1. Perception and the Fall From Eden.David J. Chalmers - 2006 - In Tamar S. Gendler & John Hawthorne (eds.), Perceptual Experience. Oxford University Press. pp. 49--125.
    In the Garden of Eden, we had unmediated contact with the world. We were directly acquainted with objects in the world and with their properties. Objects were simply presented to us without causal mediation, and properties were revealed to us in their true intrinsic glory.
    Direct download (2 more)  
    Export citation  
    My bibliography   77 citations  
Back    All discussions

Chalmers' conception of nonconceptual content of perception
Following a suggestion from Dave Chalmers, I am copying to here a query that I originally posted in the Epistemology thread.  Dave will repost his answer and then I will follow up.  I do have some questions about his answer that might be useful.

Dave, since you're still hanging on to this thread, may I change directions a bit and press the question I asked you when I was there?  What I wanted to know was whether it was fair to characterize your position (in your Eden paper) as being that perceptions have conceptual content but do not have conceptual structure.  

Clearly in that paper you are deliberating over whether perceptions have Russellian or Fregean or, your own invention, Edenic contents.  These are all species of what I am calling conceptual contents.  So as I am using the term, it seems to me clear that you think perceptions have conceptual content.  The fact that you opt for Edenic content makes that, if anything, even clearer (since it's rather easier to see the components of such contents as concepts, not properties and not modes of presentation).  

But you characterize perceptions has having "nonconceptual content" because the contents (which I am calling conceptual) might be borne by the perceptions though the perceiver does not "possess" the concepts that go into the content.  So the question is, what are you denying in denying that the perceiver must "possess" the concepts?  My hypothesis was that you meant to deny that the perceptual representation is itself conceptually structured.

To say that a representation is conceptually structured (I stipulate) is to say that it has parts or aspects such that for each conceptual component of the content of the representation there is a part or aspect of the representation that is dedicated to bearing or expressing that conceptual component.  The paradigm case of a conceptually structured representation would be a sentence.  So part of what you're saying in saying that the perceiver need not "possess" the concepts that go into the content is that the structure of a perceptual representation is not at all like the structure of a sentence.  Is that right?

Chalmers' conception of nonconceptual content of perception
Hi Chris,

The issues here depend on what one means by "conceptual content" (and "have conceptual content").  I don't know how you are using the term.  One part of your message suggests you're using it for "contents whose components are concepts".  Even on that usage, it's not obvious that Russellian, Fregean, and Edenic contents are conceptual contents.  The components of the first and the third are properties (albeit uninstantiated properties in the Edenic case).  The components of the second are intensions or Fregean senses, which I suppose that some might identify with concepts.

But in any case I'm not using "have conceptual content" in that way.  There's a standard distinction between the "state" and "content" interpretations of this talk, and where you're using the content interpretation, I'm using something closer to the state interpretation.  Here, to have conceptual content is to have content in virtue of being in a conceptual state.  The residual question is then what being in a conceptual state comes to.  This is itself frequently glossed in terms of grasping or employing concepts, and certainly a necessary although not sufficient condition is possessing concepts.  Here the notion of concept possession is tied directly to the capacity to judge, think, or believe: to possess a concept, one must be able to deploy that concept in judgment (if one sees concepts as abstract objects, one must be able to form judgment with that concept as a content).  I think that one can have perceptual states with these contents without having the capacity to form judgments with these contents.  So (by the conditions just outlined) one can have perceptual states that have these contents not in virtue of being conceptual states.  So these states have nonconceptual content in the relevant sense.

I haven't had to say anything about parts or structures of representations here.  I think the picture is neutral on whether perceptual representations have sentence-like structure.  Certainly nothing here excludes the possibility that the states have sentence-like structure.  It's the absence of a necessary link to the capacity to judge that makes them nonconceptual.

[Reposted from here.]

Chalmers' conception of nonconceptual content of perception

Thanks.  That helps.  But your answer does not entirely deter me in my interpretation, on either score, conceptual content or conceptual structure.

Regarding conceptual content.  I understand that it is awkward to describe a Russellian or Edenic content as "conceptual content", but I am supposing that since you are prepared to describe a perception with a "that"-clause, you are supposing that a perception has a conceptual content.  So if a perception merits a "that"-clause by virtue of its association with an Edenic content, which the "that"-clause denotes, then we can call that Edenic content the conceptual content of the perception.  So if an Edenic content is a thing that is built out of individuals and properties (uninstantiated, as it happens), does that mean that individuals and components are components of a conceptual content and in that way themselves are concepts?  Yes, and that may be surprising, but it's not out of the question.  (Don't get me wrong.  I'm not really speaking for myself here.  Ultimately, I want to banish contents from philosophy altogether.)

In opting for the "state" interpretation over the "content" interpretation of talk of "conceptual", you are telling me, I take it, why you use the term "nonconceptual content".  Yes, you do have to mean "conceptual" in the state sense, because you do describe perceptions as "nonconceptual" while, if I am right, you do after all ascribe conceptual content to perceptions, which shows up in your willingness to describe them by means of "that"-clauses that you interpret in terms of proposition-like entities.

Regarding concept possession:  You say that the sense in which the bearer of a perceptual representation may fail to possess the "corresponding concepts" (your phrase, p. 122) is that the bearer of the perception may fail to be capable of forming judgments by means of those concepts.  

First, could you clarify what you mean by "corresponding concepts"?  I take it you must mean concepts that are components of or at least characterize the conceptual content (my sense) of the perception, right?

But now what do you have in mind in saying that the agent may not be able to form judgments having that concept as a content?  Why is not the very perception that, in your words, has a "quite determinate content" not a judgment of the requisite kind?  The difference must lie in the difference between the kind of representation that a perception is and the kind of representation that a judgment is.  And in explaining that difference I can imagine you going in either of two directions -- functional or structural.  You might say that perceptions have a different functional role in cognition from judgments, or you might say that judgments and not perceptions have conceptual structure.  But does not the functional account eventually lead back to the structural account?  Isn't it precisely because of a structural difference that the perceptions have a different function from the judgment? (Surely it's not just a difference in location or modules.)  Anyway, that's the position I am trying to get you to own up to.

By the way, in suggesting that you deny that perceptions have conceptual structure, I did not mean that you're just denying that they have sentence-like structure.  I said that sentential structure is the paradigm of conceptual structure, but I did not mean to identify the two.  (Maybe I'm being a bit perverse.  Here you are saying that you could allow that perceptions have sentence-like structure, provided the tie to judgment were lacking, and I am still interpreting you as denying that perceptions have sentence-like or any other kind of conceptual structure.)

Chalmers' conception of nonconceptual content of perception
Re the content interpretation: it appears that you take content to be conceptual if it's the sort of thing that can be ascribed by a 'that'-clause.  I don't think that this is an especially standard usage, but I'm prepared to allow that if this is what you mean by "conceptual content", at least some of the contents I discuss are conceptual.

Re the state interpretation: yes, the concepts here would be components of the conceptual content (not necessarily in your sense, but we can go with that for now).  And yes, the picture requires the assumption that perceptual experiences and judgments are different sorts of states.  I don't think the picture requires any particular theoretical claim about just what the difference amounts to, and I don't have a well-worked-out view on that myself.  Some obvious differences include phenomenological differences and functional differences with respect to inference and behavior (e.g. judgments can be conclusions of inferences from other judgments whereas perceptual experiences can't be conclusions of inferences from other perceptual experiences).  Of course there are lots ot tricky issues here, and I don't rule out the hypothesis that the most important underlying differences might end up being something structural.  But I don't think that this hypothesis is an immediate commitment of talk about having nonconceptual content.

Chalmers' conception of nonconceptual content of perception
Dave, just one reply regarding terminology:  I don't see how it can be strange or idiosyncratic to describe the thing ascribed by a "that"-clause as a conceptual content.  Although I am not prepared to detail the history for you, I feel it's safe to say that traditionally the thing ascribed by a "that"-clause is a proposition and the components (at least the predicative components) of a proposition are concepts.  A proposition is the intension of a sentence (which is not to presume that the mind can be related to one only by means of a sentence).  And concepts are the intensions of the predicative components of a sentence (at least those).  If now the Russellians or the Edenists want to say that we don't need intensions, just the properties denoted, and that propositions can be built from those, then they should be prepared to say that concepts turn out to be properties.  Here of course we are talking about concepts in the sense of what is shared, not the mental particulars that express or bear or participate in those shared things.  If anybody out there would like to set me straight about this, I would be eager to hear what you have to say.

Chalmers' conception of nonconceptual content of perception
@ Christopher Gauker:
As far as I understand, the underlying reason for calling Russellian (and Edenic) contents nonconceptual is that philosophers involved in the nonconceptualism vs. conceptualism debate want to contrast them with Fregean contents. They equate the Fregean senses constituting these Fregean contents with concepts, and at least some of them (e.g. Peacocke and McDowell) seem to think that we should talk about Fregean senses/concepts constituting the contents of our thoughts because conceptual abilities are involved in thinking these thoughts. Since Fregean contents introduce a level of abstract objects (the Fregean senses that supposedly correspond to our conceptual abilites) in addition to the properties and objects thought about, and Russellian contents do not, I think it makes a certain amount of sense to call the former concepts and the latter just properties and objects.

I also have a general question: there are philosophers who think ((a), like David) that a perceptual state P has a Fregean content f and ((b), unlike David) that P has f in virtue of subject S somehow employing her corresponding conceptual abilities (C-abilities) in undergoing P. But one thing that I am not sure of is whether these philosophers should claim that S employs her C-abilities just like she would in thought, or whether they should hold that S's C-abilities are involved in undergoing S in some other way.

An example: When I think that the cat is on the couch, I actively exercise my concepts the cat, x is on y and couch, and thereby manage to have this thought. But should the conceptualist say the same thing of my visual perception that the cat is on the couch? A problem that I see for this position is that I don't have control over what I perceive as I have over what I think, and this control seems to be closely connected to my actively employing my concepts. The alternative would be to say something like McDowell, for example that conceptual abilities are "at play", but not actively employed, in perception. My problem with this proposal is that it is not clear at all what it would be for conceptual abilities to be at play (by contrast to being employed).

Any thoughts?

Chalmers' conception of nonconceptual content of perception
Reply to Eva Schmidt
Dear Eva Schmidt:

That's an interesting interpretive proposal, which I had not thought of.  If one held that perceptions have Russellian or Edenic content but denied that they had Fregean content, then, inasmuch as it is at least awkward to say that that the former are composed of concepts, it would be a reasonable use of terminology to say that perceptions have contents but lack conceptual content.

The trouble is that I don't see how this interpretation makes sense of what the nonconceptual content people actually say.  What they (Bermudez, Peacocke, Tye, Chalmers) always say is that perceptions have nonconceptual content in the sense that the perception may have the content though the perceiver does not possess "corresponding" concepts (Chalmers) or concepts "required to specify" the content (Bermudez).  It does not seem very likely to me that what they mean in denying that the perceiver "possesses corresponding concepts" is that the perceiver lacks a corresponding Fregean senses.  And that is not what Dave said when he addressed the question on February 19 above.  And I don't remember Bermudez or McDowell even making much of the distinction between Russellian propositions and Fregean senses.  I don't think they're in that game.

As for the rest of what you say, I am not sure that there is any good sense in which our nonperceptual thoughts are free and our perceptual thoughts are not.  OK, maybe if I decide to believe something, that's a "free choice".  But the process begins with hunches and hypotheses having the same sort of contents as my ultimate beliefs, and I think that those are any not more freely chosen than my perceptions.  So I am questioning your argument, but not because I want to attribute contents to perceptions; I don't.

--Chris Gauker

Chalmers' conception of nonconceptual content of perception
Recalling the paper, wasn't there a line in it to the effect that content could be a condition on extension? In terms of determination of associated intensions (75)? I took it that one was being encouraged to read involvements at different levels out of the contrasts offered by Chalmers - in terms of different disjunctive conceptions for instance .... but that an Edenic content was like a tracking or regulatory content, or something like a device of regulation; even as a kind of abstraction. The question in the rough vicinity that interests me, assuming I'm understanding some of this, is: if there is judgment involved as to the extent or applicability of featured properties - in the guise of conditions on content - wouldn't judgment be tracking at least conceptions of 'content'? Wouldn't conceptions of contents be the sorts of things that could be paraphrased in terms of "that"-clauses?