Notations:
^ denotes counterpoints
* denotes notes from me
Unless otherwise noted, the following are excerpts from Dreske’s “Simple Seeing”.
— — — — — — — — — — — — — — —
The point may not need emphasis, but let me stress, nonetheless, that the issue over simple seeing concerns our seeing of objects and things — not facts about these things. The terms “object” and “thing” are dummy terms intended to cover such disparate items as tables, houses, cats, people, games, sunsets, signals, tracks, shadows, movements, flashes, and specks. We see all these things and more. Since the perception of events, states of affairs, conditions, and situations introduces special complications that do not concern us here, I will confine myself to such ordinary things as rocks, robins, mountains, and people.3 Simple seeing is seeing these sorts of things.
What is not at issue is our seeing that there are rocks in the road, how many people are in the room, where the cat is, whether the clock has stopped, who is at the door, or what is happening. When the verb “to see” has a factive nominal (that. . . ) or a question word clause (who . . . , whether . . . , what . . .) as its complement, a description of what someone sees has epistemic implications. It implies something about the conceptual resources of the perceiver. One cannot see where the cat is unless one sees that she is (say) under the sofa — without therefore knowing (hence believing) that she is under the sofa.
What is controversial, and what concerns us here, is whether one can see the cat under the sofa in a way that is free of such discursive implications.
Disagreement exists over whether some belief is necessary to the seeing, whether one can seen an X without taking it to be anything whatsoever. This is not a question about whether one can see an X without any beliefs at all.6 Nor is it a question about whether, in fact, people have beliefs about the things they see. It is a question about whether their having a belief about the perceptual object is essential to its being a perceptual object — essential, that is, to its being seen.
It is easy enough to ignore this question, or to miss its significance, by changing the subject — by talking not of seeing (say) a triangle, but of perceiving triangles. On the basis, then, of a technical stipulation (or simply on the basis of prevelant usage) one notes that the term “perception” is reserved for those sensory transactions that have some kind of cognitive upshot. The subject (infant, rat, pigeon) does not perceive the triangle (among the rectangles) unless it identifies it in some way, unless it displays some appropriate discriminatory response, unless it recognizes the triangle, if not as a triangle, then at least as the sort of thing to which a particular conditioned response is appropriate. According to this usage, perception requires some degree of conceptualization, some identification or categorization of the sensory input. What one perceives is limited by the stock of concepts (or, if “concepts” is too strong a word, by the battery of disciminatory responses) one has available for sorting and organizing the flux of stimulation. If one doesn’t know what a triangle is, if one lacks the capacity to distinguish between triangles and other geometrical figures, then one does not perceive triangles. Geometrical figures, perhaps, but not triangles.
… our question about simple seeing is a question not about visual perception, but about whatever it is that we use the ordinary verb “to see” (followed by a concrete noun phrase) to describe. What are we doing when we see something? Or, if this is not something we do, what conditions must obtain for us to see a robin, a sunset, or a star? In particular, can one see a robin without perceiving it? This question may sound a bit odd, but only because one is accustomed to conflating seeing with visual perception. But if perception is (either by stipulation or common understanding) cognitively loaded, if some degree of recognition or categorization is essential to our perception of things, it is by no means obvious that one must perceive something in order to see it. Quite the contrary. One learns to perceive (i.e., recognize, identify, classify) those things that, even before learning takes place, one can see. What else, one might ask, does one learn to identify?
A second way to muddle issues is by interpreting claims about simple seeing, not as claims about our ordinary (mature) way of seeing robins, trees, and people, but as claims about an underdeveloped stage of consciousness, a dim sort of visual awareness, that lower organisms (perhaps) experience but that human beings (if they experience it at all) quickly outgrow during infancy… To say (as I did in Seeing and Knowing) that seeing a robin (nonepistemically) is belief neutral is not to say that one cannot see a robin in this way with a belief to the effect that it is a robin, a bird, or a thing. It is to say that your seeing the robin is independent of, a relationship that can obtain without, such beliefs.
(1) Simply seeing X is compatible with no beliefs about X.
There is, finally, one last source of confusion that has helped to blur whatever genuine differences may exist among philosophers about the nature of simple seeing. This is the mistake of supposing that the seer is in some special, privileged position for determining what is seen and, therefore, whether something is seen. This doctrine takes its most extreme, and least plausible, form in the view that if S does not believe she sees X, then S does not see X. A less extreme, but still implausible, version has it that if S believes she does not see X, then S does not see X. Transposition of the first yields the result that one cannot see X unless one believes one sees X — a flat-out denial of the thesis (expressed in (1)) that simple seeing is compatible with no beliefs about X. The second, less extreme, version is not a direct denial of (1), but it does have the implication that seeing X is incompatible with certain beliefs: viz., the belief that one does not see X.
^The fact is, however, that people are not authorities about what they see — not, at least, in the way suggested by the preceding two principles. Whether or not you saw my grandmother is a question that in some situations I am in a better position to determine than you. If you, with a mistaken idea of who my grandmother is, smugly assert that you did not see my grandmother today, I can correct you by pointing out that you certainly did: she was the woman to whom you gave your seat on the bus. If you respond by saying that you did not realize that that little old lady was my grandmother, the lesson is clear. The question was a question not about who you identified today, but about who you saw today.
What a person believes (about what she sees), and what she is consequently prepared to assert or deny about what she sees, is conditioned by the conceptual and cognitive resources she has available for picking out and identifying what she sees. If she does not know what a marsupial is, she isn’t likely to believe that she sees one. And if she mistakenly believes that kangaroos are the only marsupials, she might well believe she sees no marsupials when, in fact, she sees them (opossums) all over the yard
There are strong philosophical motivations for denying any beliefneutral form of seeing (simple seeing). The inspiration for such denials comes, I think, from positivistic and (more specifically) behavioristic sources. If S’s seeing X is only contingently related to S’s beliefs about X, if she could see X with no beliefs about X, then there is no secure basis in S’s behavior (linguistic or otherwise) for determining whether or not she does see X. Seeing is deprived of its logical or conceptual links with the observational data base.
This epistemological consequence is alarming enough to goad some philosophers into interpreting good eyesight as a cognitive capacity so as to secure the requisite links with behavior. If seeing X can be interpreted as a form of believing (if not believing that it is X, at least believing that it is Y where “ Y” is a description that applies to X), then, since believing has behavioral criteria, seeing does also.
One of the currently fashionable ways of achieving this linkage is by identifying psychological states with functional states of the organism. (*Functionalism) Since a functional state is one that transforms certain inputs into certain outputs, since it is defined by its associated input — output matrix (Putnam’s machine table), an organism is in a determinate mental state (e.g., seeing X) only if its input is being converted into appropriate output — only if it behaves, or tends to behave, in a certain kind of way. If seeing X is (in part at least) a psychological state, if psychological states are functional states, if functional states are defined by the output they produce, and if output is (aside from simple reflexes, tropisms, etc.) indicative of beliefs, then all seeing is believing.
There is much to be said for functional theories. From an evolutionary standpoint it is hard to see how a perceptual system could develop unless it had a role to play in an organism’s continuing adjustment to its surroundings, and this adjustment is obviously a matter of an organism’s responses to its environment.
^But a mechanism (e.g., our sensory system) can have a function without the operations of this mechanism having functional criteria for their successful performance. Cam shafts also have a function. They lift valves.11 But this doesn’t mean that cam shafts cannot operate in the way they were designed to operate without lifting valves. If the valves are not being lifted, this does not mean that the cam shaft isn’t rotating so as to lift the valves. For in order for a cam shaft to perform its function it must have the cooperation of a number of other auxiliary mechanisms (rocker arms, lifters, guides, etc.). If these other elements are not functioning properly, the cam shaft makes its appointed rounds, operating in the fashion it was designed to operate, without the usual consequences. The valves don’t get lifted. And so it is with seeing. Unless the cognitive mechanisms are in proper working order (and learning is required to put them into proper working condition) the sensory systems (those responsible for our seeing, hearing, smelling, and tasting things) can perform their role in this total operation without the kind of cognitive effects associated with a fully mature, and properly functioning, organism.
*Dreske makes it sound like functionalism is incompatible with the ‘simple seeing’ view. This seems false. Functionalists have to say that simple seeing are associated with certain input-output functions. However, they do not have to say that these input-output functions are identical to any input-output functions that are characteristic of beliefs. Indeed, functionalists can say that the outputs associated with simple seeing happen at some pre-cognitive level (unlike beliefs).
What is needed is some positive account of the role simple seeing plays in the process of gathering and utilizing sensory information. To this positive account I now turn.
Some of our beliefs are about the things we see: I believe that that is my wife, that this is a bushing, that those are oaks. Such beliefs are called de re beliefs; what they are about (that, this, those) is not determined by whatever description (if any) the believer may happen to have available for describing or picking out what it is that his belief is about. The referent of the belief, what it is a belief about, is fixed by factors other than those descriptive elements we may possess for its specification. It is fixed by what you see, not by what you believe you see or believe about what you see.
I have just copied a letter on a high-quality machine. At arm’s length in good light the copy is indistinguishable from the original. I place the copy (thoughtlessly confusing it with the original) neatly on top of the original (so that only the copy is visible), gaze down at it (them?), and notice what I take to be a smudge on the top sheet. As it turns out, I am mistaken (it was only a shadow) but, as chance would have it, the original (which I cannot see) is smudged in just the way I thought the top sheet was. I obviously have a number of false beliefs: viz., that this sheet is smudged, that this sheet is the original letter. What makes these beliefs false is the fact that they are (contrary to what I believe) beliefs about the copy. What, then, makes these beliefs beliefs about the copy and not about the original?
It will come as no surprise to find that the answer to this question is that my beliefs are about the first sheet (the copy), not the second (the original), because I see the first sheet, not the second, and my belief is about what I see. This answer isn’t very illuminating. For we are now trying to say what it is that constitutes one’s seeing the first sheet, what makes the first (not the second) sheet the perceptual object and, therefore, the thing about which I have a belief. Since what I believe (that it is smudged, that it is the original letter) about the perceptual object is true of something (the second sheet) that is not the perceptual object, what I believe about what I see does not itself determine what I see. What does determine this?
It will not do to say that I am seeing the copy because I am looking at the copy (not the original) since any sense of “looking at” that does not beg the question (e.g., you can only look at things you see) is a sense in which I am also looking at the original.14 If the copy was removed (revealing, thereby, the original letter) my experience would not change in any qualitative respect. Convergence, accommodation, and focus would remain the same since the original is (for all practical purposes) in the same place as the copy. Obviously, then, these factors, no more than belief, determine what it is that I see. What does?
I have no doubt exhausted the patience of causal theorists by this time. Isn’t it clear, they will tell us, that the perceptual object is determined by the causal antecedents of our visual experience? Light is being reflected from the copy, not the original, and since it is this object that is (causally) responsible for the experience I am having, it is this object that I see. The fact that removal of the copy (exposing, thereby, the original letter) would leave the experience qualitatively unchanged, would perhaps (if it was removed without my knowledge) leave all my beliefs unchanged, is immaterial. It would change what I see because it would change the causal origin of the resultant experience.
Whether a causal account is ultimately satisfactory or not (I shall have more to say about this in a moment), it does succeed in driving a wedge between perception and conception. It divorces questions about what we see from questions about what, if anything, we know or believe about what we see. It distinguishes questions about the etiology of our experience from questions about the effects of that experience. Insofar as it achieves this separation it succeeds in capturing the essence of simple seeing.
The causal theory gets this part of the story correct, and it is for this reason that it represents such an attractive candidate for the analysis of simple seeing. It explains, among other things, how it is possible to have all one’s beliefs about X false while still being about X. What makes my beliefs about the copy is the fact that I stand to the copy in the appropriate causal relation, the relation that, according to this view of things, constitutes my seeing the copy. What makes all my beliefs false is the fact that nothing I believe (e.g., that it is smudged, that it is the original letter) is true of the thing to which I stand in this causal relation
^The difficulties in articulating a full-dress causal analysis are well known, and I do not intend to rehearse them here. I mention two problems only for the purpose of indicating why, despite its attractive features, I do not personally subscribe to such a view.
There is, first, the problem of stating just how an object must figure in the generation of a person’s experience to qualify as the object of that experience. Typically, of course, there are a great many causal factors that cooperate to determine the character of our sensory experience. Price’s distinction between standing and differential conditions is a useful one in this regard, but, as Grice has made clear, we need more than this.15 When you hear a doorbell ring, you hear the bell ring, not the button being depressed, even though both events (the bell’s ringing and the button’s being depressed) are causally involved in the production of your auditory experience. Both are differential conditions. What, then, singles out the bell’s ringing (not the button’s being pushed) as the perceptual object? The causal theory has little to say in answer to this question.
A second problem has to do with the nature of the causal relation itself. Just what do we mean when we say that A causes B, that A is causally responsible for B, or that B is one of A’s effects? There is remarkably little agreement among philosophers about the answers to these questions. This intramural squabbling would not be particularly embarrassing to a causal analysis of perception if the disputed issues were not themselves germane to the questions we wanted answered about perception. But this is precisely what is not the case. The absorption of photons by the photosensitive pigment on the rods and cones, for example, is a quantum process with all that this implies about the indeterminacy of such events. This indeterminacy can be made to manifest itself under conditions of extremely low illumination: whether the subject will see a faint light is, in principle, unpredictable. Is this to be called a causal process? It depends on which philosopher one asks.
We get information about things by seeing, hearing, smelling, tasting, and touching them. To say that someone has seen X is to say that information about X has been delivered in a particular form, a form that differs, intrinsically, from the way information is coded when we hear smell, taste, or touch X.18 If we do not press too hard on the idea of causality, we may say that this information is delivered by means of causal mechanisms and processes. When we see X, X (or some event associated with the presence of X) initiates a sequence of events that culminates in a distinctive sort of experience, the sort we call a visual experience. Typically, this experience embodies information about the color, shape, size, position, and movement of X. The role or function of the sensory systems in the total cognitive process is to get the message in so that a properly equipped receiver can modulate her responses to the things about which she is getting information. The sensory system is the postal system in this total cognitive enterprise. It is responsible for the delivery of information, and its responsibility ends there. What we do with this information, once received, whether we are even capable of interpreting the messages so received, are questions about the cognitive-conceptual resources of the perceiver. If you don’t take the letters from the mailbox, or if you can’t understand them once you do, don’t blame the postal system. It has done its job. The trouble lies elsewhere.
This, in barest outline, is an information-theoretical account of simple seeing. It differs from a causal account not by denying that causal processes are at work in the delivery of information, but by denying that this is the essence of the matter. Seeing X is getting information (coded in a certain way) about X, and if information about X can be delivered by noncausal processes, so much the worse for causality. If the processes by means of which we see very faint stars are infected with the uncertainty, the inherent randomness, of quantum phenomena, and if such processes are not to be counted as causal in nature, then we see things to which we do not stand in the appropriate causal relation. But we still see them, and the reason we do is because the experience that is generated by the occurrence of these inherently random events embodies information about the distant stars (e.g., where they are)
In speaking of sensory systems in the way that I have, as systems responsible for the delivery of information, it should be emphasized that the term “information” is being used here in the way we speak of light (from a star) as carrying information about the chemical constitution of the star or the way the height of a mercury column (in a thermometer) carries information about the temperature. These events or states of affairs carry or embody information about something else but, of course, no one may succeed in extracting that information. It is in this sense that our visual (auditory, tactual, etc.) experience embodies information about our surroundings. It can carry this information without the subject’s (undergoing the experience) ever extracting that information for cognitive purposes. Once again, learning is required to crack the sensory codes.
If it is found that the eyes played no essential role in this process, if light is unnecessary, then so much the worse for the standard scientific accounts of how we see. As a child I never found the visual exploits of Superman (seeing through buildings) incoherent or logically paradoxical. I still don’t. This was just a fanciful (“fanciful” because, as things stand, no one can see things in this way) account of an extraordinary individual who could see things in ways that no one else could. Historians tell me that the ancients had even more bizzare conceptions of how we see things. They were wrong about how we see, but they weren’t committing logical howlers.20 They just didn’t know, as we now know, how we obtain information about distant objects and events.
It seems to me, then, that one cannot formulate a satisfactory account of simple seeing by embellishing the Causal Theory with details about the way, in fact, we see things. All one gets is a bloated concept of seeing. Seeing objects is a way of getting information about them. What makes it seeing (rather than, say, hearing) is the intrinsic character of those events occurring in us that carry the information. What makes it X (rather than Y) that we see is that the information these internal events carry is information about X (rather than Y). Everything else, I submit, is empirical icing on the conceptual cake — something the scientist, not the philosopher, should provide.
*Key point: scientific facts about how we see are not part of the concept of ‘seeing’.
Dretske, F. I. (1979). Simple seeing. In Body, mind, and method (pp. 1–15). Springer, Dordrecht.
Further notes:
Dreske’s basic idea: we can see an object without forming any beliefs on it.
Question: On Dreske’s account, does blindsight count as seeing?
Response: It seems that Dreske will have to say yes, since we are getting information in cases of blindsight. That blindsight counts as seeing is not obvious, but I don’t think it’s a problem for Dreske to commit to this. At worst, it’s just an instance of metalinguistic negotiation over the concept of ‘seeing’.
Question: How fine-grained is simple seeing?
Response: You don’t see that there are 1248 blades of grass. The question is whether you see each blade of grass. For Dreske, that is a matter of visual differentiation. It may be the case that your visual system doesn’t differentiate blades of grass at a fine-grained resolution. So it is a question of how good your eyesight is. If you can see the blades of grass at a fine-grained resolution, then you would be getting the information that there are 1248 blades of grass. You might not be able to compute that into knowledge. But that’s your cognitive capacity’s fault, not your visual system’s fault.
