The Tractatus Logico-Philosophicus, it is often and plausibly thought, is not concerned with epistemology, nor, except tangentially, with the philosophy of mind. The topic of perception thus falls well outside its remit. There is a grain – more than a grain – of truth in this view. But it is not quite right. Where Wittgenstein comments on perception in the Tractatus, he does so primarily in the service of a view of judgement. Nevertheless, the book does encompass a view of perception which is philosophically interesting and which, indeed, provides a solution (though one which is problematic in its own right) to a problem which plagued both Frege and Russell. This paper is dedicated to establishing this point.

In section one, I discuss the main passage in the Tractatus that pertains to perception: remark 5.5423. A natural reading of the passage is scouted, according to which Wittgenstein means to attribute propositional content to perceptual experiences, just as Frege attributes such content to judgement. The remainder of the paper will show that this intepretation is at best misleading. In section two Frege’s actual views on perception are examined. In sections three and four I demonstrate the non-Fregean nature of Wittgenstein’s doctrine, and show that Wittgenstein avoids a major problem that arises for Frege. In the final section I face the question of whether the Tractatus view constitutes a theory of perception, or whether on the contrary the construction of such theories is left for psychology. My answer is equivocal.

Seeing Facts and Seeing Objects

At Tractatus 5.5423, Wittgenstein gives a picture of the familiar, ambiguous Necker cube and makes the following remarks:

To perceive a complex means to perceive that its constituents are related to one another in such and such a way.

This no doubt also explains why there are two possible ways of seeing the figure as a cube; and all similar phenomena. For we really see two different facts.

(If I look in the first place at the corners marked A and only glance at the Bs, then the As appear to be in front, and vice versa.)

The problem arises because there are two different ways of seeing the same figure. One can see it, let us say, as advancing or as receding, and one’s experience changes when one suddenly sees it in a new way. What is the difference between seeing the figure as advancing and seeing it as receding? The difficulty, in part, is that the Necker diagram is simple. All of the lines that make it up are typically in plain view. What occurs when we cease to see it as advancing and begin to see it as receding is not that we notice some detail that we had not noticed before. There is no part of it which is hidden, or to which the subject does not have unimpeded visual access. Rather, the entire diagram is in full view throughout. Further, no illusion of change is created: it does not look to the viewer as though the cube has changed its shape, for example. Given these facts, the following question arises: what accounts for this change in the viewer’s experience?

One answer, which is suggested by a century’s discussion of such phenomena, is that the subject sees first one, then another, “aspect” of the cube. The moment of change in the subject’s experience may then be called an “aspect switch”. But it is certainly not the case that before and after the switch one sees different parts of the cube; nor, in the typical case, do subjects view the cube from different angles or, in the literal sense, from different standpoints. So it is not clear what an aspect is, nor what it is for a subject to undergo an aspect switch. The terms “aspect” and “aspect switch” will be adopted here, but they should be regarded as merely labelling the phenomenon and not as explaining it.

The Gestalt psychologists, from whose writings Wittgenstein’s interest in the issue may have arisen, thought of the problem in terms of the perception of “organisation”. The diagram is organised differently before and after the aspect switch. With respect to the lines on the page, this explanation is obviously false: their pattern does not change. The psychologists were, of course, aware of this. What they meant was rather that, though the diagram itself remains unchanged, the subject’s visual field changes in organisation. But, once again, it is clear that merely positing a visual field, with properties such as “organisation”, is uninformative. If we want to know what the change in our experience of the cube consists in, it does not help to be told that our visual field has changed, unless we have a grasp of what a change in the visual field is, independently of a change in our experience. A substantive theory of perception, explicating the notion of a visual field, is necessary for such an explanation to be viable.

Wittgenstein’s solution is to substitute the seeing of facts for the seeing of objects. We see different facts when we see the cube as advancing and when we see it as receding. Thus, facts are individuated more finely than objects, and in order adequately to characterise our experience we must mention the fact which is experienced. What, in the case of the cube, are these facts? On first sight, it might seem that Wittgenstein characterises them as follows. Before the aspect switch, we see that the As appear to be in front of the Bs; afterwards, that the Bs appear to be in front of the As. A closer perusal of 5.5423, however, shows that Wittgenstein does not say this. Indeed, he does not tell us what facts we see.

Whatever the facts in question are, they are facts both of which hold throughout the experience, before and after the aspect switch. As such, they cannot contradict each other. Minimally, it should be possible to formulate them in such a way that they do not appear to be in contradiction, as the formulations “the As appear to be in front of the Bs” and “the Bs appear to be in front of the As” do. It is therefore preferable, if possible, to interpret Wittgenstein as meaning instead that we see the facts in virtue of which, at different times, the points appear as they do.

If this is the right way to interpret the Tractatus discussion, we can begin to see its philosophical interest. If Wittgenstein is thinking in these terms, the passage becomes an investigation of the nature of appearances: he is not taking the notion of appearance for granted but is instead investigating what it is for something to appear in a particular way to a subject.

At first glance, and long after, a particular reading of the passage strikes the reader as overwhelmingly plausible. We are faced with the problem of how to characterise distinct perceptual experiences of the same object. The problem is resolved by attributing something like propositional content to experiences: content, that is, similar to that attributed to judgements by Frege. We find ourselves with a view of perception that might be called Fregean. On this reading, Wittgenstein looks very much like a contemporary representationalist about perception. Such theorists hold that perceptual experiences are mental states with representational content. A subject tokens a state with such content on an occasion: the state counts as accurate if the representation corresponds to some way the world is, and not if it does not. Thus to Tractarian facts, one might think, correspond different ways we can see the world as being. To characterise a perceptual experience is to say how the subject of that experience sees the world.

For a full-blooded development of a Fregean view along these lines, one might look to the third chapter of Christopher Peacocke’s Study of Concepts (Peacocke 1992, see especially 74-90). To take Peacocke’s central case, consider the difference between an experience of a square and an experience of a regular diamond. All squares are also regular diamonds, and vice versa, but an object, say a floor tile, can look to a subject like a square without looking like a regular diamond. For Peacocke, the phenomenon warrants the attribution to perception of what he calls protopropositional content. Protopropositional content need not involve the application of concepts. We need not suppose that in viewing an object we conceptualise it in one way or another, as falling under the concept square or the concept regular diamond. Instead, at a sensory and pre-conceptual level, the tile just strikes us as square-shaped or as diamond-shaped. We find the application of one or other concept “primitively compelling”, depending on the particular character of the perceptual experience we have had.

Protopropositional content is not fully propositional, at least because it does not in itself involve the application of concepts. But in it regular diamonds, for example, are distinguished from squares, though these shapes are geometrically identical, because one can take an attitude towards a regular diamond that is not at the same time an attitude towards a square. Just as in Frege’s theory of thought, the contents of experience are more fine-grained than are worldly objects, or concatenations of worldly objects.

A natural view requiring evaluation, then, is that Wittgenstein in the Tractatus puts forward a view similar to Peacocke’s. Though we see the same diagram before and after the switch, we see different facts. This response to the cube case mirrors the Fregean response to a corresponding puzzle about judgement, with facts being individuated more finely than objects. The two ways in which one can see the cube are explicated in terms of two different contents with which the expression “S sees that…” can be completed.

I hope to show that almost none of this is true, and that even the truth in it is likely to mislead. My strategy will be to discuss first what Frege actually says about perception. It will be seen that perception presents a serious problem for him, a problem which Wittgenstein escapes by distinctly non-Fregean means. A word of caution is due here, however. I do not claim that the perception remarks in the Tractatus are in any way a response to Frege’s remarks on the topic. Indeed this is highly unlikely, since Frege only discusses perception in any depth in the 1897 manuscript entitled “Logic”, which was not published until 1969, and in “The Thought”, which was published in 1918 after the Tractatus had already been completed.1

Frege on Perception

What did Frege himself think about perception? Two passages from his writing require our attention: one from the 1897 Logic and one from “The Thought”. In the Logic (Frege 1897, 137-8), Frege, in drawing a contrast between thoughts and physical bodies, writes that thoughts are not sensible objects, that is, not possible objects of sight or touch or taste.

But do I not then see that this flower has five petals? We can say this, but if we do, the word ‘see’ is not being used in the sense of having a mere visual experience [Lichtempfinden]: what we mean by it is bound up with thinking and judging. Newton did not discover the law of gravitation because his senses were especially acute.

We see the flower because it reflects light. But seeing that the flower has five petals is not a matter of Lichtempfinden: of sensitivity to light. On the contrary, it is a matter of grasping a thought, as we do in judgement, and thoughts do not emit or reflect any light for us to be sensitive to. Frege repeatedly uses the metaphor of perceiving to describe our grasp of thoughts, but makes clear in this passage that sense perception and thought have different sorts of object. Perception has objects which really do (in the visual case) emit or reflect light.

The passage from “The Thought” has been more widely discussed, because it forms part of a discussion of a topic that is key to the interpretation of Frege’s work as a whole. Frege is concerned throughout “The Thought” to argue that thoughts are not psychological entities (Frege 1918: 14 and passim). In particular, they are not what in the empiricist psychological tradition were called “ideas” [Vorstellungen]. Unlike material objects, ideas are non-perceptible: they cannot be seen, touched or tasted. But, unlike thoughts, they are had rather than grasped. They have an owner, and are private to that owner. Thoughts belong, by contrast, neither to the physical world nor, like ideas, to the mind. They belong to a third realm, mind-independent but non-physical. What, then, can be said about how we succeed in grasping them? Does Frege’s Platonism prevent him from offering a plausible account of what it is to grasp a thought?

Frege’s attempt to address this issue is puzzling. It involves drawing an analogy between perception and thought. He writes (1918: 26-7):

Having visual impressions [Gesichtseindrücke] is certainly necessary for seeing things, but not sufficient. What must still be added is not something sensible. And yet this is just what opens up the external world for us; for without this non-sensible something everyone would remain shut up in his inner world. So perhaps, since the decisive factor lies in the non-sensible, something non-sensible, even without the co-operation of sense impressions [Sinnesindrücke], could also lead us out of the inner world and enable us to grasp thoughts.

Unfortunately Frege does not specify clearly just what this “non-sensible something” is supposed to be. We can hope to get a clearer grasp of what he means only by examining his strategy in the passage. The problem, as he conceives it, is to describe how we can grasp thoughts given that thoughts are non-psychological entities. He then points out that not only thinking, but perceiving also, involves the transcendence of the merely psychological. Neither objects of perception (for example, a flower) nor objects of thought (for example, that the flower has five petals) are psychological entities. It is the objectivity – the non-psychological nature – of thoughts, not their immateriality, which poses a problem.

In the perceptual case there is, to be sure, a psychological entity involved. This is a sense impression, a sort of idea. But the occurrence of such entities is a necessary, not a sufficient, condition for perceptual experience to take place. What needs to be added to mere impressions in order to arrive at perception is “something non-sensible”. Then, Frege’s thought goes, if such a non-sensible factor can allow us to transcend psychology in the perceptual case, then perhaps it can allow us to do the ssame in the case of thought, where no sense impression is involved. But, at this point, the argument becomes frustratingly vague. What is the non-sensible something, in either case? It will be of some use to examine some attempts to answer this question.

I turn first to an interesting minority view. According to Malzkorn (2001: 45-6), what is required to transform mere subjective visual impressions into perceptual experiences is the processing of visual impressions. For Malzkorn, a visual impression is a retinal image (41-2). Looking at a tree, two tree-shaped retinal images occur in me. Somehow this is processed by me into a single visual image of a tree. An account of how this occurs (which, of course, Frege does not attempt) would provide an answer to the question of how we achieve perceptual contact with trees, and it would do so by positing something (a cognitive process) which is not itself an object of perception.

But this cannot be Frege’s view. The problem for Frege is how we move from the possession of psychological entities, conceived of as non-perceptible, private mental entities, to perceptual contact with external perceptible items. Retinal images are not private mental entities but, qua physiological, are external and perceptible. Thus, the progression from retinal images to unitary visual images is a progression in the opposite direction: from something external and perceptible to something internal and non-perceptible. Frege’s problem begins where Malzkorn’s account stops: with a visual image.

More commonly, it is thought (Dummett 1991: 273; Burge 2005: 301n5) that the “something non-sensible” must be a thought. What I must do in order to see a tree, beyond possessing a certain idea, is to grasp a thought about the tree. This interpretation makes Frege a representational theorist in the style of Peacocke: perceiving involves tokening a proposition representing the world as being a certain way. But the textual evidence for the interpretation is quite poor. Frege does not state that his “something non-sensible” is a thought, despite his using the phrase in the context of a discussion of thoughts, when it would have been natural to make the identification explicitly if that was what he had in mind. Further, it does not make exegetical sense. In the passage, Frege is attempting to defuse the worry that the objectivity of thoughts makes our grasp of them mysterious. He does this by pointing out that even in the mundane case of sense perception we apprehend something objective, that is, non-psychological. If perception had itself to involve the grasp of thought, the analogy would not serve to make thought less mysterious but on the contrary would make perception inherit the mysteriousness of thought.

One might cite the passage from the Logic discussed above in support of the Dummett-Burge interpretation. On this view, seeing that the flower has five petals involves tokening a thought content as well as having a mere visual experience. In order to take this view, one would have to identify having a visual experience with having an idea. But it seems to me that this would be a misreading of the 1897 passage. Frege’s intent there is different: his point is that a thought is not visible, and thus not an object of sense perception at all. The phrase translated by Long and White as “mere visual experience” is “bloßen Lichtempfindens“, mere sensitivity to light. The possession of ideas is not a matter of sensitivity to light, since ideas are immaterial and thus do not emit or reflect light. It is physical objects like flowers, the objects of perception, which require such sensitivity of us for their apprehension.

In any case, the Dummett-Burge interpretation does not make philosophical sense. Frege’s problem is how we transcend the psychological in either perception or thought: how we apprehend the non-psychological. It would be no answer to this question for the perceptual case to say that we do so by grasping something in the realm of thought. One cannot explain our apprehension of non-psychological items by merely positing further non-psychological items for us to apprehend. We are left in something of an impasse. It may help to remember that Frege’s purpose in this passage is to demystify thought by an analogy with perception rather than to explain perception: thinking of the non-psychological is no more inherently mysterious than is perceiving the non-psychological. Still, it raises a question for Frege’s view of perception which he fails to answer. How do we in perceptual experience transcend merely private sense impressions to apprehend external objects? With this question in mind, we can now return to the Tractatus.

Judgement, Perception and Appearance

The immediate context of remark 5.5423 is a discussion of the nature of judgement. Its antecedent in the Tractarian hierarchy is 5.542, which reads:

It is clear, however, that ‘A believes that p‘, ‘A has the thought p‘, and ‘A says p‘ are of the form ‘”p” says p‘: and this does not involve a correlation of a fact with an object, but rather the correlation of facts by means of the correlation of their objects.

Jose Zalabardo [this volume] reads this passage as qualifying Wittgenstein’s rejection of Russell’s multiple relation theory of judgement. Let some object a stand in a relation R to some further object b. What is it to judge that this is the case? Russell understands this question in a particular way, which clearly influences Wittgenstein. Russell asks: what is the nature of the complex fact created by a subject S judging that a stands in relation R to b? The burden of the multiple relation theory is that this complex should not be understood as a relation between S, on the one hand, and a unit made up of a and b, related by R, on the other. Rather, it consists in a relation between S, a, b and R, where these elements are bound together by the relation of judgement, rather than whatever relation is named by R.

According to Zalabardo, Wittgenstein objects to any account of judgement on which relations appear in the analysis without doing their relating work. On the Russellian analysis, R is a constituent of the fact created by my judgement, but does not work to relate the constituents of the fact to each other. Instead, the only relation that actually relates the constituents of that fact is the judging relation, and R is among the constituents which that relation relates. For Wittgenstein, this is incoherent, and the multiple relation theory fails.

In rejecting the multiple relation theory, does Wittgenstein return to a dual relation theory? If he did, his position might justifiably be called Fregean, since Frege conceives of judgement as a relation between a judging subject and an inhabitant of the realm of thoughts. According to Zalabardo, Tractatus 5.542 shows how Wittgenstein avoids this. On his analysis of the fact created by a subject’s judging that something is the case, the subject is not a constituent of that fact. In other words, the judging subject drops out of his analysis. Consequently, so do relations between subjects and anything else.

At this point it is worth considering Tractatus 5.633-5.6331, where Wittgenstein points to a dangerously misleading picture of the visual field, on which the eye is included at the limit of the field. Just as the eye is not seen in the visual field, the “metaphysical subject” is not found in the world. Wittgenstein is drawing an analogy between the visual field and the world, for the purposes of illustrating a certain point about the latter. But the point still holds for the visual field itself.

This suggests an analysis of appearances on which the perceiving subject disappears from the analysis. The cube appearing as it does to a subject is no longer what is in question. The visual field is completely describable by mentioning the objects referred to in our reports of propositional attitudes, and these do not include the subject. Further, once the visual field has been described, there is nothing that remains to be done by way of explaining appearances.

This may seem straightforwardly incoherent. The objection I have in mind would run: the notion of appearance is inherently subject-relative. There cannot be an appearance without someone to whom it appears. An account that ignores subjectivity is thus precisely not an account of appearances. Even if we grant the premise of this objection, it is not a valid objection to Wittgenstein’s account, but rather clarifies his point. What it shows is that, once we fully understand what is occurs when we view the cube, the notion of appearance will drop out of our analysis along with the subject.

In Wittgenstein’s terms, the question can be put in this way. Whether I see a square or a regular diamond depends on which fact (Sachverhalt) is seen. How then, should the visual field be characterised? Is it open to being described in more than one way, so that the same field may be described as containing a square or, alternately, a regular diamond? Or is it rather simply given by a particular description?

To take the first option would be effectively to revert to something like Peacocke’s Fregean view. One would be taking the visual field itself to be an opaque object which may appear in different ways. Wittgenstein’s appeal to the visual field was intended instead to explain opacity and transparency. It is clear, then, that we should take the visual field to be given and not merely described by the perceived Sachverhalten. Yet Wittgenstein’s anti-psychologism prevents us from taking this in what is perhaps the most innocuous and natural way. We often take visual fields to be psychological entities distinct from the external-world objects of our perceptual experiences. But, as we have seen, Wittgenstein’s notion of the visual field is precisely not psychological.

More radically still, Wittgenstein ultimately refuses the distinction between the visual field and the visible object. This can seem to lead to a sort of idealism or, perhaps better, solipsism. The world is as I perceive it to be. And, of course, this consequence is recognised or even embraced by Wittgenstein himself at Tractatus 5.64, at which he takes himself to have arrived at a form of solipsism in which “the self … shrinks to a point without extension”. The disappearance of the subject from the analysis of judgement and perception both vindicates the solipsist view and strips it of its substance.

We should also note the other point Wittgenstein makes at 5.64. His peculiar form of solipsism “coincides with pure realism”. The shrinking of the self leaves only “the reality co-ordinated with it”. This topic is much too large to be treated here, though some of its aspects will be discussed in section four. But one consequence of Wittgenstein’s doctrine might relevantly be mentioned. While according to the Tractatus the world is as I perceive it to be, it would be wrong to conclude that reality is constituted by appearances. The mistake is in presupposing an explanatory relation between appearances and reality, such that the latter is explained in terms of the former. Such a direction of explanation is ruled out by the “pure realism” of the Tractatus.

We still have not been told what sort of fact is in question here. Indeed, Wittgenstein is silent on this issue, for reasons I discuss in the final section. Some speculation may assist the reader in understanding the interpretation on offer here, however, so long as it is kept in mind that Wittgenstein himself refrains on principle from engaging in such speculation.

It could plausibly be supposed that the facts in question will be geometrical facts about points on the two-dimensional cube diagram. Since the diagram is two-dimensional, these will of course not include any facts about which points are in front of which others, or about how the diagram appears to a viewing subject. But they will be facts to which the subject is responsive, in that they help to determine how the subject views the cube.

Does Wittgenstein Face a Fregean Gap?

We saw in section two above that Frege is faced with a problem: given the externality of the objects of perception, what psychological facts about the subject of a perceptual experience can explain the subject’s perception of those objects? A trite but not inaccurate way to characterise the position of the Tractatus with respect to this problem is that it does not arise, since the subject drops out of the Tractarian analysis of perception.

We can do a little better than this trite response, however. One way to see the point is to draw a contrast between the views of Russell and Wittgenstein on our cognitive limits. For Russell (1984: 11-12), each of us can only understand those singular thoughts whose constituents belong to the domain of his or her acquaintance. Nevertheless, he thought, we can understand (and, indeed, know) a great deal about what lies beyond the domain of our acquaintance. This is because we can understand (and even know) general truths about the world, truths whose particular instantiations we cannot count as understanding.

Russell sees the solipsist as fallaciously moving from the (true) premise that one can only know those particular things which are within the domain of one’s acquaintance, to the (false) conclusion that one cannot know that there are things outside that domain. The fallacy depends on a failure to appreciate the possibility of purely general thoughts. We can know that there is no highest prime, and thus that there is a prime number greater than any given number, even if we do not know what number is both prime and greater. Thus the distinction between singular and general thought is for Russell an escape route from solipsism.

Wittgenstein denies this use of the distinction between singular and general thoughts. According to the Tractatus (see, for example, 5.3), all statements that I can understand are constructed truth-functionally from statements whose constituents are objects I know [kennen]. General thoughts do not range over any domain of objects which do not fall within my cognitive limits. (It should however not be assumed that Russell and Wittgenstein draw my cognitive limits at the same place: in particular it should not be assumed that Wittgenstein agrees that all that I can know are “sense data”.) Thus the ascent from the singular to the general cannot have the epistemological importance which Russell attributes to it. It appears, then, that for Wittgenstein, since the Russellian escape route has been blocked, we are each of us indeed caught within his or her own cognitive domain.

Wittgenstein faces this consequence with equanimity. At 5.61, he glosses the truth in solipsism as follows: “The world is my world: this is manifest in the fact that the limits of language (of that language which alone I understand) mean the limits of my world.” Logic does not lend us the resources to lift ourselves out of our cognitive domain: rather, the range of propositions I can understand is limited by the domain of objects which I can know.

Although an assessment of Wittgensteinian solipsism lies beyond the scope of this paper, there is one lesson we can draw. Because the subject is not set over against a domain of perceptible objects, there is no question of a gap between the psychological and the non-psychological domains. There is, however, a distinction which in a manner goes proxy for that distinction: that between the opaque and the transparent. This is a logical distinction and not a (epistemological) distinction between domains of objects.

Here I draw on some aspects of Peter Sullivan’s essay “Identity theories of truth and the Tractatus” (Sullivan 2005, see especially 58-60). Sullivan interprets 5.5423 as providing the Tractarian account of externality. It is characteristic of external objects that they display what Sullivan calls “cussedness”. They are not transparent to the apprehending subject; instead, one can be acquainted with them without knowing them through and through. External objects present unexpected aspects to the viewer. By contrast, the Tractarian simples precisely are transparent to the subject, and so cannot constitute genuinely external objects.

The facts which are seen when one sees the cube are, in Wittgenstein’s terminology, Tatsachen. Tatsachen are complex facts built truth-functionally from Sachverhalten, which are themselves concatenations of simples. Sachverhalten themselves completely specify the content of any thoughts about them. They are not the sorts of thing on which one can have different perspectives. Sullivan’s thesis is that the transparency of Sachverhalten does not transfer to Tatsachen.

To see a complex object like the cube is to see some Tatsache or other. But there is more than one Tatsache which is such that, in seeing it, I count as seeing the cube. This is precisely what makes the cube complex. Further, the possibility of seeing different Tatsachen is explained by the multiplicity of Sachverhalten at a lower level of analysis. Thus, relatively simple facts, at one level of analysis, explain the appearance, at another level, of complex objects, and thus the possibility of taking different perspectives on things. We can take different perspectives on cussed objects like the cube because of the existence of different cognitively transparent facts at the lower level.

If we take seriously Sullivan’s association of complexity with externality, we can see the truth-functional relation between Sachverhalt and Tatsache as offering a bridge between the internal and the external. In sum: though for Wittgenstein there is no gap between the psychological and non-psychological, there is an analogous distinction to be drawn. This is the distinction between the transparent and the opaque or multi-faceted. The work that in Russell’s account is done by the distinction between singular and general thought is done for Wittgenstein by the distinction between elementary propositions and propositions which are truth-functionally complex.

Does the Tractatus Contain a Theory of Perception?

William Child (2013: 150) writes, “It is agreed on all sides that the Tractatus is not concerned with epistemology: it contains no theory of perception, for example … and it says explicitly that theory of knowledge has no special or foundational place in philosophy (4.1121).” Child’s claim is largely correct: my intention here is not to dispute it but rather to test its limits. As I see it, the Tractatus does contain a substantive philosophical view on the nature of perception, though not one that deserves to be called a “theory”.

Remark 4.1121 states both that “theory of knowledge is the philosophy of psychology” and that “psychology is no more closely related to philosophy than any other natural science”. As Child says, Wittgenstein’s purpose here is to deny that epistemological questions are foundational for philosophy. Epistemological questions fall within the domain of philosophy of psychology, and to whatever extent philosophers may take an interest in psychology, in doing so they are not addressing philosophical questions which are any more fundamental than those addressed by philosophers interested in other scientific disciplines. But to say this is not to say that epistemology (or perception) does not fall within the domain of legitimate philosophical interest. Thus, we can legitimately ask whether the Tractatus encompasses a philosophy of perception, though we must admit that its view of perception will not be fundamental to its philosophical doctrine taken as a whole.

There is another reason to suppose that Wittgenstein may have conceived of perception as outside the reach of the Tractatus, however. In correspondence during August 1919, Russell asks what the constituents of thoughts and facts are. Wittgenstein (L 72) responds: “I don’t know what the constituents of a thought are but I know that it must have such constituents which correspond to the words of Language. Again the kind of relation of the constituents of thought and of the pictured fact is irrelevant. It would be a matter of psychology to find it out.”

In this carefully worded response, Wittgenstein makes two distinct points. The first is that he takes himself to know, on a priori grounds, that thoughts have constituents which map onto items in the world. The second is that it is the task of psychology, not philosophy to discover just what these constituents of thought and worldly entities are, and it is further a psychological matter to discover how the mapping is established. One of the lessons of remark 5.5423 is that the point goes for perceptual experience as well as for thought. Perceptual experiences have constituents which map onto worldly entities (we do not know which worldly entities), and the establishment of the mapping is a question for psychology.

Let us apply this thought to the analysis of appearance, as discussed in section three above. On Wittgenstein’s account, I argued there, appearances are to be explained in terms of the mapping of mental states onto items in the world. No reference is to be made, in the analysis of appearances, to ways in which worldly entities appear. The present point is that how this mapping is established is a psychological matter.

Whether this analysis amounts to a theory of perception depends on what one requires of such a theory. Plausibly, such a theory should tell us why things appear to us as they do. By this plausible criterion, Wittgenstein neither offers nor attempts a theory. Nevertheless, his remarks do entail a substantive philosophical view about perception. The quickest way to see this is to notice the philosophical views which Wittgenstein’s account rules out.

One of these is the view that appearances are themselves objects of perception, mediating between us and worldly entities. To think this would be to make a mistake about the notion of appearance. Appearances are explained by pointing to perceived facts; appeal to appearance plays no part in the explanation of perceptual phenomena. A related consequence of the account is that, as we have seen, Wittgenstein escapes the problem that in section two I attributed to Frege: the problem of explaining our capacity to bridge the gap between appearance and reality.

His account has, as a further consequence, a more general philosophical thesis. This is the view that a treatment of the visual field can be approached via the theory of judgement. If the argument of this paper has been anywhere near correct, Wittgenstein’s explanation of the possibility of perceiving external objects runs precisely parallel to his explanation of the possibility of thinking complex thoughts, or making complex judgements. Part of the Tractatus view is that there is no room for a distinct philosophical theory of perception: an adequate philosophy of perception is in a sense already present in a theory of judgement.

This attitude towards perception is entirely overthrown in Wittgenstein’s later work, in which he recognises that “The concept of ‘seeing’ makes a tangled impression” (PI ii 200). Part of the tangle arises from the conceptual richness of the perceptual phenomena he discusses there. We see a posture ,as hesitant or a face as timid (209), or hear a sequence of notes as a variation on a theme (213). It is clear that the explication of such phenomena will have to delve deeply into human psychology, in ways that the Tractatus account of the Necker cube does not. It may be that in the later as in the earlier work Wittgenstein stops short of offering a theory of perception. But he no longer does so because he supposes that the work of a theory of perception has already been done by the theory of judgement.

Burge, T. (2010) The Origins of Objectivity, Oxford, Oxford University Press.

Child, W. (2013). “Language argument?” in Peter Sullivan, Michael Potter (eds), Wittgenstein’s Tractatus: History and Interpretation, Oxford: Oxford University Press.

Dummett, M. (1991) “Thought and perception: the views of two philosophical innovators” in Frege and Other Philosophers, Oxford: Clarendon Press.

Frege, G. (1892) “On sense and reference”, reprinted in Michael Beaney (ed.) The Frege Reader. Oxford: Blackwell, 1997.

Frege, G. (1893) “Logic”, reprinted in Beaney.

Frege, G. (1918) “The thought”, reprinted in Beaney.

Malzkorn, W. (2001) “Demythologising the third realm” in A. Newen, U. Nortmann and R. Stuhlmann-Laeisz (eds) Building on Frege, Stanford: CSLI Publications

Peacocke, C. (1992) A Study of Concepts, Cambridge, Mass.: MIT Press.

Russell, B. (1984) Theory of Knowledge: The 1913 Manuscript, London: Routledge.

  1. Sullivan, P. (2005) “Identity theories of truth and the Tractatus”, Philosophical Investigations vol. 28 no. 2, pp. 43-62.


In any case Wittgenstein apparently held a low opinion of “The Thought”, remarking to Peter Geach that it was not worth publishing. See Geach’s preface to Frege 1918, vii.