David Chalmers helpfully pointed out that the first relevant use of “first-person data” appears to be due to Herbert Feigl (in “The ‘Mental’ and the ‘Physical'”, 1958). This brings up what seems to me a confusion in the current literature between first-person data as a kind of scientific data and something like sense data.
Sense data are a kind of mental entity postulated by a now largely discredited theory of perception to serve a certain epistemological purpose. They were intended to constitute the foundation of empirical knowledge. They shouldn’t be confused with scientific data, like those scientists collect with their apparata and present in their tables and figures in scientific papers.
Some of the people who talk about first-person data slide between talking about them as the data of a science of mind (data about the mind), and talking about them as if they were the mental states themselves. I don’t see how these two notions can go together. Either the first-person data are mental states, i.e. the object of investigation, or they are facts about the mental states, which can be used in a science of mind.
Compare what Feigl said:
“
Clearly Feigl was in the grips of the classical empiricist view of evidence, according to which all empirical evidence is reducible to something like sense data (which he calls “first person data”). After we update our philosophy of evidence (following Sellars 1956, Bogen and Woodward 1988, etc.), we no longer postulate sense data, let alone follow Feigl in believing that they constitute the confirmation basis for all factual knowledge claims. At any rate, even Feigl didn’t confuse first-person data in his sense with scientific data (in the present sense).
I vaguely remember reading somewhere, in quite old papers on Wittgenstein’s private language argument, a distinction between first-person verbal reports and third-person verbal reports. I think it was in Tugendhat somewhere, so it’s from 60s or 70s. Anyway, first-person verbal reports and first-person data are similar concepts and are still something which is a third-person accessible data…
Way to go, Gualtiero.
I like what you have to say here. I would like to see some further support for the following claim (but not because I dispute it, but because they it so awesome):
“Some of the people who talk about first-person data slide between talking about them as the data of a science of mind (data about the mind), and talking about them as if they were the mental states themselves.”
Also, you write:
“I don’t see how these two notions can go together. Either the first-person data are mental states, i.e. the object of investigation, or they are facts about the mental states, which can be used in a science of mind.”
I’ve been thinking a lot recently about what the so-called data are that are supposed to be explained by various philosophical property dualisms. As best as I can tell, the model of theorizing at play is that of conceptual analysis whereby the “data” are intuitions and the theories work by showing how the intuitions could be true.
The form of the argument, then, goes something like this:
Premise: Intuitively, it seems like there are qualia.
Conclusion: There are qualia.
My big question, and I take this to be Dennett’s question too, is that given that qualia are quintessentially appearances, why doesn’t it suffice to simply identify qualia with the intuition that it seems like there are qualia?
Re: your request for evidence of people sliding between first-person reports as scientific data and as mental states/events.
Chalmers 2004
defines first-person data as data about conscious experience but then talks about first-person data being only directly available to the introspecting subject, and contrasts them with behavioral data.
Horst forthcoming
argues that qualia, or phenomenological properties, play the role of data in psychophysics.
Re: my claim that first-person data can’t be both mental states and the data about mental states. Well, this is more or less an analytic truth, as far as I’m concerned. Scientific data are pieces of information about some thing. They are not the thing itself. To think otherwise is a category mistake.
Re: your big question. I think you are right that the argument proceeds from naive intuitions, and I agree that there is a problem there, provided that “qualia” means something with certain special characteristics (which entail that qualia are non-physical entities or properties). If “qualia” is just a theoretical term in a science of mind, then there may well be qualia, but naive intuitions that they have some property or another cut no scientific ice, and their properties should be investigated empirically “in the third-person” no more and no less than anything else, and first-person data about qualia are no less public than any other scientific data.
Hmm. Certainly my own use of ‘first-person data’ applies in the first instance to certain sorts of information about conscious experiences, and not to the experiences themselves. While of course the term, like any term, can be used in multiple ways, I don’t see any “slide” in the combination of views you note in the comment above. It simply involves the claim that the information in question is only directly available to the introspecting subject (though it may be indirectly available to others). Of course someone might argue with this claim, and of course a lot depends on just what ‘direct’ comes to, but that’s par for the course in philosophy.
Thanks for the (partial) clarification. You say your “use of ‘first-person data’ applies in the first instance to certain sorts of information about conscious experience”. Your use of “in the first instance” suggests that you recognize some ambiguity in your use of the term. That impression is the source of my concern. What does your use of ‘first-person data’ refer to in the second instance?
The slide I’m talking about is part of what gives rise to the claim that first-person data are private. The mental states are certainly private. But unless first-person data are illegimitately identified with the mental states (and unless the introspecting subject is mistakenly seen as the scientific observer of the mind, a la classical introspectionism), I see no reason to say that first-person data are private. (Sure, the information is directly available only to the introspecting subject in some sense of ‘directly available’, but that doesn’t make it private.)
So what’s the difference between third-person data as Dennett understands it (=information about consciousness experiences which are of course directly known to the subject that reports them) and first-person data in your sense? Simply two misleading ways of referring to the same thing?
I take the expression ‘first-person data’ from the literature. In my opinion, it should be used to talk about any data about mental states extracted from first-person reports and similar behaviors. First-person reports are reports like, “I feel tired,” “I’m thinking about Baghdad,” and “I see red.”
Dennett has a view of first-person data. He thinks scientists should suspend judgments on the truth value of first-person reports and instead interpret first-person reports as expressions of subjects’ beliefs about their mental states. I do agree with Dennett that the science of mind–including consciousness–is third-person science. But I think we can do better than what Dennett suggests as a methodology of first-person data, by studying empirically which first-person reports can be used to obtain which information about which mental states (cf. my “Data from Introspective Reports“, J. Consciousness Studies, 2003).
Well, I remember Dennett saying that the heterophenomenology should suspend judgment etc. and that heterophenomenology isn’t all that there is to the science of consciousness, as it is just the only way to begin with. So after you’ve got the reports you can compare them to behavioral cues you get, to fMRI, etc. So I don’t think there is any controversy between you and him. You just seem to say that the heterophenomenological data should be used for further research. But this is clear, nobody would garner such data if they weren’t useful for further research on consciousness.
Do you have in mind any specific passages from Dennett? I don’t subscribe to heterophenomenology, so I wouldn’t say that the heterophenomenological data should be used for further research. I would say that first-person reports can often be used to find out important facts about the mind of the subject, including but not limited to the subject’s beliefs about her mind. Furthermore, I would add (contra heterophenomenology) that first-person reports are not always a good indicator of the subject’s beliefs about her mind. The subject might be lying, or engaging in wishful thinking, or confabulating, etc. It’s no more straightforward to infer beliefs from first-person reports than to infer any other mental states.
Here is what strikes me as a helpful quote:
“When a conscious system is observed from the third-person point of view, a range of specific behavioral and neural phenomena present themselves. When a conscious system is observed from the first-person point of view, a range of specific subjective phenomena present themselves. I think both sorts of phenomena have the status of data for a science of consciousness” (Chalmers 2004).
As I understand it, the last sentence explicitly says that the subjective phenomena (i.e., the experiences) are the data. That’s what I have a problem with. (Of course, I have a problem with identifying even intersubjectively observable phenomena with scientific data. Cf. Bogen and Woodward’s classic 1988 article, “Saving the Phenomena,” in Phil Review.)
Here is what strikes me as a helpful quote:
“When a conscious system is observed from the third-person point of view, a range of specific behavioral and neural phenomena present themselves. When a conscious system is observed from the first-person point of view, a range of specific subjective phenomena present themselves. I think both sorts of phenomena have the status of data for a science of consciousness” (Chalmers 2004).
As I understand it, the last sentence explicitly says that the subjective phenomena (i.e., the experiences) are the data. That’s what I have a problem with. (Of course, I have a problem with identifying even intersubjectively observable phenomena with scientific data. Cf. Bogen and Woodward’s classic 1988 article, “Saving the Phenomena,” in Phil Review.)
G,
“If “qualia” is just a theoretical term in a science of mind… and first-person data about qualia are no less public than any other scientific data.”
There are problems, if “qualia” is going to be treated as “just a theoretical term”. Can it be a theoretical term in the first place? Even the theoretical terms must be at least “in principle” observational in scientific theories. If so, it does not help to appeal theoretical terms a bit.
The second part of your claim. Can the introspective terms be – even in principle – observational? The traditional notion of publicity is based on idea that inter-subjective (=objective) comparability of reports (or, more precisely, of the observation method the reports are generated with) is required if one wants the terms to be observational in the sense required of scientific observation.
Even if a scientist could by introspection produce true beliefs or true reports of her private mental states (= according to Goldman the necessary condition for scientific evidence producing), she would not be making any scientific observation, unless she can express her thoughts i.e. report them publicly, in a such way that others will be able to interpret and understand what she means by her “observational terms”. Hence, unless we are given some justification to assume this is so, we have no reason to assume introspection is a public observation method in the relevant sense. This epistemic constraint of public interpretability of reports is the heart of the publicity constraint. In the case of introspection, this constraint is violated, and because it is violated, introspection cannot be a scientific observation method. Traditionally public validation constraint is formulated as an ability to use some independent means to weed out systematic error as the basis of agreement; it`s purpose is to evaluate the observation method. Introspection fails this (cf. Wittgenstein’s private language argument) and cannot be considered a scientific observation method, maybe not observation method at all.
According to the private language argument we cannot give any verbal report of the content of private phenomenal or cognitive states. We can express our internal state with verbal (and other) behavior, but this behavior cannot be considered as one of making publicly available some information in the form of a public report, if THERE IS NO PUBLIC that could become informed, in the appropriate way, by that report.
So, how could the public be informed?
Thus; if an observational term is a term that is applicable only to publicly observable phenomena, and if introspection is construed as observing privately, but not publicly observable phenomena, introspection (i.e. first-person data) is not a public method of observation.
a
Gualtiero,
Whatever problems Chalmers has, I don’t see that this is a particularly big one.
Suppose that I’m doing an experiment on the relations between temperature and pressure in a gas and I collect the relevant data. One might ask whether the data are (a) a bunch of temperatures and pressures at times or (b) a bunch of measurements of temperatures and pressures at times or (c) a bunch of representations of temperatures and pressures at times. One might also ask why it matters to describe it one way or the other.
Gualtiero,
Part two:
First, some general clarifications. I do not think that there is a huge disagreement between us here. I do think that one can use “introspective report” as a source of data in different kind of experimental designs. However, I am still not convinced, how genuinely “introspective” they are in a classical sense of “introspection”.
As far as I can see, you have two options:
(i) The data your experimental design produces can capture some “introspective” features as understood as private, first person data and the introspective method is observational, no matter what. But then you have to face the publicity constraint- argument. Of course, you can probably avoid Goldman`s formulation of publicity constraint, but the publicity constraint formulated as a problem of public validation of reports is much more vicious one (see the earlier post).
(i) The data your experimental desing produces is about the neuropsychological mechanism that are responsible for introspective behavior (verbal or whatever). Then theoretical terms are inferred to explain observations, but in that case the data is just like any other data in any other standard experimental design. The observational and theoretical terms (of that theory) do not directly to refer to introspective states as mental, private states, right? On the other words; what the terms are inferred to explain observations, and they are not very “private” any more, but it is just like any other psychological terms.
Of course, you could say, that we “could save the phenomena” and try to do a woodwardian-bogenian trick and say that “yes, yes, but the phenomena are still introspection-phenomena”. But then it could be asked, whether or not the phenomena to be saved is actually a third-perspective “phenomena” without any “private flavor”, and hence of course introspective, but not private.
Your decision, dottore.
anna-mari
Well, I can find specific page numbers from his Synthese paper where the idea of HF was first presented, but he repeats it in many places.
Anyway, HF is not about inferring beliefs from verbal behavior in any setting. It’s about verbal reports as produced during psychological research when the subjects aren’t lying, ironising etc. but just, say, pressing a red button or a green one. I cannot find any place where Dennett would say that verbal reports is the only evidence we have about beliefs. He would say (and I agree) that this is the best evidence we could get (as there are no mindreading machines (yet?)) – fMRI is hopelessly crude about my beliefs about, say, heterophenomenology.
I think the best context for HF is the introspection-based research in empirical psychology. In the Polish Lvov-Warsaw school, there was a psychologist Mieczyslaw Krenz who was supporting it methodologically in 50s when behaviorism was the dominant school of thought (in USSR even more than in the US). Well, Krenz was suggesting exactly the type of research Dennett is trying to describe as HF.
If you go the behaviorist way, you simply cut out all data about psychological variances between persons, and a lot of interesting evidence. And if you don’t want to use HF data for research on consciousness, what would you use?
Well, perhaps (b) count as a representations, so they might count as a case of (c). But certainly, there is a big difference between (c) and (a), don’t you agree? (a) are instantiations of real physical properties, whereas (c) are representations of them. Conflating the two amounts to idealism, or naive empiricism, or instrumentalism, or otherwise bad philosophy of science.
In the case of first-person data, there is a further problem. Mental events are private. But scientific data must be public. So if mental events are data, then it turns out that we have a case of private data. So conflating experiences and data forces us either to reject first-person data (as behaviorists did) or to reject one of the most fundamental principles of scientific methodology (as Alvin Goldman argues we should do in a 1997 Phil of Science paper, precisely on the grounds of private data). Neither of these conclusions is desirable. If we distinguish correctly between mental events and data about them, however, we don’t need to draw either conclusion.
Hi, “in the first instance” was meant to leave room for the phenomenon that Pete points to: there’s a common secondary usage of ‘data’ to apply to the entities measured rather than to representations thereof. The passage you quote from me is an instance of that (you can find others in any scientific journal). This sort of variation of usage shouldn’t lead to confusion as long as one doesn’t exploit it in an argument by conflation. This passage obviously doesn’t involve any such conflation — the relevant point could have been made equally well, in your preferred language, by saying “observations of both sorts of phenomena have the status of data for a science of consciousness”.
As for whether first-person data (in the strict sense) are public or private — this depends on what it means for data to be public or private. If it is a matter of being communicable, then I think first-person data are public. If it is a matter of a certain epistemological status, though, one can argue that first-person data about the experiences of a given subject have a special epistemological status for that subject that they don’t have for other subjects (e.g., perhaps they are directly known by that subject, and indirectly known by others). If so, one could argue that they are “private” at least in a sense tied to this epistemological status. I don’t think this is a sense that rules out doing science with these data, though.
I certainly would not take your first option. Your second option sounds more appealing, minus your claim that first-person data would be about the mechanisms responsible for introspective behavior. I think first-person data are, typically, about the mental states or processes that are being introspected. I don’t think there is a problem with saying that mental states and processes are private, in the sense that they belong to one and only one subject and if they are experienced, they are experienced only by that subject. I think a sensible science of mind should be able to say that.
Re: ‘data’ used to mean phenomena. Of course you can use any word to mean anything you wish, and some scientists are notoriously sloppy in their use of words. But to avoid misunderstanding or worse, I think we should be as careful and precise as possible about these matters, especially when we discuss methodology. As Bogen and Woodward have argued, a lot hinges on recognizing the distinction between data and phenomena.
Re: privacy of first-person data. I’m not sure I understand what it means to say that first-person data about a subject’s experience have a special epistemological status for the subject who has the experience. To the extent that I understand, I disagree. Mental states are private. Data are public. In a nutshell, here is how the collection of first-person data works: (1) a subject issues first-person reports, (2) an observer records the reports, (3) an observer (possibly identical to the observer in (2)) extracts some data from the reports, based on relevant assumptions about the reports and what the observer thinks they can tell about the subject’s mind. The subject may or may not be informed of what the data say. The data may agree or disagree to varying extents with what the subject actually said or thinks about himself, depending on the circumstances. I don’t see that the subject has any special epistemological status with respect to the data from her reports. The subject, however, does play a special epistemological role in generating the reports. That is because the subject is the one who has to generate the first-person reports, and is the only one who can generate them. I’ll say more about the epistemological role of the subject next week at the PSA Meeting.
“minus your claim that first-person data would be about the mechanisms responsible for introspective behavior.”
Whoops. Yes, you are right. What was I thinking? I was probably in my typical monday condition; fuzzy and out of control.
” I don’t think there is a problem with saying that mental states and processes are private, in the sense that they belong to one and only one subject and if they are experienced, they are experienced only by that subject. I think a sensible science of mind should be able to say that.”
Yes, of course they are “private”. However, the method of gathering information about the data from the verbal or other introspective behavior cannot be private, because it has to be public in order to be scientific.I think there is no disagreement about that between you and me, since we both seem to be saying the same thing all over and over again.
However, I hold the position that the “science of mind” will loose the privacy, no matter what. There is no way to capture the (essential) features of privacy by any public method of observation, it will be lost. (This is where we disagree.)
But, this is what is bothering me: You start, for example, your 2004 paper by saying that your position does not make you an intropective agnostic. I am not convinced about _that claim_. I guess it may well lead to it.
What is, at the end of the day, exactly the difference between your position and agnostics? This is the question I have been posing all the time…
I
have to add
something to make my point clear. I wrote earlier that the main
difference between your and my position is this: ” There is no way to
capture…”. We do not disagree about that one either, what we REALLY
disagree is the definition of “public method” or “publicity
constraint”, right?
It seems to me that you have been following the goldmanian analysis of
publicity constraint. I follow the classical analysis (for
example Frege, Wittgenstein, Hempel among many others). In that
tradition the notion of publicity constraint is defined via
reportability. In goldmanian it is not. Hence: A lot of
misunderstandings.
My apologies, I should have been more clear.