my strategy for explaining away the sort of unease you are expressing by way of my distinction between justification and exculpation?”
      I’m worried that it doesn’t adequately address the case I mentioned in my comments: the one in which Sam is looking but believes that she is dreaming because of trustworthy testimony. This strikes me as a paradigm case of rational belief formation– which suggests that justification, not exculpation, is what’s wanted.***


      I hate to say it but I find that this begs the question.

      Don’t want to go out on a sour note so let me say thanks a heptillion for the exchange, I’m finding it extremely valuable.


    29. Thanks again for your replies, Benj. Your responses make sense.

      I guess I have a little question about your saying that “the notion of exculpation applies not to transitions we are /justified/ in sympathizing with but to transitions we /can/ sympathize with.”

      I’m wondering how we would determine what we can and can’t sympathize with, if we can’t appeal to anything in intentional psychology, but perhaps that’s not something you feel like you have to explain.

    30. Thanks very much for the clarifications, Benj–they were very helpful. I think that we can accommodate the phenomenological data without going all the way to (3); perhaps by saying just that in the good case we have infallible perceptual evidence. But I’ll have to think more about that.

      I’m still stuck on the mismatch case I mentioned in my comments, though (the one in which Sam is looking but believes she’s dreaming on the basis of trustworthy testimony). Intuitively, she would be *irrational* if she went ahead and believed that there was a red widget before her in spite of this testimony. You suggested that there was something question-begging about this case, but I don’t see why (unless appealing to intuitions about cases to generate counterexamples is illegitimate in general, but that can’t be right). It would be great if you could elaborate on this before the conference ends; but if not, hopefully we can continue the discussion outside the conference confines…

      Anyway, thanks again for your paper! I’ve enjoyed thinking about the issues you’ve raised.

    31. Hi Heather, two questions:

      1) The point on which I say the question is being begged is not on the judgement of rationality but on whether the judgement can be accommodated by declaring the rationality a variety of ‘doing what we would normally do’ rather than ‘acting within the descriptive capacity of intentional psychology’.

      2) Your recent reply left something out. I’d like to know what your position on it is, since I think it is the most interesting philosophical issue raised in this exchange. Namely, the point I address in the second-last block in my previous reply: the one beginning ‘I’m going to think out loud a bit’.

      This to me seems like the fundamental point, so I remain quite interested to learn your thoughts on it.

    32. Since the conference is drawing to close, I just want to say thank you again to everyone who participated in this absolutely wonderful session of the conference!

      Many thanks to Richard Brown for making this conference happen, to Heather Logue, Susanna Schellenberg, and Jeff Speaks for their wonderful commentaries and discussion here, and, of course, to Benj Hellie for the incredibly rich paper and terrific responses to the commentaries in the discussion threads.

      I for one have learned a great deal from this session, and have no doubt that many of the issues discussed here will be at the center of debates about perceptual justification for some time to come. And I hope we can discuss these things further offline.

      Thanks again!

    Comments are closed.