We must not suppose that we first have a disembodied feeling, and then set out to find an embodiment adequate to it. In a word, imaginative expression creates the feeling in creating its embodiment, and the feeling so created cannot be otherwise expressed, but cannot otherwise exist, that in and through the embodiment which imagination has found for it. (Bosanquet 1915 p. 34)

I think that the recent theory of artistic expression has displayed a somewhat misleading tendency of concentrating on facts or statements of the form ‘the work expresses E’, where ‘E’ is an abstract singular term indicating an emotion. The problem is not that the thing expressed may be something altogether different from a mental state or process, as when it is said that the famous picture by Delacroix expresses the collective triumph of political liberty. Nor is the problem that in many cases, indeed many of the most captivating cases, the content expressed is too rarefied or idiosyncratic to admit of verbal capture, or that in many cases E must be particular rather than general, or that successful reference to the emotional content of a work requires an indexical or demonstrative referring to the work. These things are often true, and no one, except perhaps a follower of Nelson Goodman (1968), would deny them; versions of these points recur often in writing on artistic expression since Benedetto Croce’s Aesthetic of 1902 and presumably earlier.

The problem is that this tendency—even where the expressive content of a work of art is restricted to affective mental states and processes—has led to an unwarranted constriction of the topic of expression. Although ordinary language does not neatly distinguish them, a rough and ready sorting of emotions, moods, and feelings is familiar. Setting mood aside for the moment, we sometimes say such things as that an jazz saxophonist expresses his or her feelings, that a certain gesture drawing was done with great feeling. And it would be decidedly revisionary to deny that such remarks represent central cases of expression. I believe that the idea is sharply distinct from the idea of the expression of emotion, and that it is at the centre of John Dewey’s Art as Experience— and also to Bernard Bosanquet’s Three Lectures on Aesthetic, Croce’s Aesthetic, Susanne Langer’s Feeling and Form, and Guy Sircello’s Mind and Art.

I will discuss only the arts of painting and drawing.Footnote 1 Section 1 considers and explicates the three-way distinction between feeling, emotion and mood. Section 2 considers the expression in painting and drawing of feelings; I allude not only to older philosophers such as Dewey but also to representatives of recent psychology including Vittorio Gallese and J. J. Gibson, and tellingly to critics of the past such as Bernard Berenson, Water Pater and Meyer Shapiro to drive home the point. Section 3 considers matters of ontology, semantics and logical form, as intimated in the epigraph.

1 A Taxonomy of Emotion, Feeling, and Mood

Neither commonsense lore or ordinary language sharply distinguishes feeling from emotion and mood. But the distinction is sharper from a theoretical or analytical point of view. What follows is a more or less orthodox classification, more or less current among the many philosophers and psychologists who have perceived shortcomings in the James-Lange account according to which emotions are at bottom mere feelings; emotions must have, in some sense, an object, as in the object of fear. Such a classification inevitably will not be consistent with every such theory of the relevant phenomena, but I wish to remain as neutral as possible with respect to the ever-burgeoning psychological and philosophical literature on these topics. I will not take a stand, for example, on the viability of perceptual accounts as opposed to judgemental accounts (as in Solomon 1988, Nussbaum 2004, Prinz 2006, Brady 2013); on the viability of evolutionary accounts (in particular on the pan-cultural ‘affect program’ of Ledoux 1998 and Ekman 2003); on whether or not emotions are natural kinds (Barrett 2006), on whether or not they are socially constructed (Averill 1980); or whether the intuitive classification of emotions should be retained in a mature theory (doubted by Griffiths 1997). For my purposes, all that is needed is that the scheme separating feeling from emotion and mood be rigid enough to make it plausible that there is some viable theoretical classification along the proposed lines.

Two more points of ground-clearing.

First, it is sometimes important to distinguish mental dispositions from mental states, events and processes. With respect to emotion, this distinction is sometimes obscured by ordinary language—‘His jealousy’, for example, can be used for either the state or the disposition of that name. Paradigmatic mental states can occur concurrently with others, and the phenomenological ones are episodes with a certain duration about which one typically can ask what it was like to undergo it (some mental states are near-instantaneous but still phenomenologically manifest). A mental disposition, in the sense intended, is more than the mere logical fact that such states come and go; it is the factual support or undergirding for the recurring, corresponding mental state: like the state, it is psychologically real, if not phenomenologically so.

Second, there is no thought of strict definition in what follows. The classification does not strictly partition the realm of affective phenomena (some affective phenomena can take any of the three forms: there is for example the emotion of anger, the feeling of anger, and the angry mood).


  • An emotion is an affective mental state (or substantive disposition to have the corresponding affective mental state) with a singular thought (involving an intentional object) at its centre (or even simply an intentional object at its centre, as when one is said to fear cats); it is correlated with a specific range of behavioural dispositions and feelings, and with general types of thoughts related in some way—by inference, and by association or imagination—to the thought at its centre. The thought need not be the cause, or the psychological cause, of the emotion, but typically it is. Standardly as mental states rather than dispositions, they are relatively short-lived, as in an episode of anger.

  • A mood is an affective mental state or disposition without a thought at it centre (thoughts can be a cause of them, however; and the distinction between states and dispositions in this case is comparatively fluid). It encompasses a cluster of behavioural dispositions and specific feelings, and typically general types of cognitive states, but there need not be thought or desire at its origin. For example sadness and happiness can be moods as well as emotions. Whether as states or dispositions, they are non-episodic; they more long-lived than either emotions or (the first kind of) feelings. We speak of being grumpy all day, for example.

  • A feeling is one of two basic varieties (there might be an element of linguistic happenstance in there being a single term for both).

  • The first kind of feelings are various somatic phenomena: particular mental events or processes typically involving some region of the body, according to some accounts non-intentional or non-representational in themselves (but others assume or argue that at least some of them intrinsically represent states or events of the body or the proximate environment), and paradigmatically if not always conscious. In particular, each is such that is it is potentially conscious; it has a distinctive phenomenology, in that what is like to undergo it when it is conscious is sufficient to individuate it. All feelings of this kind are sensations, but not all sensations are feelings. They divide into four types (these may overlap, perhaps necessarily in some cases, and the list is not exhaustive):

  • Exteroceptual feelings: sensations via the sense of touch of impact, resistance, smoothness and hardness, wetness and dryness, heat and cold, and so on, as of the environment or objects in the environment. I call the analogues for the visual and auditory systems and the other systems sensations but not feelings.

  • Interoceptual feelings: internal sensations including those of the organs, for example feelings of fatigue, energy, vertigo, of having a pit in one’s stomach, of shivers down the spine, of goose bumps or frisson or hair-standing-on-end, of elimination, of inhalation and exhalation of breath (including the rapid inhalation of breath), of eyes watering, ‘choking up’ or crying, of sneezing and coughing, of brisling of the skin, of hunger, of sexual excitement and orgasm, of blushing, of laughter, nausea, and pain (the startling response described in Robinson’s 1995 is an interoceptual feeling, in this terminology).

  • Proprioceptual feelings: sensations of the movement, resistance, acceleration and deceleration, and layout, of the head, limbs and body; the sensations are kinaesthetic and internal.

  • Imaginative feelings, including empathetic feelings: the feelings one has as-if one were to undergo a proprioception, exteroception, or interoception. For example, the flinch one feels when one sees a boxer take a heavy punch. They can be partially characterised by those myriad tiny tactile sensations, those imagined kinaesthetic flexes and pulls, that one experiences when perceiving the muscular and vocal antics of others, or more subtly, the non-human scene before one. One imagines feeling the smoothness of a marble fireplace, of running on a windswept beach. Imaginative feelings are often subliminal, or in the penumbra of consciousness.

  • The second kind of feelings are dispositional and have an intentional object. For example, we speak of ‘a feeling for jazz’, ‘a feeling for landscape’, or ‘feelings for her’.Footnote 2 They are dispositions to undergo either (a) certain emotions with respect to the intentional object, or (b) certain heightened feelings of the first kind which involve the intentional object—for example, when thinking of the intentional object, one sometimes has certain feelings (of the other kind). It is commonly thought that the distinction between a dispositional emotion involving X and a feeling-for X is that feelings are more settled or permanent, but otherwise the distinction is somewhat obscure; happily the issue will not obtrude in what follows and in any case it is feeling of the other kind which will bear the weight, either in type (b) of the present kind or of the other kind.

Two further remarks.

First, about belief and desire. All feelings, emotions, and moods come in degrees. Cognitive mental states—belief, questioning, etc.—are not feelings or sensations; similarly neither are conative states (of desire, of aversion, of effort, and of willing). These typically are parts or aspects of the stream of consciousness even if some of them have no phenomenology in themselves. But certain conative states—desire included—are reliably found in conjunction with interoceptual feelings—intense thirst for example is correlated with sensations in the larynx and mouth.

Second, the description of feelings is meant to cover, for example, feelings of sadness. That feeling is intimately connected with or part of the emotion and the mood of sadness, but sadness cannot exist merely as a feeling—it must exist also either as an emotion or a mood. However it can be characterised further, if not defined, in terms of more particular feelings—of heaviness, slowness, increased susceptibility to cry, and so on. Similar things go for the feeling-sides of anger, of joy, of fear, and so on (not that there are no problem cases, as in the response of blushing). These tend to be the aspects of emotional life that were seized upon by James (1884) as being constitutive of it, and will tend to be of special interest where the expressiveness of works of art is concerned.

2 The Expression of Feelings

With the explicated distinction just described between feeling and emotion in hand, I now want to sketch the beginnings of a psychological model for the expression of feeling.Footnote 3 It is not the only conceivable model, and it is perhaps a partial model, for it should not be ruled out that different forms of expression require different models. But it is a model with some independent plausibility.

In addition, I want to stress a rather powerful historical consideration. For that works of art express feelings as well as emotions was the way of speaking which had equal pride of place in aesthetics until perhaps the 1960’s or 1970’s. Though they did not use the terms quite as I do, and their usage of the terms was far from rigorous, Clive Bell, Roger Fry, Croce, Bosanquet, Langer and Dewey in various ways wrote, and not always interchangeably, of feeling being expressed as well as emotion.Footnote 4 Croce maintained that feeling is a necessary ingredient of mental activity in all its forms, a life-force fuelling its many shapes; art is the expression of intuition, the objectification of mental activity (Croce 1909 pp. 1–23, 74–81). Bosanquet deliberately places feeling and its expression at the centre of his concerns in the Three Lectures on Aesthetic; he thought of art essentially as that mode of expression in which ‘feeling becomes incarnate’ (1915 p. 7). Langer’s carefully considered and memorable title was Feeling and Form (1953; Chs. 3–5, 20). I’ll comment briefly on Bell and Fry below. Dewey—whose views I shall dwell upon at more length as many of his specific doctrines accord surprisingly closely with the psychological model I shall advance—placed feeling and its expression at the centre of his more general view of human life and culture in his Art and Experience (1936). The assumption that feelings are expressed is also more or less explicit in various degrees in many of the titans of criticism of the period—for example in Pater (1980 [1893]), Henrich Wölfflin (1950 [1932]), Edwin Panofsky (1955), Meyer Schapiro (1988 [1962]) and notably Berenson (1948), whose views I shall single out as singularly prescient.Footnote 5 Some of this might be passed off as involving feeling-for of the second type, in the sense of the last section, but only some.

2a Gallese and Gibson

By the expression of feelings I certainly do not mean the communication of any of the sundry feelings mentioned in the last section. The artistic expression of feelings is confined to those that can be communicated by the given medium (although as we’ll see presently the range of such feelings is wider than one might initially suppose).

In the simplest sort of case, think of a line drawn on paper. We know intuitively what would be meant by saying that the line lacks feeling—it is mechanical, haphazard, or unfocused. The opposite might said of a line drawn by Picasso or Botticelli. (Similarly, we say that a played violin line lacks feeling—it is timid, out-of-tune or rushed; or that it has great feeling, say one played by Menuhin or Oistrakh). The visual or auditory sensation enters into the lines’ having feeling, and the felt muscular action of producing the line—in delicate and rapid feedback with the visual (or auditory) stimulation produced—also enters into the line as produced by Picasso or Botticelli (or by Menuhin or Oistrakh). And here I simply assume an hypothesis advanced by many recent psychologists led by Gallese, and not a few philosophers including Croce, Collingwood and Dewey in their vaguer ways but also more meticulously by more recent figures, that the attentive perception of such gestures or actions—or of their traces, as in an inscribed line—feeds directly if subtly into a cotemporaneous imaginative simulation of the action. Croce, Collingwood and Dewey sensed it, but Gallese and others have mounted a formidable, conceptually rigorous case, grounded in judiciously considered empirical evidence, that this ‘off-line’ mimicry, the motor imagining of such actions (the neural basis of which for Gallese crucially involves the ‘mirror neurons’), may typically occur at the margins of consciousness, but is normally present nonetheless whenever such actions or their traces are attended to (see Gallese 2005; Freedberg and Gallese 2007; Ferretti 2017). Speaking specifically of art, Freedberg and Gallese write ‘[t]he artist’s gestures in producing the art work induce the empathetic engagement of the observer, by activating simulation of the motor program that corresponds to the gesture implied by the trace.’ (2007 p. 202).

The ‘motor imagining’ of drawing a line might itself seem a dubious and otherwise flimsy basis for expression of any sort. But two further points can now be made. First, the imagined interoceptual feeling of muscular effort is typically not simple. It has many dimensions or properties—or it is inseparable from other feelings, or closely connected to them—such as the shape, smoothness, curvaceousness, or jaggedness of the line, its degree of thickness, hardness, ease, sinuosity, strength, weight, tone, limpidity, slope, vigour, liveliness, force, lightness or darkness, tautness or flaccidity, where these are imaginatively felt or sensed qualities. The artist creates with such naturally-equipped spectators implicitly in mind. When a simple line is felt, the gesture, the act of inscribing the line, is typically felt in a certain way instantaneously, without immediate conscious distinction between a visual or auditory sensation and a separate psychic phenomenon. Sometimes the latter can be recovered only by reflection, and indeed could rarely if ever be of independent aesthetic interest (indeed it is a mark of expressive feeling that the gesture is experienced by the artist as fused with its product, something often achieved only by practice; a great deal more on this in closing). Yet it remains complex, multi-faceted and often extremely delicate. Similarly for the spectator, even if the awareness of those qualities proceeds by different and more ambiguous channels and is not nearly so sharp as it is for the artist. Nor need the uptake on the part of the spectator be at all technically advanced or exact; one’s perceiving the expressiveness of Picasso’s drawing need not involve any detailed knowledge of Picasso’s technique, even if it may be so enhanced (for the point with respect to music, see Alperson 2008 pp. 46–50).Footnote 6

Second, motor imagining spreads further to many other creative procedures, which in turn make other varieties of expression possible. To achieve this in the drawing case—via combinations of lines, patterns of shading and hatching, bringing with them various further felt qualities—the artist presents those things as they are experienced, as they are felt; the attentive spectator simulates—however clumsily, vaguely or inexactly—the actions insofar as he or she feels the expressive item.

Crucially, a quality of this kind may have a semantic, functional or representational dimension. How exactly these dimensions or qualities enter into the experience of creating and experiencing the work of art, and what exactly in various cases constitute such qualities, are good questions but I will not enter into them here because of their sheer size (see Tilgman 1970 for his general treatment; Wollheim 1973 [1965] pp. 21–30, 1987 pp. 19–25, 2005 pp. 1-10 in particular for painting and drawing). Except for a brief remark below, I will simply assume at least of some of them to be amongst the qualities to which the model applies. Thus amongst the immediately felt qualities of a drawn line is its representing the outline of a shoulder, the shape of a vase, the edge of a table (one ‘sees it as’ a shoulder, a vase, a table); or, in the case of a gesture drawing, the central movement of a torso (one ‘sees it as’ the central movement or centre of gravity of the torso).

I now insert two parenthetic remarks which concern issues which the model might seem to render pressing. Both complicate the picture.

First, a way in which feeling-expression might depart from the paradigm is that—of course—an artist need not overtly feel everything that he or she does in order for it to be expressive; and similarly the spectator, despite his or her best efforts, needn’t feel the appropriate feeling whenever faced with an expressive work that he or she recognises as expressive. These points are unquestionable but they may simply be conceded while insisting on the centrality of the paradigm, on the character of central cases. Much creative activity will only implicitly be guided by cotemporaneous feelings, and will often be, in a given case, automatic or habitual; feelings may enter into the practice of the artist only subliminally, by way of implicit knowledge, normally thanks to training and practice. The only rock-bottom requirement is that such knowledge, whether implicit or explicit, must be experiential knowledge, of the ways in which one would feel when experiencing the vehicle if one’s frame of mind were more suggestible (for the same problem arising for the expression of emotion see Wollheim 1980 pp. 22–33, especially p. 27). Likewise, both with respect to a single spectator from time to time and with respect to different spectators, responses are variable. The point is not to form exceptionless generalisations but to capture the centre of the paradigm; with the paradigm in place, departures from it can be tolerated, but still it is the paradigm which makes the category what it is.

Second, the expressive character of a given element may be context-dependent; indeed its expressiveness is often holistic. Especially in more complex cases, the artist is not merely assembling marks or noises piecemeal, each with a self-sufficient expressive meaning. The expression may emerge more gradually from amidst many such constituents, in the course of positive feedback between the perception of what one is doing or has done and one’s physical or imaginative work with the medium, as Croce described in a memorable passage (1966 pp. 227–8; it was also stressed by Ducasse 1964 p. 111 and see also Fry 1920 p. 58). The artist, in many cases, is engaged with representation which is fully realised only in the relations of artistic elements. For the artist or the spectator, the significance of a given mark often depends on its representational properties, possibly developing or future; the context, sometimes only a very extensive context, makes the exact expressive character of a given gesture possible. And the context is not a bloodless abstract structure; it is a structure of expressive marks, and grasp of an expressive picture, or an expressive constituent of same, is not mere calculation but involves feeling it in a certain way. The ‘expressed content’, as I will explain further below, is not separable from the action or actions, or their product, which embody the expression. (In order to get that expressive effect, you must make those gestures, make those marks.)

I shall complete this sub-section by specifying two further ways in which simulation may be involved.

First, on the basis of further empirical evidence, simulation has plausibly been held to happen—notably and appositely by Greg Currie (2011)—even with respect to the ordinary non-animal and non-artefactual environment. From the phenomenological point of view, it is simply a matter of fact that one can subliminally or imaginatively feel the wind through looking at a wind-blown tree, the heat of a desert landscape or the roughness of the stones through looking at them (this was smuggled into the list of feelings in Section 1). Simulation may thus be involved in this further way in the perception of scenes and objects represented in pictures (and it would also be involved in this way in the perception of sculpture and architecture).

Second, simulation theory dovetails with J.J. Gibson’s celebrated theory of ‘Affordances’ as a theory of vision (see Thill et al. 2013; Noë 2004 p. 21ff; Garbarini and Adenzato 2004). In his 1979 book Gibson writes: ‘The affordances of the environment are what it offers the animal, what it provides or furnishes, either for good or ill … it implies the complementarity of the animal and the environment.’ (p. 127) The theory ‘implies that to see things is to see how to get about among them and what to do or not do with them. If this is true, visual perception serves behaviour, and behaviour is controlled by perception.’ (p. 223) In seeing a significant ledge dropping directly before one as one walks, for example, one immediately visualises and feels, if subliminally, one’s falling off the ledge if one continued walking in the same direction. Visual perception has its raison d’etre by being shot through with awareness of potential muscular action, awareness that is made more or less salient by the subject’s having the relevant desires and aversions: the visual system is interwoven with those mechanisms of potential action that are directed at behavioural opportunities, and those directed at avoiding dangers, in the immediately perceived environment.

The special relevance of affordances to aesthetics will emerge in connection with the older figures. As I said, I venture only the beginnings of a psychological model, just enough to see how it might go in a few relatively simple cases.

2b Historical Precedents

We have now seen a rough picture of the artistic expression of feeling as opposed to emotion, focussing on drawing and painting; I turn now to some historical precedents. The involvement of the visual apparatus with the systems of simulation and affordances was strikingly explicit, if understandably not expressed in those terms, in Dewey’s Art as Experience, and he connects them with artistic expression. He wrote in the expected sense of emotions as being expressed (1934 pp. 64–9), but more generally, at greater length and perhaps more interestingly, of the myriad qualities found in perceptual experience as receiving the imprint of the artist via expression. He wrote of the artist as mastering a medium and thereby achieving expression: ‘Sensitivity to a medium as a medium is the very heart of all artistic creation and esthetic perception’ (p. 199); such sensitivity—a capacity for feeling in my sense— ‘converts it into an authentic medium of expression’ (p. 200). He speaks of lines and shapes as expressive, of varieties of their dynamics as ‘feel[ing] differently’ (p. 207). He describes in extraordinary detail how he thinks this conversion takes place, which might be summed up as the imposition of form—form in its infinitely various manifestations—onto the material (more on this below).

Dewey’s allowance for simulation and affordances emerges in response to the hyper-formalist theory that ‘[t]he expressiveness of line as mere line is … proof that esthetic value belongs to sense qualities in and of themselves’, when he writes that on the contrary:

… the optical apparatus … never functions in isolation. It operates in connection with the hand in reaching for things and in exploring their surface, in guiding manipulation of things, in directing locomotion. This fact has for its consequence the other fact that the sense-qualities coming to us by means of the optical apparatus are simultaneously bound up with those that come to us from objects through collateral activities. The roundness seen is that of balls; angles perceived are the result not just of switches in the eye-movements but are properties of books and boxes handled; curves are the arch of the sky, the dome of a building; horizontal lines are seen as the spread of the ground, the edges of things around us. This factor is so continually and so unfailingly involved in every use of the eyes that the visual experience of lines cannot possibly be referred to the action of the eyes alone. (pp. 99-100, second emphasis added; see also pp. 255, 256, 259)Footnote 7

He spoke elsewhere of the dynamic ‘qualities of experience’—spatial and temporal characteristics, ‘directional tendencies’ that ‘signify loosening and tightening, expanding and contracting, separating and compacting, soaring and drooping, rising and falling; the dispersive, scattering, and the hovering and brooding, unsubstantial and massive flow’ as not being metaphorical descriptions, rather as being ‘the stuff out of which the objects and events we experience are made’ (p. 207), which ‘in works of art are expressed’ (p. 208; see also Bosanquet 1915 p. 20–2, and Langer 1957 p. 15).

Turning to the great critics roughly contemporaneous with Dewey, we find that Pater, Berenson, Schapiro and Wölfflin spoke repeatedly and unreservedly of artists expressing their feeling for the things around them, of their delight in the world, of their love of the world—of landscape, of nature, of the human body, of water, colour and light—through their portrayal of those things (this is not love, of course, in the full-fledged romantic sense; it is simple but intense delight, liking of things as seen or desire for them). Fry wrote ‘The drawn line is the record of a gesture, and that gesture is modified by the artist’s feeling which is thus communicated to us directly’ (Fry 1920 pp. 23; see also p. 57, p. 108, p. 112, and p. 121). He goes on to describe more complex varieties of expression, somewhat as I do, though naturally the distinction between emotion and feeling is unclear in his discussion. Schapiro speaks of Cezanne as if it were common knowledge that ‘the later works allow us to see the range of his art which transposes in magnificent images so vast a world of feeling’ (1988 p. 11; see also Schapiro 1961).

Berenson’s views deserve special mention, especially as expressed in his more theoretical Aesthetics and History in the Visual Arts. As Collingwood in effect stressed in a wider philosophical sense, and as art students learn in a narrow practical sense, gesture drawing as opposed to contour drawing is the way to convey, to enhance ones feeling for movement, energy, solidity, gravity, mass, and essential form, without concern for the exact shape or outlines of an object (Berenson 1948 p. 75; also see Dewey, p. 92ff). It is drawing ‘from within’ rather than ‘from without’; indeed it is routine for gesture drawing to be described as expressive, is comparison with contour drawing, with its relative emphasis on representational accuracy. Consider then Berenson’s celebrated doctrine of ‘Ideated Sensations’ (pp. 24ff, 73–8). An ideated sensation is a vivid item of visual awareness that portends the features just mentioned—the heightened sense of where an object has just been, where it is going, its immediate potentiality for action or being acted upon, and so on (see p. 79). It is an enhanced sense of the ‘vitality’ of the item, in Berenson’s terms. And via another well-known concept he has allowed roughly for the idea of Gibson’s affordances in his notion of ‘tactile values’ (p. 69f; another hint of the idea of affordances is at p. 72); these refer to the sense that one could pass behind a depicted item, that one could go through a doorway, that one could feel the solidity of the stone; it is the bringing together of vision with responsiveness to touch and movement. Giotto was the first to significantly display these (not to be equalled until Masaccio over a hundred years later, says Berenson):

His first business…is to rouse the tactile sense, for I must have the illusion of being able to touch a figure, I must have the illusion of varying muscular sensations inside my palm and fingers … it was of the power to stimulate the tactile consciousness of the essential, as I have ventured to call it, in the art of painting that Giotto was supreme master. This is his everlasting claim to greatness … (Berenson 1952 [1896] p. 41/p. 2)

Giotto was by far the most conspicuous of those who first had the knack of systematically representing in two dimensions what is intrinsically in three, namely the possibility of recession, of one thing being behind another, of a believable space into which one might enter. This heightened the sense of the real presence of objects, with all their conative potentialities, and thus the feelings were engaged as never before. Vision supplies the brute data, and one feels, proprioceptually and kinaesthetically, through a thousand little tugs and tingles in one’s nerves, what would be the possible, appropriate, or desired actions in response.

Collingwood cites Berenson approvingly:

When Mr. Berenson speaks of tactile values, he is not thinking of things like the texture of fur and cloth, the cool roughness of bark, the smoothness or grittiness of a stone, and other qualities which things exhibit to our sensitive finger-tips. As his own statements abundantly show, he is thinking, or thinking in the main, of distance and space and mass: not of touch sensations, but of motor sensations such as we experience by using our muscles and moving our limbs. But these are not actual motor sensations, they are imaginary motor sensations. In order to enjoy them when looking at a Masaccio we need not walk straight through the picture, or even stride about the gallery; what we are doing is to imagine ourselves as moving in these ways. In short: what we get from looking at a picture is not merely the experience of seeing, or even partly seeing and partly imagining, certain visible objects; it is also, and in Mr. Berenson's opinion more importantly, the imaginary experience of certain complicated muscular movements.(Collingwood 1938 p.147)

Berenson ties these things together with idea of the ‘spiritual significance’ of a work of art, which is the product of the ideated sensations and tactile values conditioned and reinforced by the work’s overall form as experienced (Berenson 1948 pp. 73f, 124, 131). This hints at a solution to a certain problem that was felt to be left hanging in Bell’s What is Art? (1914) and Fry’s various statements of the view. According to Bell and Fry, the monarch of aesthetic values is ‘significant form.’Footnote 8 What is that? We cannot define it, but, like Beauty, we know it when we see it, by the presence of the ‘aesthetic emotion’. And what is that? If it is the emotion awakened by significant form as Bell at one point suggests, then a likely complaint is that the circle is too close, too close at any rate for explanatory or analytic purposes. But this is to mistake the aim; the aim rather is elucidatory. The way out is not to expand the circle but to deny that, at least phenomenologically speaking, there are two things in the first place, the form and the feeling. Form in the aesthetic sense is not a mathematical or otherwise scientific notion, but an experiential notion. It is the deepest or most primitive experience of the form of things, of the shape of things. To become aware of the aesthetic emotion just is to experience the work as possessing coherent and arresting form, of the form as ‘moving through’ the work (compare Croce 1909 pp. 74–6). The more pronounced it is, the more ‘significant’ it is; there is no form that is not, to some degree or other, significant form (the opposite of significant form is not bad form but formlessness).

The result is perhaps a contentiously narrow circumscription of the domain of the aesthetic, but here Dewey can provide some relief. For in the middle chapters of his Art and Experience, Dewey said a great deal about what dynamics of perception are singled out by the notion of form, and he explicitly goes against Fry in accepting a much greater range of qualities—including certain representational qualities—as aspects of form (p. 86ff). I refer the reader to the text (i.e. pp. 130–3, 185–6), but very briefly, phenomenologically speaking, form is not an abstraction from matter but the way experienced matter is arranged in actual examples: it is the unification and ordering of the experience of objects, events and scenes, and the balance and rhythm of energies and substance of various kinds that constitutes experience (see Rose 2019, pp. 18–47). For her part, Langer wrote: “The word ‘feeling’ must be taken here in its broadest sense, meaning everything that can be felt, from physical sensation, pain and comfort, excitement and repose, to the most complex emotions, intellectual tensions, or the steady feeling-tones of a conscious human life.” (1957, p. 15, author's emphasis). Thus Dewey, Langer and Berenson, with Bell and Fry at least glancing over their shoulders, look to be singing from similar hymn sheets.

Needless to say I am not trying to defend or analyse Dewey, Croce, Langer and Berenson and their contemporaries word for word—nor positively to argue for the views of Gallese or Gibson—but to suggest that many of the minds who have thought most penetratingly about expression and its importance in art in the past one hundred years have had similar thoughts, and to suggest that at the centre of their thinking, which I have suggested is allied roughly with more modern and perhaps more believable doctrines, is the expression of feeling.Footnote 9 Of course—not least because of the aforesaid semantic uncertainty surrounding their use of the key terms—I cannot prove that these figures were sensitive to the issue, and that they would be receptive to the approach taken in this article. But that is very unlikely to be so in every case.

3 Ontology, Semantics, and Logical Form

In the epigraph to this piece from Bosanquet, the claim is implicit that the paradigm for the expression of feeling involves the expressive character of the expressive vehicle being experienced by both the artist and spectator as fused with the vehicle, as an intrinsic property of the vehicle. The medium is perceived as inseparable from the message, in Marshall McLuhan’s words. Bosanquet speaks of feelings as being ‘embodied’ (1915 pp. 7–9); Ducasse of feelings as being ‘objectified’ in the material work of art (1929passim;1964 pp. 110–1). For example that singular feeling of gravity, of downward momentum centred on Christ’s body, in Rubens’ Descent from the Cross, is made out unequivocally as something perceived rather than inferred. If we consider music for a moment, some such thought is behind Debussy’s provocative remark that music ‘is not even the expression of feeling, it’s the feeling itself’ (quoted by Taruskin 2009 p. 197). Likewise the disarming appeal of Carroll Pratt’s oft-quoted aphorism that ‘music sounds the way moods feel’ (1931 p. 203) or that music ‘does not suggest mood and feeling; it is mood and feeling’ (p. 290), might also be recognized. What these views have in common is the denial of any experiental bifurcation of the expressive content of a work into its physical properties and its emotional or feeling content.

However experiental or epistemic inseparability does not imply ontological inseparability. We can allow Pratt his metaphorical license for his first, aphoristic statement, but as pernickety philosophers we must find against Debussy, and against Pratt in his second statement. Feeling is not identical to music, nor is the feeling identical to the music in the particular instance.Footnote 10 What is true is that epistemologically, the feeling cannot properly be known—because it cannot be experienced—apart from the gesture. The feeling is not a self-sufficient object, perhaps not even a property of the gesture, but an experiential aspect of the gesture—an instance of a certain nebulous kind, or perhaps a trope—which indeed cannot exist without the gesture, and to which one has access only by experiencing the gesture, either by being the agent of the gesture or by means of empathy, the ‘off-line’ capacity described earlier. They are not re-identifiable except via the gesture, and perhaps indeed no sharp criteria of identity apply to them.

The matter of the seeming ineffability of artistic content of many works of art, or of the difficulty in linguistically specifying their exact artistic content, was explored by Arnold Isenberg in his 1949 paper ‘Critical Communication’. I shall restrict his point to ascriptions of expressive character rather than artistic content generally. Well-formed, ordinary descriptive statements—‘My hammer is broken’ for example—have truth-conditions which can be determined purely linguistically, given the referent of the indexical ‘my’ and allowing for ordinary vagueness. One can adequately understand a statement made by an utterance of such a sentence without bearing any perceptual relation to the relevant agent or hammer. By contrast, a singular predication of expressive character to a work of art, as meant in critical discourse—for example ‘It expresses despair’ said of Van Gogh’s Night Café—does not normally function in that way. What the critic means can only be understood with anything approaching exactitude if one sees or has seen the work oneself (I assume seeing a reproduction counts as seeing the work, for these purposes). The words themselves give only a rough schema of what to expect when viewing the picture: Epistemically speaking, the intended meaning is not wholly linguistically determined, but awaits perceptual determination, must be ‘filled-in’ by means of acquaintance with the picture (Isenberg 1949, p. 336). This is not to deny that those words can also be meant in a way that says no more than that the picture can roughly be categorised in a certain way, and thus expresses no more than its linguistically determined truth-condition. But in the standard critical use, or as used by a normal gallery-goer when looking at the picture, or later when describing what he or she saw, what is meant is indefinitely more fine-grained; it requires perception to complete, fully to grasp what the speaker intended to communicate. That is what is distinctive about ‘Critical Communication’ (the point was anticipated by Dewey 1934, p. 215).

Isenberg is not terribly explicit about ontology or semantical form. We should not assume that feeling-expression is simply a dyadic relation, as is suggested by ‘W expresses F’, with W the work of art and F the feeling. Perhaps we should understand the form ‘__ expresses F’ as tacitly abbreviating further structure, or as its being a complex monadic predicate (like ‘__ runs quickly’), not one which signifies a genuine relation (for a parallel one might think of the adverbial approach to perception, as will emerge shortly). However in many cases we cannot comfortably locate F with terms for particular feelings; although I think it would be wrong to do so, one can see the point of treating such cases of ‘__ is expressive’ as epistemically and semantically basic, not to be thought of as practical shorthand for ‘__ is expressive of F’.

I said that we might look to the adverbial theory of perception for a parallel. This is the view that to see something red, whether or not the seeing is veridical, is to ‘experience redly’, without a percept—for example a batch of sense-data—being a separate object figuring in the act itself. But although the adverbial theory does continue to have its defenders, it is distinctly a minority view (Ducasse himself was a supporter in his 1942). Among other things, the view appears to be allied to the highly questionable idea that adverbs themselves can be explained without positing objects—that for example, ‘__ runs quickly’ could be explained simply as a restriction of the simple predicate ‘__ runs’, and that so it is with ‘__ sees redly’ and the predicate ‘__ sees’. I shall look instead to a certain account of adverbs themselves, namely Davidson’s well-known technique of positing a tacit quantifier ranging over events, with adverbs having the effect of predicates of the events. I will suggest one way of fine-tuning Isenberg’s account of what is going on when a spectator grasps the feeling-content of a work of art, one that dovetails both with what he explicitly says and with what I described three paragraphs ago, and which involves quantification over gestures. It does so in a way which does not exactly parallel Davidson 1967 account of adverbs, but which is inspired by it.

If the feeling is indicated by an ordinary word of public language as in ‘W expresses F’, with ‘W’ indicating a work of art and ‘F’ a category of feelings, it must be that in favourable cases there is some range of particular experiences that a spectator undergoes or empathetically undergoes when perceiving the gesture that embodies the feeling. The gesture must be part of the work of art if it is not the entirety of the work (and I shall set aside for the moment the complexities due to the holistic and context-dependent nature of expression, and to the cultural relativity of at least some forms of feeling-expression). These particular features of experience are ontologically perhaps tenuous—I characterized them above as ‘aspects’ or ‘tropes’, and grant that they are without sharp identity conditions—but they are often pellucid experientially. I will indicate them by the schematic Greek letter ‘Γ’, and characterize the manner in which the subject experiences a gesture as the spectator’s perceiving the gesture under Γ (perhaps more articulately, under the aspect Γ; equally: perceives it as Γ). The relation between these half-entities and the categories of feeling indicated by ordinary terms is a type of logical sufficiency: If a gesture expresses a certain feeling, then competent and receptive spectators perceiving the gesture will undergo some experiential aspect Γ such that perceiving the gesture under Γ is sufficient for the gesture’s expressing the feeling (with the various caveats mentioned above in Section 2a.; I speak only of paradigm cases).

It will be remembered that in many cases of feeling-expression there will not be any statement of the form ‘W expresses F’ that is adequate beyond ‘W expresses feeling’. On the present picture, such a statement should be understood as a restricted existential quantification into the place marked by ‘feeling’, i.e. ‘W expresses some feeling or other’ (where the quantifier is objectual, not substitutional). With that in mind, an informal way of representing the facts underlying ‘W expresses F’ is as follows, where to be the agent of a gesture counts as being a competent spectator of the gesture: ‘W embodies or contains a significant trace of a certain gesture, and any competent and receptive spectator perceives or undergoes the gesture under a certain experiental aspect Γ’. This is not itself intended as a conceptual analysis or definition. It is an explanation of one sort of fact in terms of another, or, if that strikes one as too metaphysically grand, it is a more revealing statement of what must be true when and only when the statement ‘W expresses (the feeling) F’ is true. In order fully to grasp such a statement, one must have experiential access to Γ.

A logical paraphrase of our factual equivalent of ‘a expresses (the feeling) b’—with ‘a’ replacing ‘W’, ‘b’ replacing the schematic abstract noun F, and writing ‘z’ for ‘Γ’—would be ‘∃x(a embodies x & x is a gesture &∀y(y is a competent spectator → ∃z(y experiences x under z))). Abstracting to logical form yields ‘∃x(Rax & Gx & ∀y(Fy → ∃z(Syxz)))’.Footnote 11

That all feeling-expression is of this simple atomistic form—one gesture, one feeling—is of course implausible. To allow for more complex cases, to get the effect of several gestures figuring in the expression, the initial quantifier may range over combinations of gestures, or we can allow several initial quantifiers.Footnote 12 To allow for the holistic nature of feeling expression, we can require the truth of whole collections of these statements—perhaps other kinds too, such as statements of emotional expression—all with the same spectator and the same work of art. To allow for the context-dependence and cultural relativity of feeling expression, we can add a parameter to the antecedent of the conditional, ‘yc is a competent spectator’ or ‘Fyc’, explained as ‘in circumstances c’ or ‘of culture c’, or ‘of artworld c’. Needless to say the topic represented by this new parameter is eminently disputable; I will rest here with treating it simply as an existentially quantified variable, the scope of the accompanying quantifier being the narrowest possible.Footnote 13

4 Concluding Remarks

I began with Bosanquet’s idea that the expressed feeling cannot be separated from the embodiment. I’ve treated this in the first instance as an experiential or epistemic claim—that the feeling cannot be known apart from the embodiment, that it cannot be known apart from the embodying gesture—but the claim has ontological repercussions as just described. Indeed the points are connected: not only are feeling-aspects obvious candidates for Berkeley’s esse est percipi, the ontological half-entity-like status of the aspects Γ that figure centrally in the account is one symptom of the epistemological gulf between art and language, between the contents of works of art and the ordinary contents of ordinary sentences, which Isenberg and so many others have stressed.

What I have said about ontology is restricted to the expression of feelings, and is not meant as extending to the expression of emotion. I have concentrated largely on the more difficult area of the plastic arts, but as more than hinted above something very like this distinction applies to music: one wants to aver that, say, Sonny Rollins played his solo on ‘You Don’t Know What Love Is’ with great feeling, that he thereby ‘expressed himself,’ but one shies from saying that he ‘expressed his emotions’ in his solo—where emotions are the sorts of intentional mental states I have assumed, and which feature for example in Robinson’s (2005, 2017) account of musical expression. Many have made somewhat similar points with respect to music including such central figures as Kivy (1980) and Stephen Davies (1994), but not, so far as I know, as part of a general account of the artistic expression of feelings. The art of sculpture, interestingly, appears for the most part to work differently: in standard figurative sculpture, feeling can sometimes be conveyed by the artist’s motor actions or for example by the texture of the stone, but overwhelmingly they are conveyed by the representational features of the work—in particular by the attitudes of the figures themselves.