One of the common mysteries of art, and a barrier to our understanding of Coleridge's claim that natural form lies at the heart of knowledge, is the problem of how it is that art can embody meaning in its sensuous forms—for instance, in visual, tactile or auditory "images." To our common way of thinking, this seems mysterious, for we usually think of thinking as something which is propositional and linguistic, or (though less popular these days) as being in some sense "pure" and above both the senses and language. But while propositional thinking is certainly a dimension of our mental experience, we should not let it blind us to the more fundamental ways in which we perceive and understand the world through the senses. Thus we should not think of syllogistic argument as the paradigm for thought, as the Anglo-American philosophical world has tended to, nor should we think of thought as paradigmatically linguistic in the way the literary theory of the last thirty years has suggested. Rather, we should find that paradigm in those moments when we are looking at the world (looking out of the window at a tree, for instance). Concrete sensuous form, or image, I shall argue, provides a more fundamental paradigm for thought—a paradigm which art brings to the fore, and which is also fundamentally Coleridgean.
Corps de l’article
One of the common mysteries of art, and a barrier to our understanding of Coleridge's claim that natural form lies at the heart of knowledge, is the problem of how it is that art can embody meaning in its sensuous forms—for instance, in visual, tactile or auditory "images." To our common way of thinking, this seems mysterious, for we usually think of thinking as something which is propositional and linguistic, or (though less popular these days) as being in some sense "pure" and above both the senses and language. But while propositional thinking is certainly a dimension of our mental experience, we should not let it blind us to the more fundamental ways in which we perceive and understand the world through the senses. Thus we should not think of syllogistic argument as the paradigm for thought, as the Anglo-American philosophical world has tended to, nor should we think of thought as paradigmatically linguistic in the way the literary theory of the last thirty years has suggested. Rather, we should find that paradigm in those moments when we are looking at the world (looking out of the window at a tree, for instance). Concrete sensuous form, or image, I shall argue, provides a more fundamental paradigm for thought—a paradigm which art brings to the fore, and which is also fundamentally Coleridgean. In this essay, then, I do not wish to look in close detail at Coleridge's own views on the constructed nature of the perceptual image, for that work has been done by many earlier commentators, for instance by Prickett and Engell. Rather, I wish to put Coleridge's views in the context of the work of Louis Arnaud Reid and Susanne Langer, whose writings were influential in the middle third of the last century. I also want to introduce a context in the Artificial Intelligence (AI) debate, particularly in the recent writings of Antonio Damasio (whose influence within the AI community has become enormous during the last five or six years), Ralph Ellis and George Lakoff. These writers have moved back towards what is essentially a Coleridgean view of the place of the place of imaging (or imagination) at the heart of cognition. I thus don't so much wish to look closely at Coleridge himself as to construct a new paradigm for understanding Coleridge. For while much recent criticism has treated the Coleridgean symbols as what John Hodson wittily calls "transcendental tropes," figures which should be understood primarily in linguistic terms, I want to suggest that Coleridge's views are much more intimately tied to the perceptual imagination—and, perhaps more significantly, that Coleridge's actual views (albeit shorn of their untenable idealist foundations) have much to recommend them.
To make this case adequately would require a book, for I would need to defend my use of "psychologistic," and more broadly "mentalist," language of the kind Kant, Frege and Wittgenstein rejected.  I shall refer briefly to these matters below, but more generally would wish to point to my recent essay, "Coleridge, Language and the Imagination," which discusses psychologism and suggests that for Coleridge linguistic concerns are less fundamental than imagination. And I can only suggest that if the kinds of account I give below are at all persuasive, then many of the underpinnings of twentieth-century linguistic anti-psychologism begin to look less stable. But even if the reader is not persuaded to abandon the anti-psychologism of the last century, what I offer may well still be a reasonable model for understanding Coleridge, a thinker whose philosophy is fundamentally psychologistic. A more complete account of the role of imagination would also require considerable discussion of those essentially Coleridgean topics, "feeling" and "form." But they must be discussed elsewhere, since here I want to write a preliminary and short essay about the status of "image." 
2. L.A. Reid: A Heuristic Model
Louis Arnaud Reid's contributions to mainstream epistemology are not now well known, though they were widely respected in the 1920s and 30s. And though he was not a Coleridgean, nor an idealist, he was a student of Muirhead, writer of what is still the best basic introduction to Coleridge's thought. Reid's name will be familiar to those acquainted with the controversy over A.J. Ayer's appointment to the chair of philosophy at London, for Reid was the candidate strongly preferred by the philosophers on the selection committee (Ayer 308). Reid's writings on aesthetics were influential in the 1930s and 40s, in America where they contributed to a growing debate, and in Britain where they helped to revive interest in a branch of philosophy which at the time was scarcely considered respectable. A Study in Aesthetics, Reid's major work, was first published in 1931 and, though not now widely known outside the circles of professional aestheticians, drew favourable reviews from such critics as Collingwood, Trilling and Barzun and has remained in print to this day. The foundations of Reid's thought, however, were laid in his 1923 book, Knowledge and Truth, where he considered the status of the sense datum and established the epistemology for which he was still arguing in his last book, Ways of Understanding and Education, in 1986.
Reid believed that the various and unsatisfactory idealisms and realisms of the early years of the twentieth century derived from attempts to grasp one horn or the other of the dilemma of representative perception. The representative theory of perception is usually traced back to Locke, who argued (though not quite explicitly) that when we see an object we in fact see an image in our mind which is the product of the stimulation of our optic nerves by light. Locke claims that light bounces off the perceived object and sets up a "motion," or what later theorists like Hartley were to call a "vibration," on the retina, which motion or vibration is conveyed to the brain where an image is constructed.  The problem with this account is that it makes the image the immediate object of perception, and thus leaves the physical world in a curiously vulnerable position. Berkeley of course stepped into the breach by claiming that there was in fact no more to the story than the image—that our conceptions of an independently existing material world are pure prejudice. God apparently feeds into our minds (for our notions of having bodies and brains are also mere prejudices) a steady stream of images. And thus was born idealism, in the perceptual sense,  in its original and crudest form.
Berkeley is a primary exemplum of Reid's argument that, given the schism between idea and reality (the "representation" and the represented), the common practice has been to seize on the one and to deny the significance of the other. But Reid argued, in terms which largely anticipate Mary Warnock's useful book, Imagination,  that the representative theory was itself the product of a faulty analysis of perception. The sense datum, he claimed, is not what is known but is an (active) way of knowing the world. Moreover, the sense datum per se is an abstraction rather than a substantial entity, though sensation (conceived as an act) is very real. In Locke's account, above, the "seeing" should not be taken, as Locke appears to take it, as a seeing of an image which appears automatically before a private and internal viewer—what Daniel Dennett calls (and the description is rightly dismissive) the homunculus in the theatre of the mind.  The "seeing" is in fact the act by which the image is constructed in the first place. The "image" is thus not a substantive object (a Polaroid picture free-floating in our visual cortex) but a dynamic entity—an object-directed, mental act. 
This emphasis upon the image as an act of seeing is not to deny the story about the mediating function of light, nor is it the kind of extreme realism (with corollaries within the phenomenological tradition and positivist phenomenalism) which denies the phenomenal/noumenal distinction (Warnock 141-149). It does, however, reject the empiricist account of sense-perception, for empiricism assumes that sense data arise in our senses in a relatively passive and unproblematic way. Reid, on the contrary, asserts that the act of "seeing" (or "seeing" considered as a mode) is essentially mental, an act by which mind directs itself to ("knows," "feels") the physical evidence about the external world available to it. As Reid puts it, "Imaging [. . .] is just thinking of the real world, and images are the real world as appearing in a certain way" (Reid, Knowledge 105). And "in the end there are no images, but only imaging of real things" (113).
We should note in passing that this emphasis on the object-directedness of the act gives imaging the kind of purchase on the "objective" that Frege denied to merely psychological "images." For if we allow that the "image" is not what is known, but is a way of knowing the world, then Frege's worries about the mere subjectivity of the image or idea, and Wittgenstein's worries about its privacy, are somewhat lessened, for it now seems at least to be in the very nature of the image or idea that they are object-directed.  And if we allow that there is in fact no image (or reified idea) as such, then there is no purely subjective "thing" which could exist in the psyche, disconnected from the world or object, and cause the kinds of errors Kant, Frege and Wittgenstein worried about. I only wish to add that if Frege and Wittgenstein exaggerated the worries about the subjectivity of "ideas," Frege at least also exaggerated the objectivity of his putatively non-mental "thoughts," for despite worries about the "parallel postulate," people regarded Euclidean geometry as an exemplum of the kind of objectivity to which thinking should aspire—whereas, after Einstein, we now know that it is merely an approximation which works well on the human scale of experience (Currie 28-33). Frege, in arguing against "psychologism," was simply demanding too much of mind, and we should retort that a degree of subjective fallibilism is an unavoidable fact of life.
To return to the main subject, then, my emphasis upon the active nature of imaging has important consequences for our theories of meaning and mind. And it finds support in the evidence from marginal perceptual phenomena, for optical illusions catch the mind in the act of (mis)construction. The well known examples of the vase/two faces and Wittgenstein's rabbit/duck illustrate this, as do Escher's drawings, for we do not see these pictures (nor the world more generally) immediately and without interpretation.  We can at will see such pictures one way or the other, but we cannot see them as both at once. Similarly, dim light or an over-hasty glance may lead us to "see" (for example) the friend we have been expecting—to "see," that is, until the disconfirmatory cues become overwhelming and the "image" dissolves before our eyes.  (As is well known, Coleridge was acutely interested in these phenomena, particularly as he moved away from the relative passivity of associationalist, and Lockean, views of the image. The shift seems to date from late 1803, and is reflected in Notebook entries 1668, 1751, 2080, 2191, 2555, 2583, 2619, 2720, etc..) 
Evidence such as the kind I have raised in the previous paragraphs is of course phenomenological, and of a kind which until very recently would have been dismissed as "mere folk-psychology." And indeed to talk of imaging as a "mental" act would, until the mid-1990s, have been unacceptable. This is not the place to go into the history of the twentieth century's rejection of talk of mind, though I shall return to the question briefly in my conclusion. Here I can only insist that such talk need not imply the dualism which Ryle so rightly rejected, and point to the flood of books and articles which have followed Searle's The Rediscovery of the Mind. As Ralph Ellis shows, the Artificial Intelligence (AI) community has long since lost its faith in the possibility of any easy or quasi-empiricist move from "sensory" input (for example by video camera) to computational algorithm. And much the most promising developments in AI are modeled on Johnson-Laird's "constructivism," a view which parallels Reid's in a significant way. 
I do wish to point to recent empirical and neurophysiological work which supports Reid's views. Ellis, for instance, points to studies which show that it takes a third of a second for increased electrical activity to move from the cortex (where sensation is first registered) to the parietal and prefrontal areas whose activation is necessary for conscious visual, auditory or tactile "images" (13). The puzzle here is why it takes so long, in relative terms, for a movement of only a few millimetres. It turns out, however, that conscious perception involves a global, whole brain process, a process which includes the initial registration of sensory stimulation (in the cortex) but also an essential motivational element in which the mind decides which parts of the almost infinite stream of incoming sensory stimuli are to be attended to (14). Here, already, perception turns out not to be direct and unmediated: it depends on something as subjective as motivation; and it depends on an act of attention.
But we also know that the active nature of perception goes further than this, for again the brain does not simply register an incoming image. Rather, Aurell, Ellis and other neuropsychologists argue that the brain engages in what they call an "efferent" process ("efferent" means "conducting outwards"; cf. "to ferry") in which it begins to construct possible images, as yet not consciously perceived, which might account for the evidence arriving in the form of sensory stimulation (127). According to Aurell's 1983 study, if the kind of object being looked for does not present itself in the afferent activity of the occipital lobe ("afferent" means "conducting inwards)—i.e., in the pattern of stimulation being received—then the parietal lobe at first exhibits a negative electrical potential 200 milliseconds after the stimulus, apparently corresponding to the negation of the original image being looked for.  Then the efferent system adjusts to the discrepancy and produces models which more closely fit the incoming sensory evidence, until a satisfactory match is found. By about the 300-millisecond point, a Positive potential arises in the upper and posterior lobe, corresponding to the consciousness of the object which is now perceived. And in a similar vein, Libet et al. have recorded "expectancy waves" during the period in which we are searching for a match (Ellis 193).
The interpretation of such "evidence" (indeed, the very descriptions given) is of course controversial. But as an observer of this debate over the last fifteen years, my sense is of a growing willingness to recognise the active nature of imaging, as opposed to the notion of an immediate and passive image. Damasio, for instance, has called attention to Bartlett's similar and widely accepted theory that memory and recall are reconstructions, interpretative acts (100). But what is the significance of all this? I began this essay with the problem of how art can embody meaning within an image, and I have by now given half the answer to that question. For Reid saw the active nature of perception as paradigmatic of the active and constructive nature of mind more generally. Images are not merely passively "received" from the outside world: they are active mental constructs, and hence precisely the kind of thing which can embody meaning.
3. Imaging, and Thinking
Aristotle long ago established what until the twentieth century was the standard view of thinking: that "Without an image thinking is impossible" (Aristotle, On Memory 450a1, 427b29-429a9 and De Interpretatione 16a3-8.). For Reid (as, in very different language, for Damasio), the kind of "Gestalt," or what Langer would call "presentational whole," which is exemplified in sense perception is also the ground of all cognition (Langer 92, 97, 21, 42, 64).  As Reid says, concepts ("dog," "cat," "containment," "addition," "cruelty," etc.) are "derived, ultimately, from sense perception and imagination," something Damasio similarly affirms when he says "dispositional representations constitute our full repository of knowledge" (Reid, Ways 6; Damasio 104). Knowing, says Reid, "must involve something other than propositions. [. . .W]e experience wholes through sensation long before we make explicit judgements about their separate qualities" (Knowledge 218). 
Here we come to the crux of the argument against those who refuse to ground "meaning" in images. How, they ask, can we possibly extract the abstract concept "horse" from the "image" of a particular horse? The answer is firstly that there are no images, and that nothing is passively received in the way the empiricist account would suggest. There is no passive image of the horse in which (per impossible) we might seek to find something general like "horse-in-the-abstract." For the whole point is that our imaging of the horse is an act, and one which is dependent on our whole cognitive structure. If we see a horse as such, rather than a large and undefined quadruped (as the horse would have seemed to the American Indian who had never seen such a creature before), we see "horse" because our concept of "horseness" is present in the very seeing or imaging. It didn't get there passively: we put it there.
Does this mean, asks the linguist, that the linguistic category "horse" is prior to our seeing of the horse as such? The answer is that this is often the case in the experience of individual humans, for much of our early childhood is taken up in learning the names of objects which we are shown. But our concepts cannot have originated in a linguistic structure, for that would lead to an infinite regress (there would have to have been someone already in possession of the word "horse" to explain it to our earliest ancestor). 
A much more persuasive evolutionary story suggests that the ability to divide the world into at least some basic kinds is prelinguistic—and that this ability is the basis of our concepts. Thomas Metzinger emphasises the priority of our visual capabilities in his claim that:
Human beings are visual creatures and evolution has invested much more neurocomputational resources in our conscious visual model of reality than it has in our recent ability to internally simulate quasi-syntactic operations with discrete symbol-tokens. We are experts at grasping visual environments in a fraction of a second, whereas we are clumsy thinkers, who take much longer to reconstruct logical environments mentally.
The suggestion, then, is that concrete, sensuous experience came first in evolutionary terms. Animals are certainly capable of dividing the world into kinds (berries of this sort taste bad and cause illness; creatures of that sort are prey, but creatures of this sort are dangerous and not to be tangled with). This ability to present the world to our selves visually is also often (arguably always) the basis of novel linguistic concepts. Botanical taxonomies, for instance, rest on close observations of similarity of structure in different subspecies. It is only after that moment when the relevant similarity is "seen" that a name is assigned. 
Moreover, such moments of "seeing" are interesting phenomenologically in that the perceived object (the way we image) changes almost irrevocably at such moments. We don't retain the same Polaroid snapshot before our mind; rather, new understanding transforms an appearance in ways which can range from the subtle to the radical. The person who has not seen a horse before does not know what is essential and what is incidental to the species (American Indians are alleged to have thought that the human rider was part of the horse, at first sight). And we all experience similar changes in the way we perceive an object as, for instance, we come to know a stranger (their large nose becomes naturalised, and no longer sticks out like a sore thumb), or when we become familiar with a new city, or when we look at a painting while someone explains its features to us.
The point, then, is that our imaging depends upon our understanding. This understanding is arguably what Kant called "schematic," for when we have met more than one horse, our understanding of "horseness" is no longer based on any single act of imaging, but rather on a constantly changing sense of what is common (the schematic element) to all of our imagings. We should admit straight away that we do not know what a schematum is: it is certainly not a reified entity, and nor do I wish to adopt Kant's faculty psychology, with its rigid separation of the faculties. But "schematism" is a useful theoretical term which points to an aspect of the imaging process.  It is clearly central to our concrete apprehensions of the world, is pre-linguistic, and provides the basis for "meaning" which the "image" could not. When I said earlier that "horseness" is something which we put into the "image" (i.e., the efferent act of imaging), its source is to be found in this underlying schematic element.
A final thought: the reader may well object that it is hard to think of images which ground words like "of," "which," and "the." And perhaps the answer to the objection is that all theories of meaning face difficulties, so that this kind of objection does not amount to a knock-down refutation. But I think there is a more profound answer, for to seek "images" for words like "of" is to fall foul of a major misapprehension. I have not argued that the word "horse" is indexed to an image of a horse, for I do not believe that "images" as such exist. I have instead argued that words are indexed to the schemata which also underlie the imaging process, a very different matter. Schemata are essentially mental, but their "language" is not that of the kind of deeper symbolic "mentalese" advocated by Jerry Fodor and Stephen Pinker (Pinker 47; Fodor passim). Rather, and as I shall try to show in the following section, schematic understandings are concrete. And my guess is that words like "of," "which" and "the" (function words, as opposed to content words), as grammatical markers, are understood concretely. We understand the word "of" as a byproduct of our concrete or experiential grasp in early childhood that "the red of the rose" is different from "the red of the fire-engine." "Of" is a way of linking the quality redness to an object—and that centrality of concrete objects lies behind our grasp of the functioning even of factors as abstract as function words or grammatical operators.
4. Abstract Thinking
It follows from the discussion above that propositional or linguistic knowledge is a form of abstraction. As Reid says, "[a]bstract thinking is abstract thinking, in that it uses language which picks out from a concrete particular experience" (Ways 7). We can see this phenomenologically. When, for instance, I was a small child I was introduced to the basic concepts of arithmetic by being allowed to play with cuisenaire rods, pieces of wood varying in length from one to ten centimetres. I learned to understand, to visualise, that a clump of five single unit blocks could divided into two groups of three and two respectively. And (pace Frege) when any competent mathematician, at least early in her career, understands the equation 5 – 3 = 2, she does so in similarly concrete terms—whether by cuisenaire rods or by visualising the fingers on her hand, or by some analogous concrete experience. 
This can be put more generally. When we think abstractly we apply rules first derived through a concrete, or what Langer calls a "presentational," intuition. If the rule to be followed is not so familiar as to be automatic we shall, in applying the rule, have to think hard and reconstruct the original presentational intuition (and rules become "automatic" only after repeated reconstructions of this type). Mind then is not fundamentally a logic-calculating machine, manipulating propositions. It is fundamentally concrete in its operations. Johnson, for instance, has shown this in his demonstration that our ways of categorising things (i.e., our concepts) originate in bodily experience, which leads to the "part-whole" schema, the "container" schema, the "source," "path," and "goal" schemas, etc. (Ellis 199). 
Ellis also points to good empirical reasons for believing in the concrete basis of abstract thought—and also, incidentally, of language. He points, for instance, to the finding of Johnson-Laird and of Kaufman that spatial strategies are used to facilitate reasoning in unfamiliar contexts, but that as the contexts become familiar we switch to linguistic strategies (96). He also points to neurophysiological studies (e.g., brain scans) that show that the meanings of words are first registered primarily as images (right brain) and then gradually taken over by the left brain as they are naturalised. And he refers to Newton's four reasons for preferring the "image" view of thinking as opposed to computational or algorithmic or strongly linguistic views of what is "really going on" when we think (93).  Newton's reasons are:
even after success with realistic or concrete examples, people fail to transfer the reasoning pattern to more abstract examples (which they ought to find easier rather than more difficult, if thinking were in fact computational);
the language system in the human brain is closely associated with certain motor processing systems. This is the empirical basis of Husserl's claim long ago that the name of a concept brings up a complex internal sensation (121). For neurophysiological evidence suggests that to think of a word is largely to imagine ourselves saying it—a concrete and bodily, sensorimotor experience (122);
apes are probably capable of limited abstract thinking, despite the absence of language;
that humans consciously experience themselves this way.
It is time to draw some interim conclusions. The account which I have given here of images as activities caters to the twentieth-century demand that intermediate terms like "image" or "mind" should not necessarily be treated as substantive (what Ryle famously called the category mistake), but is superior to the accounts provided by Brentano and Merleau-Ponty on the one hand, and Ryle on the other, in preserving the sensuous qualia—the subjective, or qualitative side of perception (again understood as object-directed acts and not as reified entities) which is a corollary of Reid's insistence that perception is a mental act. It also allows a more satisfactory account to be given of the relation between direct perception and its echoes in imagination and memory (things Ryle was forced to treat, roughly, as dispositions, but which—on Reid's account—turn out to be acts very similar in kind to that involved in seeing an object in the world). 
At a broad level, Reid does not simply defend the view that "meaning" is grounded in imaging. Such a view might attempt to locate meaning in some kind of pure, non-qualitative, schemata, a language perhaps of reified shapes. But Reid insists on a richer phenomenology, beginning with a very Coleridgean emphasis on the role of feeling (which can again be defended on neurological and evolutionary grounds). He then insists that imaging is a species of feeling, not a passive reception, but a qualitative construction. The richness of colour (and of sound, taste, etc), for instance, reminds us that colour as experienced is a secondary quality, the way in which mind (to use both Coleridge's and Langer's term) presents to itself the relevant features of the external world (Langer 92; Coleridge Statesman's 113). Imaging is the basis of meaning, but it is thoroughly qualitative in its mode—for on Reid's argument, mind operates fundamentally in ways which are concrete and sensuous.
This emphasis on the concreteness of the mind underlies the aesthetic in Reid's best known book, A Study in Aesthetics—first published in 1931—for there Reid was interested in the question of how "meaning" comes to be embodied in aesthetic form, how for instance Schuman's sadness in thinking of Clara comes to be embodied in the tone of the instrument. Much of the energy in that book is directed at establishing the simple fact that the sensuous qualities of art do embody meaning, for he was responding in part to the tendency to idealise art — to think, as Keats playfully suggests of melodies, that "those unheard are sweeter," or to think with Shelley that art realised (as opposed to conceptualised) is like a fading coal (Keats "Ode" stanza 2; Shelley 504). But if the qualities of sensuousness ("qualia") do not reflect an immediate likeness to the qualities of objects in the real world, but are thoroughly cognitive acts, then qualia are already cognitive or interpretative — that is, bearers of meaning. Thus there is no mystery to the way in which the concrete or phenomenal qualities of art can embody meaning, for meaning is "pure" only in an abstract sense: it always originates in concrete intuitions.
I began this essay by asking how sensuous form, or image, could embody "meaning." My answer has fallen into two parts, the first a claim that imaging and all of its associated qualia (qualities of colour, tone, tactile sense, etc.) are paradigmatically mental, and therefore at least in the same realm as that of meaning. The second part of my answer is to go a little further and to suggest that abstract and linguistic knowledge is based on those qualitative images, and therefore that sophisticated and complex intentions and meanings can be embodied in art. Indeed, L.A. Reid suggested that art is itself a form of knowledge.
I wish, however, to neglect that suggestion here and to finish instead with three points:
Firstly, in this essay I have adopted a mentalist vocabulary, using terms like "mind," "meaning" and "intention." Many readers will be uncomfortable with such a vocabulary, for it goes against the predominant twentieth-century approach established by Frege—and transmitted by him into continental philosophy through Husserl, and through Russell and Wittgenstein into Anglo-American analytic philosophy. The rise of behaviourism in the middle part of the century, and Ryle's famous attack on "mind," made talk of mentalism unacceptable during most of the period, while in the humanities post-structuralist argument was similarly hostile. The popularisation of computationalist models of cognition in the late 70s replaced earlier talk of clocks, levers and vibrations, convincing many in the AI community that they had at last found the mechanism behind intelligence, and that mentalist talk could be dismissed as mere "folk psychology." But as the 90s arrived without any sign of anything approaching computer intelligence, the tide began to turn. Gordon and Goldman began to ask whether we "know" by simulating concretely or by theorising (quasi-)propositionally, while other philosophers explored more deeply the implications of knowing "how" as opposed to knowing "that." Searle and Penrose (the latter reviving Lucas's Gödelian arguments)  launched influential attacks on the very notion of AI, while in neurophysiology a number of new imaging techniques like CAT and PET scans opened new windows on the brain, while curiously appearing to support a number of the so-called folk-psychology myths; and consciousness became a newly respectable area of study.
My point here is that we are in the midst of a sea-change in our understanding of mind. Admittedly there is no sign of a new consensus, but neither can we safely rely on the older consensus, and literary theory can no longer ignore the question. Let me add that in speaking of mind we do not need to revert to any kind of dualism, nor to idealism. When I speak of mind, I wish to commit myself to no more than the view that in understanding the world, mind is doing something interesting (something transcending passive behaviourist conceptions, and not accounted for in the computational model)—a view which, in the field of epistemology, Susan Haack's "foundherentism" seems to acknowledge, albeit nervously. If we ever do understand mind, I believe that it will be in terms of a model which goes beyond the scope of reductive nineteenth-century physicalism, though it will probably emerge as a physicalist account based on a broader conception of matter.
The ideas which I have explored in this essay have most recently been explored by George Lakoff and Mark Johnson in their book, Philosophy in the Flesh: The Embodied Mind and Its Challenge to Western Thought. They repeat the argument that evolution is unlikely to have produced first a complete sensory motor system, and then, in humans, a completely separate "cognitive" system (43); and nicely convict the holders of such views of an unwitting commitment to faculty psychology (37). Calling instead, and polemically, for an "empirically responsible philosophy" (13), they point to psychological research which shows that most of our thought is "unconscious"—not in the repressed, Freudian sense but because, for example, most of the neural processing underlying not only vision but also cognition is not available to introspection (10). By failing to acknowledge this evidence, they claim, twentieth-century philosophy has manufactured the fictions of the Fregean person, dealing in disembodied, objective "thoughts" (6), and the post-structuralist text consisting of a disembodied language. Instead, they point to the way the primary metaphors for reason (e.g., grasping an argument; an argument "going over our heads," etc.) derive from sensori-motor experience—an argument which draws on empirical work by Christopher Johnson, Grady, Narayanan, Turner and Fauconnier (45).  They also claim that the most promising computational models of language acquisition use basic perceptual and motor computation as their substrate, and point to the work of Reiger, Bailey and Narayanan (569).
Thirdly, the model I have outlined here places the imagination, rather than language, squarely back at the heart of aesthetics. I am not for a minute suggesting that we can dispense with the understanding which literary theory has offered us in the last three decades, for language as a structure and a medium undoubtedly has its place in our explanation of literary works. Nonetheless, imagination in the Coleridgean sense needs to be restored to its central place within aesthetics, without turning "imagination" into a merely linguistic metaphor.
Fourthly, if imaging is the ground of meaning, we can understand not only Coleridge's view, consistent throughout his career, that the world of natural form is "that eternal language which thy God/Utters" ("Frost at Midnight" 60), but also his more fundamental insistence that the self-knowledge through which God instantiates Himself is a knowledge of the Father in the image or "adequate idea" which is the Son (Coleridge Opus 203). Just as "images" are the constructions by which we know the world, so the Son is the constructed image through which the Father knows Himself.
But I must leave the Coleridgean Trinity, and the role of form within it, for the essay on "Form in Coleridge" listed below among the works cited.
- Kant, for instance, tells us that “No image could ever be adequate to the concept of a triangle in general. It would never attain that universality of the concept which renders it” (B 180). See also Wittgenstein pars. 6, 329 and Frege 87-96, 224-226.
- See my “Form in Coleridge, and in Perception and Art more generally,” which was originally intended to follow up this article.
- Locke book 2, ch. 8, sec. 11-12. It has recently been denied that Locke had a representative view of perception, and it must be admitted that he is not entirely explicit on the subject. However, Berkeley and most readers since have read him that way. Locke uses the term “represent” in book 2, ch. 31, sec. 1, 6, 12-14 and elsewhere; and his consideration of association, adequate and inadequate ideas, true and false ideas, simple and complex ideas, etc., only seems to make sense in terms of a representative theory. Locke recognises the difficulties of the representative theory of perception (albeit obscurely) in book 4, ch. 4, sec. 3 and book 4, ch. 11 (passim). See also Hilary Putnam's comments (58).
- In a broader sense Plato's philosophy is sometimes described as idealist in thinking that the archetypes of all objects in the world are to be found in a quasi-mental world of the forms.
- Warnock 156, 158,172, 182. Though Warnock is not a reliable guide to the details of Coleridge's philosophy (she has not read his philosophical manuscripts), her general treatment of the question of imagination is helpful.
- Daniel Dennett of course (and rightly) argues against the existence of a homunculus in the theatre of the mind in his Consciousness Explained. Damasio does argue that “the entire prefrontal region seems dedicated to categorising contingencies in the perspective of personal relevance” (182), but he sees this region as part of a whole and thus not commensurate with the homunculus.
- The sensuous qualities of the act (for the act is qualitative) constitute the “qualia” which sceptics like John Searle and Roger Penrose claim are the vital key to awareness which is not captured in computer simulations of cognitive processes. I shall argue elsewhere that these sensuous qualities (considered in themselves) are types of feeling, a fact recognised by anyone who has seen what is meant by “secondary qualities.” This follows because they are not in any direct sense qualities of the external object in the way in which primary qualities may be thought to be. They thus owe their qualitative dimension to mind, and fall under the more general considerations which lead me to postulate an ultimate unity between feeling and thinking. Given Reid's usage of the term “feeling,” there seems no reason to treat sensation as ultimately a distinct category.
- My position mediates between internalism (the view that “meaning” is grounded in internal “mental” symbols) and externalism, the view that “what a person's thought "means" depends on how things are in the world outside the thinker's head.” See Greenberg and Fodor for Fodor's conversion to an “externalism” which is no more satisfactory than his earlier internalism had been.
- Philosophical Investigations, 194. Wittgenstein borrowed the duck-rabbit from Jastrow.
- Coleridge was particularly interested in facial recognition. See Notebooks 1751, 2080, 2583, 2619. Curiously, modern research not only confirms the “constructivist” view, but suggests the sexes image faces differently, with boys forming a “holistic” impression, while girls apparently pay more attention to specific features which convey mood and feeling. See Everhart et al.
- My “Form in Coleridge” discusses these entries (pars. 24-26).
- Philip N. Johnson-Laird has been arguing for this view for over thirty years. A useful overview of his thought is given in his Human and Machine Thinking.
- Aurell also adduces evidence from patients with brain lesions for his thesis that two distinct processes are involved in imaging.
- Langer’s description of sense percepts as “symbols” (21), or presentational forms, consciously builds on Reid’s 1923 book, Knowledge and Truth.
- Reid actually speaks of “truth” rather than “knowing” here, but later abandoned the term “truth” because of its possible transcendent implications.
- Curiously, Coleridge advances just such an argument: that there is indeed a regress, though it is circumvented by a God who originates language (Opus Maximum 14). Perkins points to James Burnett, Lord Monbodo, as a possible source here (17).
- Linguists will reply that the notion of “relevance” here is linguistically relative, for one needs a theory about what is relevant and what is not. But, I claim, that theory was in turn based on observation—a moment when a similarity of kind was “seen.” On a different tack, the emergence of DNA analysis has complicated, but not essentially changed, the story.
- This is no more illicit than speaking of “concepts.” Damasio uses the more fashionable term, “dispositional representations” (which he tells us are not topographically organised) (103). The schematic element will also correlate with the role played by Johnson-Laird's “mental models.” See Ellis 71 and Kant 181.
- This account of arithmetical knowledge is, of course, highly contentious, for mainstream twentieth-century philosophers have argued that it confuses a psychological process (how we acquire knowledge) with an ontological question (why the thing known is true). But Quine’s naturalised epistemology (a highly qualified empiricism) shows us the way around this apparent impasse. See Galloway for a sympathetic exposition of Kitcher’s version of this.
- When Terry Eagleton criticises the idea that meaning is “an affair of consciousness, rather than of words“ and invites the reader to “experiment ... by looking up from the book for a moment and 'meaning' something silently in his or her head“ one would wish to draw attention to these kinds of evidence (67).
- Ellis also replies to Wittgenstein's objections to the idea that concepts must always be related to images. Wittgenstein says that when we develop a concept of the “null case” we cannot possibly be using images since it is impossible to imagine something that isn't there. But Ellis replies that we can be ready to imagine without actually doing so (108-110).
- Ryle is wrong to think that there is a profound difference in kind between hearing a sound and imagining one has heard a sound. The former involves an act of imagination in interpreting the physical stimuli; the latter involves much the same act of imagination, without the stimuli. Phenomenologically, there is at times little difference between the two cases. It is quite possible to be unsure whether the piece of music which has just attracted your attention is being played quietly next door, or whether it has been playing (relatively unnoticed) in the back of one's head. Ellis produces evidence that both the imagined and the dreamed are products of the brain's efferent system, when disconnected from the afferent system (57). On a different tack, Ryle's mistake here owes much to Hume, who wrongly thought that imagined and remembered images are paler copies of an original and passively received sense impression.
- Lucas suggested that Gödel’s incompleteness theorem can be extended to show that computers cannot transcend their immediate frames of reference, in ways which humans can. This is highly controversial, but like Penrose I view the purported refutations of Lucas as question begging.
- Lakoff and Johnson argue that Language is primarily metaphoric rather than literal. By this they don’t mean to claim that language is merely metaphoric, for though they qualify unbridled realism (341), they do argue for what is essentially scientific or empirical realism (88, 244). And they argue that the basic sensorimotor concepts are in one sense literal (58). By claiming that language is “metaphorical,” they thus mean that it is grounded in the concrete or experiential modes of mind.
- Aristotle. The Complete Works of Aristotle. Ed. Jonathan Barnes. Princeton: Princeton UP, 1984.
- Aurell, Carl G. "Perception: A Model Comprising Two Modes of Consciousness." Perceptual and Motor Skills 56: 211-220.
- Ayer, A.J. Part of My Life. London: Collins, 1977.
- Barzun, Jacques. Berlioz and the Romantic Century. New York: Columbia UP, 1969.
- Coleridge, Samuel Taylor. The Notebooks of Samuel Taylor Coleridge. Ed. Kathleen Coburn. London: Routledge and Kegan Paul, 1957.
- ———. Opus Maximum. Ed. Thomas McFarland and Nicholas Halmi. Princeton: Princeton UP, 2002.
- ———. The Statesman's Manual, reprinted in Lay Sermons. Ed. R.J. White. London: Routledge and Kegan Paul, 1972.
- Collingwood, R.G. "Review." Philosophy 8 (1932): 335-337.
- Damasio, Antonio R. Descartes' Error: Emotion, Reason and the Human Brain. New York: Putnam, 1994.
- Currie, Gregory. Frege: An Introduction to his Philosophy. Totowa NJ: Barnes and Noble, 1982.
- Dennett, Daniel. Consciousness Explained. Boston: Little Brown, 1991.
- Engell, James. The Creative Imagination: Enlightenment to Romanticism. Cambridge Mass.: Harvard UP, 1981.
- Ellis, Ralph D. Questioning Consciousness: The Interplay of Imagery, Cognition, and Emotion in the Human Brain. Amsterdam: John Benjamin, 1995.
- Everhart, D. Erik, Janet L. Shucard, Teresa Quatrin, and David W. Shucard. "Sex-Related Differences in Event-Related Potentials, Face Recognition, and Facial Affect Processing in Prepubertal Children." Neuropsychology 15.3: 329-341.
- Fodor, Jerry A. The Elm and the Expert: Mentalese and its Semantics. Cambridge Mass.: Bradford Books/MIT Press, 1995.
- Frege, Gottlob. The Frege Reader. Ed. Michael Beaney. Oxford: Blackwell, 1997.
- Galloway, David. "Wynn on Mathematical Empiricism." Mind and Language 7.4 (1992): 333-358.
- Goldman, Alvin. "In Defense of Simulation Theory." Mind and Language 7.1 (1992): 104-119.
- Gordon, Robert M. "The Simulation Theory: Objections and Misconceptions." Mind and Language 7.1 (1992): 11-34.
- Greenberg, Mark. "What Connects Thought and Action?" Times Literary Supplement (June 23 1995): 8.
- Haack, Susan. Evidence and Inquiry. Oxford: Blackwell, 1995.
- Hodson, John A. "Transcendental Tropes: Coleridge's Rhetoric of Allegory and Symbol." Allegory, Myth and Symbol. Ed. Morton W. Bloomfield. Cambridge Mass.: Harvard UP, 1981. 273-292.
- Johnson-Laird, Philip N. Human and Machine Thinking. Hillsdale, NJ: Erlbaum, 1993.
- Kant, Immanuel. Critique of Pure Reason. Trans. Norman Kemp Smith. London: Macmillan, 1980.
- Keats, John. "Ode on a Grecian Urn," The Norton Anthology of Poetry. Ed. A.W. Allison et al. 3rd ed. New York: Norton, 1983.
- Lakoff, George and Mark Johnson. Philosophy in the Flesh: The Embodied Mind and its Challenge to Western Thought. New York: Basic Books, 1999.
- Langer, Susanne K. Philosophy in a New Key. London: OUP, 1951.
- Locke, John. An Essay Concerning Human Understanding. Ed. John W. Yolton. Abr. ed. London: Dent, 1977.
- Lucas, J.R. "Minds, Machines and Gödel." Philosophy 36 (1961): 120-124.
- Metzinger, Thomas. "Teaching Philosophy with Argumentation Maps: Review Of Can Computers Think? The Debate, Robert E. Horn." Psyche 5.30 (October 1999) <http://psyche.cs.monash.edu.au/v5/psyche-5-30-metzinger.html>.
- Penrose, Roger. Shadows of the Mind. London: Vintage, 1995; first published 1994.
- Perkins, Mary Anne. Coleridge's Philosophy. Oxford: Clarendon, 1994.
- Pinker, Stephen. The Language Instinct. London: Penguin, 1994.
- Prickett, Stephen. Coleridge and Wordsworth: The Poetry of Growth. Cambridge: Cambridge UP, 1970.
- Putnam, Hilary. Reason, Truth and History. Cambridge: Cambridge UP, 1981.
- Reid, L.A. A Study in Aesthetics. London: Allen and Unwin, 1931.
- ———. Knowledge and Truth. London: Macmillan, 1923.
- ———.Ways of Understanding and Education. London: Heinemann, 1986.
- Reid, Nicholas. "Form in Coleridge; and in Perception and Art more generally", Romanticism On the Net, No. 26 (May 2002) http:www.erudit.org/revue/ron/2002/v/n26/005699ar.html.
- Ryle, Gilbert. The Concept of Mind. New York: Barnes and Noble, 1949.
- Searle, John R. The Rediscovery of the Mind. Cambridge Mass.: Harvard UP, c1992.
- Shelley, P.B. "A Defence of Poetry." Shelley's Poetry and Prose. Ed. D.H. Reiman and S.B. Powers. New York: Norton, 1977.
- Terry Eagleton, Literary Theory: An Introduction. Oxford: Blackwell, 1983/1985.
- Trilling, Lionel. "Art and the Philosopher." Griffin 3.8 (August 1954): 5-13.
- Warnock, Mary. Imagination. London: Faber, 1976.
- Wittgenstein, Ludwig. Philosophical Investigations. Trans. G.E.M. Anscombe. Oxford: Blackwell, 1997.