The Real Trouble for Phenomenal Externalists: The Science of Taste, Smell and Pain

Presenter: Adam Pautz, University of Texas at Austin

Commentator 1: Ruth Millikan, University of Connecticut

Commentator 2: David Hilbert, University of Illinois Chicago

Advertisements

12 Comments

  1. I am very grateful to David Hilbert and Ruth Garrett Millikan for taking the time to produce their very helpful and interesting comments on my paper.

    I will begin with Hilbert’s comments. In my paper, I gave three arguments against tracking intentionalism: the internal-dependence argument, the structure argument, and the percipi argument. The first two of these are empirical, relying on “good internal correlation” and “bad external correlation”.

    Hilbert first (in his section “Perception is not magic”) makes some interesting general points. However he concedes that they do not “directly address the detailed arguments that Pautz constructs”. I agree with this, so I will try to be brief in response. For one thing, he says that even tracking intentionalists can say that the brain plays *some* role in making experience possible. Of course I agree: it is common ground that we need a brain to have experience – perception is not magic! But I did not suggest otherwise in any of my three arguments, so this is not a criticism of any of those arguments. For another thing, Hilbert suggests that tracking intentionalism actually predicts good internal correlation and bad external correlation. But even if this is true, it is again not a criticism of any of my arguments, because in none of the arguments do I move directly from good internal correlation and bad external correlation to the failure of tracking intentionalism. My arguments concern certain cases and examples that the tracking intentionalist must say something about. (Hilbert quotes a passage in which I say good internal correlation and bad external correlation should make us *suspect* tracking intentionalism is wrong; but immediately afterwards in the paper I say that this is not itself an argument, and that the arguments come later.)

    (In addition, it is not true that tracking intentionalism actually predicts good internal correlation and bad external correlation. I think Hilbert might have missed what I mean by ‘good internal correlation’ and ‘bad external correlation’. He seems to think that by ‘good internal correlation’ I just mean that our experiences have neural correlates and that by ‘bad external correlation’ I just mean that experiences can occur in the absence of their normal causes. These things are indeed predictions of tracking intentionalism but they are not what I mean. I say this in the paper at p. 10 but I will have to be even more clear about this in the future. As I say there, I mean different things, things that are definitely not predicted by tracking intentionalism.)

    Hilbert then (in “Perception is not atomistic”) turns to one of my actual arguments – the internal-dependence argument. He does not address my other two arguments.

    My internal-dependence argument is a general argument-schema that can be illustrated with many cases. Hilbert focuses on one: the case of Yuck and Yum. They taste some berries that are poisonous to Yuck but an important foodsource to Yum. They differ in their taste processing and behavior yet I stipulated that they track the same external chemical property of the berries. The argument has two premises: First, *if* tracking intentionalism is true, the right verdict is Same Content (Yuck and Yum sensorily represent the same chemical property etc.) and hence (given intentionalism) Same Experience. Second, it’s more reasonable to suppose that the right verdict is Different Experiences. So, tracking intentionalism is false.

    Hilbert thinks this is a bad argument against tracking intentionalism. He accepts my second premise: he accepts that the right verdict is Different Experiences (“internal-dependence”). But he seems to reject my first premise, the one saying that if tracking intentionalism is true, the right verdict is instead Same Content and hence Same Experience.

    What Hilbert says at first confused me a bit. His initial objection to my first premise is that it’s reasonable to accept Different Contents, not Same Content. This would indeed undermine my first premise if my first premise were Same Content. But it is not. It is merely conditional: *if tracking intentionalism is true*, then the right verdict is Same Content and hence Same Experience. (Indeed, I myself think that the right verdict is Different Contents and Different Experiences.)

    Accordingly, to answer my argument against tracking intentionalism, it’s not enough to say that Different Contents is reasonable – with this I entirely agree. Hilbert must also show how Different Contents (and hence Different Experiences) can be squared with *tracking intentionalism*. I think Hilbert cannot do this and has not done this. Yuck and Yum track exactly the *same* external conditions. How then could the *tracking intentionalist* say that they sensorily represent different contents, so that (given his intentionalism) they have different taste experiences?

    Hilbert does mention some interesting possibilities on behalf of the tracking intentionalist, without developing any of them in detail. They are as follows:

    The Evaluative Account: Yuck sensorily represents the proposition that the berries are poisonous, while Yum does not; this representational difference constitutes their different taste experiences of the berries.

    The Similarity Account: Yuck sensorily represents some proposition about taste similarity, and Yum represents a different such proposition; this representational difference constitutes their different taste experiences of the berries. And their sensorily representing these different propositions is somehow (?) determined “holistically”.

    The Digestive Account: Yuck and Yum sensorily represent different propositions about their digestive systems; this representational difference constitutes their different taste experiences of the berries.

    I propose to ignore the digestive account. It doesn’t seem consistent with tracking intentionalism (Yuck and Yum’s taste systems don’t track anything about their own digestive systems – they track conditions involving external foodstuffs). And it certainly does not generalize to my other counterexamples to tracking intentionalism involving smell and pain. But I would like to briefly address the other accounts.

    I already addressed the evaluative account in connection with the fifth objection in the objection-reply section of my paper (p. 51), so I will be brief. Of course, it can save tracking intentionalism only if it is consistent with tracking intentionalism. But I argue on a couple of grounds that it isn’t. (Contrary to Hilbert, I argue that it is even incompatible with non-causal versions of tracking intentionalism like Dretske’s.) In any case, I argue that the evaluative account is a non-starter on independent grounds, and that it does not generalize to my other cases involving smell (Harold-Henry), pain (Mild-Severe), color (Max-Twin Max) and sound (Loud-Soft).

    As for the similarity account, I find it hard to say much about it because Hilbert does not develop it. But I address something in the vicinity in connection with an objection in my paper (p. 50). One question is: what *exactly* are the different contents Yuck and Yum sensorily represent, on this account? Hilbert does not say. (In my paper, I consider one option, and show that it is undermined by an argument due to Alex Byrne.) More importantly, the similarity account can save tracking intentionalism only if it is consistent with tracking intentionalism. So another question is: how exactly does tracking intentionalism deliver the verdict that Yuck and Yum sensorily represent these different contents about similarity, whatever they might be? This seems impossible, because Yuck and Yum’s brain states track exactly the *same* external conditions. Hilbert mentions ‘holism’ but does not develop the thought. What exactly is the holistic theory Hilbert had in mind? How would the proponent of such a theory complete ‘X sensorily represents Y iff X . . .Y’? (A terminological point: contrary to Hilbert, I’m not sure how such a theory might be regarded as a tracking theory of the sort that was the target of my paper, for tracking theories are typically regarded as atomistic.)

    Actually, the internal-dependence argument (and the structure argument) doesn’t just create a problem for tracking intentionalism. It creates a problem for anyone who, like Hilbert, accepts intentionalism plus what he calls ‘external world content’ (he does not say what he means but I take it he means that the contents of our experiences concern reflectance properties, chemical properties, etc.). What exactly are the *different* external world contents of Yuck and Yum’s experiences, and how did they get them? It’s very hard to come up with a good answer to this question, given that Yuck and Yum’s states track exactly the *same* external world conditions.

    Hilbert mentions holism. I do think that there are holistic theories of sensory representation that might seem to entail that Yuck and Yum sensorily represent different contents, and hence (given intentionalism) have different experiences. For instance, at the end of the paper I mention David Lewis’s interpretationist-functionalist theory. But, whereas Hilbert says his holistic idea is consistent with tracking intentionalism, Lewis’s theory is not consistent with tracking intentionalism. By contrast to tracking intentionalism, Lewis’s theory is output-based. It would appeal to the behavioral differences between Yuck and Yum in order to pin down the different contents of their experiences. (And presumably they will not be ‘external world contents’ about response-independent environments features, contrary to Hilbert’s view.) So Hilbert presumably didn’t have in mind anything like Lewis’s version of holism. (And, in any case, I argue that Lewis’s version of holism cannot be applied to the content of experience in my paper ‘A Simple View of Consciousness’.) So it is still unclear to me what he had in mind.

    So, as far as I can see, although what he says is very interesting, Hilbert hasn’t shown how to defend tracking intentionalism (or indeed his own combination of intentionalism and external world content) from my internal-dependence argument, illustrated with the case of Yuck and Yum and other such cases. In particular, he doesn’t answer the following crucial questions:

    Q1 What exactly are the different contents that Yuck and Yum sensorily represent, on the view he has in mind?
    Q2 How exactly does “holistic” tracking intentionalism (or indeed any theory) deliver the verdict that Yuck and Yum sensorily represent these different contents about similarity, whatever they might be, even though Yuck and Yum’s brain states track exactly the *same* external conditions? How would the “holist” complete ‘X sensorily represents Y iff X . . .Y’?
    Q2 How might the account go in the other cases I discuss in the paper involving smell (Harold-Henry) and pain (Mild-Severe) – not to mention cases I discuss elsewhere involving color vision and sound?

    To conclude. Given the complexity of the issues Dave focuses on the internal-dependence argument. That is fair – I am grateful for his feedback. But what he says, while intriguing, seems to me incomplete and problematic. There’s also the structure argument to contend with. How might the tracking intentionalist (or the intentionalist of Hilbert’s persuasion) accommodate Harold’s judgment (discussed in the paper) that his first two smell experiences resemble more than his second two, given the psychophysical facts of the situation (“bad external correlation”)? Byrne and Hilbert have a story about color resemblance but it does not seem to carry over here. (And what to say about pain? Does the kind of externalist intentionalism Hilbert and others favor in the case of color vision apply to the case of pain? I argue no in the paper. What’s the alternative? Hilbert candidly admits he doesn’t know but it seems like something that ought to be discussed.) So, while Hilbert titles his comments ‘No Problem’, I continue to think that my arguments create serious unanswered problems for certain reductive intentionalist programs (including Hilbert’s own).

    As I said in the paper, one of the things I’m doing in the paper is just making a plea to try to come up with *detailed* reductive versions of intentionalism which answer the empirical arguments I construct, not just in the case of color vision (the sole focus of many) but across the sense-modalities. I myself think that in the end it cannot be done. The externalist reductive program was a wrong turn. It’s time to throw in the towel and look for alternatives. In the objection-reply section I consider responses from Byrne, Tye, Hill and Dretske but I argue that they do not work. I myself stick with intentionalism but I advocate a non-reductive, internalist and projectivist-Galilean version of intentionalism. But I also welcome detailed suggestions about how the tracking intentionalist or indeed any sort of reductive intentionalist might answer the internal-dependence argument, the structure argument and the percipi argument.

    Now I turn to Millikan’s interesting comments. Millikan does not take issue with my arguments against tracking intentionalism (phew!). Instead she puts forward some interesting alternative suggestions about sense experience and our thought about it. As I saw it, her main claims were as follows:

    1 “Basic perception does not involve perception of color and odors and sounds as such at all, but only perception of the objects and kind they help to signify” (p.2)

    2 “The relations, say, among the neural representations of colors, among the neural representations of odors, and so forth — the dimensions and distances in this or that neural similarity space — though they might in some cases carry a certain amount of natural information about relations among the properties discriminated, would carry no intentional information. They would not represent any relations, any more than the relations among “cat” and “bat” and “rat” and “sat” represent relations.” (p.2)

    3 “Similar points apply to concepts developed of the relations between colors. These relations are, seemingly, systematically reidentified, and confirmed by other people, but they are not relations among any univocal properties of external objects. Thought of that way they are chimerical.” (.5)

    4 “This would explain how our concepts of colors and tastes and odors might come to be equivocal, a single concept confusing together a diversity of distinct actual properties.” (p.5)

    5 “This realm [that portrayed by a phenomenological description of a visual scene] is no more real than the realms containing the four humors or caloric. . . . However one [describes other experiences like smells], the descriptions express representations, concepts, of the same sort associated with any other deeply mistaken scientific or lay theory”. (pp.5-6)

    Some of these claims I’m very sympathetic with – especially 5. I am a kind of error theorist about the content of color experience, smell experience, and so on. (I have a kind of Galilean-projectivist view.) My view resembles David Chalmers’ view in ‘Perception and the Fall from Eden, only it’s based on somewhat different considerations, in particular, the sort of empirical considerations I discuss in my paper. But it’s a long story – I certainly do not think that there is any quick and easy argument from the sorts of empirical facts I discuss to a kind of error theory. (And I am not sure exactly on what grounds Millikan advocates such an error theory – it seems to me that there are many non-error theoretic theories she hasn’t ruled out – more on this below.)

    But I do have questions about some of these claims. For instance, 1: obviously, ordinary perception involves the perception of colors and odors and so on. So by ‘basic perception’ Millikan must mean something different. This is a kind of technical term. So I am not sure I have a firm grip on the claim being made here – or the argument for it.

    As for 3: does 3 amount to an error theory of ordinary claims like ‘blue resembles purple more than green’ and Harold’s ‘these two smells are more like each other than those two smells’ (which I consider in the paper at p.34)? (She does say ‘chimerical’.) If not, what does 3 amount to? If so, what exactly is the case for 3? After all, there are accounts of these claims on which they come out true, despite the “bad external correlation” that I discuss in the paper. For instance, in the paper, I discuss the Shoemaker-Lewis-McLaughlin response-dependent account (p.37). What is Millikan’s argument against this kind of account?

    As for 4: Millikan’s claim here is that concepts of colors and tastes and odors are equivocal, a single concept confusing together a diversity of distinct actual properties. What exactly are the different actual properties they’re equivocal between? Exactly what’s the case for equivocation? Does her view entail ambiguity at the level of language in English terms for colors, tastes, and odors? Is there any real semantic evidence for ambiguity? Also: in 5 Millikan seems to say that these concepts are empty [“the sort associated with any other deeply mistaken scientific or lay theory”] but in 4 she seems to say that they are merely equivocal. Which is her true view? (There are, by the way, parallels between Millikan’s views here and Hill’s recent views on the concept of pain.)

  2. In his conference paper “What we Hear” Jason Leddington quotes Heidegger: “We never really first perceive a throng of sensations, e.g., tones and noises, in the appearance of things…; rather we hear the storm whistling in the chimney, we hear the three-motored plane, we hear the Mercedes in immediate distinction from the Volkswagen. Much closer to us than all sensations are the things themselves. We hear the door shut in the house and never hear acoustical sensations or even mere sounds. In order to hear a bare sound we have to listen away from things, divert our ear from them, i.e., listen abstractly.” Compare Pautz: “…obviously, ordinary perception involves the perception of colors and odors and so on” — in his paper he also says sounds. Is Heidegger right here, right that if one listens in different ways or for different purposes one perceives different things? Or is Pautz right, that there is just “ordinary perception,” that one always perceives in the same way or at the same level? I am with Heidegger. I am suggesting that perception for guidance of immediate action is different from perception for making judgments about the properties of things. Also, that perception for action is evolutionarily earlier, perception for theoretical judgment in humans being built on top of this in the customary way that evolution proceeds. Pautz rightfully asks for an argument. It is in Varieties of Meaning, especially Chapters 18 and 19, where I argue that it is likely that only Humans indulge in theoretical judgment. The evidence that human action is not guided in immediate detail by the operations of belief (theoretical judgment) plus desire but instead by something like Gibsonian perception of affordances does not, I hope, need to be documented for this audience. My argument is also philosophical, based on a theory of representation that says, what is represented is a function of what needs to be represented given the kind of uses the representation is designed to have. Representation is not a mere matter of causal history but of use.
    Yes, (3) amounts to an error theory of ordinary claims like ‘blue resembles purple more than green.’ Notice first that on a consumer-involved theory of representation, a property X can be represented and a property Y can be represented at the same time without any relations between X and Y being represented (even the relation of identity or difference — On Clear and Confused Ideas). Similarly. a relation between X and Y can be represented without the absolute value of either X or of Y being represented. (Perceiving gradients and differences is a good example. For most of us who possess only relative pitch, hearing thirds, fourths, fifths etc. without being able to identify absolute pitches is another.) In the case of colors, absolute colors, as non-relational properties of objects, (roughly reflectances) can be perceived within a range of error, although also in some cases equivocally (metamers). (Yes the equivocation is in the language as well as in perception and thought, in the same way that an equivocation between weight and mass was in the word “weight” as well as in people’s thoughts of weight before Newton.) In many or most cases, absolute differences of color in objects are correctly perceived. But the apparent dimensions of color, the overall topological similarity space in which colors seem to be perceived, does not characterize actual relations among the colors of objects. These relations are mistakenly projected as relations among colors in the objects. But they actually depend on the fact that our neurological representations of colors at a certain level are natural signs relations not among colors alone, but of relations with more addicity including elements of the human visual systems as a relata. Used as intentional signs of relations among colors in objects they represent chimerical relations. All this remembering that representing X and Y is not, just as such, representing any relation between X and Y.

  3. In reply to Adam Pautz:

    I appreciated your very interesting article. Getting serious about exactly which perceptible properties are supposed to correlate with the qualitative contents for each modality is an important project. I’d like to focus on pain.

    I agree with you on a number of points. Most importantly, I fully agree that there is “bad external correlation” between pain and any external, sensed property. More particularly, I fully agree that there is very poor correlation between the perception of tissue damage and pain experiences, which is where your discussion begins. (Though I think more needs to be said to dispose of a view like Tye’s that takes the representation of tissue damage insufficient for pain unless represented “in a pain experience.”)

    What strikes me as very odd, however, is your use of the pain science literature. Focusing on intensity, presumably because it is the most manageable, you appeal to these studies to show that there is good “internal correlation” and bad “external correlation”. That is, you seem to think that the cited studies on pain intensity establish that subjective pain intensity ratings are well correlated with activation in S1, but poorly correlated with the intensity of the stimulus.

    Some of these studies do correlate reported pain intensity with firing in S1 (though I think you overstretch when you call these correlations “perfect”). When they do, however, they also further correlate the reported pain intensity and S1 firing with the intensity of the stimulus. In those cases we’ll have good internal correlation, but also good external correlation. I’ll consider just two examples from the literature to which you refer.

    First, consider the Timmerman et al (2001). The abstract itself connects the pain ratings with the stimulus intensity: “Subjects’ pain ratings correlated highly with the applied stimulus intensity.” The discussion connects the applied stimulus intensity with the S1 activation: “The present study demonstrates a high correlation of nociceptive stimulus intensity with activity in contralateral SI and bilateral SII” (1501). Most importantly, the data presented in figure 2 shows impressive correlation not just between pain ratings and S1 activation, but with these two and stimulus intensity.

    Next, consider Coghill (1999). Perhaps just one quote this time will suffice to note the author’s assertion of the correlation between reported intensity and stimulus intensity: “Subjects perceptions of pain intensity increased significantly during graded increases in stimulus temperature…” (1936). Coghill is quite explicit throughout that the stimulus intensity is correlated with the reported intensity. That sure seems like good external correlation to me.

    Perhaps more interesting about the Coghill article in conjunction with your argument, however, is that his central point is that pain intensity is not just correlated with activation in one area in the brain; consider his very title: “Pain Intensity Processing within the Human Brain: A Bilateral, Distributed Mechanism.” More specifically, he sums up his findings:

    The present findings confirm in a fully quantitative manner that pain intensity is processed in a highly distributed fashion. This distributed mechanism encompasses a number of functionally distinct regions that all exhibit activation that is closely related to perceived stimulus intensity. These include brain areas typically thought to be important in 1) somatosensory processing: SI, SII, and the posterior insular cortex; 2) motor processing: cerebellum, putamen/globus pallidus, supplementary motor cortex, ventral premotor cortex; and 5) autonomic funtion: anterior cingulated cortex and anterior insular cortex … (1939).

    Indeed, of the main brain regions involved in pain intensity processing, on Coghill’s view, S1 is not the most important: “Of all brain regions examined, the ventral premotor cortex exhibited the most pronounced increases in activation as intensity ratings increased” (1941). These results, if anything, would seem to suggest good external correlation and bad internal correlation if the relevant correlate is supposed to be S1 activation.

    Squaring the findings of these two articles is important. Pain science, at the moment, seems to be experiencing profound growth which, as perhaps always, comes with many theoretical decisions. Put aside, however, whether or not we can broker some agreement between these findings about what the internal correlates of subjective pain ratings are (I think, for the record, that we can). That is, set aside the question of good internal correlation. What both articles clearly conclude in common is that subjective pain ratings are well correlated with the stimulus intensity. That sure seems like what you were calling “good external correlation” to me. If so, then this research simply doesn’t support your claim that there is good internal correlation and bad external correlation for pain intensity.

    I wonder if I am missing something in your interpretation of these studies or if you could say a bit more about how you thought that they supported your argument.

    I should note that I think you are likely right, in general, that subjective pain intensity ratings are poorly correlated with stimulus intensity. Data supporting that contention, however, are likely to require some additional variables; e.g., the individual is attending to something other than the stimulus, or is primed to interpret the stimulus in a particular way. In real life, as these sorts of additional variables are constantly present, I wager they dissociate all the time. Standard clinical trials, however, where healthy individuals have been instructed that they will experience pain, and in which they then attend to the stimulus, are not likely to present dissociations. The studies you cite, at the very least, do not seem to me to support that contention.

  4. Thanks very much, Ruth, for your helpful follow-up. You pit me against Heidegger (yikes!). I actually agree with one of Heidegger’s claims, namely that we typically don’t perceive our own sensations, if that’s just taken to be a kind of weak (fairly uncontroversial) transparency claim. I also agree with your claim that one doesn’t always perceive in the same way or at the same level (you seemed to think I denied this). My claim was just that, in *some* sense or other, perception involves the perception of colors and odors and sounds and etc., even if it also involves the perception of objects and it is the objects that we are typically most interested in. (I was reacting to your claim that ‘basic perception does not involve perception of color and odors and sounds as such at all, but only perception of the objects and kind they help to signify’.) But I’m not sure how big a disagreement this is or even that there is a real disagreement here – and, although it’s an interesting issue, it’s somewhat far removed from the sort of issues that are the focus of my paper. On matters that are closer to the concerns of the paper, we agree on a fundamental point. In particular, we agree that at some level things aren’t the way they seem – at some level perception portrays things otherwise than they are. As David Chalmers would put it (or Hardin – see his ‘Qualia and Materialism’), we’ve “fallen from Eden”. I didn’t discuss this positive, error view in my paper. The aim of my paper was just negative – to rule out tracking intentionalism. I argue for the error view elsewhere. (I might mention here that a paper by Kathleen Akins, ‘Sensory Systems and the Aboutness of Mental States’, is relevant to my concerns, although her arguments are different and are in fact not directed at tracking intentionalism about phenomenology, and her positive view is quite different from mine. I hope to discuss this paper in future work.) I don’t think the error view follows immediately from the sorts of things I discuss in the paper – for instance, “bad external correlation” when it comes to smell resemblance or taste resemblance. Rather, I think it takes a lot of argument. There are views that accommodate all the empirical stuff but also avoid error that we have to rule out (for instance, the Kriegel-Prinz view that I discuss in the paper, or a view that combines the identity theory of experience with some kind of response-dependent view of sensible properties, etc.). In any case, we agree on this fundamental point – that at some level perception portrays things otherwise than they are. There are some differences. You say claims like ‘blue resembles purple more than green’ and ‘these smells resemble each other more than those’ are false (thanks for clarifying that). I would not say this. I’d locate the error elsewhere – as I explain in some of my other papers. (I think that the sensible qualities have the resemblance structure they seem to have but that they are not instantiated by external items.) I’m also still not sure what your argument is for thinking that claims like ‘blue resembles purple more than green’ and ‘these smells resemble each other more than those’ are false. I think this does not immediately follow from the kind of “bad external correlation” I discuss in the paper. For instance, as I mentioned in my first post, on the Shoemaker-Lewis-McLaughlin response-dependent view which I discuss in the paper, these claims come out true, despite “bad external correlation”. So an argument for your error view would have to rule out the Shoemaker-Lewis-McLaughlin view. (I do think there are arguments against this view – I develop them in my paper ‘Can the Physicalist Explain Colour Structure in terms of Colour Experience?’.) You also say that English names for colors, smells, and so on are equivocal (“the equivocation is in the language as well as in perception and thought, in the same way that an equivocation between weight and mass was in the word “weight” as well as in people’s thoughts of weight before Newton”). Thanks for clarifying that. But I am still not sure what properties you think for names for colors, smells, and so on are equivocal between, or why you think they’re equivocal. However, we at least agree on some fundamental matters – a rare thing in philosophy!

  5. Very interesting paper and discussion!
    I am very sympathetic to Adam Pautz views. However, I have the impression that some of the examples that he provides can be replied by appealing to imperative content (Millikan 1984 ch.6). For instance in the case of Severe and Mild, they differ in this imperative content, and similarly in the case of taste, something like ‘Don’t have this bodily disturbance!’ (Manolo Martinez 2010 “Imperative content and the painfulness of pain”). Mild and Severe differ in the force of this imperative content.
    Of course this account is not merely information-tracking.

    I also would like to hear your opinion about combining Prinz-Kriegel views with theories of function that do not depend on the subjects environment. These theories are highly problematic for naturalizing semantics, but I am not so sure in the case of the content of experience.

  6. Thanks very much, Jennifer, for your very helpful and interesting comments about pain. Let me start with two general points about what you say.

    First: In section 3, I provide some empirical background about taste, smell and pain. You make some interesting points about what I say in this section specifically about pain (which, as I will explain, I basically agreed with – in fact I already discussed some in the paper). But I wondered whether you think these points somehow create trouble for my arguments about pain – the internal-dependence argument, the structure argument, and the percipi argument – and the specific cases (Mild-Severe, Max) I use to illustrate those arguments. Those arguments do not take place in my section 3 which you focus on but in my sections 4-6. Maybe you do not think what you say creates any trouble for those arguments. For instance, you say you “fully agree” with my basic claim that in the domain of pain there is “bad external correlation” (given the empirical evidence it would be hard not to). As I explain in the paper, one of my arguments, my structure argument about pain (section 4), depends on only bad external correlation (at premise 5 at p. 36). So it is hard to see how what you say might make trouble for this particular argument. (Maybe you do not think otherwise – in your comments you did not say it makes trouble for this particular argument.) As for my internal-dependence argument about pain (section 4; laid out in premise-conclusion form at p. 31), again I’m not sure if you think what you say makes trouble for this argument or the Mild-Severe case I use to illustrate it. I guess I find it hard to see how what you say might make trouble for this argument, because it is hard to see how what you say might motivate rejecting one of the premises (which one would it be, and what would be the motivation?). For instance, as I say in the paper, that argument is neutral on the distributed-local controversy about pain intensity coding (more on this below).

    Second, issues specifically about pain cannot in any case undermine my general internal-dependence and structure arguments. (Again, maybe you don’t think otherwise.) The reason is simply that these arguments are general and do not crucially depend on anything specifically about pain. For instance, for the internal-dependence argument to work, only one counterexample is required, as I point out at p. 31. It does not have to be a counterexample involving pain like the Mild-Severe case I discuss; it could be the Yuck-Yum case involving taste or the Harold-Henry case involving smell (which I discuss in the paper), or one of the cases involving color vision or sound that I discuss in other papers. Likewise for the structure argument. The tracking intentionalist would need general responses to these arguments, not just a response that focuses on one sense modality.

    With those general points out of the way, let me address some of the specific points you make.

    (1) You say “Coghill’s central point is that pain intensity is not just correlated with activation in one area in the brain”. I agree – in fact I pointed this out in my paper at p. 21. In fact, I quote him as saying, “Many cortical areas exhibit significant, graded changes in activation *linearly related* to pain intensity”. For this and other reasons, he takes a distributed view. Another, perhaps more common view says S1 plays a special role. As you note, that’s what the Timmerman paper argues (see also the 1999 Price paper which I cite in my paper, esp. what he says at 396-7, which summarizes the evidence that S1 plays a special role). But, as I also say in my paper (p.21), my internal-dependence and structure arguments are neutral on this issue.

    (2) You say the Timmerman and Coghill experiments do not demonstrate bad external correlation. I agree. To establish bad external correlation, I cite other papers, papers by Stevens and Price. Price summarizes his experiments in the early chapters of his very helpful *Psychological Mechanisms of Pain and Analgesia*. Let me make a clarification that might be helpful. Of course I agree that pain intensity is correlated with the physical properties optimally tracked: it’s correlated (in a complex, non-linear, difficult to codify way) with noxious temperature level and stimulus size and duration, it’s correlated with size of lesion (in a way that probably can’t be codified at all), and so on. What I mean by saying that in the domain of pain there is bad external correlation and good internal correlation is just that it is much *better correlated* with firing rates in various regions of the pain matrix (I focus on S1 just because as I have said the dominant view is that it plays a special role here). In fact, Coghill reports linear correlations. Perhaps I should make it more clear in future presentations of this material that this is what I have in mind. This is pretty clear and is even granted by one of the philosophers who accept the views of the kind that I am targeting in the paper. I have in mind Chris Hill (see e. g. his interesting recent book *Consciousness* at p. 178 and forthcoming work by him on these issues; I reply to what Chris has said about my arguments at pp. 51-3 of my paper). This is all that is needed for my internal-dependence argument about pain. By contrast, as I have already mentioned, my structure argument does not even require good internal correlation and bad external correlation; it just requires bad external correlation.

    (3) You say that all cases bad external correlation involve “some additional variables; e.g., the individual is attending to something other than the stimulus, or is primed to interpret the stimulus in a particular way”, which you say are not present in the studies I cite by Stevens and Price (summarized in *Psychological Mechanisms of Pain and Analgesia*). A more general claim would be that all cases of bad external correlation in the domain of pain are due to interfering, exogeneous agencies. I guess I have three points. First, you do not quite specifically say how this, if true, bears on my arguments about pain, if at all. I wasn’t sure if you thought this undermines one of my premises or why. (Maybe you don’t think it does undermine any premise.) Second, as I in effect point out in premise 5 at p. 36, what you say is actually not true. There are cases of bad external correlation in which there is no top down influence, hyperalgesia or hypoalgesia, or other interference. Indeed, in the cases studied in the experiments I cited, such factors are not present. For instance, in the Stevens experiment I cite as providing an example of bad external correlation, it is not the case (as you say) that the subjects are “attending to something other than the stimulus, or is primed to interpret the stimulus in a particular way”. His experiment demonstrates what practitioners of psychophysics call ‘response expansion’, which is simply part of the default, wired-in operation of my sense modalities. Likewise, in the many experiments by Price (summarized in the book I cite, *Psychological Mechanisms of Pain and Analgesia*) on VAS and temporal summation and spatial summation when it comes to noxious temperature, it is not the case that subjects are attending to something other than the stimulus, or is primed to interpret the stimulus in a particular way. In general, no interfering agencies are at work. His work on pain psychophysics is about how we experience pain in response to noxious temperatures under normal or optimal conditions. Third, as I say in the paper, “bad external correlation” is a quite general thing. (Again, my arguments are general and don’t turn on anything specifically about pain.) In the paper I also talk about smell and taste resemblance. In another paper I talk about how the relationship between sound intensity and sound pressure is an example of response compression. (Zwislocki, in *Sensory Neuroscience: Four Laws of Psychophysics* summarizes the psychophysical facts and also summarizes experiments showing that sound intensity is much better correlated with the total neural activity evoked by a sound-wave.) These cases of “bad external correlation” are not due to interfering, exogeneous agencies: they are the result of the normal, default operation of the sensory systems.

    Hope this clear some things up and thanks again for taking the time.

  7. Just caught a typo in my discussion of Jennifer’s post. I said according to Jennifer interfering factors are *not* present in the studies I cite. What I should have said is that according to her the experiments I cite do not undermine her claim that in cases of bad external correlation inferring factors *are* generally present (she cites distraction and priming), so that cases where they are absent “are not likely to present dissociations [by which I took her to mean cases of bad external correlation because that is what is at issue]”. My second point under 3 is just that in the experiments I cite by Stevens and Price such interfering factors are not present and yet there is bad external correlation in my sense.

  8. Thanks very much, Miguel, for your helpful comments. Glad to hear you are sympathetic to my views.

    On imperative intentionalism about pain: You are right that it is not a version of the tracking intentionalism which is the target of my paper, so it could not be used to defend tracking intentionalism against my arguments. Still it is an interesting view, so thanks for bringing it up. (Of course it goes beyond Millikan’s claim that pains have imperative contents; it makes the far stronger claim that all phenomenal differences among pains are grounded in differences in such content, so that all there is to pain is imperative content – a claim Millikan nowhere endorses as far as I know. Below I will be raising some problems, but I only think they’re problems for the stronger claim, not Millikan’s weaker claim.)

    I discuss imperative intentionalism elsewhere (‘Do Theories of Consciousness Rest on a Mistake?’, fn. 36 – on my website). There I am discussing Klein, but I think the main points carry over to Martinez’s somewhat different version (disturbance, stop!). (Thanks, by the way, for referring me to the Martinez paper.) The basic points are these.

    First, it is hard to see how imperative intentionalism could answer the internal-dependence argument, illustrated by the case of Mild and Severe. What exactly are the different imperative contents? You say “Mild and Severe differ in the force of this imperative content”. Is the idea that Mild sensorily represents ‘stop, disturbance, soon’ and Severe represents ‘stop, disturbance, really, really soon’? But that’s still not clear. How exactly is the difference in content to be cashed out? And how might a psychosemantic theory entail that Mild and Severe sensorily represent those different contents?

    Second, and relatedly, it is hard to see how imperative intentionalism might answer my structure argument, illustrated with the case of Max (section 5.2, p. 35). On imperative intentionalism, what are the truth-conditions of Max’s report ‘my second pain was roughly twice greater than my first in intensity’? In general, on imperative intentionalism, what in the imperative content corresponds to degree, such that such ratio judgments make sense? (An identity theorist might appeal to firing rates here.) (As far as I know, neither Klein nor Martinez discusses degree but as I recall Klein told me he addresses it in future work.)

    Third, imperative intentionalism seems to entail, implausibly, that if one goes gradually from having a non-painful experience (e. g. of temperature) to a painful one, then the type of content possessed by the experience at some (maybe indeterminate) point radically shifts, from wholly descriptive to wholly (or at least partly) imperative.

    Fourth, even if, despite these problems, imperative intentionalism can answer my internal-dependence argument and my structure argument when it comes to my cases involving pain (Mild-Severe, Max), it certainly cannot answer these arguments when it comes to my cases involving taste, smell, sound and color vision. For instance, I cannot see how it might handle the Harold-Henry case involving smell (section 4.2 of my Consciousness Online paper), or Harold’s judgment ‘my first two smell experiences resemble more than my second two’ (section 5.1). So it cannot provide a general answer to the problems I think my arguments raise not just for tracking intentionalists (the focus of my paper) but for reductive intentionalists more generally.

    Btw, these problems with imperative intentionalism parallel the problems I raise for ‘evaluative intentionalism’ at pp. 51-2.

    On combining Kriegel-Prinz views with theories of function that do not depend on the subject’s environment: Kriegel and Prinz try to combine a standard externalist Dretskean psychosemantics with a Shoemaker-like view of the represented properties. In the paper I argue that this view is mistaken for the simple reason that the two views they try to combine are inconsistent. (As I also point out in the paper, Levine and Hill argue for the same thing, although their arguments are somewhat different.) You seem to suggest an alternative idea: combining a Kriegel-Prinz-Shoemaker-like response-dependent view of the represented sensible properties with theories of function that do not depend on the subject’s environment. That sounds intriguing to me, but I wasn’t quite sure what you had in mind here. (In my paper “A Simple View of Consciousness”, at. p.48, I do discuss combing a Prinz-Kriegel-Shoemaker-like response-dependent view of the represented properties with a theory of experiential intentionality different from tracking intentionalism – one that invokes what I there call ‘the manifestation relation’). But I argue it is a non-starter, and anyway it cannot be what you have in mind.)

    Thanks again for the helpful comments.

  9. I was thinking of non-etiological theories of function. This theories can hardly provide an explanation of why the mental state of a frog is about flies and not about black specks, but I am not so sure about naturalizing the normal conditions in something like your “manifestation relation”. Maybe something along this lines: Mossio et al. (2009) “An Organizational Account of Biological Functions” (they are more ambitious but I am restricting the use to the case of the content of experiences)

    When I see a red apple, an hologram of the apple or when I take a pill that make me hallucinate an apple I have a redness experience. They all have the disposition to cause the experience. In the first two cases the stimuli causes the mental state in “normal conditions” , I attribute redness to the apple and to the hologram, but not in the third case. The state has the function of indicating what has the disposition to activate the state in normal circumstances (in this case we need a theory of function that tells us why the visual system and not the vascular system is normal).

    I think that the content of experience that constitutes the phenomenology is de se, not functions from world to extensions but from world-individual to extensions (something close to Egan’s Appearance properties or Chalmer’s fregean content). I do not deny that there is another content that brings me to something like surface reflectance or colors (a là Chalmers), in fact there might well be a sense in which my experience of the hologram is false, but I prefer the dispositional version to the Fregean one. I think I reject your constraints but I have just taken a quick look to your “A simple view on Consciousness” so it might be that I have missed many points.

    I would like to hear you opinion on these issues. Some of them are far away from the topic of your paper (it is not an externalist position) but in any case, if you are interested, it would be very nice to have the opportunity to talk with you about them here or in another place.
    I come back to the issues on imperative content.

  10. Hello, Adam, Miguel has let me know that, in the course of discussing your paper, you have gone through an imperative-content interpretation of some of your examples. I thought it might be useful to provide quick answers to some of your criticisms to my version of the imperative account of pain phenomenology.

    >First, it is hard to see how imperative >intentionalism could answer the >internal->dependence argument, illustrated >by the case of Mild and Severe. What >exactly are the different imperative >contents? You say “Mild and Severe differ >in the force of this imperative content”. >Is the idea that Mild sensorily represents >‘stop, disturbance, soon’ and Severe >represents ‘stop, disturbance, really, >really soon’? But that’s still not clear. >How exactly is the difference in content >to be cashed out?

    There is no guarantee that we have adequate English counterparts to the imperative contents that underlie painful phenomenology, and I agree that the “really, really soon” version sounds a bit silly 🙂 It might be better to introduce a semi-technical content schema, such as “Disturbance, stop with urgency U”, where U would be higher in Severe than in Mild. What this would amount to could be better seen by answering your next question:

    >And how might a psychosemantic theory >entail that Mild and Severe sensorily >represent those different contents?

    One roughly teleosemantic story about Urgency could be this: The pain-state creator, PS, has used several somatic cues -that have normally covaried with the severity of the lesion, but need not do so- to create mental states that cause the cognitive system to spend a variable amount of resources in avoidance behaviour. For example, for severe lesions, PS has normally created states that prioritise avoidance, overriding any other short-term plans, increase dramatically hearbeat rate, etc. Less severe lesions have normally resulted in PS creating mental states which advocate for a less extreme resource expenditure in avoidance. This kind of behaviour has made PS fitness-conducive, and provides a basis for crediting its product mental states with contents involving the “… with urgency U” modifier: urgencies correspond to the priority that the mental state that PS creates accords to avoidance -btw, that the mental state advocates for a quicker or slower avoidance does not mean that the subject will eventually follow the advocate. It only means (remember, this is teleosemantics,) that in a sufficient number of past instances it has done so, and this has improved prospects for the avoiding subject. All of this is awfully rough, of course, but I hope it does not sound completely implausible.

    >Second, and relatedly, it is hard to see >how imperative intentionalism might answer >my structure argument, illustrated with >the case of Max (section 5.2, p. 35). On >imperative intentionalism, what are the >truth-conditions of Max’s report ‘my >second pain was roughly twice greater than >my first in intensity’? In general, on >imperative intentionalism, what in the >imperative content corresponds to degree, >such that such ratio judgments make sense? >(An identity theorist might appeal to >firing rates here.) (As far as I know, >neither Klein nor Martinez discusses >degree but as I recall Klein told me he >addresses it in future work.)

    If urgencies (remember that this is a semi-technical sense of “urgency”) makes sense, prospects for dealing with the case of Max look better. Two pains with contents “Disturbance, stop with urgency U” and “Disturbance, stop with urgency Udouble”, where Udouble is roughly double than U, could well warrant Max’s report. Urgencies correspond, I have speculated, with the degree of resource expenditure in avoidance that the mental state advocates for. To be sure, it is an open question how good will any cognitive system be in comparing these two degrees. I, for one, am unsure that I can make ratio comparisons among pains with anything more than mediocre reliability.

    >Third, imperative intentionalism seems to >entail, implausibly, that if one goes >gradually from having a non-painful >experience (e. g. of temperature) to a >painful one, then the type of content >possessed by the experience at some (maybe >indeterminate) point radically shifts, >from wholly descriptive to wholly (or at >least partly) imperative.

    I don’t see that as implausible at all -although this carries little conviction: I have some sympathy for my theory 🙂 Anyway, in the case of thermal pain, I think the idea that imperative contents appear only halfway through a process of increasing temperature does conform with our experience of pain. The first pains in the series can be really mild and, as you say, it is indeterminate which of a number of borderline cases counts as the first to carry imperative content. I don’t see anything wildly implausible about this description of the situation with thermal pains.
    Anyway, thanks for the very nice paper and your careful responses to commenters and audience. I’ve enjoying a lot reading this thread.

  11. Thanks very much, Manolo, for your responses to the problems I raise for your imperative intentionalism about pain. Although I reject imperative intentionalism about pain, I’m glad we agree that tracking intentionalism is mistaken (showing this was all I was up to in the paper), and that we need some alternative kind of intentionalism.

    I plan to respond to your responses in a subsequent post when I have a bit more time. I see some problems. For instance, I don’t understand what it would be for a bodily disturbance to stop with urgency, or urgency of degree D. ‘Urgently’ (as opposed to say ‘at 12 noon’ or ‘really, really soon’) doesn’t seem like an adverb that can truly modify ‘stopping’. (I can understand wanting urgently that it stop, or behaving urgently so that it stop, but not its stopping urgently.) I also didn’t follow your psychosemantic proposal about how these contents get represented. But, as I said, I’ll develop my worries more fully in a future post. For now I just wanted to reiterate what I took to be the most important of my points:

    As I just said, I still tend to think imperative intentionalism about pain (while very interesting and worthy of very serious consideration) doesn’t work in the end and cannot answer my internal-dependence argument and my structure argument when it comes to my cases involving pain (Mild-Severe, Max). But even if I am wrong about this, the imperative contents idea certainly cannot be used to answer these arguments when it comes to my all of cases involving taste, smell, sound and color vision. For instance, I cannot see how it might handle the Harold-Henry case involving smell (section 4.2 of my Consciousness Online paper), or Harold’s judgment ‘my first two smell experiences resemble more than my second two’ (section 5.1). So it cannot provide a general answer to the problems I think my arguments raise not just for tracking intentionalists (the focus of my paper) but for reductive intentionalists more generally.

    Miguel: thanks for the interesting follow-up. If we reject tracking intentionalism but retain intentionalism, as I think we should, we need an alternative form of intentionalism. I’m not sure I quite understand yet the particular view you are proposing – at least I don’t have a complete picture of it – but I’m intrigued. I have some questions, which I’ll post when I get a chance. Must go teach!

  12. Adam, thanks for your reply.

    Of course, I realise my post was mostly off-topic 🙂 We can maybe move the discussion elsewhere.
    I think I may have something to say about the smell examples -although it’s unrelated to anything imperative. I’ll try to find time to post about that later.

    Cheers
    Manolo

Comments are closed.