From the ALWS archives: A selection of papers from the International Wittgenstein Symposia in Kirchberg am Wechsel

1. Seeing-as

Despite their disparities, most contemporary image theories seem to agree on the fact that the constitution of the image’s meaning is fundamentally codetermined by the gaze directed towards it. Images – thus – do not have a single sense, but can have plural meanings, depending on the perspective from which one looks at them. To illustrate this fact, one image example has been frequently invoked: the duck-rabbit-picture of which Wittgenstein makes use in his later philosophy (Wittgenstein 1993, 204sq., §118sq.). The ambiguous picture, drawn from the Polish-American psychologist Joseph Jastrow, either shows a duck or a rabbit, but never both at once. Now: what changes in the switch from one to another? Certainly not the lines on the paper. Rather (as Wittgenstein would say) their meaning, as we never just see lines, but organized lines which we see as objects. In a phenomenological vocabulary, the as-structure is made possible by the intentional structure: we do not only have optic impressions on our retina, but perceive trees, houses, objects. Applied to Jastrow’s ambiguous figure, this implies that we either see the drawing as the drawing of a duck or the drawing as the drawing of a rabbit. Tertium non datur. The ambiguity of the “flip flop image” (Kippbild) can thus be literally brought back to ambi-valence, insofar as it alternates between two (and only two) possible values.

While Wittgenstein used a pictorial example to exemplify a feature of perception – its intentional as-structure –, many image theorists, on the contrary, applied this feature of perception to images. For Virgil Aldrich, the possibility of an image rests on the capacity of aspect seeing (Aldrich 1958), for Gombrich, the question “rabbit or duck?” is the “key to the whole problem of image reading” (Gombrich 1960: 238), for Wollheim’s Art and Its Objects, the structure of seeing-as is sufficient for understanding pictorial representation (Wollheim 1968). Or as Goodman puts it, rather than saying that the picture of Pickwick represents Pickwick, we should say that we see the picture as a Pickwick-picture (Goodman 1968, ch. 5). What is more, we do not simply see Pickwick as such, we see him under a certain aspect, such as Pickwick-as-a-clown (Goodman 1968, ch. 6). If we wanted to sum up the image conception common to those theories, we could formalize pictorial perception as follows: we see images as an ‘x’ depicting a ‘y’.

Analytic aestheticians have widely discussed the aporias of such an approach and Wollheim has revised his own position defended in the first edition of Art and Its Objects by replacing the concept of “seeing-as” by the concept of “seeing-in” (Wollheim 1980) which shall be analyzed in the second part of the paper. While this discussion concerns mainly aesthetic issues, it however hints at a more general problem: the argumentational fallacy which consists in identifying intentionality and propositionality. John Searle’s theory of intentionality is biased by this fallacy, when from a seeing-that, he concludes to a seeing-something: “From the point of view of Intentionality, all seeing is seeing that: whenever it is true to say that x sees y it must be true that x sees that such and such is the case” (Searle 1983: 40). Such an assumption leads to stating that every seeing-something (p) must be understood as seeing-that-p: every time I see a red ball, I see that the ball is red. In this respect, any perception will have a propositional content which can be expressed as the ascription of identity or as a predicament.

In many cases, it may be true that perception corresponds to a propositional perception of the type seeing-that-p. Such a description, however, obliterates the fact that a) there are intentional forms of seeing that cannot be immediately translated into ascription of identity and b) that there are forms of seeing that do not have any intentional content at all.

Intentional, but non-propositional seeing: In many cases, seeing is directed to an object, although we would be incapable of unambiguously telling the nature of the object. We might be aware that we are in front of something without being able to name it. This is particularly true of images, which do not give us the possibility to vary our angle of perspective. Confronted with certain portraits or caricatures, we might say that the face “reminds” or “looks like” x, without enabling Searle’s conclusion that every seeing-that is a seeing-that-p. Or, in other terms: every seeing-something is not necessarily a seeing-something-as-something.

Non-propositional seeing without an object: Many types of vision such as peripheral vision or perception in a state of fatigue, for instance, are not intentionally directed towards an object nor to its recognition. Rather, they allow a particular attention to qualitative or atmospheric dimensions. Many artists have delved into this kind of vision in their image installations, from the American Color Field Painters up to James Turrell boxes or Olafur Eliasson’s Black Horizon. Such a visual experience in the immersive space of the image allows for a reflection on the visual process itself and to its performative dimension. Rather than the what, it is the how that comes into focus. The quality of vision – the “how” – does not allow for any propositional or existential claim.

However, such non-propositional and non-object-oriented types of vision are far from being unfathomable or mysterious: they rather correspond to modes of distinction that attributive logics of the concept do not adequately account for. Or as Wittgenstein would say, the “grammar” of the visual field, unlike that of language, is not based on unambiguously discrete elements. The Big Typescript is explicit in this respect: “the visual field does clearly not consist of discrete parts” (Der Gesichtsraum besteht offenbar nicht aus diskreten Teilen; Wittgenstein 2000, 2.243.4.1).

When refering to a nonconceptual gaze, classical aesthetics have often invoked the proverbial “je-ne-sçais-quoi” or “non so chè”. This gaze, however, is not a farewell to intelligence, but the opening of another, of a visual intelligence, crucial in approaching images. Such an attention to the “how” or – in other terms – to the style of the visually organized field, is albeit not restricted to the gaze of the art critic or the connoisseur. Experiments with pigeons (i.e. birds with a high capacity of orientation in landscapes seen from above) have shown that through specific training, the pigeons are able to distinguish between cubist and impressionist paintings (Watanabe et al. 1995). It would be hard to attribute a notion of “cubism” or “impressionism” to the birds; and it is improbable that they recognize women, fruits or rags or the fact that their representation is twisted. Nevertheless, the pigeon’s identification of the style of painting is almost flawless. Drawing on similar experiments, Arthur Danto thus concluded in his essay Animals as Art Historians: “Pictures as such are not like propositions, nor can we speak of a pictorial language, as Wittgenstein endeavored to do in his Tractatus, since animals demonstrably have pictorial competence while animal propositional – or sententional – competence remains undemonstrated” (Danto 1992: 20).

2. Seeing-in

The specific pictorial competence that can be acquired or trained is, however, different from the seeing-as insofar as it cannot be taught independently of the perceptive situation. While seeing-as can easily be translated into similar expressions devoid of any sensory dimension such as “interpreting-as” or “understanding-as”, the situated visual discrimination can only be made in front of the object. As opposed to linguistically mediated learning of the propositional content of the ‘as’, the discrimination is made along lines within the artifact. Or as Danto formulates it, beings without propositional competence but with pictorial competence like pigeons are, though not capable of seeing-as, capable of seeing-in (Danto 1992: 28).

The category of „seeing-in“ has been introduced by Richard Wollheim in order to address the double problem that a) seeing-as is not specific to pictorial perception and b) a general structure of perception has been wrongly applied to pictorial perception. A striking case of a flawed generalization of the duck-rabbit-example to pictorial vision as such is Ernst Gombrich’s image theory. For Gombrich, the disjunctive structure of Jastrow’s figure is that of images in general: we may either see what is represented or be attentive to the canvas, but we can never see both at the same time: “To understand the battle horse is for a moment to disregard the plane surface. We cannot have it both ways” (Gombrich 1960: 279). Michael Polanyi contested Gombrich’s disjunctive logic, inasmuch as he showed that the seeing-what and seeing-in do not operate on the same level but correspond to a “focal” and to a “subsidiary” or “peripheral awareness” (Polanyi 1970: 153). Wollheim, in turn, not only contests the claim that “we cannot have it both ways”, he moreover maintains that images require “simultaneous attention to what is seen and to the features of the medium” (Wollheim 1980: 212). Images are neither fully transparent with respect to their referential object nor totally opaque, exposing their material qualities of the medium: according to Wollheim, images always imply an attentional “twofoldness” (a trompe l’oeil would thus not meet the requirements for being an image).

Hence, Wollheim’s concept of seeing-in firstly aims at readjusting the conceptualist bias of the seeing-in logic, which focuses on the fleshed out “recognitional” aspect, in order to rehabilitate a “configurational aspect”. Secondly, it aims at rehabilitating the material, objective qualities of the image’s medium, in which something is seen. This second point, although claimed by Wollheim, can be doubted, however. By insisting on the creational aspect of seeing-in, referring to our capacity to seeing dragons’ heads in clouds and castles in a Rorschach inkblot, Wollheim reduces the “recognitional” dimension intrinsic to seeing-as. But can we distinguish seeing-in from a seeing-into? In other words: can we distinguish the perception of a form emerging from a canvas and an arbitrary projection onto a surface, regardless of its configuration? To avoid the impression of arbitrariness, Wollheim is required to introduce a further element: while in standard perception, we may virtually project everything into everything, pictorial seeing-in is only successful, when we see in the image what the artist wanted us to see in it (Wollheim 1980: 207).

It is somewhat curious how Wollheim, who affirms to be advocating for an “object theory” of images, counterweighs the excessive subjectivity of the spectator’s gaze by the subjectivity of an artist’s intentional gaze. But a theory of the gaze does not provide us yet with a theory of the image. Once again, the co-constitutive function of the material medium of the image is eluded. Moreover, the introduction of the artist’s intention is announced as an effort of disambiguating the multiple possible perceptions of an image. While arguments can be brought forth questioning the possibility of such a disambiguation (Lopes 1996, ch. 8.3), one could raise a further question: why does the ambiguity of images have to be reduced to the twofoldness of denotate and medium? Isn’t Wollheim’s “bivalence” theory yet another reduction to a static simultaneity of what is, phenomenologically speaking, constantly oscillating? Can we exclude trompe l’oeil’s from the domain of images straight away, simply because they do not meet the requirements of the simultaneous perception of figure and medium? Wollheim’s formalization of imageity must inevitably lead to what Merleau-Ponty termed as “bad ambiguity”. Can an image theory be developed which would not think images in terms either of a disjunctive logic (like Gombrich) or of simultaneous twofoldness (like Wollheim), but rather in their very manifoldness?

3. Seeing-with

In L’Œil et l’esprit, Merleau-Ponty affirms that we “do not look at [a painting] as one looks at a thing […] Rather than seeing it, I see according to, or with it” (Merleau-Ponty 1964: 126). This seeing-with underscored by Merleau-Ponty has long been underestimated in contemporary image theories, which either excessively focus on images as mere things or, on the contrary, on the constitutive force of the gaze. While Merleau-Ponty elsewhere criticizes the idea of images as “second things” (choses secondes), devoid of any own efficacy, in this statement, he implicitly targets the dominance of a gaze theory of images, in particular that of Sartre. Sartre’s L’Imaginaire is thoroughly based on a concept of consciousness which can be compared to that of Wittgenstein’s “change of aspects” (Aspektwechsel). In order to see an image, I need, according to Sartre, to “deny” the materiality of the painting. We may either look at the material qualities of the image-object in a “perceptive attitude” (attitude perceptive) or, by changing our consciousness state and negating the material world, we may have an image emerging in an “imaging attitude” (attitude imageante) (Sartre 1943). For Merleau-Ponty on the contrary, an image does not emerge despite its material support, but thanks to it. In an unpublished manuscript, Merleau-Ponty notes: “What is a Bild? It is manifest that we do not look at a Bild the way we look at an object. We look according to the Bild [selon le Bild]” (Fonds Merleau-Ponty, BNF, vol. VIII: 346). In other words, we do not only see in images, rather seldom as images, never despite them but always with them and through them.

Seeing through images does not mean that images are transparent windows onto reality: as Wittgenstein says in another context, one thinks “that one is retracing nature over and over again, and one is merely tracing round the form through which we see it“ (Wittgenstein 1993, §114, modified trans.). Stating that we see through images means that, rather than neutral surfaces of the beholder’s projection, images generate gazes which, although never ultimately fixed, are by no means arbitrary. The form of the image, its figural organization, its material ridges, dales and crests, open up a space for potential vision. Between the unambiguousness of a communicational message or an artist’s intention inscribed into the object and the image as a space of free variation of consciousness, it appears that the density of images, their material stratification and their phenomenological overdetermination demands a specific time of contemplation.

Seeing with images then means that the evidence they provide the spectator resists generalization without further ado: iconic evidence is not a ladder that could be thrown away after we have climbed it, but remains inherently situation-dependent, case-sensitive and thus, ultimately, precarious. Images help drawing distinctions, but these distinctions do not exist beyond the material medium which they organize from inside. Images thus yield a potential, but neither in the sense of a mere indetermination (the pura potentia of matter) nor of a preexistent form or meaning which the gaze would have to reveal, just as the sculptor releases the already inherent form from within the marble. Rather, seeing with images entails following those veins in the marble of which Leibniz said that they signify a tension inherent to matter towards certain unfoldings and individuations.