Sensing as Non-Epistemic
Dr Edmond Wright
Sensing as non-epistemic: 'Our sensing is real in itself, apart from the percepts we select from it'
A sensory receptor, in any organism anywhere, is sensitive through time to some distribution - energy, motion, molecular shape - indeed, anything that can produce an effect. The sensitivity is rarely direct: for example, it may track changes in relative variation rather than the absolute change of state (as when the skin responds to colder and hotter instead of to cold and hot as such); it may track differing variations under different conditions (the eyes' dark-adaptation; adaptation to sound frequencies can lower the difference threshold; the kinesthetic sense will shut down if a limb is held in a stationary position too long - the limb 'going to sleep'); it may be subject to distortion of the input from overloading (dazzle producing strong-after-images); it may not be confined to one channel of sensitivity (the retina is sensitive to pressure; the hands can feel some strong sound-vibrations, the tympanum of the ear records touch). Strictly speaking there is no limit as to what intensities and what ranges receptors could be sensitive. Sharks are sensitive to electrostatic fields, homing pigeons to magnetic fields; snakes to infra-red rays; bacteria to acid concentrations; perhaps there has even been a mutant organism sensitive to the passage of cosmic rays, even though that would hardly have bestowed any conceivable survival value. What is irrefutable is that individual receptors differ markedly from organism to organism, between different members of the species (one dog being better at tracing smells than another; one person being able to sense light-waves of 375 nanometres, another not; children able to hear 20,000 Hz, older persons not), and between receptors of the same kind within one organism (one eye being sensitive to 765 nm and the other not; one ear deaf to 15,000 Hz and over, the other not). There are also just-noticeable-differences (JND's), in that one person can see two shades of a colour where another sees only one; similarly with sounds.
From all receptors the input is transmitted via nervous impulses to parts of the cortex the sensorium, where sensory realization is produced. On the way, processing of that input is possible to produce a more adaptive realisation (two examples - inputs may be combined to facilitate a stereo-effect (as from the two ears in hearing and from the two eyes in sight); portions of the input may be subjected to differential intensification so that transition regions between one area of intensity and another be rendered narrower and the differences between the two areas become artificially heightened). None of this processing of the input, however, alters its brute quality. At this stage the realization in the sensorium contains no information; it is no more than bare evidence. To draw an analogy: a slightly increased temperature in the ceiling over your head is a natural sign that you are in the room, but no one would say that the ceiling literally contained information about your presence.
It has already been argued (Sellars, R. W., 1922, 37; Sellars, W., 1956, 324) that, as regards external causation, the relation is not one of pictorial resemblance, but of structural isomorphism, where there is a proportional concomitance, not necessarily direct, between (a) intensity-states over an internal matrix of immediately sensed experience and (b) intensity-states at a receptor-matrix . A logical parallel for this isomorphism would be between the grooves on an old 78 record and the variations in the laser beam in a CD player producing the same music: there is no direct qualitative resemblance - it is just that the one has variations which can be indirectly mapped onto the other. Note that this implies that the internal colour-experiences are completely unlike what they change in harness with externally. Neither need there be any direct ratio between the two: all kinds of automatic modification of the input can take place (such as evolved Gibsonian facilitations like contrast enhancement), but none of these will affect the complete absence of qualitative resemblance.
John McDowell has clearly not considered this mode of resemblance, for he asks 'How can there be a resemblance between a colour and something we can't characterize in terms of how it would look?' (McDowell, 1985, 113), yet a structurally isomorphic concomitance of field variation is precisely such a resemblance. Robert Kirk (1994, 9-10) is similarly blocked, for he tries to dismiss inner sensing on the ground of there being no pictorial resemblance, but the argument here for inner sensing does not claim a pictorial similarity between input at the receptors and cortical realization, as Frank Jackson's does (Jackson, 1977, 74).
One interesting failure of pictorial resemblance never considered by anti-experientialists can be shown by an analogy Sellars suggested. You will have seen 'Movitype' strips in advertisements and railway carriages on which running announcements are made. They consist of a matrix of tiny lights which are triggered in series. Notice the fact that over a Movitype matrix nothing actually moves , just as nothing moves on a cinema or TV screen. It seems plausible therefore that, similarly, upon the cortical matrix nothing actually moves - there are perhaps merely successions of pixel-firings across the matrix converted by the neural circuits responsible for the Phi Phenomenon into an apparent movement. Whatever actual motion in the external continuum may be it will in no way pictorially resemble what happens within the cortical matrix or even what the Phi Phenomenon produces: the only match it will be of highly indirect proportional variations between two utterly distinct series of changes.
One can see here how two of the more familiar objections to inner sensory experience fail: (1) there being no claim for pictorial resemblance, the Regress Argument loses its purchase, for whatever cortical visual experience may be it cannot involve another set of eyes since there is no picture there to look at with them; (2) the claim that the Indirect Realist has no way of comparing 'inner' with outer' also has no point, since no comparison other than an isomorphic one is claimed, and that may yet be empirically proved. Visual sensing is a direct experience completely unlike external reception of light-rays, except for the indirectly traceable proportional variations, and furthermore, it is only contingently connected to it.
That there is no qualitative resemblance is a clue that upsets a very strong prejudice. Alan Millar (in discussion after his talk, 'The Idea of Experience' at the Cambridge University Moral Sciences Club, 31 January, 1995) can be taken as representative of one who finds it difficult to shake off the notion that what the Indirect Realist is arguing for is some kind of picture being viewed with eyes from a distance, for he rejected the idea of inspecting a mental image on an inner field on the ground that one could not move one's eyes here and there as over a picture; he demonstrated the movement by looking at a glass in front of him at different angles and distances. But this is to confuse what the eye does, namely, pick up a distribution of uncoloured light rays from differing angles and distances in order to change perspective on an external region, with what the internal mental-image system can do, namely, bring a whole series of distributions onto the cortical matrix for gestalt-inspection, distributions that could display an object at different angles and distances. If a computer-graphics system can manipulate shapes from all angles without any eye having to move over the TV screen, there is no empirical difficulty in having a mental-image system that can turn objects to any angle, shrink the image, do close-ups etc. As is well known, Daniel Dennett challenged good visualizers to turn a shape he drew, and then say whether a face of a particular cube would be visible through a hole. I found no difficulty in performing this task and assuring myself that the face would be visible (Dennett, 1991, 289). My mental-image system can perform far more difficult tasks than that, such as showing me a Rolls-Royce car on a moving turntable.
Therefore, with regard to vision, cortical colour is not seen with eyes, but is directly sensed in the brain. It must be strictly distinguished from the external causes to which we attribute 'objective colour'. The phenomenon of stereoscopic space, which is itself a cortically produced feature, is what allows the development of a judgement of external space, but the same applies in that external space and phenomenal space have no qualitative resemblance.[1].
These proportionately matching intensity-variations provide the needful constraint that McDowell and others are in search of. Evolution has provided this Gricean natural sign (Grice, 1967) upon which, within the present theory, organisms with motivation systems can project gestalts in order to guide future action. A dark cloud on the horizon is a 'natural sign' of rain, that is, evidence for it, a repeatedly observed causal concomitance from which we have learned to make a fairly reliable inference. Within this theory, as will be explained at greater length below in the 'Perceiving as Epistemic' section, the epistemic level is taken to be produced by this gestalt-projecting module, excited by pleasure and pain to place a gestalt in the memory, tabbing it there with desire and fear respectively. To subvert a term from Dennett, the 'intentional stance' is something we cannot choose but take up, given that we are lucky enough to have it. It is quite distinct from the sensory non-epistemic level; the two systems exist independently of each other. The existence of agnosics is good empirical indication of this: they are those who, although they sense perfectly well, have unluckily lost the ability to select from their visual fields in that their experience is purely knowledgeless- they are able to see but not to look. In extreme pathological cases there may be sensory experience without any possibility of the gestalt-module working appropriately, that is, there could be no self at all and yet colour, for example, could be being experienced. This feature of the argument, as well as that of the isomorphic nature of the resemblance, counters Richard Rorty's claim that such an inner experience is claimed to be a 'mirror' of objective facts (Rorty, 1980). It also counters the arguments of Gerald Vision (Vision, 1997) for his theory depends on the visual system delivering information to the eye. The isomorphic relation is one that provides only evidence that may or may not be interpreted, certainly not 'information' that is already sub-edited for the brain into given recognizable external entities and properties.
The idea that sensory phenomena are natural signs, no more than 'pure sensations' is traceable to Hermann von Helmholtz:
Our ideas of things cannot be anything but symbols, natural signs for things which we learn how to use in order to regulate our movements and actions. Having learned how to read those symbols, we are enabled by their help to adjust our actions so as to bring about the desired result; that is, so the expected new sensations will arise (Helmholtz, 1968, 179, 188)
Helmholtz's only error here is to use the term 'symbol' and thus mistakenly associate the sensory evidence with the practice of human linguistic symbolization. The sensory fields being structurally isomorphic effects of the overall input are not in themselves signs of anything, certainly not any thing, no more than the dark cloud on the horizon is in itself a sign of anything. It is our learning via the strictures of the motivation of pleasure and pain that enable us to use portions of the fields as signs. He is actually being inconsistent, for if he allows that we can 'adjust' our responses, he must also concede that we can adjust our reading of the field, that is, change the 'sign', for later experience may lead us to take that selection from the sensory field another way - we are even at liberty to change the boundaries of the selection itself. 'The' dark cloud may turn out to have been a mountain, and one extending further in our view than our 'dark cloud' (since portions of the mountain were in sunlight and we had not included them in what we took to be a singular 'thing' before.
Different parts of the sensorium produce different sensory experiences, but the connection between receptors and those parts of the sensorium is contingent (Gregory, 1983, 451). For example, there is no necessary link between what comes into the eyes and a visual experience: coughing in the dark produces pressure-phosphenes for some people; subjects with a probe in the visual cortex report light-experiences when the probe is electrically activated. In the MIT psychological laboratories recently a ferret had the connections of its optic and auditory nerves interchanged, so that it heard light-waves and saw sound-vibrations. To make the point of this contingency between input and presentation clearer: a human subject could have the probe in his visual cortex connected to a microphone and he could be a source of reliable information about whether sound was being picked up by that microphone. A practical demonstration of this could be arranged in analogical form: the output of two stereophonic microphones could be led to two miniature TV screens in a hood in front of the eyes (transforming pitch into the colour hue, loudness into brightness, tone variations into colour saturation, all 3-D presented), with the output of two TV cameras at the front of the hood transformed contrariwise into aural effects in stereo-headphones. A subject could learn to act in the world to some degree of efficiency with such a transformation (Wright, 1993, 370-71) and communicate with others about what is to be considered the best form of referent. One clear consequence of this contingency is that there is no given information in a sensory field; there are merely concomitant causal variations that may or may not be found significant for action.
This point has been elaborated because it supports that vital characteristic of all sensed fields already indicated above, namely, that they contain no information whatsoever, that they are 'raw feels' (Feigl, 1958), 'non-epistemic' (Collins, 1967), 'non-doxastic' (Crumley II, 1991), or 'non-conceptual' (Evans, 1982). The word 'data' cannot strictly be used of them at all, an error made by both the Sense-'Datum' theorists of the early part of this century and their opponents of the mid-century. A typical Sense-Datum theorist such as Roy Wood Sellars endeavoured to resist the notion that the fields were 'anoetic', as he called it. He said that he was distinctly unhappy with the idea that the field itself before the 'configurational wholes' of cognition are selected exists without any given information upon it (Sellars, 1932, 88), that is, . Yet he still wanted to maintain an intuition/judgement distinction (Sellars, 1965, 236), with the intuition 'simpler' than the perception. It was because of this, he said, that artists were able to disturb 'inferential elements' (1916, 18) since they were more responsive to unnoticed sensory features. He had defined his point more exactly two years earlier in saying that he did not believe that there was a 'chaos' of sensations but a 'patterned field' controlled by 'the stimuli coming to the organism' (1930, 268). However, the patterned field does not contain any given information. If, as Sellars correctly insisted, one is to allow for shift of attention over the field to improve reference, one cannot in the same breath believe that there are already a set of given selections awaiting choice. He should have seen that the terms in which one could describe the state of a field are remote from the terms one uses to describe what is seen on that screen. For example, the digital state of the phosphor cells of a TV screen is on a different logical and empirical level from what is judged to be present on the screen.
However, many philosophers are opposed to the idea of the sensory being no more than brute evidence. John McDowell expresses the fear of the non-epistemic thus: believing in its 'bare presence' threatens 'to dislodge our grip on the requirement that empirical thinking be under constraint from the world itself' (McDowell, 1994, 42). The source of this fear lies in the failure to understand the relation of a sensory field to its causation, internal or external. He repeatedly admits that he finds the whole notion 'unintelligible' and 'mysterious', sometimes arguing as if the mere fact of it being a mystery to him were sufficient for its non-existence. As an example from McDowell: 'If the "inner" role of colour concepts were a self-standing starting-point, "outer experience" of colour would become impossible to comprehend' (1994, 31); and also, 'we cannot make sense of thought's bearing on the world in terms of an interaction between spontaneity and receptivity' (139). It is fair to ask who the 'we' are that cannot make sense of it.
Even the Idealist philosopher finds it difficult to yield to the notion of the sensory phenomena being no more than a non-epistemic field. John Foster in his recent book on perception (Foster, 2000, 116-20)) can only characterize the sensory at the simplest level as what he calls 'imagistic', but his definition still has the observer picking out perceptible aspects. He cannot make the move to accepting the sensory as bare, non-mental eividence.
The non-epistemic can in principle be described (it would be a matter of listing all the states of the cells of the matrix) in what can be called a 'field-determinate' description, but the terms applied to the objects seen are inapplicable to it - it would be like trying to say of a TV picture of the Statue of Liberty that a computer print-out of all the cell-states was a description of the statue. Referring to the results of the object-selection process, we make an 'object-determinate' description. R. W. Sellars himself had used the example of a 'Movitype' panel mentioned earlier (1969, 217; compare Maund, 1993, 177), showing himself aware of the distinction between the state of the matrix of electric bulbs and the various interpretations that passers-by might make from that matrix. Sellars yields to the same temptation as James J. Gibson, who, correctly impressed by the pre-phenomenal processing that goes on before actual sensation, was mistakenly led to think that these provided some confirmation of there being pre-existing entities to be observed, 'invariants', as Gibson wishfully called them (Gibson, 1968). But this is equivalent to saying that on a TV screen turning up the Contrast knob provides a guide to the perception of what the screen may show. It is wholly false, since recognition has no logical relation to any such contrast: a skilled entomologist who happens to be somewhat short-sighted may recognize the camouflaged moth on the tree before the tyro who has perfect eyesight. There are even occasions when a blurred view is more helpful in picking out an object than a clear one, as in looking at a large-pixel picture; there is a large modern painting ('Janet' by Chuck Close) in the Albright-Knox art museum in Buffalo which shows a face seen through cellular glass - when one is close to this picture it is virtually impossible to see the face, not until one screws up one's eyes to blur the image ( i.e. reduce one's level of JND discrimination), does it gather into recognition. So there is no necessary relation between the state of the field as regards distinctness and the knowledge that may or may not be drawn from it. It can be put in the most general way by saying that discrimination at the sensory level is not to be equated with interpretation at the perceptual level; it would be the equivalent of asserting that the more sharp-sighted one was the more one knew; one might say that acuity is not the same as acuteness. A discrimination at the non-epistemic, sensory level is not the same as a distinction at the epistemic, intentional level. This in itself is a proof that sensing and perceiving are utterly different processes. The Gibsonian systems can only be said to be apt for perception, automatic features that are useful on many occasions, but they do not of themselves bestow any knowledge, and indeed on some occasions may be unalterably maladaptive, actually misleading, which has been the fate of many automatic processes in evolution. In a particular evolutionary niche they may provide some current facilitation in the gathering of knowledge, but they can in no way be a scientific or metaphysical foundation for it. This has also been the error made by David Marr (1982), who was led into using the term 'edge' to describe what a visual recognition program was aiming to define; an edge is a term applied on the interpretation side of the description of a matrix - Marr had only the right to speak at the level of the matrix, that is, he was really discussing transition regions where the rate of transition was rapid, not 'edges' at all. So the 'patterned stimulus' is not patterned with given entity outlines. Hypothetical constructs are made by a species of gestalt-projection upon the sensory field and therefore there must be the possibility of analogue shifts of attention over the field; they cannot be limited to given stopping-points - otherwise there would be no possibility of adaptation.
The implications are significant. First, if the field can exist in a purely non-epistemic state, it is empirically possible for there to be no self observing at all, that is, that there is no selection of action-relevant gestalts taking place at all. There is the obvious example of the brain-damaged patients described as 'agnosics' who, still possessing the ability to sense have lost the ability to perceive. Someone who studies severely autistic children, Beate Hermelin, has said of them that 'they can hear and see, but they cannot listen and look' (Hermelin, 1976). Under the present theory, they have been born without the ability to bring to bear the memory-deliverances of the pleasure-pain system upon the sensory fields (for, in a moment, it will be claimed that no flexible cognition can take place without such a system). It is obviously empirically possible that a mutation in some animal occurs without any neural connection between the sensory fields and the pleasure-pain system; in such a case no cognition of any kind could come about. There would be just blank experience going on without any consciousness. Note, then, that this theory denies that sensory experience in itself is conscious. Because of this, it escapes the familiar objection to the postulation of inner sensory experience that it logically implies solipsism, for there is no solus-ipse, 'single self' present in that experience.
If the neurophysiologists actually discover how a sensory field is instantiated in the brain, the Faustian possibility opens up of being able to create an active sensory field on the laboratory bench. Such non-epistemic 'experience' would not be mental at all, which leads us back to the normal human and animal case and enables us to say that the field in itself is not 'mental'. As has already been noted, it is no more than an evolved and very elaborate 'natural sign'. But this should have been obvious from the start, as it is plain that nothing we can do can prevent us having these experiences when we open our eyes and uncover our ears; the process is outside our control.
This is not Phenomenalism, because it claims that sensing is a part of the Real, whether or not it is being used to form perceptions. Sensory experiences do not 'represent' the world of reality, for that is merely what is guessed out of it. Interestingly, there was in Descartes himself an indication of this. Having wondered whether all phenomena were a dream, he did confusedly on the very next page of Section I of his Meditations toy with the notion that we might have to, as he says, 'admit the reality ... of certain real colours'. If he had pursued that thought, he would have seen that there was something that resisted the notion that all internal presentations could be dismissed as a dream. Because a cartoon appears on the TV screen we are not thereby led to think that the phosphor glowings have suddenly become illusory. There is a systematic ambiguity here that has been overlooked by the too-logically minded, and there is a special reason for their ignorance, explored in the Faith section.
If the field itself is not mental, it is therefore part of nature, a portion of the Real, of brute existence. It is certainly brute: we may be able to open and close our eyes but we cannot stop the visual field reacting to what stimulates it, even what we experience with our eyes closed (and that includes dreams, mental imagery voluntary or involuntary, after-images and the like). Just as the screen of a TV set remains real whether it shows a 'live' broadcast, a video recording, a cartoon, or even just interference, the visual array in the brain, its raster, as the neurophysiologists call it, whether it shows us the outer world, a dream, an after-image, mental imagery or an hallucination, remains real. The conclusion is that we sense the Real, whether or not we are deriving any knowledge from that evidence. To be more accurate, one should say that in sensing there is a part of the Real active within us, not that there is something through which we become aware of the Real; that is, we are not using the word 'sensing' here as one uses it in the phrase 'to sense the fragrance of a rose'. It is devoid of representation, of meaning, of 'appearance' at this level since it is not even to be called 'evidence' until we work upon it. It would in fact be nonsense to say that it was an appearance of itself. It can exist without any conceptualizing being in progress at all, but for all that it remains a part of the Real, of being, of the flux, of the continuum, of what exists (for further on this claim, see Wright, 1996, 27-8). The argument that permits us to make this inference arises out of the human process of intersubjective concept correction, for which see the section on the Joke and the Story. This is the basis of this form of Indirect Realism.
It is worth giving a special prominence to a consequence for a familiar philosophical metaphor. Those who oppose the notion of qualia always use the image 'the veil of perception' when they are attacking Indirect Realism. The most recent book has it in its very title: Skepticism and the Veil of Perception (Huemer, 2001), and its argument is devoted throughout to trying to overthrow that claim. It shows, for example, in Huemer's treatment of hallucination, for he is quite unaware that hallucination can be non-objectified, that is, 'abstract' in the painter's sense, consisting of phantasmagoric changes that present no object whatsoever, but he founds his belief in Direct Realism on 'being non-inferentially aware of objects', which he claims is a mistaken feature of all hallucinations. It should now be seen that this Veil metaphor has no purchase on the theory being presented here, for sensing is identified, not as a part of some immaterial consciousness, but not mental at all, a part of the brute Real itself, which for much of the time varies in rough ratio with real input at the sensors, but which can have more immediate neural causes (producing dreams, mental imagery, after-images, hallucinations, phosphenes, migraine 'fortification' patterns, etc.). To repeat the point made above, just as a TV screen remains real whether it is showing a cartoon, a video recording, a computerized scene, a screen-saver, or a live broadcast, so too the inner presentation remains real whatever we select from it. So the metaphor of a 'veil' is entirely inapplicable to sensing as described here, and those who might attempt to use it of the present theory are guilty of the fallacy of ignoratio elenchi, that is, endeavouring to refute what is not implied.
So even in the middle of a dream, the visual field remains perfectly real, so that, of a dream you had about unicorns, you can still be sure that the visual field presenting those unicorns was real in itself; furthermore, since we do sometimes in the middle of a dream become aware that we are dreaming, at such moments you can observe the distribution of those colours and accede to its reality without believing for a moment in 'what' you see, the unicorns.
It is noteworthy that the experiences of pain, pleasure, desire and fear also have this character. As regards pleasure and pain, it might be countered that they are more than mere evidence since they impel to particular action. But the learning that guides that action they motivate is not given; indeed, it has constantly to be refined and updated. A newborn baby can be thirsty without knowing that it is, that is, essentially, not know what to do about it. In addition, as we have already seen, it is that empirically possible for a mutation to be born without any neural connection between the sensory fields and the motivational system. Advanced organisms are also able to ignore the impulsions of pleasure and pain in pursuit of some other goal so the impulsions themselves are not determinately bound to what will satisfy them. Take C. B. Martin's example of the valet who used to test the temperature of his master's bath with his elbow, a master who liked it painfully hot according to the valet's judgement (Martin, personal communication). As regards desires and fears, the very fact that we can change them so that the underlying feeling becomes transferred to another object or removed altogether indicates that the feelings in themselves are not absolutely bound to the thoughts that accompany them; otherwise, we would never be able to change them. The same feeling of fear, for example, can attach itself to an endless range of objects and situations.
There are other considerations that lead us to the conclusion that the field remains basically non-epistemic even for the mature observer.
(1) Take, first, the difference between the fields of the two eyes. The perspectival difference, as those computer manipulators of the new 'Magic Eye' 3-D patterns well know (Horubichi and Inoue, 1994), is not specifiable in the epistemic terms we apply to the entities we select. Only at the level of the pixel is the difference describable. This renders inapplicable the objection of such philosophers as Ryle (1966, 107) and Pitcher (1971, 41), who tried to claim that when looking at a candle we squint and see two candles, we are, to quote Pitcher, only 'seeing one thing twice'. It is, however, very difficult to find a Sense-Datum philosopher of the old school who, unlike Ryle and Pitcher, is not aware that the two 'candles' are different; as Henry H. Price points out: he notes that the stereoscopic quality disappears and the images become 'flattened' (Price, 1961/1932, 28) and each image is of the object from a different place, producing 'two qualified entities' (57). To talk of 'entities' is actually jumping the gun: what Price should have realized, as Helmholtz did (Helmholtz, 1901/1868, 293) is that each eye's field is detectably different, actually at every point, and not at the level of 'entity' recognition.
Again, the neurophysiologist, like the computer programmer, may be able to give a precise field-determinate description of the state of the point-states of the cortical field traceable to each eye, but that epistemic description by the neurophysiologist at the level of the field is to be distinguished from the epistemic description that the experiencing subject may give, an object-determinate one. I have just looked at the front cover of the 'Magic Eye' book, squinting in the required manner: I saw a stereoscopic bas-relief of the number '3' and the letter 'D', but that description is utterly useless if what was wanted as a description of the detail in the pixels of the picture which allowed my internal cortical circuits to project the stereoscopic experience. That level remains non -epistemic to me. So too the difference in the fields of the two eyes.
There is a further point about double vision: because of the constant readjustment of the focusing angle of our two eyes, parts of our visual field are always overlapped with double images. It takes considerable introspective attention to become aware of them, but it is the same effect that shows up plainly when one holds a finger close to the eyes and then focuses on a distant point. None of these double images are cognized, so the conclusion is that much of one's vision is normally in a state of confusion, always non-epistemic for this reason alone. A simple experiment you can do now will also prove the point: without moving your head, try to recognise what is to the left and right of you; you will easily discover that all round the periphery of your vision is an undoubtedly sensed region which is outside your ability to recognise. An objector, on hearing this, tried to claim that those regions were merely ones of what he called 'faint belief', but that falls down because it assumes that those regions of the continuum causing those distributions have already been inspected and recognized, which is far from the case. It has the absurd implication that all sighted organisms throughout evolution have been safely protected from ambush from left and right by the 'faint belief' that predators were there. Sensing and perception exist to protect and maintain the organism and to further the reproduction of its species.
(2) The nineteenth-century psychologist Hermann von Helmholtz pointed out that the whole of one's vision is always affected by the after-image effects of what one was last looking at (von Helmholtz, 1901, 255). Butchers take advantage of this effect when they surround the meat in their windows with green paper, for the complementary colour of green is red, and the result is that the meat appears redder to the eyes of prospective customers than it otherwise would. Thus one's vision is continuously altered by a feature that plays no part in the normal assumptions of recognition. Perhaps that illusory effect tipped the balance on whether a piece of meat was bought or not - a woman who had been looking in the window for some time buys the meat against her husband's wishes, he having in boredom been gazing round at the traffic. The difference remained entirely unknown to either of them, outside the range of their mutual cognition, and yet part of the intrinsic feature of their experience.
(3) Some after-images also have the feature of existing first in a non-epistemic state and subsequently their evidence becoming interpretable, that is, allowing the derivation of knowledge from them (Wright, 1983). An after-image which until time t1 had been utterly unnoticed, that is, entirely non-epistemic, can at time t2 be noticed as having been caused by something of significance (e.g. I become aware of an after-image which I realise has been for a few moments just off the centre of my visual field; I have no recollection of how it got there, for I may have glanced in the direction of the dazzling light and had to glance away again so immediately and automatically that I had no opportunity to register what caused it; its characteristic shape makes me now aware that my wife has brought me up a cup of tea on our hexagonal silver tray which she had placed on the windowsill in the sunlight. This is an undeniable counter-example that proves (a) that parts of our visual field can be non-epistemic; (b) that they may subsequently become inspected as evidence; and (c) the experience from which I obtained this information about an external object was entirely internal, for everyone admits the internality of after-images.
(4) It is quite easy to arrange for someone to experience the sensory non-epistemically. If the subject with a probe in his visual sensorium does not know what is exciting the electric impulse at the other end, he cannot attribute any epistemic interpretation to it. First, he may not even know that he has a probe in his brain. Secondly, he may know that he has but he does not know what it is connected to - perhaps, a photoelectric cell in another room, such that, if he did know, he could tell us when a light came on, or perhaps it is connected to a Geiger counter, so that he might be able to warn us of dangerous radiation, but he does not know this, and so the experience is quite non-epistemic. But one need not go to the trouble of probes. Fit a virtual-reality hood on someone's head while they are sleeping and, as they are waking up, unaware of what has happened to them, make the input to the screens an entirely random one produced by some computer such that all that is on the screens is a phantasmagoria of rapidly changing shapes and colours, which undoubtedly would be purely non-epistemic, devoid of knowledge.
But one need not go to the trouble of virtual-reality hoods, for a large proportion of people have 'hypnagogic' imagery as they go to sleep and 'hypnapompic' imagery as they wake up, and the best description of these experiences is to say that they correspond to a computer randomized output on a monitor screen, rather like a screen-saver - indeed, it is possible that they may have something of that very function, so it is a perfectly common and normal experience to sense and not to perceive. Of course, as one wakes one may begin to play a sort of 'faces-in-the-fire' game with them, as I often do, as I have them regularly, but it is not necessary, and often the speed of transformation defeats one's attempts to fix a percept on the changing field. Nor is it necessary that memory is already at work in these transformations. It is a subtle but I think definite point: if I happened to catch a glimpse, say, in a turmoil of purple convexities, of a portion that looked like the face of Samuel Johnson, it would not necessarily mean that an image of Samuel Johnson had emerged from my iconic memory that I might have missed perceiving - no, Samuel Johnson is no more in that passing show than he would be if I caught a likeness of him in a chance distribution of pebbles on a beach. As I wrote this, I caught sight of an expressive face among the leaves of the clematis outside my window - no one is going to say I that the face was there before my gestalt module picked it out. Of course, I can call up an image of Dr. Johnson in my mental imagery if I wish to - I can do it now, since, having studied at Pembroke College, Oxford, I am familiar with his face from the portrait of him there, but that is a different kind of mental imagery from the involuntary hypnagogic sequences.
But one can experience a more commonplace example than this. On waking in an unfamiliar room one may not be able for a while to make any sense of what one sees before one's eyes: one eye may be partially overlapped by a portion of a sheet; your focus may be doubling all within the nearest three feet; and beyond is a wall with unrecognisable shapes on it on it (that in normal conditions one might say are actually made up of a coat on a peg, shadows from a blind, patterns on wallpaper, an abstract painting, and a carved wood stand) all seen sideways.
All these are plain empirical counter-examples to the claim of many philosophers (one can take Gilbert Harman and John McDowell as typical, McDowell, 1985; Harman, 1985, 112-13) that it is impossible to have a non-epistemic experience. In their view, one can have no access at all to any intrinsic features of a presented object, such as a tree: Harman boldly predicts that if you look at a tree 'you will find that the only features there to turn your attention to will be features of the presented tree' (1985, 39). This seems a very foolhardy claim for it would disallow any new discovery about that portion of the continuum: for example, that what person A had taken as part of the tree was actually part of a bird sitting in the branches, while person B had noticed the bird but not thought it worthwhile to mention the fact to A, assuming that he saw the bird too. But the example of the bird has wider, evolutionary implications: what if the new element had never been cognized by anyone before? Harman comfortingly assumes that no adjustments would ever be necessary, but the continuum is full of surprises which organisms must have some means of learning about. But one can ask whether Harman has ever sat stared at a tree and tried to view it with a painter's or even a poet's eye. The latter can become so attentive to the sensuous features of the field that they begin to see quite different gestalts in the distributions before their eyes. It is in fact precisely what Leonardo da Vinci counselled young painters to do, to stare at an old wall covered with patches of damp and moss and lichen until they begin to see the outlines, say, of a battle. It does not seem to be counter-intuitive to believe that in their emotional excitement they might forget that they were looking at a wall. Many people who are under emotional stress to just this with faces in the fire. In Dickens's Great Expectations Pip as he went downstairs to steal the pie heard the wooden steps of the stairs say 'Stop thief!' and 'Get up, Mrs. Joe!' Some philosophers - J. L. Austin is a good example - always imagine that perception is taking place within very limited conditions, devoid of any motivation. Their error is to presuppose that the difficult business of learning has taken place, and everyone is safely and securely agreed to the point of the impossibility of any future disagreement on an unthreatening, unexciting, already defined portion of the continuum. In their view of the world, everything has already been learned about and agreed upon.
This leads to the key point about the trust in others which underpins all agreed objectifications (see the section on Faith), that it is fatally easy to believe that one's own favoured version of the commonly agreed objectification is shared in all details by everyone else. The all-too-frequent consequence is that any challenge to the privately favoured version is at once seen by the authoritarian as a form of rebellion and by the subjected as oppression. What cannot be escaped, even in the most well-confirmed of objectifications, the 'truest', is the element of risk that 'nature' - the flux, being, the Real - can present at any time.
References
[For references see at the end of 'Perception as epistemic']
[1] There is of course, first, no guarantee that external space is three-dimensional, and, second, much speculation among physicists that it is not (Stewart, 1986). See also Yoshitaka Nakajima and Shinsuke Shimojo's experiments with turning the 3-D phenomenal space inside out, 1981)