First-Person Data, Publicity, and Self-Measurement

Chair: Jordan Dodd

Presenter: Gualtiero Piccinini, University of Missouri – St. Louis

Gualtiero’s paper 


Commentator 1: Eric Schwitzgebel, University of California -Riverside


A larger version of Eric’s video

Commentator 2: Jordan Dodd, Syracuse University

Jordan’s paper or a larger version of his video



  1. Thanks to Eric Schwitzgebel for his insightful comments.

    1. I agree that conscious mental states are private in the sense that a subject is the only person to experience her mental states. (I said so in my paper.) As a consequence, I agree that by and large, subjects have epistemic access to their conscious mental states that is more direct than anyone else. I don’t object to including this second claim as part of the sense in which mental states are private.

    2. I disagree that in general, subjects’ epistemic access to their mental states is more reliable than that of others. Eric lumps directness and reliability together, but I am not convinced. True, external observers lack direct access to the mental states. But it doesn’t follow that their access is less reliable. For starters, they can rely on subjects’ first-person reports (as well as other sources of evidence). Insofar as first-person reports are reliable, external observers inherit the reports’ reliability. In addition, external observers may lack some of the biases that subjects have towards themselves, so they may be able to correct for some of those biases in evaluating first-person reports. Therefore, depending on the situation, external observers may have more reliable access to other people’s mental states. That is one reason many people go to counselors: to have their misconceptions about themselves corrected by an authoritative external observer (who in turn relies on the subject’s first-person reports). All of this is relevant because Eric suggests that more reliable access to one’s mental states is part of the sense in which mental states are private. I don’t think it is.

    3. The topic of my paper is not whether mental states are private, but whether *first-person data* are private (i.e., only the subject of the mental states has access to the relevant first-person data). I deny that first-person data are private. Why does this matter? Because if first-person data are private, they cannot be validated by public means, and if so, their use conflicts with standard scientific methodology. So either first-person data need to be rejected (because private), or standard scientific methodology needs to be rejected. I argue that we can keep both first-person data and standard scientific methodology, because first-person data are not, in fact, private.

    4. Eric suggests that no philosopher believes that first-person data are private in my sense. In my paper, I give references to the literature, including some quotes from Chalmers. Many of these authors explicitly say that first-person data are private in my sense, infer from this that first-person data cannot be publically validated, and conclude that the study of the mind falls outside standard scientific methodology. This is what they call “first-person methods” (as opposed to the standard, third-person scientific methods). Many of these people cite each other and seem to be pretty excited about their subversion of standard scientific methodology. This rejection of standard scientific methodology is the main impulse behind my paper – I find it both bothersome and unwarranted. I give plenty of references and there are many more to find. Check them out!

  2. Thanks for the replies, Gualtiero!

    On 2: Would you disagree that I have better access (in the current state of science) to my current imagery experience than you do? Here’s a parallel: The contents of my room are private because it’s locked. You can only know what’s in there by asking me, or by theorizing and making generalizations. My access is some important sense direct and better. But this needn’t be permanent and in principle, if a room-contents detector (that bounces ultrasound through the room, for example) is invented.

    On 3: I think the phrase “first-person data” invites a little confusion — or confused me a little, at least. If the data are the reports of course they’re not private. If the data are the mental states, then there’s an interesting issue. However, I did like your argument that the data should not be thought of as the mental states themselves. So I guess in my commentary I was shifting the issue a little bit. However, I think this shift gets to the real heart of the matter, which is the extent to which a science of the mind needs to think of subjects as in some special and privileged epistemic relationship to their mental states, a relationship that makes the methodology different in kind than in other sciences. On this large point, as I said, I am inclined to agree with you.

    On 4: I think the problematic claim is (4) (“first person data cannot be validated by public means” — and by “validated” I take it you mean confirmed, disconfirmed, tested, called into doubt, etc.), which only follows from (1)-(3) if those are taken very strongly. Your two main opponents appear to be Goldman and Chalmers. Goldman does not accept (4), at least not recently, as you point out. And although Chalmers says some (4)-ish things at other points (as you also point out) he seems to deny (4). I think the key here is in his 2003 “Content and Epistemology of Phenomenal Belief” where he very narrowly limits the sort of beliefs he thinks are infallible or incorrigible.

  3. Thanks to Jordan Dodd for his detailed and provocative comments.

    1. Jordan suggests that privatists might justify the reliability of introspection (in the absence of public evidence) by inference to the best explanation. He draws an analogy to the way some physicists justify grand unified theories by the explanatory power they have. I am not sure how this analogy is supposed to work. For starters, what is supposed to be explained by the “theory” that introspection is reliable?

    2. Jordan also suggests that privatists might justify the reliability of introspection (in the absence of public evidence) on the grounds that introspection is basic and indispensable to the study of the mind, and if we reject the reliability of introspection, then we have to be skeptics about the mental. Consider memory: we can’t justify the reliability of memory without using some form of memory; thus, memory is basic and indispensable to the study of the past; therefore, we should accept that memory is reliable. But there is a big difference: when we establish the reliability of someone’s memory, we either rely on other people’ memory of the same events or on public recordings of the events (or both). So although memory is basic in that we need memory to justify the reliability of memory, the reliability of memory is still established on public grounds. Not so for introspection as privatists conceive of it. Introspection – as the privatist conceives of it – is problematic not because it’s basic, but because it supposedly yields private evidence – evidence that cannot be validated by public means. Therefore, the analogy with memory and other basic sources of evidence does not help the privatist.

    3. Jordan points out that even though we might be able to tell when someone is giving inaccurate reports about their mind (as I suggest in my paper), it doesn’t follow that we are doing so on the basis of evidence independent of (privately conceived) first-person reports. True. My argument does require that we have independent evidence, and I did intend to make that claim.

    4. So do we have evidence about the content of people’s minds independent of their (privately conceived) first-person reports? I think it’s easy to see that we do; Jordan disputes this. I think his reply is partially based on an equivocation. He develops an example in which A hears B giving several reports and uses some of the reports as evidence against the others. Jordan objects that A is not relying on independent evidence, because A is still relying on B’s reports, which in turn are based on introspection. First, I dispute that all verbal reports are based on introspection, and even that all first-person reports are based on introspection, unless introspection is construed so weakly that it can no longer do any reliable epistemic work for the privatist. I think the argument over the validity of first-person data is best conducted leaving introspection aside. Second, all that matters for A to make a legitimate argument is that her evidence be independent of the reports under dispute, not that the evidence be non-first-person. (I think first-person data are public anyway, and that’s precisely why it’s legitimate to use them.) Third, in addition to first-person reports and non-first-person reports, there are plenty of public sources of evidence about people’s minds. Jordan himself mentions “reflexes and other non-intentional behaviors”, which are not so much the “exception” but the rule. There is the tone of voice in giving the reports, the speed of the voice, the facial expression, the gestures, other relevant behaviors, hormone levels, and neural events (which may be recorded). I talk a bit more about our ability to gather independent evidence about other people’s minds in Section 2 of a previous paper (“Data from Introspective Reports: Upgrading from Commonsense to Science,” Journal of Consciousness Studies, 10.9-10 (2003), pp. 141-156). But if Jordan is trying to say that more deserves to be said on this topic, I agree.

  4. Thanks for the further comments, Eric!

    Re: On 2. Would I disagree that you have better access (in the current state of science) to my current imagery experience than I do? If by ‘better’ you mean more direct, I agree. If by ‘better’ you mean more reliable (i.e., such that your judgment about it is more likely to be true), then it depends on whether I’m allowed to ask you and rely on your answer. If I’m not allowed to ask you and rely on your answer, then I agree that your access is more reliable. If I’m allowed to ask you, then I disagree. Since we are discussing situations in which scientists are allowed to ask subjects about the contents of their mind and rely on their answers, I think this is the relevant case.

    Re: On 4. My opponents are all those who argue that there is such a thing as first-person science that is methodologically different from ordinary, third-person science. For instance, many “neurophenomenologists”. I use Goldman (1997) and Chalmers (2004) as my primary foils because they are among the most explicit and clear defenders of privatism. Incidentally, Goldman shifted his position on these matters after he read my 2003 paper on this topic. Whether there is a causal connection, I don’t know. At any rate, I am not primarily interested in what any particular philosopher believes at a given time. There is a view out there, which I call privatism. Many people are tempted by it and defend versions of it. Even some psychologists seem to take privatism seriously these days. I think privatism is influential enough and misleading enough to be worth refuting and replacing with something better.

  5. Hi Gualtiero,

    Thanks for the responses. I found each of them really engaging to think about. Indeed I found myself mulling them even while surfing this weekend. Some thoughts re: 1, 2, and 4:

    1- You’re certainly right that more needs saying here. More needs saying to develop in detail the analogy that I mentioned with unifiers in physics. More generally more needs saying to develop the idea of a privatist-friendly abductive argument that introspection is reliable. But I do find the initial prospects of what I said in my comments tempting. So I’ll try to say something more.

    You ask: ‘What is supposed to be explained by the “theory” that introspection is reliable?’ I think there are a few ways that things could be developed here. One candidate is that such a theory could hope to explain a variety of robust intuitions and epistemic seemings. The intuitions could be ones like the intuitions that we generally retrieve our beliefs, assess our desires, etc. fairly accurately. The epistemic seemings could be ones like it seeming to me that I know that I’m Tofino, B.C. right now. These sorts of intuitions and epistemic seemings look like major features of our mental lives and they seem meriting of explanation. A theory that introspection is reliable couldn’t do all the explanatory work here. But it would seem to pack a lot of explanatory power on these topics.

    This point might become clearer once we emphasize that a lot of work would have to be done to give a rigorous formulation to any theory that we could refer to, in short, as a theory that introspection is reliable. A lot of work would need doing to nail down what the theory means by ‘introspection’ and ‘reliability’, for example. But if this can be done, then we could get seemingly powerful explanations like the following:

    We have particular robust intuitions about the general sucessfullness of belief retrieval (in part) because we have a particular faculty [i.e., introspection, rigorously delineated] that we use (e.g.) to retrieve beliefs and that is reliable [according to a rigorous delineation of the term] when employed under certain ‘ordinary’ conditions (e.g., we’re not drunk, etc.).

    This just says a bit more on the topic at hand. But I think it’s a sufficiently interesting teaser to think that there are prospects for privatists here. Indeed I think that in the end the big questions here might not so much concern ‘could we blend privatism and explanationism to get a privatist-friendly argument that introspection is reliable?, but rather ‘is explanationism really an acceptable route to theory justification?’.

    2- I suspect your response here doesn’t work. Suppose that we can, as you suggest and as seems reasonable, appeal to public recordings, etc. to test the reliability of someone’s memory. Privatists can, it seems to me, happily grant that this marks a disanalogy with introspection. But they can then say that what the disanalogy indicates is (a) that introspection is an even more robustly basic source of evidence than memory (as we can’t, according to privatists, bring public evidence to bear to test the reliability of introspection) and (b) that this suggests that the threat of skepticism really is looming if we don’t assume that introspection is reliable.

    In short the disanalogy you note seems to not undermine the potential for privatists to appeal to a ‘basic beliefs’ reply, but rather to underscore that introspection (at least for privatists) should (or at least can) be taken to be an even more robustly ‘basic’ source of evidence than memory. Another way to look at this reply is to ask which of the three substantive premises of the Basic Beliefs and Independent Evidence Argument that I mentioned in my comments is undermined by the disanalogy you note. It’s not clear that any is – partly for the reasons just mentioned.

    4. You make several good points here. You’re right that the main thrust of my response here was just to push that more needs saying on the topic in order for the self-measuring instrument view to fly. In that vein I still have several questions. It seems to me that several of them might be answered indirectly if you could help me out by walking through an example. I’m particularly interested in the ‘everyday’ cases (e.g., teenage bullshit cases) that you mentioned – as it seems to be that the problem of independent evidence might be a bit easier to solve (at least sometimes) in experimental settings.

    Take the Amanda and Florence case that I used. Is the independent evidence that Florence can bring to bear, on your view, best taken to be things that may be less than intentional – e.g., maybe Amanda’s voice speeding up, maybe Amanda making aggressive gestures, etc.? Or does the second part of Amanda’s utterance count as independent evidence for Florence to use here? If it does, what’s the relevant sense of ‘independent evidence’ at work here – the sense according to which one part of an utterance can easily be ‘independent evidence’ to bring on another part of that utterance? If it doesn’t, what’s the relevant sense of ‘independent evidence’ according to which it turns out that one part of Amanda’s utterance isn’t independent evidence to bring to bear on another part of that utterance?

  6. I find it curious that my two kind commentators have, in one respect, diametrically opposed responses. Eric finds privatism (as I define it) so implausible that he doubts anyone believes it. Meanwhile, Jordan doubts that I have done enough to refute privatism and to propose an alternative; he seems to believe that privatism is unavoidable. In this respect, I think my commentators’ responses cancel each other out. I’m happy to remain in the middle: privatism is a tempting view, but it’s wrong and there is a better alternative — the self-measurement account of first-person data that I defend in my article.

  7. Jordan, many thanks for your further comments.

    Re 1: I don’t see that “robust intuitions and epistemic seeming” are on a par with the physical phenomena to be explained by grand unified theories, so I’m still skeptical that there is a genuine analogy with physics. But I’ll be curious to see if anyone takes up your suggestion and develops it in detail. Are you planning to mount a full blown defense of privatism?

    Re 2: Two things. First, with respect to your “Basic Beliefs and Independent Evidence Argument”, I reject both P1 and P2 (for reasons I give in the paper). Second, basic sources of evidence are scientifically legitimate only if they can be validated by public means. Memory can be so validated, but introspection (as the privatist conceives of it) cannot; that’s why introspection (as the privatist conceives of it) is scientifically illegitimate even while memory is legit. If the privatist were right, then we should be skeptical about the mind. In fact, that is what many behaviorists argued. The reason behaviorists were wrong about this is not that introspection is legitimate even though private/basic; the reason is that first-person data are public data that can be validated by public methods.

    Re 4: If Amanda makes different assertions, each assertion may be used as evidence against the other assertions. Even Goldman would grant that (this point is behind his no-self-contradiction test for introspection). Each assertion is a separate piece of data whose informational value deserves to be evaluated by the observer. Assertions are independent of each other insofar as their truth conditions are independent of each other. Leaving Amanda aside, consider a simpler case: you are very late for an important appointment with someone, you apologize, and she (honestly) says it’s ok. But her facial expression is upset, her voice is edgy, her body is tense, she is chain-smoking, and she acts passive-aggressively towards you. If you could measure her pulse and perspiration, you would detect a full-blown fight or flight response. Contrary to what she said, it’s not ok: she is furious! But she won’t admit it even to herself, perhaps because it would be impolite and she is trying really hard to be civil and feel ok. We all go through situations like this on a regular basis – someone tells us how they feel or what they think, and they may even be sincere, but we know better.

  8. Gualtiero, here’s just two quick things:

    “Jordan doubts that I have done enough to refute privatism and to propose an alternative; he seems to believe that privatism is unavoidable.”

    Everything before the semi-colon is right here. But what follows the semi-colon isn’t. I’ve been trying to tease out ways to support privatism. I think there are some interesting paths available. But I actually don’t have any strong views about whether we should be privatists or not.

    “Are you planning to mount a full blown defense of privatism?”

    It’s a tempting thought. But my work will more or less just be on my dissertation – which involves some discussion of introspection but not privatism – for as far into the future as I’m letting myself plan things.

  9. Gualtiero appears to have three main points against privatism. (1) If “privatism” about first-person data is accepted, a science involving these data is problematic or impossible. (2) Privatism commits a category mistake, as the relevant data aren’t mental states but records of mental states. (3) Given (2), one can have a perfectly adequate objective science using these data.

    Now, “privatism” is a neologism and “private” is ambiguous, and as Eric notes, these can be understood in many ways. Gualtiero defines “privatism” in terms of “private”, and doesn’t explicitly define “private” except to give a brief gloss as “accessible only to the subjects”. So I suppose the core of “privatism” as Gualtiero understands it is that first-person data about conscious states are accessible only to subjects of those states. Let’s call this “strong privatism”.

    It’s true that in my paper “How Can We Construct a Science of Consciousness”, I say that first-person data are private. But I immediately gloss this in a way quite different from Gualtiero: “first-person data concerning subjective experiences are directly available only to the subject having those experiences. To others, these first-person data are only indirectly available, mediated by observation of the subject’s behavior or brain processes.” So the key claim here is that first-person data about conscious states directly accessible only to the subjects. Call this “weak privatism”.

    It’s obvious that weak privatism does not entail strong privatism. In fact, I go on to state explicitly that each of has has indirect access to first-person data concerning others’ experiences, by relying on behavioral indicators of those data. So I explicitly reject strong privatism.

    Now to Gualtiero’s point (1). I agree that strong privatism would make a science of consciousness problematic. Perhaps not impossible, for reasons like those that Jordan gives, but still quite unlike the sort of science of consciousness that has been developed in recent decades, as this view would appear to be incompatible with our having collective justification for claims about others’ conscious states. (At various points in Gualtiero’s paper there’s a danger of getting into verbal issues about what counts as “science”. I don’t much care about that, but I do care that we have a robust method for arriving at collectively well-justified beliefs about the subject matter.) But weak privatism does not make a science of consciousness problematic in the same way — or at least, Gualtiero doesn’t offer any arguments that it does. So I don’t think the sort of privatism that I advocate is threatened by the arguments in Gualtiero’s paper. The same goes for the sort of “privatism” advocated by Alvin Goldman, Gualtiero’s other main target.

    I suppose that there’s also an “intermediate privatism” that allows that we have access to others’ conscious states but that our evidence regarding these mental states consists wholly in their verbal reports. I don’t see why this sort of privatism isn’t compatible with a science of consciousness (it’s consistent with a science of stars that our only evidence for them consists in the data of telescopes), but in any case I don’t accept it. While I think that verbal reports play a fundamental epistemic role, I think that brain measurements and all sorts of behavioral evidence can provide evidence regarding others mental states. And certainly, there are many checks on reliability of others’ reports over and above straight self-contradiction (Eric’s work provides many nice examples).

    Regarding (2), this is really a terminological point about the use of “data”. Of course this term can be used in many ways, and there’s no consensus on the correct use. When Darwin’s observes that a certain finch has a triangular beak, is the datum (1) the triangularity of the finch’s beak, (2) the fact that the finch has a triangular beak, (3) Darwin’s judgment that the finch has a triangular beak or (4) Darwin’s record in his notebook that the finch has a triangular beak? I take it that the ordinary and the scientific use of the term doesn’t dictate an answer here (though (1) and (2) seem natural to me). In my paper, I’m using “data” in a manner analogous to either (1) or (2), where a datum is a directly observed phenomenon or fact, rather than a judgment or a record of that phenomenon or fact. Then in the case of consciousness, it is plausible that experiences or facts about them often serve as data. But obviously nothing turns on this verbal point. Likewise, nothing substantive turns on the different verbal usage that Gualtiero suggests.

    Regarding (3), I think the point in the previous paragraph applies here. One can use “data” for objective records if one likes, but it doesn’t change the substantive issues. As far as I can tell, the core of Gualtiero’s suggestion is that a science of consciousness should take these objective records as its primary explananda. This sort of move is familiar from Dennett’s heterophenomenology (though of course there are important differences between Gualtiero’s view and Dennett’s), and I reject it for familiar reasons. I think that to hold that verbal reports and other objective records are what a science of consciousness should explain is potentially to miss a further class of explananda, perhaps the most crucial in the science of consciousness — namely, experiences themselves. This is not to say that our judgments about experiences are unquestionable, but it is to say that respecting these judgments where possible serves as an important constraint on the science, over and above simply explaining the production of the judgments. (So we have good reason to reject an eliminativist theory that explains the production of reports while denying that there are any experiences, for example.) Of course there is a lot to say about the debate about what constitute the primary explananda here. But I don’t think the verbal point about the use of “data” does much to resolve the issue, and I don’t see much else in Gualtiero’s paper that bears against the view that experiences serve as primary explananda.

    That said, Gualtiero’s self-measurement view serves as an interesting alternative model, one that should be set alongside Dennett’s heterophenomenology in providing a somewhat deflationary reading of what’s going on in the science. Even if one thinks that Gualtiero’s criticisms of more robustly “first-person” approaches fall short of the mark, it is good to have this positive alternative proposal on the table.

  10. I imagine that the worry about weak-privatism concerns the indirectness in the way that an investigator can gain access to the first-person data of others. Start with verbal reports. Suppose we reject the view that verbal reports should be treated as meaningless noises, and suppose we allow that others have conscious experiences and that their introspective judgments are reliable (which seems fair enough; of course we could not *prove* this, in much the way that we may not think a physicist can *prove* that an external world exists). The problem is in giving some reason for thinking the verbal reports convey what the interpreter takes them to convey. After all, we are supposed to admit that the conscious experiences of others may diverge radically from our own, even if we are otherwise similar (functionally). One move might be to appeal to some sort of shared, public meaning for the words we use to describe our conscious experiences, but this is problematic. First, if such a shared, public meaning was not about the experiences themselves (but rather about, e.g., external conditions under which reports were elicited), then we would not be getting the relevant data. But if the verbal reports were about the experiences, we would be begging the question in supposing there was a shared meaning to them (after all, perhaps no two people share experiences, even if everyone has experience). Third, if we allow for a divergence in content between thoughts and words, it may turn out that a person simply could not report their experience in their natural language. For example, if ‘green’ means phenomenal green, but a particular subject uses ‘green’ in response to an experience of red, then the investigator will get the data wrong.

    Now, we might try to get around this by assuming, not only that others have conscious experiences, but that their experiences are more or less like mine (when they are in similar physical states as me, or when they use the words I use to report my experiences). But this is a much stronger assumption, and I don’t think it has an analog in the case of physics mentioned above. Sure, let’s grant that the physicists data is data concerning an external world. But in so granting that, we do not grant the *data*, as this is used in science. That there is an external world is not data; that there is consciousness as such is not data. What are data are the particular experiences (or, in the Darwin case, the shapes of beaks, etc.). To assume that other’s experiences are much like my own seems illegitimate, in this context. But without it, one might wonder whether we are just fooling around in doing the science.

    Here’s is a way one might respond to the above, though. Suppose we assume that nature, generally, is law governed. This seems harmless enough, or, if it isn’t harmless, all the sciences are in trouble. Then perhaps we might say that, since there are robust correlations between physical/functional states and phenomenal states in MY case, it is not unreasonable to assume that they apply to others as well. The problem with this is that it looks rather ad hoc. After all, if we posit a law from which we can derive the correlation in my case, we can posit any of number of laws that will do the same. In addition, it is not as if we could get further confirmation of the law by studying others, because in their case we would have to use the law to justify the conclusion that they were having such-and-such experience (no independent access). In this way, such a law would be very different from other well established laws. In Newton’s case, for example, new experiments added credence to his laws.

    A final out might be to appeal to some sort of principle of simplicity to choose among candidate laws (in an effort to rule out laws which brought in, say, being born in September into the statement of the law). The problem here, I think, is that we have no good prior reason to suppose that being born in September is irrelevant to the phenomenal character of our experience, if we reject supervenience. Sure, there are strong correlations between cognitive (understood functionally) states and experiences, but we are to reject the thought that cognitive states explain the experiences, in any robust sense. Prior to having gathered evidence to support a particular law, it is just as mysterious why certain cognitive states would be correlated with experiences as it is why being born in September would be correlated with having particular experiences. And since (in my case), I was born in September, I might reasonably wonder whether that fact is relevant to the kinds of experiences I have. I cannot test the relevance of being born in September by examining those not born in September because, well, how would I tell in their case if it made a difference?

    Don’t get me wrong: I think it is just silly to suppose that being born in September should matter to the character of experience. But I am also not so sure I can explain why it is silly, if I accept that the physical facts do not determine the phenomenal facts.

  11. Thanks to David Chalmers and Martin Roth for their kind and helpful comments.

    1. The distinction between weak and strong privatism seems to me merely terminological and doesn’t make a substantive difference to my argument. When I formulated (“strong”) privatism by saying that only the subject of experience has access to first-person data, I meant approximately the same as what David means by saying that only the subject of experience has *direct* access to first-person data. The rest of my criticism works the same way, for reasons nicely pointed out and further explored by Martin in his insightful discussion. I will close this loophole when I revise the paper.

    2. I disagree with David that “each of [us] has indirect access to first-person data concerning others’ experiences, by relying on behavioral indicators of those data”. My point is precisely that each of us has *direct* access to first-person data concerning others’ experiences. If we only had indirect access, we would be unable to establish that the first-person data correspond to their supposed indicators.

    3. The point about the meaning of ‘data’ may be terminological but it also has consequences unnoticed by David on what we should say with it. If by ‘data’ we choose to mean the objects of investigation (in our case, someone’s experiences), then what we should say, contrary to privatism (whether weak or strong), is that scientists may directly observe other people’s experiences. This may sound paradoxical, but only if we confuse the notion of direct scientific observation with the notion of directly experiencing our own mental states. It seems that privatism is partly motivated by such confusion. For instance, physicists talk of directly observing the center of the sun (e.g., by detecting neutrinos coming from the sun), even though there are no instruments (let alone unaided human senses) that can go to the center of the sun to examine what is there. This example is discussed at length by Dudley Shapere in a great paper (“The Concept of Observation in Science and Philosophy,” Philosophy of Science, 1982). Shapere argues convincingly that talk of directly observing the center of the sun rightly belongs in the notion of scientific observation. By the same token, psychologists and neuroscientists may be able to directly observe other people’s experiences (even though they can’t *experience* other people’s experiences). Upshot: even if by ‘data’ we choose to mean the experiences, privatism is wrong: in the sense that is relevant to scientific methodology, we may directly observe other people’s experiences.

    4. David suggests that “the core of Gualtiero’s suggestion is that a science of consciousness should take these objective records as its primary explananda,” thereby assimilating my proposal to Dennett’s heterophenomenology (in this respect). If I ever said that a science of consciousness should take first-person data as the primary explananda, please tell me where so I can take it back. The primary scientific explananda are not data but phenomena (cf. Jim Bogen and Jim Woodward, “Saving the Phenomena,” Phil Review, 1988). In this case, the phenomena include the experiences, so those are definitely the primary explananda. This is an important respect in which I, like many privatists, find heterophenomenology unsatisfactory. One important advantage of my proposal over heterophenomenology is that it does not have the kind of deflationary implications that heterophenomenology has. I guess I should point this out explicitly in the paper. So thanks, David, for bringing it up.

  12. Gualtiero: I think there’s now a big lacuna concerning the argument that weak privatism makes a science of consciousness problematic or impossible. The lacuna is even bigger once one makes the adjustment in the use of “data” that you note in 3, so that weak privatism just comes to the claim that only the subject can directly access their experiences, with others restricted to indirect access. There is an obvious argument from “no access” to no science, but there is no such obvious argument from “no direct access” to “no science”. Obviously, there can be sciences of phenomena to which we have indirect access. So much needs to be said to fill in the argument.

    Martin: OK, this challenge is grounded in issues about the meaning of verbal reports, based in the possibility that others have inverted experiences and therefore mean different things with their experiential expressions. I take it that this semantic challenge is quite different from the challenge in Gualtiero’s paper. But it’s an interesting challenge all the same.

    Some thoughts in response: (i) As this stands it isn’t really a challenge to weak privatism (which gives no special role to verbal reports) but to intermediate privatism (which does). (ii) It’s unclear why the possibility that others have inverted experience is more of a challenge to the science of consciousness here than are other skeptical possibilities in science: for example, the possibility that the data of telescopes are generated by some bizarre source, and are therefore unreliable guides to astronomic reality, or even the possibility that much of our perceptual experience is generated by an evil genius. (Note that the mere assumption that “there is an external world” isn’t nearly enough to rule out these possibilities.) (iii) In all these cases, I think standard scientific practice is to make background assumptions that exclude these possibilities, until there is some positive reason to take them seriously. (iv) It may be that simplicity plays a role in these background assumptions: I think that Martin’s doubts about simplicity can be answered, but in any case they would seem to apply equally to the evil genius possibilities. (v) The challenge applies most directly to non-structural features of experiences such as the intrinsic quality of an experience of red, and less easily to structural features of experiences such as spatial features and differences in color experience space. The latter are less easily invertible, and are naturally accommodated by the sort of weak background assumption of structural isomorphism that some scientists make explicitly. So even if the challenge worked for red and green, there’s a fallback position according to which there can be a science of the structural features of experience (a not uncommon view).

  13. I certainly agree with David that a science of consciousness should not be held to a higher standard than any other science, so if an argument against the possibility of a science of consciousness rests on premises that can be used to undermine, say, physics or chemistry, then so much the worse for those premises. With that said, there should be no special pleading for a science of consciousness, and if it conflicts with methodological principles found in the other sciences, this is a problem.

    Regarding weak privatism, the worries about verbal reports apply to brain scans as well–what are we to make of them? Sure, in *my* case, being in a particular functional/physical state is correlated with conscious experiences of red. But why think that is true of other creatures? In other words, how would including brain scans, etc. help at all with the worries about verbal reports?

    The response here is to point out that the same worries apply to telescopes, evil geniuses, etc. But those cases seem different. Take the telescope case. Here, the reason we rule out certain scenarios is that they tend to be completely ad hoc, and explain less than alternatives. If I have an alternative account that explains more data, makes novel predictions that come out true, coheres with the rest of science, and so forth, those are good reasons to accept the alternative. In the case of consciousness, this does not appear to be the situation (again, it follows from how we conceive of inverts, zombies, etc.). The evil genius case is harder, and it looks like the case of trying to distinguish zombies/invert hypotheses from the alternatives. And if I am willing to grant that we should put the evil genius scenarios aside when it comes to science generally, then it looks like I should do the same with the invert/zombie alternatives. But there is a difference. Imagine a theoretical physicist wondering about whether he is the victim of an evil genius. I think it would not be inappropriate for him to say “settling this question one way or another is not going to help me figure out quantum gravity.” The point would be that, even if the truth of the evil demon hypothesis would undermine science somehow, rejecting the hypothesis does not help the physicist much when it comes to settling the questions he is interested in. But in the case of consciousness, it is different. If I assume that others have experiences more or less like mine, then I am actually helping myself to data (or making data easier to come by) that bear on the science. After all, why all the fuss about NCCs? I don’t simply want to know that certain of my brain states are correlated with specific experiences, but whether this is a robust phenomenon. And this, scientists think, requires some real empirical work. But if I assume that others are like me, experientially speaking, then that assumption is doing a lot of the heavy lifting in establishing the correlations. Should not we be neutral on this point? Here is where I think the considerations of simplicity and so forth will need to do a lot of work.

    (d) Another worry in this vicinity concerns the issue of relevance, as mentioned in my last post. Let’s set aside the worry about whether being born in September matters to consciousness. There is the more pressing fact that brains are not identical, so even if I could establish correlations in my case, I could legitimately wonder whether those neural differences mattered to the experiences of others (and to what extent). How could I ever tell, short of simply assuming the facts about the relationship between the physical/functional and experience? And I am not just being picky. Sometimes small differences (e.g., genetic differences) matter to whether behavior/activity of a certain kind (e.g., drinking alcohol) will be correlated with something else (addiction).

    (e) We might think it is reasonable to make certain assumptions until we have positive reasons to consider seriously alternatives. In the case of consciousness, though, I’m not even sure what a positive reason could be to consider the alternatives (inverts, zombies). What test could I perform that would make it reasonable to suppose that my subject was really a zombie, after all?

    (f) With all this said, I don’t think the situation is hopeless. And I grant David’s point that these worries are more relevant to intrinsic character than structural features of consciousness. Obviously a lot more can and should be said (both for and against).

  14. David: I am not sure I am following your last comment. Of course there is no problem with the claim that only the subject can directly access their experience, if by that you mean that only the subject has the experience. That is not privatism: that is a truism, which was never in dispute!

    What is problematic is the move from “only the subject has the experience” to “only the subject has direct access to first-person data”. That is the core of privatism, and it conflicts with standard scientific methodology, according to which scientific methods and data must be public. And again, privatists themselves have pointed out this conflict between their view and standard scientific methodology.

    All I’m saying is that if privatism is in genuine conflict with scientific methodology, then so much the worse for so called first-person “science” (echoing Martin’s last comment). If privatists are willing to rephrase/clarify/modify their view to the effect that first-person data are not private, great! That is what I’m advocating.

    But I doubt that this is what you are suggesting, since you also insist that “once one makes the adjustment in the use of “data” that you note in 3, […] weak privatism just comes to the claim that only the subject can directly access their experiences, with others restricted to indirect access”. Now this statement seems to equivocate between ‘accessing’ in the sense of experiencing and ‘accessing’ in the sense of making a direct scientific observation of. In the first sense, the first clause is true but the second is irrelevant. In the second sense, both clauses are false, but in any case their impact on scientific methodology is less than straightfarward.

    What I said is that once one makes the adjustment in the use of “data” that I note in 3, for the purpose of scientific observation, subjects have no more direct observational access to their experiences than external observers do. Now the burden is on the weak privatist to (1) give reasons to believe otherwise, and (2) show that this generates a special difference between first-person science and the rest of science.

    Privatists are the ones claiming to be doing a different kind of science based on private data. That’s what I have a problem with. If the privatist is simply claiming that scientists have access to some mental phenomena indirectly in the same sense in which other natural phenomena are accessible only indirectly, then the only problem I have left with privatism is their insistence that their science is different from the rest of science.

  15. Gualtiero: “Only the subject can directly access the experience” is not the same as “Only the subject has the experience”. It’s an epistemological claim: others access the experience by accessing something else (say, brain or behavior), the subject doesn’t. Of course just what this comes to depends on just what “access” comes to, but for present purposes it is probably fine to understand access as knowledge and to understand the thesis as the claim that the subject can have knowledge of the experience through direct observation, while others can only have knowledge through indirect observation (that is, by observing something else).

    Now, maybe you hold that people other than the subject can directly observe experiences. That’s a respectable philosophical view, although I think it’s a minority view. In any case that’s a huge issue that I don’t really want to get into here. I’m more interested in arguments that the alternative view is somehow incompatible with science, or (more importantly) with well-justified collective beliefs. Right now I’ve lost track of how those arguments are meant to go.

    Your argument in the preceding comment appeals to the claim that science requires that data are public. But of course you are understanding “data” to be something like records of observations. Of course the view above is entirely compatible with the claim that records of the relevant observations are public, and I’m happy to accept that claim (that is, I’m certainly not a privatist about first-person data in your sense of “data”). So there’s no argument against the view here.

    As for differences with the rest of science: the relevant difference is that there is asymmetric observational access to the key phenomena. In most areas of science, either many people can directly observe a given instance of the phenomenon, or no-one can. In this area, on this view, any given instance of the phenomenon can be directly observed by one observer but not by others. That’s different from what happens in other areas of science, but not in a way that makes it incompatible with science.

  16. David: Thanks for the clarification. I will clarify something too. I agree with you to this extent: people have a kind of direct epistemic access to the qualitative aspect of their experience that they don’t have to the qualitative aspect of other people’s experience (or of anything else, for that matter). It doesn’t follow, though, that in the sense relevant to science, people cannot directly observe other people’s (non-qualitative aspects of) experience.

    I also think that this difference in epistemic access makes consciousness different from other objects of scientific investigation. (NB: I am not saying that it makes the scientific study of consciousness methodologically different from the scientific study of other things.) In other scientific domains, no one has direct access to the qualitative aspect of the phenomena (in the sense in which they have direct access to the qualitative aspects of their experience). We access ordinary phenomena either through our experience of them or through other experimental and observational apparatuses (rather than “directly”). This is a profound difference, but it’s not and does not entail a methodological difference. And the difference is not quite as you describe it.

    You say that in the rest of science, “either many people can directly observe a given instance of the phenomenon, or no-one can.” In the rest of science, no-one ever has the kind of direct access that we have to the qualitative aspects of our experience. In the rest of science, we have direct access in another sense (the sense in which we may directly observe the center of the sun); but we may have that kind of direct access to other people’s experience too. Bottom line: there is still no methodological difference (i.e., difference in how you collect, process, and validate data) between the science of mind and the rest of science.

    Maybe now you will tell me that you are using “methodology” in a different sense, but I don’t think that would be fair. Here is a quote from you that I give in my paper: “Our access to [first-person data] depends on our making certain assumptions: in particular, the assumption that other subjects really are having conscious experiences, and that by and large their verbal reports reflect these conscious experiences. We cannot directly test this assumption: instead, it serves as a sort of background assumption for research in the field (Chalmers 2004, 1117)”. This is the kind of claim that I find objectionable, because by your own admission, this is a background assumption that cannot be directly (?!) tested. This is what I disagree with. Just because people can directly access the qualitative aspects of their experience (but cannot directly access the qualitative aspects of other people’s experiences, or of the center of the sun), it doesn’t follow that there is any more of a principled difficulty in establishing that people’s verbal reports correspond to their experiences than there is in establishing that certain rates of neutrino detection correspond to certain properties of the center of the sun.

Comments are closed.