The Uninviting Room:  Representations without Contents

 

 

 

 

 

 

Anne Jaap Jacobson

Department of Philosophy

University of Houston

Houston, TX 77204-3002

ajjacobson@uh.edu

 

 

 

 


 

The Uninviting Room:  Representations without Contents

 

A room can look uninviting.  In some cases the reason is quite simple:  One imagines walking through the room as one looks into it, and the imagining gives one a sense of unease.  For most cognitive theorists, such an imagining would involve a representation with intentional content.[1]  Looking at recent literature, we might indeed expect the debate to center around whether the content is conceptual or not.[2]

I will argue that such imagining need not involve states with content at all; in so doing, I extend the arguments and considerations presented earlier work.[3]  The implications of this thesis, and the arguments for it, are notable.  First of all, we will be examining a kind of representation which is almost entirely overlooked in the philosophical literature.  As I indicate below, the conception of representation connects in interesting ways with questions about compositionality[4], and with questions about the disjunction problem.  Further, imagining involves possibilities, and hence the case discussed presents a counterexample to the idea that we can represent possibilities only through intentional content.[5]  Secondly, the discussion is concerned with an area of mental life that is often treated as an instance of relatively higher cognitive functioning at the personal level.  However, as we will see, the case as described appears to consist largely or entirely in the exercise of capacities we clearly share with many other species, including rats and cats, and that is continuous with functioning on the sub-personal level.

Thirdly and finally, the discussion sets the stage for an investigation, not pursued at all fully in the present paper, of relationships between perception and action, and the implication of such relationships for the debate about  conceptual versus nonconceptual content.  The fact that the case in question can be accommodated without any appeal to content shows that two problematic and ill-defined notions[6]  – “conceptual” and “content”– can be dispensed with, at least for an important range of cases.  It additionally suggests that a philosophical investigation of perceptual experience conducted, as it is too often, by a philosopher alone with a desk is concerned only with a very partial paradigm of such experience.[7]  

In what follows, I will use the case of imagined walking to introduce the overlooked kind of representation.  The discussion leaves open the question of how the imagined movement gets linked to one’s environment in the sort of case being considered.  I will use Milner and Goodale’s discussion of dorsal stream processing to provide a link without appealing to intentional contents (Milner and Goodale, 1995).

 

1.  Imagined Movement

            Recent research strongly suggests a particular account of what goes on when we imagine movement “from the inside.”  Such imagining contrasts with a more external and “third person” imagining; that is, it contrasts with cases where one imagines oneself moving in the same way as one might imagine someone else moving.  The more first person kind of imagining occurs when one imagines one’s own body from the inside making those movements; this is a kind of imagining that one cannot do about others. Imagining of the first person sort is a kind of rehearsal, and it seems to be used by many species as a way of finding out about some of the consequences of an action before it is performed. Among other things, it can give us information about how long the action will take.  

The representations involved in such cases are interestingly described in the recent literature.  In “The Mental Representation of Hand Movements after Parietal Cortex Damage,” (Sirigu et al., 1996) examine how one has knowledge of spatial and temporal features of one's actions before they are performed.  Subjects are asked to imagine themselves making some motions.  The authors talk, as the title indicates, about the resulting mental representations.  These representations have a crucial feature:

[Previous researchers] have shown that motor imagery can be used to predict the time needed to complete a movement, and that the mental reenactment of an effortful exercise causes the same vegetative changes as its actual performance.  Studies of cerebral metabolic activity have demonstrated that most of the regions that are active during overt movement execution such as the parietal and premotor cortices, the basal ganglia, and the cerebellum are active during mental simulation as well.

            In the case of imagining from the inside, then, research shows that what goes on is a neural enactment of a part of the action(s).  It is not that one has some representation that is about one’s moving; rather, one runs the neural routine off-line. Thus, writing on the representation of action in rats, (Leggio et al., 2000) maintain,

Action learning by observation has been considered somehow related to motor physiology, because it provides a way of learning performances that is almost as effective as the actual execution of actions.  Neuroimaging studies demonstrate that observation of movements performed by others, imagination of actions, and actual execution of motor performances share common neural substrates .

These representations are a special case of what I have called elsewhere “token-realizations.”  Token-realizations represent by realizing features, and not through some content that is about such features.  The above neural substrates are counted as representations of movements because they realize the neural parts of the actions. 

            The conception can be made clearer through analogy with some everyday examples.  For example, there are two ways of informing you of what someone said.  I can describe it or I can repeat it by producing tokens of the same type.  Similarly, I might describe someone’s gesture or I might repeat.  In the later case, one’s gesture is not about the other person’s gesture; rather, it displays what was done.  Another example:  one might try to describe a particular color or one might display a sample of it.  The color sample does not have any intentional content linking it to, for example, the paint on the new doors, but it can nonetheless represent their color.  In all these cases, something is represented, but not by one’s using something with content that is about the thing.

            I have said that token-realizations are merely analogous to these ordinary examples to forestall an objection.  The objection claims that intentional content is still part of the representation because all these ordinary cases rely on a conversational context.[8]  Since the cases principally under discussion in this paper will be brain representations, the issue of conversational context is really irrelevant.  However, even in conversational contexts, it is far from clear that the spoken context is in some sense part of the representation; we can see this by considering the very closely related case of examples.  Thus, it may take a verbal context to make someone aware of an example, but that does not mean that the verbal context is part of the example.  Thus a particular bird may be an example of a curlew sandpiper without anyone’s saying anything; there are indeed many good examples of Curlew sandpipers upon which no one has ever remarked.[9]  Hence, in public contexts it may be that an example will not function as an example for my interlocutors unless I say something, but that does not make the saying part of the example.

The case of examples can also be used to bring out a much more serious objection which we will discuss in the final section of this paper.  Notice that one’s saying something does point out some one respect in which the bird is an example. Without some sort of restriction, the bird is an example of everything that has mass, everything that has eaten today, and so on, ad inifinitum..

            As I have just indicated, token-realizations as representations bring in problems, but before we take a look at these problems, we should be aware of how token-realizations will figure in the central project of this paper, giving an account of a particular kind of imagining without invoking representations with intentional content.  We have put in place one ingredient in such an account:  imagining walking involves token-realizations in so far as the representation of the action consists in a neural routine that instantiates some of the neural parts of the action.  However, we are still far from explaining how one imagines entering a particular room.  What licenses the appearance of the room in the description of what is imagined?  The role of a doorway is very different in an actual walking through it and in a merely imagined walking through it.  How does the doorway connect up to the running of the neural routine? 

Of course, a standard answer starts by saying one has an intentional representation of the doorway and the room beyond; presumably, the standard answer would continue that the imagining is of the room because of various causal relations.  The account I am developing here will provide an answer that similarly invokes causal relations, but it will not bring in contentful representations.  Rather, we will understand “Joan imagined walking through the room” solely in terms of token-realizations.

 

2.  Connecting the Room

            It has been know for a while that visual input goes to the occipital cortex and then is further carried in two streams, a ventral one and a dorsal one; we have also had a general account of the difference between them.  Milner and Goodale have proposed a much more detailed account of what happens with the two streams.  On their theory, the ventral stream yields the conscious perception.  The dorsal stream, on the other hand, has an unconscious product which is concerned with the spatial features of the environment from the perceiver’s perspective.  The conscious perception has an impersonal lack of specificity about it; it will leave one without precise knowledge of the relative distance and relative orientation, for example.  Not so the dorsal stream, which provides information needed to, for example, rotate a key to get it into a lock.

            If one is going to imagine walking through a room, then the most relevant information is carried by the dorsal stream process.  This information just is that which is used in orienting one’s body spatially.  But the representations produced by this phase are token-realizations.  This is so because the dorsal stream gives us, in Goodale’s word, “isomorphism.”  Let me explain.

            Token-realizations are at bottom instantiations.  They represent something because they co-instantiate a property with it.  And isomorphism gives us co-instantiation.  That is, if A is isomorphic to B, then there is some general property which A and B share.  According to Goodale, what dorsal stream processing gives us is isomorphism.[10]  Hence, it gives us token-realizations (though, given the use of isomorphism in various accounts of intentionality, it is important to stress that isomorphism gives us only one class of token-realizations).

            The picture that emerges from this is the following:  Imagining moving through a room may consist in a causal sequence of token-realizations.  A neural array isomorphic to the room is one token-realization and a motor routine of walking run offline is another. 
The second may in fact be produced automatically in response to the features of the first.[11]  That is, the imagined walking is imagined walking-through-the-room because it is caused by a representation of the room.

The sorts of imagining I am describing may be largely or entirely out of awareness.  What reaches consciousness may be only the sense of unease that leads one to judge the room to be uninviting.  The point is that to explain the sense of unease we do not need to bring in intentional contents in prior representations.

 

3.  A different kind of imagining

            An account of imagining moving through a room in terms of token-realizations gives us a better picture of what is going on in such cases than does an account which appeals to representations with intentional content.  There are two ways in which it does so; first of all, it emphasizes the important and contrasting role of first person imaginings.  Secondly, it bring out the fact that such imaginings are not compositional in an important way.

            It is tempting to suppose that the description “Joan imagined walking in the room” applies because of a contentful representation, or series of such, that Joan had, either “I am walking through the room” or “Joan is walking through the room.”  In many such cases, Joan can believe she is walking through the room, or imagine it, or wish it were not so, just as one can oneself believe Joan is walking through the room, imagine Joan is walking through the room, wish Joan were not walking through the room, and so on.  In each series, the subject is related to the same propositional object, but with a difference force.  And it seems easy to assume that one should look for some event in the brain that is the bearer of such a content, or at least is what puts one into a relation with the abstract object. 

            The account I have offered above points to a very different sort of imagining.  The difference between Joan’s first person imagining when it involves a neural routine and an imagining of herself in the third person is much more than a difference to be captured as a different in propositional content.  Imagining of this sort is, as I mentioned above, a kind of rehearsal.

            There are other interesting features.  This physically based imagining does not appear to be compositional. The same term, “walking,” occurs in many different contexts; one can indeed hope that one will walk on the moon and, in the same sense of “walk”, believe one will walk on the moon. However, imagining walking that involves invoking a neural routine is very different; I might, for example, be able to imagine myself walking through a room, but be unable in the same way to imagine myself walking on the moon. One could, of course, imagine walking ordinarily in a very bare place and call it “imagining walking on the moon,” but most of us just do not have the appropriate neural routine to invoke.[12]

 

4.  Representation and Restriction

         We will now look at the central problem facing token-realizations as representations.  It will turn out that looking at the various alternative solutions will raise far reaching questions that cannot be settled in one paper.  Consequently, it is important to stress that we can nonetheless show that the central problem for token-realizations can be solved, at least as well as bringing in a conception of intentional content would.

The problems with token-realizations can in general be though of as falling into two groups:  (1) those involved in keeping intentionality out, and (2) those that are more concerned with getting the logic and extension right.  Objections of the first sort concern issues about whether intentionality is covertly at work; we saw one possible objection that falls into this group earlier when we considered conversational context and examples.  Those of the second sort focus more on problems such as the reflexivity that is apparently possessed by “coinstantiating” and not by “representing.”   I’ve discussed these problems in forthcoming work, and in this paper I will restrict myself to what we can call “the promiscuity problem”, in order to introduce the distinction between descriptive and referential approaches to token-realizations and to introduce a link with the disjunction problem (or the problem of misrepresentation).[13] 

         Let me state the promiscuity problem as it applies to the case with which we are dealing.  The output of the dorsal stream processing represents a room, it is claimed, because it is a token-realization of the room, and it is a token-realization because it is isomorphic to the room.  The promiscuity problem here arises from the fact that the output is also isomorphic to many other things and it is indeed a token-realization of just about everything.  Because the semantics of “token-realization” is very like that of “example” - almost anything is an example of almost everything - a token-realization represents just about everything.  But, it might well be claimed, such a notion of representation is useless.

            Such a reaction would be premature.  The notion of an example is much more useful than its promiscuity suggests and, similarly, the notion of a token realization can be much more useful.  It is interesting to see how this is so.  There are problems in two different dimensions the answers to which provide the account we need.  In describing each problem, I will talk about “restrictions,” to signal the idea that we need restrictions so that a token realization has a determinate representation. 

            The first question is whether the restrictions should be pragmatic or semantic in nature.  If we choose the former, we would say that everything really is a token realization of just about everything else, but we can nonetheless understand the import of explanations employing token realizations quite differently.  Just as an event may be one of many different causal factors but still describable as “the cause,” so some one thing can be understood to be what is represented by a token realization. 

Typically, any one event, E, will have many factors in its causal history, and each of these factors will have many effects; citing one F as the cause of the E requires normally a considerable conversational setting and presumed convergence of interests.  However, we do not have to understand any of this as part of the semantics of “cause,” nor do we have to think of the actual cause – perhaps the striking of a match – as itself replete with intentionality.   

Similarly, as I have argued elsewhere (forthcoming), the restrictions on what is represented by a token-realization can be understood to function as pragmatic conditions.  Among other things, this move preserves the analogy of token realizations to “example.”  Thus it may be that some at least quasi-linguistic context is needed before anything can be a useful example, but this is not, I think, at all part of the semantics of “example.”  The world contains examples of curlew sandpipers that we may never see or name.

            To say this, of course, is not to say that the alternative of writing the restrictions into the definition of “token realization” has been ruled out.  Instead of simply defining “token-realization” in terms of co-instantiation, we could do so in terms of, for example, co-instantiation and cause, or cause-plus-other-factors.  However, the point of this section is to show that there are paths to a solution to the promiscuity solution, and not to argue that one alone is the best.  It may also be the case that the semantic alternative would produce a useful notion for some cases.

            The second question is whether we will understand the restricted token realization relation as more like a “bare” referential relation or more like a descriptive relation.  Is what makes the token realization represent the room a relatively simple causal relation?  If so, we will say that the relation is like a bare referential relation.  Or does it need more?  For example, if causation is not enough, and we need to add in something like the role the token realization  has for the organism, then I will say that it is more like a descriptive relation.

            For now we can understand the bare referential account as falling into the class of causal theories in philosophy of mind; it would say that a token-realization represents some thing, F, of which it is a token-realization, just in case F causes the token-realization.  What would a descriptive account look like?  There is a wide array of resources available to provide the factors needed for a descriptive account; namely, those that have been developed and classified in many naturalized accounts of intentionality.  Accordingly, we can take the elements of a naturalized account of intentional content and use them to narrow the class of relevant representeds.  There is no bar to putting any or all of the elements of any one consistent account into an account of token-realizations.  Doing so would enable us to fix the representeds of token-realizations as well as any naturalized account invoking intentional content can. 

            But do we need to use the technical resources of naturalized accounts of content?  Could it be the case that the token-realization is a token-realization of a room simply because it is isomorphic to the room and caused by it?  If this is so, then the token realization is semantic-referential or pragmatic-referential. 

            The deciding issue may be also one of the central issues in the development of accounts of intentional content.  Do token-realizations fit into explanations in ways that depend on a solution to the problem of misrepresentation?  When a neuroscientist points to a neural firing pattern and says that it is the banana’s movement in the monkey’s brain, does the success of the explanation depend on anything’s legitimizing the neuroscientists use of “banana,” rather than “banana or something else rather like a banana.”  If the success of the explanation depends on our locating such legitimizing resources somewhere in the conceptual mechanism employed in the explanans, then it looks as though we will need a descriptive account of token-realizations. 

            While the problem of misrepresentation is a problem for naturalized accounts of intentional content, it is hard to see that it is a problem for an explanation that employs a contentless representation.  Hence, the evidence favors a pragmatic-referential account, which could be extended to a number of cases in the literature, including the bb-swallowing frog.  But the important point here is that naturalized accounts of intentionality nonetheless present resources that we can use if we decide that a descriptive account is needed, and we can use them without claiming that the naturalized account has actually succeeded in explaining intentionality.  If the explanation of the monkey’s movement needs to be deeply connected to such things as the role of the token-realization in the monkey’s mental economy, or what the monkey has evolved to react to, then the descriptive account may well be appropriate.  But whether this is so is independent of the question of whether the promiscuity problem can be solved, because both the descriptive account and the referential account are available to an advocate of token-realizations. 

             

 

1.      AKINS, K. 1996. Of sensory systems and the aboutness of mental states. The Journal of Philosophy, 93(7), 337-372.

2.    Author. (2002).  Ed special issue of Phenomenology and the Cognitive Sciences.

3.    _____.  forthcoming. “Mental pepresentations:  What philosophy leaves out and neuroscience puts in,” Philosophical Psychology

4.       

5.      Churchland, Paul. 1995. The Engine of Reason, The Seat of the Soul. Cambridge: MIT Press.

6.      _____  . 1998. Conceptual similarity across sensory and neural diversity: the Fodor/Lepore challenge answered. Journal of Philosophy, 95, 5-32.

7.      CLARK, Andy. (2001).  Mindware  Oxford: Oxford University Press.

8.      Crane, Tim. (1995). The Mechanical Mind. London: Penguin Books.

9.      _____. 2002.The hot theory of sensations. The Times Literary Supplement, 5761, 9-10.

10.   Cummins, R. 1996. Representations, Targets, and Attitudes. Cambridge: MIT Press.

11.   Dretske, F. (1995). Naturalizing the Mind. Cambridge: MIT Press.

12.   Fodor, Jerry. (1975). The Language of Thought. New York: Crowell.

13.   _______ . (1990). A Theory of Content and Other Essays. Cambridge: MIT Press.

14.   Gazzaniga, M. S., Ivry, R. B. and Mangun, G. R.. 1998. Cognitive Neuroscience: The Biology of the Mind. New York: WW Norton.

15.   GUNTHER, Y.H.  (2003(.  Essays on Nonconceptual Content..  Cambridge:  MIT Press.

16. Hardcastle, V.G. (1999). The Myth of Pain. Cambridge: MIT Press.

17. HECK, JR.,  Richard G.  2000.  “Non-conceptual Content and the ‘Space of Reasons’”, Philosophical Review 109,  483-523.

18. Kandel, E. R., Schwartz, J. H., And Jessell, T. M. (1995). Essentials of Neural Science and Behavior. Norwalk, CT: Appleton & Lange.

19. Leggio, M.B., Molinari, M., Neri, P., Graziano, A., Mandolesi, L., and Petrosini, L. (2000). Representation of actions in rats: the role of cerebellum in learning spatial performances by observation. Proceedings of the National Academy of Sciences of the United States of America, 97(5), 2320-2325.

20. Lycan, W. (1993). A deductive argument for the representational theory of thinking. Mind and Language, 8, 404-430.

21. Millikan, R.G. (1989). Truth-rules, hoverflies and the Kripke-Wittgenstein paradox. Philosophical Review, 99, 323-353.

22.   Milner, A. D., & Goodale, M.A. (1995). The Visual Brain in Action. Oxford: Oxford University Press, 248 pp. (paperback 1996).

23. NICHOLS, S.  (2003).   Imagination and the puzzles of iteration.  (forthcoming in Analysis.)

24. Peacocke, C. (1994). Content, computation and externalism. Mind and Language, 9, 303-335.’[[[[[[[0-------

25. ___________ (2001) Does perception have a nonconceptual content?, Journal of Philosophy 98, 239-264.

26. Sirigu, A., Duhamel, J.-R., Cohen, L., Pillon, B., Dubois, B., and Agid, Y. (1996). The mental representation of hand movements after parietal cortex damage. Science, 273 (5281), 1564-1568.

 

 



[1]  See  (Cummins, 1996), (Dretske, 1995, p. 28), (Akins, 1996), (Lycan, 1993), (Peacocke, 1994), (Fodor, 1976), (Hardcastle, 1999), (Churchland, 1995), (Millikan, 1989), and (Crane, 1995); for a fuller description of this thesis and a rejection of it, see  (Author, forthcoming).  For a more extended discussion on the part of a number of authors, see  (Author, 2002). Though it is not described fully enough to determine conclusively, the relocation strategy in (Nichols, 2003)’s discussion of imagination may involve a special case of the sort of contentless representation I describe.

[2]   See (Gunther, 2003).

[3]   (Author, forthcoming and 2002).

[4]   See discussions of the Generality Constraint in (Gunther, 2003)).

[5]    See (Clark, 2001) and also arguments by various authors in (Author, 2002).

[6]   On the problematic status of these concepts, see (Crane, 2002).

[7]   The discussion of the conceptual versus nonconceptual content of perceptual experience is very often very seriously vitiated by the supposition that the fundamental nature of perceptual experience is revealed to solitary philosophical inspection; though such an implicit supposition characterizes much in (Gunther, 2003), solitary philosophical inspection of the experience of looking at a desk is explicitly part of the methodology of (Heck, Jr, 2000).

[8]   Nakita Newton and Jim Garson have both presented this objection, as did several participants in the Phenomenology and Cognitive Science Conference, July, 2002.

[9]   One can follow up on curlew sandpipers at:  http://www.mbr-pwrc.usgs.gov/id/framlst/i2440id.html

[10]  According to Goodale, in his remarks following his Society for Philosophy and Psychology conference presentation, 2002.

[11] See Clark (2001), especially chapter five, for the idea that vision produces action plans; in effect, this paper can be seen as arguing that the plans are token-realizations.

[12]   In her Oxford classes, Elizabeth Anscombe would sometimes remark that something was merely in the title one gave to the imagining, and this seems an apt case for that sort of expression.

[13]   As of the spring of 2003, there is a useful discussion of this by Chris Elias at http://www.artsci.wustl.edu/~philos/MindDict/representation.html.