MULTIPLE DRAFTS: PERCEPTION WITHOUT REPRESENTATION

   outline    theory index   Ellie Epp worksite


Part I. Assumptions

i.

The materialist assumption is that every difference is a physical difference: so the difference between perceiving one thing and perceiving another will be a physical difference in the perceiver, as will the difference between conscious perception and unconscious registering or discriminating.

The same will be true of the processes of reporting perception; the difference between reporting one thing and reporting another will also be a physical difference, both in the person reporting and in the persons, including the reporter, who receive the report. For the materialist there can be no exits or lapses or arcs up and out of materiality.

But we are Cartesian materialists, Dennett says; our present discourse about cognitive things is transitional, a mesh of mentalist and physicalist metaphors and vocabularies -- hard to work with. The remedy Dennett suggests is to match our manner of speaking to a more fully imagined vision of the spatial and temporal layout of cognitive events.

ii.

First imagine a person or an animal in some part of a world, doing something. A bear turning its head to listen to a splash in a stream, a person watching a semitransport braking slowly at a crosswalk. Perception is a relation between two objects in a world: bear and fish, person and truck. This is an assumption -- the realist assumption that there are such objects.

Second, imagine the brain of a person watching a truck. There is some sort of structure that changes as the truck looms. That ability to change is how the person is able to see the truck: is the person seeing the truck.

When we imagine the structure of the brain changing, are we still imagining the person and the truck/ Usually not; it is as if we enclose ourselves in the brain, seeing its structure.

iii.

The structure by means of which the person is seeing the truck is not something the person can see, because its purpose is to make the person able to see the truck. The means of perception are perceptually transparent.

By its nature -- this is the evolutionary assumption -- perception is the ability to detect those things in the world that are relevant to the self. So perception, by its nature, is transparent to itself. It perceives something other than itself by means of a modification of itself it does not perceive as such.

There are media of perception other than the body -- the air, light, chemical molecules, etc. -- and perception ignores them too. It simply ignores them. It does not discriminate them. Ignoring them makes them transparent, in effect.

Perception picks out what interests it and ignores the rest. Simply ignores it. The fact that the means by which perception is being effected are transparent to perception in the very act of perceiving has the effect of surrounding the perceiver with an empty zone, a diaphanous terrain.

There is something inherently fictional about perception: the imperceptibility of the things that are actually doing the work. Like Beauty in the house of the Beast, being provided for by invisible servants.

'Consciousness' is a name for this fictionality.

Or say it this way: 'consciousness' as we imagine it -- or feel it -- or sense, or intuit it -- is the presence of (all) the actual, material, effective, but unperceived means of cognition.

So 'consciousness' can be a name for the real means. Though we shouldn't call the body, the nervous system, a medium, since it does not, like air or light, transmit.

'Consciousness' as a fantasy structure is the fantasy of "knowledge acquired without means of cognition."

Ayn Rand, quoted in Kelley, p.40.

iv.

We forget what it is we're explaining.

We want to know how the things we do consciously are done. That's what Dennett wants to explain, and that's why notions of macro-organization, the virtual machine, are part of the explanation.

If we think we are explaining consciousness, we might mean that. But we might mean the particular quality consciousness has that the world doesn't have: that glassiness, insubstantiality; the insubstantial material we see when we imagine or imagine we remember an episode of being aware of something.

We might explain why we imagine it that way. Dennett wants to do that too.

v.

Gibson understood that the brain is the real eye: it is fed by the eyes, but we don't see what they eyes would see if they were seeing.

It makes sense to think of the brain as itself the organ of vision, because it is the organ evolved to contact objects and events as such, discerning them in noise and change, "attending to, keeping in touch with, tracking and trailing". The brain integrates what they eye picks up, compensates as it must, completes and deletes, "feedback-guided, error-corrected, gain-adjusted'.

Lucretius thought vision is the eyes feeling the blows struck by the streams of thin images flowing from objects and pushing air ahead of them, so "the longer the breeze that bushes our eyes, the more distant the thing seems to be": "We see at once both what it is and how far away."

The ancients and earlies took the eye as accomplishing vision because when the eye is injured the vision is lost. But when we take the eye to be the organ of vision the stabilized integrated seeing of objects seems to be something the 'higher centers' have to collate from the raw, jumpy, noisy two-dimensional images which can be the only products of seeing thus imagined.

But if we think of the brain as the real eye, the temptation to think of the eyes' product as images vanishes: there is no vision or visibility earlier than the whole brain; the brain's global work of integrating known, identified, stable, three-dimensional objects is vision, not inference.

And then it is less mysterious that vision could occur by means of actors on the skin of the back, and that such prosthetic vision should reproduce the point of view which is, after all, that of the brain not the eye.

Similarly, if we take the brain as the real organ of touch, the fact that, when we touch surface with the tip of a pencil rather than with our fingertips, we continue to seem to be touching the object, makes sense: the brain is properly interested in the properties of the object, not (in this instance) in our skin's lack of contact with it.

Dennett, CE 33-335.
Lucretius, IV 230.
tactors: Dennett CE 338-42
pencil: Dennett CE 47.

vi.

Certain sorts of mental talk are rightly applied at the level of the creature or the person, and are misapplied when attributed metaphorically to anything inner, whether brain or 'mind'.

Nothing internal to the system should be said to perceive.

'Direct perception' means this: there is no internal seeing. Any seeing is a relation of a creature to a perceivable situation. If there is no internal seeing (or imagining, reading, understanding, remembering, sensing, feeling), then talk of internal representations is disabled.

Similarly we shouldn't talk about internal inference.

This does not mean that any perceptual moment isn't supplemented by structures enabled at other times -- that it is not made possible by training, experience, naming practices, etc. It does mean that concepts like memory and inference are not properly applied to this structural potentiation.

 

Part II. Dennett

i.

There are three key assumptions kept quite inexplicit in Consciousness Explained. The book does not make sense without them, but they are not set out in plain view in plain speech. Careful reading finds them in turns of a phrase, and they can be felt throughout, but Dennett -- shyly or discretely? -- hides them where they can be overlooked by the unsympathetic. They are the radical heart of the book, a positive account of material sentience hidden behind (inside, under) a swashbuckling go at Cartesian metaphor.

The three assumptions are related aspects of Dennett's microhomuncularism. I will call them microsentience, microsentient simultaneity and sensory engagement. When they are made explicit Dennett looks anti-representationalist rather than behaviorist.

ii.

Imagine it this way, Dennett says: any act or moment of perception is a brain set up in one way rather than another, and there will be as many kinds of difference as there are differences being registered, right down into whatever size of grain you like. The sites of these differences are neural processors minutely and innately responsible for aspects of the larger discriminations relevant to living creatures. Dennett calls them "pattern-recognition mechanisms", "fixed semantic elements", "IF-THEN primitives", "operators in conditional branching circuits", and "events of content-fixations in the brain." He also calls them homunculi.

Some of these micro-discriminators will have to do with properties we think of as sensed properties of objects -- location, size, smell, and the rest -- and some will be working on reactive discriminations -- what to do next, how to find out more, what that thing is called. So the cortex of any perceiver will be a "complicated slew" of "concurrent contentful events", active states of many micro-deciders and -detectors multiple-y connected and feeding back on one another. This much of Dennett's picture is familiar from the connectionists: an anchored semantic constituency with microelements of many functional flavors.

Dennett's positive aim in Consciousness Explained is to use this model to imagine how we might learn to talk about cognitive states in thoroughly materialist ways. "My fundamental strategy has always been the same," he says, "first, to develop an account of content that is independent of and more fundamental than consciousness -- an account that treats equally of all unconscious content-fixations ... and second. to build am account of consciousness on that foundation."

There is more than one reason for Dennett's strategy. In all our many centuries we have not succeeded in making "consciousness' an explanatory posit: that's one. A second is that we seem to be perceptually related, for instance, in ways that are not all or always conscious, or are conscious in varying degrees. The experimental evidence is that we are right when we sense an (as if) sub-platform of effective neural relatedness to things that is deeper and wider and better informed and sometimes differently motivated than perceiving is. We aren't perceiving everything we register.

So a neurally-based picture of perception would want to account for the "background of conscious experience" too -- activation effects that are ballistic, like shooting stars, and leave no trace; potentiated states, like node changes that make certain patterns of activation more likely to recur; circuits in chaotic search for an attractor -- as well as transient stable states and what we might call hyperstable states, circuit states being kept going by verbal or other self-prompting.

It might be that hyperstable states are those we'd want to say contribute to perception as opposed to 'background' registering and sorting. Dennett is anatomically noncommittal, he says, about whether, for instance "branches radiating from the thalamus to all parts of the cortex fits it for the role of 'searchlight', differentially arousing or enhancing particular specialist areas, recruiting them to current purposes". But what is apparent to him is that "the things we are most definitely conscious of are the items we frankly and unhurriedly observe, gathering in and integrating the fruits of many saccades, building up an acquaintance over time while keeping the object located in personal space".

What we can conclude for sure, from these and similar passages, is that Dennett's homunculi are not always on, but that when they are on, they are contributing members of conscious states. By Dennett's homuncular hypothesis they are the very grain of sentient states, they are microsentient. Homunculi in a waking state will be perceivers, and they will be direct-perceivers, object-discriminators, not reads or writers of micro-instantiations of symbols.

Dennett, CE 433.
CE 457
CE 279.
CE 274.
CE 335.

iii.

"Our senses do not bring exhibits into 'court' to show us, but rather tell us of things', Dennett says. How is perception like utterance? Or, alternatively, like writing multiple drafts? Perception seems as close to showing as any cognitive process could be -- what is evidence but "the evidence of the senses"?

Dennett's description of perception as report is corrective not to the notion that we see things (in the sense any of us would ordinarily understand) but to the notion that the senses, rather than being shown evidence, themselves show evidence to some inner perceiver. His main point is anti-Cartesian.

If we remember what sort of microhomunculi are doing the telling in Dennett's account, it is clear that perception is like a speech act mostly in being an "event of content-fixation" in the way we imagine an utterance to be an event of content-fixation. A microhomunculus tells what it knows (in its little way): that there is some object or some microaspect of an object in some relation to the person perceiving. (Or it tells what it thinks it knows, tells falsely.) In any case the contrast between telling and showing is a contrast between exhibits, evidence, representations brought inside and a perceiver who stands in place saying "there it is, there, out there."

Dennett finds the metaphor apt in various ways: the microconstituents compete for influence like academics speaking at a conference; but less polite: they all yell at once (a multiprocessor trading floor) changing each other's bids -- or are at work like production systems in a computer's global workspace "where various demons could write messages for all others to read, which in turn provokes another wave of writing and reading".

On a larger scale, perception as telling is story-telling, a narrative stream whose shapes of flow are plotted in ways that pick out the characters, prizes and adventures relevant to culturally transmitted and animal motive. We perceive as we need to perceive and as we are taught to perceive.

The narrative stream is not a stream of sentences. It is a stream which is simultaneously a stream of microsentient knowings and a processing stream of activation being passes. If this stream involves language-processors, hard-copy or imagined sentences may result.

CE 316.
CE 264.

iv.

A judgment in Dennett isn't a sentence either. The judgment metaphor for perception is relevant primarily in relation to decision. A processor/microhomunculus "discerns, discriminates, judges, registers, concludes, thinks", that list being a list of equivalent misdescriptions of the act as sui generis which is elsewhere called having a presentiment. Judging, like telling, is something a microhomunculus does on the spot and aims toward objects. Judging as perception is judging that ... something. And since different homunculi may be supporting different claims about what's there, processor settling can also be called finding out or pronouncing. What we perceive is thus the import or 'content' of a judgment.

There are ways the legal contexts imported by the term are not wrong. A homuncular pronouncement requires expertise, is concerned with public matters, is consequential, leans on precedent, and is arrived at by an evolutionarily authorized due process. Unlike a legal decision, though, a microhomuncular judgment can be enacted without going on record, and " the content of the judgment does not have to be expressible in propositional form". States of perceptual belief, decision or opinion, then, are not states that intervene between the receiving of evidence and the giving of a report: they are the perceiving and (in functional terms, as interacting processes) also the reporting.

As an attempt at neutral terminology which would at the same time be a provocative antidote to our always -renewed urge to think of perception as perception of appearances, 'judgment' has good uses ("not made of figment but made of judgment"), but it also contributes to a behaviorist misreading. There is an emphasis of flavor in his metaphor, which in the end is not neutral enough -- a rationalist and even patriarchal flavor of rule-enforcing and handing-down which makes of the cortex a society of little judges rather than a community with every flavor of homuncular citizen -- including little courtiers and courtesans whose role it is to flatter and enjoy. Because his metaphor leaves out the more sensual flavors of homuncular decision he is read as saving there is no sensory engagement in perceiving. That would make him a barefoot behaviorist, and he knows he isn't that.

A representationalist reading, on the other hand, takes 'judgment' as implying that Dennett believes perceptual representation to be propositional in format. This reading leads to a wonder how an abstract or propositional format for perceptual representation may "lead to" the "phenomenology we experience". But 'judgment' in Dennett is not meant to imply any format whatever: it is an effort toward a manner of speaking that bypasses the sorts of metaphor that -- like "evidence of the senses" -- seem to suggest inner courtrooms where judges never see daylight.

Better they should be riding the circuits.

CE ?
CE 365,.
CE 350.
Atkins and Winger, pp. 1, 15, 16.

v.

Dennett's second assumption, then, is that a conscious state is a simultaneity of distributed microprocessor states, some of which are microsentient spots of information processing that are "doing the same work" as consciousness because they are themselves the grain of conscious states.

Perceiving something happens all over the brain: "there need be no time and place where 'it all comes together' for the benefit of a single, unified discriminator; the discrimination can be accomplished in a distributed, asynchronous, multilevel fashion."

One implication is that the linked standing-together of various bits of a perceptual state is already an integration of the results of those microdecisions -- there need be no further and final tying together. Processing could in this way be seen as both modular and "open", with the conscious state itself being the binding of modular results.

Another implication is that sensory structures and probe structures and report structures would contribute simultaneously to moments of perceiving, as copresent parts of the conscious state. Faculty distinctions between sensing and thinking would be unuseful. There would be no pure states either in processing terms pr in experience.

More, felt simultaneity of many sorts of discriminations could do away with the form/content and processing/representations distinctions, since conscious states could include as 'content' the processing interactions themselves, with their "additions, incorporations, emendations, overwitings in various orders".

Perceiving could thus be thought of as a standing texture of on-the-spot knowings, unstable in part, and held together by simple copresence in time and space: a rope of sand. "When a portion of the world comes in this way to compose a skein of narratives, that portion of the world is an observer. That is what it is for there to be an observer in the world, a something it is like to be."

Dennett FT 8.
CE 297.
CE 135.
CE 137.

vi.

Dennett explains sensory engagement in what we call thinking by analogy with kludges in a program, indirect forms of processing access, like patch-cords on an audio bay: internal circuits accessing each other by means of detours through output and back again into input. We can sustain attention to something by talking aloud to ourselves -- "Look at that, there, see, that little screw must have fallen out of that thing behind the thing there". We hear what we say and it boosts some ability of circuits to stabilize differentially, the way our mum's voice telling us to look at the squirrel directed our ability to see it against the leaves.

Since we imagine hearing things we can imaging hearing ourselves speak, and so we can direct our use of one sense by simulating the use of another -- "Is that lavender, a bit of grapefruit ... rose geranium, maybe, we think (in words), sniffing. Or we can do it the other way around, simulating a scent as best we can, calling it up so thought-words will follow. Sensory simulating of this kind is not externally patched (by voice through air), but it is presumably still an odd hack, since it sets up internal access by simulating external access.

Dennett speaks of this sort of kludging as a kind of broadcasting, whereby discriminations of certain kinds can call all hands. Inner speech recruits sensory simulation and vice versa, the whole microsentient complex with its large overall shapes of narrative flow being what we call thought.

So thinking isn't primarily linguistic, and its linguistic components can be a form of sensory engagement too, circuited as they are through simulated audition. Dennett uses this fact to undo snarls about belief as sentential: "Cut through the tangle this way: the experience itself is the thought about itself": "My thought (or believe) about how it seems to me is just the same thing as what my expedience actually is."

There is a further conclusion we could enjoy arriving at, though Dennett doesn't go there: if it is the brain -- the common sense -- that sees and smells and hears and feels and thinks -- and if thinking is a complex of simulated perceivings -- then it is perceiving that is the great evolutionary achievement, the great central state of which thinking is a region.

CE 278.
CE 318.

vii.

A thinking state might be a conscious state just to the extent that it is sensorily engaged -- though we are far from knowing just what that might mean. But Dennett follows Ryle and Wittgenstein in reminding us that what we mean by certain cognitive terms like 'seeing' and 'thinking' extends past occurrent states to our dispositions to act and to say. His attack on the notion of qualia (mainly directed against Cartesian materialist representationalism which talks as though perceiving is made of appearance rather than being made of seeing -- figment rather than judgment) decomposes the notion of 'vision' into some of the reactive dispositions we would identify with seeing, for instance being able to act effortlessly and instantly on information about location and color, and being able to talk about doing so.

But Dennett carries Ryle and Wittgenstein into the brain by tying the notion of reactive disposition into physiology, and so talking about complex dispositional states which are activations of the "wiring that determines mere behaviors". And then 'reactive dispositions' in Dennett's handling becomes the brain's present disposition of reactions to a discrimination made.

So reactive dispositions can be among the microconstituents of a perceiving state; Dennett talks about the "innate and learned associations and reactive dispositions triggered by" the particular way someone is sensorily informed, and thus can identify "the way it is with me" with the "sum total of all the idiosyncratic reactive dispositions inherent in my nervous system as a result of my being confronted by a certain pattern of stimulation."

What is implied is that the distinction between perceiving and behavior is also unfixed to the extent that behaving -- the process of behaving -- has to be a slew of processor activations copresent with sensory activations, where the distinction between 'sensory' and 'other' -- on the ground, among the troops -- may itself be hard to make when microhomuncular components are individually driven by combinations of sensory and other input.

CE 431.
CE 388.
CE 387.

 

Part III. Myths

i.

what kind of body this mind is

interlaced through veins, flesh and sinews

nothing exists more easily moved and more thin than this

exceedingly delicate and formed of exceedingly minute particles

dispersed only at rare intervals through the frame

such intervals apart as equal the smallest things which falling upon us are able to awaken sense-bringing motions

sensation is kindled for us in our flesh

that motion which we name feeling

many particles must be moved in us, before those first seeds of spirit can be touched and felt which are intermingled through our frame, and before they can go buffeting over these intervals, run together, melt together, and leap apart in turn

Lucretius III. j94-340.

ii.

these images which pass off from things like films drawn from the outermost surface of things

constantly flowing off from things and gliding away

fixed outlines of shapes and of finest texture which flit about everywhere, but singly and separately cannot be seen

there are others which arise of themselves or are formed by themselves in this part of the sky called the air

more thin in texture than those which take the eyes since they penetrate through the interstices of the body and awake the thing substance of the mind within

many images of things are moving about in many ways and in all directions

in that one moment of time many times are lurking

what we see with the mind is like what we see with the eye - it must come about in a like way

and because they are thin, the mind cannot perceive any sharply except those which it strains itself to see

A myth is, of course, not a fairy story. It is the presentation of facts belonging to one category in the idioms appropriate to another. To explode a myth is accordingly not to deny the facts but to re-allocate them. And this is what I am trying to do.

Lucretius IV. 26-818.
Ryle p.9.

iii.

The instabilities of reference we feel reading the classical representationalists comes from an overlay of metaphoric systems that go in and out of register.

Hume, for instance, intends a psychology. He wants causal explanation, and so attempts something like a dynamics of cognition, using concepts such as intensity, flow, speed and inertial moment.

There is a difficulty about the locus of his explanatory dynamics. he is sometimes, infrequently, physiological in the manner of his day: "A different perception or idea may require a different direction of the spirits in order to its conception" he says (194). But he does not strictly speaking have access to the body as an explanatory ground, since our belief in the identity and continued existence of our bodies is one of the things he thinks needs psychological explanation.

A psychologist who believes that "Nothing is ever really present to the mind besides its own perceptions" (1970 will have to explain perception in terms of perceptions. Hume thus tries to bootstrap an introspected dynamics -- a dynamics phenomenologically derived.

I've said that in perceiving we perceive objects other than ourselves by means inside ourselves which are not themselves perceived. Those unperceived means must be the ultimate locus of causal explanation, but since they are not perceived, a perceptual theory that lacks physiology and experimental psychophysics will have to do its best with what it seems to have: observations of the perceiving of objects.

A phenomenological psychology will try to (as if) see through the intentionality of perception to the organization of its means. These means are, and must be imagined as, spatiotemporal.

Hume p *

iv.

Thus Hume's explanatory dynamics easily reads physiologically, as if he were a sort of neural connectionist.

He seems to talk about energy transfer and activation levels: perceptions have 'force and vivacity" which "may be communicated by relation" (208). There is talk of transience or stability of structure: impressions may be "fleeting & perishing" (195) or "firm and steddy" (197). Transitions between such states are facilitated by customary conjunction, and a more energized state "gives propensity to that passage" (208) so that the thought "glides along the succession" (204). once stabilized, a structure perseverates: "is apt to continue, even when its object fails it ... like a galley put in motion by the oars " (198).

So one of Hume's metaphoric systems is organized by means of concepts relevant to imagining the physical processes by means of which we perceive. The other is organized by means of concepts relevant to objects perceived

Perceptions and impressions are spoken of perceptible: they "appear as (194) fleeting and perishing. They may be copied, and we know the difference between impressions and their copies by looking at them: the copies are dull. When a perception is absent we can see that it's not there: "It is impossible to overlook the interruptions in the appearances of the perceptions" (206).

They resemble each other and may be mistaken for one another: "We are not apt to regard these interrupted perceptions as different (which they really are) but on the contrary consider them individually the same, on account of their resemblance" (199).

The may be copresent and related in different ways. "What we call a mind, is nothing but a heap or collection of different perceptions united together by certain relations" (207). When we consider them we are "augmenting their number by present reflexions and passions". and "storing the memory with ideas (207).

iv.

Central terms, such as 'a perception', can read alternatively either as a placeholder for whatever unknown physical structure is the event of perceiving, or they can read in terms of the relation to an object perceived: "my perception of the bear".

There is no problem with this alternation or coincidence of readings, since the unknown neural structure is being taken as the unperceived means by which the perceiving relation is being accomplished.

The twist in Hume (and others) is that he superimposes these two senses of 'a perception" without distinguishing them, so "my perception of the bear" is spoken of as causal.

Since the causal locus of perceiving is inside the head, 'a perception' in the second sense cannot but be imagined as also being in the head, and then we are on the way to imagining the trains of similar and dissimilar time-sliced bit-perioded little transparencies that Hume describes moving rapidly or languidly, vanishing and perishing, vivid or dull, "in the mind" -- like the thin films of Lucretius following one another ("as light comes up behind light") into the diaphanous body of the mind: buffeting, running together and leaping apart.

Lucretius IV. 176; III. 370.

 

Part IV. Representational Metaphor

i.

A representation is a representing object - a perceptible object of some kind - which we use to trigger or evoke brain states in ourselves and others.

Its function as a representational object depends on our taking it in a certain way that has to do with what it represents, depends on our brain's responding to it in ways that are more relevant to the object it represents than to the represented object. The representing object has to be physically present, has to affect us physically, but need not affect us in a way that would constitute our perceiving it. It merely needs to trigger a differential response. We have to be able to ignore it by responding as if to something else, in its presence.

We call the thing we are attending to by means of the representational object its 'content', but the term is a symptom of our misunderstanding of the way we use representations. There is strictly speaking nothing contained in a representing object.

A written sentence, which goes on existing on the page when no one is reading it, seems to contain whatever I or any stranger find when we come to read it -- the way a closed box may contain a ring that anyone who opens it will find there.

if we remember how a representational object does its work, we see that this is a manner of speaking. A representational object evokes neural structure of one kind or another, and this evocation is our understanding of the representation, our use of the representation.

This doesn't imply that we contain what we have called the sentence's content. Our skull contains the neural structures by which we understand the sentence, but the thing we are related to by means of that structure is not contained at all. It is either outside us -- a thing or episode in the world -- or else it is nowhere.

ii.

Representation has no business in the brain.

Confusion about what sort of representational format the brain works with -- whether visual experience must arise from iconic representations or could arise from propositional representations, for instance -- is a malady that comes from misunderstanding the metaphor -- from not understanding that 'neural representation' is a metaphor.

The metaphor is applicable in this way: a representational object makes us see or understand something other than itself, as the configured brain does. In both brains and representing objects, what we see or understand depends on how a material is organized.

The metaphor is inapplicable in this way: the relation between the representing object and the person using the representation is a relation between two separated entities. The person is using the representing object by means of its whole human ability to respond to the representing object by ignoring it relevantly.

Neural representationalism is Cartesian insofar as it presupposes these two inapplicable aspects of the metaphor: the opposed presence of two separate entities, and the whole human ability to respond to the presence of one thing as if it were the presence of something else.

It is very difficult to use the represenationalist metaphor without importing the structure of our practices in using and talking about public representational medial if we talk about neural configuration s inner representation, we immediately imagine it as separate from whoever is using it -- we immediately imagine it as separate from whoever is using it -- we imagine someone either a) perceiving it as we do perceive sentences when we want to; or b) perceiving something else (its content) in its presence.

A neural structure is a physical entity (of a somewhat indeterminate sort) but there is no sense in which it is opposed to the person using it: it is of course part of the person who is among other things the use of it.

And since the person is not separate from the neural structure it does not make sense to say that its presence evokes a neural response in the person -- that is, functions as representations do.

iii.

There is a way to use 'representation' carefully and harmlessly, and Dennett does sometimes use it in this way, as shorthand simply for neural means, the structures/processes by which we see, think, etc.

Then we can say 'representational format' is just all the ways brain processing is laid out spatially and temporally when we are perceiving or thinking something: the way it does what it's doing in order to be doing what we're doing.

Then 'representational content' is just the neural system's relevantly differential response in the world -- a flux of many simultaneous responses some of which are sensory and some of which are not.

iv.

Yolton makes the interesting suggestion that Descartes is not Cartesian in the way we have thought, and that he and other classical representationalists are not to be taken literally when they use representational metaphors to begin to try to talk about the relation of physiology and perception.

"A revision of the standard interpretation is required," he says, although there are "many supporting features for that stereotype in the language about ideas and knowledge found in the writers." he concedes that the terminologies we find fin Descartes, Locke and Hume suggest that "none of these writers fully articulated the perceptual theory required by the revision I am suggesting", and that there is evidence in the texts of a "tension ... between literalness and cognitivity.;'

The literal reading of these writers is Reid's: "they all suppose that we perceive not external objects immediately, and that the immediate objects of perception are only certain shadows of the external objects". A Cartesian reading of Descartes thus brings with it an inner perceiver of a perceptible representational object: the Cartesian homunculus.

Yolton calls his alternative reading a "psychological or cognitive reading", where "cognitive" can be taken as linking the classical representationalists to more sophisticated perceptual theory. His linking device is a version of the notion of code structures which he calls natural signs.

He cites an unusual passage in ?Descartes: "But, if words, which signify nothing except by convention of men, are sufficient to make us conceive of things with which they have no resemblance, why is Nature not able to have also established a certain sign which makes us have the sensation of light; even though this sign has no feature which is similar to that sensation".

the relation between physiological processes and perception is made more explicit in another passage where Descartes note that "by moving the pen in one way, we can trace letters that make us imagine battles and storms; but by moving the pen in other ways, we can cause quite different thoughts".

Yolton's suggestion is that we can imagine "motion in the body", i.e. perceptual physiology, as being a kind of geometrical language, a language of motion, the "reading" of which, by "the rational mind", is the perception of "things themselves."

Perceptual reading, in this version, never regards the representational medium and so is not said to be an inner perception of representational objects. We don't read the presence of the outer object off the presence of an inner object which is "our sensation" or "our perception". Instead, physiology, the language of natural signs, gives us the things themselves, made visible by means of our bodies.

The 'look' of the format of the representational medium is thus irrelevant, because "the rational mind" is not looking at it. It need not resemble, just as sentences do not resemble the understanding they evoke.

The perceptual theory inherent in this reading resembles a direct perception theory such as Gibson's insofar as it eliminates the notion of a perceived inner object by a "movement away from the literalness of the language" toward a "firm distinction ... between ideas as objects or entities, and ideas as perceptions, as acts with contents."

Whether it is in fact still a representationalist theory seems to depend on what is meant by a perceptual act -- on who it is who is thought to be acting. The perception-as-reading metaphor can of course be taken literally too. I have said that a sentence has 'content' by being read, and that reading is an evocation of brain structure by means of a physically present representational object. We can read in such a way that we are triggered by the presence of the representational object without doing much in the way of perceiving it. We are instead engaged in a kind of sensory simulation of what we are reading about.

So there are two ways we can take Yolton's/Descartes' reading metaphor. If we take it literally, the perceptual theory is still representationalist, because the 'reading' of brain writing which disregards its physical presence while being triggered by it into a train of perceptual experience requires the presence of a person, with full cognitive capacities, whose act this reading is: a person which a brain whose perceptual states can be evoked by the curves and waves of brain motion.

If we take it metaphorically we seem to arrive at a description that eliminates the need for the metaphor, or for any representationalist metaphor. If brain motion is not a language, but simply the complex structure of the creature's manifold relatedness in the world, then it does not need to be read or de-coded. There is no extra act required: the creature is the brain motion which is its relatedness.

I am taking it that Yolton's reading is the former, mainly because his text seems to retain the sorts of ambiguity I think representationalism cannot avoid. His reading of Descartes (and Locke and Hume) may or may not be too generous when it interprets him as the more modern sort of Cartesian that a computational representationalist is, rather than the more easily repudiated sort of Cartesian the computational representationalist isn't.

Yolton, p.14.
Reid, Essays, in Yolton, p.3.
Yolton P.6.
Descartes, in Le Monde, ou Traite de la Lumiere, in Yolton p.24.
Descartes in Principles, quoted in Yolton p.23.
Yolton p.14.

v.

Akins says Dennett's question is "What is the relationship between representational vehicles (formats, types, levels), representational content and conscious experience?"

She guesses that Dennett's answer is that the form -- format, structure -- of perceptual representation is not perceived: that the physicality of the representation is transparent: that it is the organization by which we perceive the content of the representation, this content being our "conscious experience", the "phenomenal feel" of it.

This construal of Dennett is as if an importation into the head of the basic perceptual relation -- that we perceive that thing there by being these means not themselves perceived.

The representational vehicle is the transparent means by which we are the perceiving of -- our experience? Which in turn is acting as the representational vehicle by which we come to know about the world?

No. The representational metaphor is too much trouble. Leave it out. No representational vehicles, no representational content. Just this: neural structures are the transparent means by which we are being our experience -- perceiving the world

Akins and Winger, P.1; p.15.

vi.

How to talk about color phi is evident when we move out of a representationalist frame. The question is whether nerves have to take events in the order they are given.

The answer is that they don't. Circuits discriminating microaspects of something can be faster or slower than other circuits working on perceiving the same event. Their lag-times in relation to the time of the event can be different: their discriminations can be differently offset. When the event perceived has a very small duration we may be able to notice temporal anomalies.

Dennett is not clear in the way he says this because he is using representationalist terminology that implicitly separates neural events from "our experience".

We have to be in the same sequence as neural events, since we are those neural events; a color-detector discriminating color NOW is our experiencing color now. But we can get it wrong: we can experience it after it stops happening. We can experience it in the wrong order, or partly in the wrong order. We experience it when we're ready, not when it happens. That is called reaction time. It's normal. it has to be assumed by any materialist theory of perception.

If you are thinking of neural events as representations you can imagine a neural structure being like the words "a second ago" being read now: a representational possibility like the word "green" being printed in red ink.

Like any representing objects, these worlds work as representations by evoking neural structure relevant to a second ago, or to red. That is, they are representations because they are separate from their user, and because they trigger a structure relevant to something other than their own properties as objects: their existing now, or the color of their ink.

But a neural circuit whose task is to discriminate red can only be doing what it does: discriminating red. It is not representing red because it is not triggering a discrimination of red in someone else: the buck has stopped, this discriminating is (sometimes) also the experiencing.

If neural events are seen as representational they can be thought to say 'then' now, but to whom?

vii.

Does this mean we can never think about a second ago? no, but it means that all the micro-discriminations by means of which we do so will be discriminations we are making now: will be experiential microconstituents of now This is the same as saying that remembering how Wren looked ten years ago requires a slew of presently active micro-discriminators whose activity my experiencing its.

This is obvious. We can's remember without neural means.

The implication that is not obvious in Dennett's explanation of color phi is that "subjective sequence", when we mean by this the micro-order of experiential microconstituents, is and has to be the same as the order of neural events.

Neural structures, by being active now, cannot evoke an experience in some time other than now.

First, because unlike representational objects, they do not evoke at all. They are what is evoked.

Second, because we misunderstand the use of representational objects when we think of them as evoking experience in another time.

A representation does freeze something, temporally. It is a material inscription: the forming of a material that will tend to hold that form. Thus the handwriting in my mother's letter retains all the stops and starts and tensions of her act of writing it.

But it does not freeze the time in which she wrote it; it merely retains the form she gave it in that time. Because it does so, I can read it again, and when I do, it may trigger neural structure which is somewhat like the structtere it triggered the first time. Or it may trigger structures which include the structures that constitute 'remembering when I read this before.'

Neural structures which are re-evoked are like letters, in a sense. They do not go on existing as structures, but to the extent that they are re-evocations they do depend on material changes in a medium. Synapses have to have been modified, presumably. They will retain something of the form given them in another time.

 

PartV. "Conscious Awareness, as People Redundantly Say"

i.

We could say 'experience' is this flux of many simultaneous differential responses but that is not what people seem to mean when they talk about their "conscious experiences", because the way they talk about them resembles the way they talk about the things they experience: as things they can observe and report.

Perhaps what they report -- "conscious experiences" -- is shaped by the way they report them . Perhaps when you report something you cannot observe, you are imagining instead of observing.

Then, if you are a philosopher, a folk philosopher maybe, you will be in the difficult position of looking for ways to explain 'consciousness' as you imagine it, rather than ways to explain the flux of multifunctional differential mini-responses -- 'rolling consciousness' -- which could be in easier explanatory contact with neural models.

ii.

To say 'consciousness' is imaginary is not to say there is some nothing we are talking about.

It is to say we are talking about it in ways we are not aware are metaphorical -- ways that are structured by metaphors we are not aware of.

To say our talk is structured by a metaphor is just to say that it is structured in ways that both miss and fit.

Ryle, Wittgenstein and Dennett are all saying: yes, there is a difference between being aware of something and being unaware of it -- "Admit it? What greater difference could there be?" -- but the way we talk about that difference implicates us in a structure that fictionalizes it. "We have only rejected the grammar which tries to force itself on us here."

Wittgenstein, PI 314.

.iii.

"Always get rid of the idea of a private object in this way: assume that it constantly changes, but that you do not notice the change because your memory constantly deceives y ou ."

Consider the fact that, as they are in the process of making micro-discriminations circuits go through a process of settling. And consider the fact that the settling process can be unsettled or supported by activity in other circuits. A pattern amounting to a larger discrimination will take time to stabilize, and will be vulnerable as it gels. What we perceive may waver: and the physical event that is the perceiving may not last long enough to make the sort of physical difference that would constitute the possibility of remembering it.

Consider further that reporting a perception will require connections among circuits that are the event of perception and circuits that are the event of reporting. If the act of reporting, and if hearing one's own report, feed back onto the cortical areas that are doing the perceiving, we may destabilize, or we may restabilize, the discriminative decision we were making as we began to report. Reporting might help us sustain a pattern; it might extend it by turning on other sorts of circuits compatible with it: it right alter it in small ways ; or -- although this seems unlikely for evolutionary reasons--- reporting could set you up to see something else altogether.

And it could do so, again, in a way that doesn't make the sort of difference that would constitute being able to compare. And now imagine the perceiver again reporting, from this altered discrimination, this time in such a way that the probe, the process of reporting, the process of hearing the report, all work toward stabilizing the new pattern.

Dennett's point isn't that perception is built on sand, but that talking about it is. Notice how careful I had to be where I put my feet, writing these paragraphs. "What we perceive may waver" I wrote, and that wasn't careful enough, because, reading it, I 'see' a diaphanous sheet of immaterial color hang in midair wavering: something between waving and fading.

Two things: first, this construction suggests that it was my perception I saw, and that I saw it waver, when in fact the bear that was there to be seen did not waver, though I wavered in my seeing of it. Second, when I reconsidered 'what we perceive may waver' -- when I considered its effect -- I was doing something whose structure resembles the structure of looking and reporting: some kind of sensory pondering along with a process of finding words -- in that way it was as if I were seeing; I was set up as if I were seeing; I was simulating seeing; I seemed to be seeing.

I really seemed to be seeing but if there is nothing to see (and there wasn't) we shouldn't call it seeing, really. Call it seeming to see; but let's be more specific. What I was seeming to see was my wavering perception -- which doesn't exist to be seen (but which did exist as a momentary relation between me and the bear I almost saw, or didn't see, or registered without seeing). Because that was what I seemed to see, it was also what I seemed to report.

Ordinarily, when I report a perception, say of the bear in the presence of the bear, I am reporting from the perception. I am set up so the structures that are the reporting are running off, driven by, the structures that are the perceiving. But when I was reporting from whatever circuits were the considering of the effect of 'what we perceive may waver', I was reporting from circuits that were imagining not perceiving.

I imagined I was reporting the perception of a perception; I was reporting a fantasy of a perceptible perception.

Cartesian materialism is materialism interwoven with unnoticed fantasy of this kind.

More: when I said I saw a diaphanous sheet of immaterial color hung in mid-air wavering, you saw it too. Didn't you? Well, no. You probably imagined something, but you didn't imagine seeing it, because that it is another trick.

Wittgenstein, PI, xi.

iv.

Imagine a philosopher trying not to forget about brains, in the very event of doing philosophy.

Here I am talking about perceiving. How is perceiving done? I'll call up an episode of perceiving and see what it's like. The introspective method.

But now try to imagine how perceiving is really done ...

Circuits configure in my brain in one way rather than another, the way they must configure in order for me to see the particular thing I'm seeing.

Further: when I was trying to introspect an episode of perceiving, how was that introspecting done? When I imagine perceiving, circuits configure in my brain in something like the way they configure when I'm perceiving. That's all. Introspecting perception is trying to seem to perceive.

But I wanted to see how perceiving was done.

You can't see it by looking. Your brain is designed not to be able to perceive itself. The way yourr nose is designed not to smell itself. It would put you at a disadvantage.

But then how am I to talk about perception? By imagining the thing that's happening that you can't see happening.

When I am imagining that-- the configuration of circuits-- the event of imagining it will be ... itself some configuration of circuits. Yes. But you won't be seeing it, or imagining it right. It will (not) be a coincidence.

The introspecting process sets itself up.

v.

When I say, what is memory like, what is remembering like, I imagine something.

I never have 'a memory' in front of me (or behind me). A memory is not something I am examining: it is something I am being.

Different questions:

1) What does 'a past perception' mean, in effect? It means the state that constructs itself when I ask what happened. The state I am in now.

Operationalism: that the phrase means what we mean by it, not what we imagine we mean by it.

2) What do we imagine we mean by 'a past perception'? The state we were in in the past, still there to be looked at again.

vi

"The very 'feel' of our phenomenology. in all its apparent complexity and variety, in some sense 'is' the representational content of a 'narrative stream', a stream that has been precipitated by particular 'probes' or questions."

If by 'phenemenology' Akins means what I called the flux of differential microresponses, then the phenomenology is the feel -- the rolling multiple microfeel -- which not only seems but is complex and varied-- but it is not 'representational content' because it is not representational -- it is the perceiving of the thing. No representational metaphor or we'll be seeming to use our phenomenology as our means of perception.

If by 'phenomenology' she had meant what Dennett often means by it -- 'conscious contents' or 'conscious episodes' as we describe them if asked -- what we imagine when the story is told, the transparent flow of things we perceive dematerialized into 'perceptions' and 'memories' -- then, yes, phenomenology and especially its feel of complex and varied apparency, is the content (nit precipitated but) stabilized as neural eddies that result from a process of probing. It then is phenomenology in the technical sense of experience as it seems to be experienced when we seem to observe it.

Akins and Winger, p.15.
Dennett CE p.365.

vii.

"Narrative stream": there is a real stream and an imagined one. The real one is the temporal stream of the brain's continuous formings and dissolvings (more cauldron than stream) which are the person as flux of differential micro-response. This is the stream that among other things forms narratives -- and when it does so is the narrative stream.

The unreal stream is that stream -- the brain streaming in the process of making up in response to probes -- as it is being imagined by itself.

Dennett, CE 113,
 
 


MULTIPLE DRAFTS: PERCEPTION WITHOUT REPRESENTATION - bibliography

Akins, K. and S.Winger   " Ships in the Night: Churchland and Ramachandran on Dennett's Theory of Consciousness"; forthcoming.
Dennett, D.  1991 Consciousness Explained, Little Brown & Co.
    "Consciousness: More Like Fame Than Television", forthcoming.
Gibson J.  1966 The Senses Considered As Perceptual Systems, Houghton Mifflin.
Hume  1888 Treatise of Human Nature ed. L.Selby-Bigge, Oxford University Press.
Kelley, D.  1986 The Evidence of the Senses: A Realist Theory of Perception, Louisiana State University Press.
Lucretius  1924 De Rerum Natura ed. T.Page, Harvard University Press.
Pitcher, G.  1971 A Theory of Perception, Princeton University Press.
Ramberg, B.  1994 "'Dennett's Stance: Is Consciousness Explained a Theory of Consciousness?", unpublished.
Ryle, G.  1949 The Concept of Mind, Hutchinson and Co.
Wittgenstein, L.  1967 Philosophical Investigations, Basil Blackwell.
Yolton, J.  1984 Perceptual Acquaintance from Descartes to Reid, University of Minnesota Press.