The notion of the “visual field”, though useful in everyday talk and perhaps essential to psychology, is a difficult one to understand. Here I attempt to throw some light on the notion by exploring the similarities and contrasts between two sets of remarks on the notion of the visual field: one offered by Bertrand Russell in the abandoned 1913 Theory of Knowledge manuscript; the other by Wittgenstein in the Tractatus.

As is well known, the 1913 manuscript was abandoned by Russell under the pressure of Wittgenstein’s criticisms. The precise content of those criticisms, however, is a matter for scholarly speculation. They appear to have centred around the notion of judgement. I make no large claim for the significance of the contrast I draw here for this historical issue. But the contrast does relate very closely to Russell and Wittgenstein’s differences on the issue of judgement. Even if the contrast between their views on the visual field did not itself form part of Wittgenstein’s criticisms, it seems reasonable to surmise that it can inform our view of their disagreement.

My thesis is as follows: Wittgenstein’s account of the visual field resolves tensions which are internal to Russell’s account. Wittgenstein’s view can thus be seen as a development of Russell’s position.

One can speak of auditory and tactile fields, in addition to the visual field. Like Russell and Wittgenstein themselves, I shall concentrate primarily on the visual case. Russell tends to assume, like many other philosophers, that his conclusions about the visual field will simply transfer to other sensory modalities. He does make occasional reference to the auditory and tactile fields, but the emphasis is on the visual. Whether or not this assumption is warranted, however, the visual field is of independent interest, and it is also the case in which Wittgenstein exhibits the most interest.

It may be doubted, however, that the visual field was a topic of particular interest to Russell at all, or that he was in any sense attempting to construct a theory of the visual field. In section one, in order to address this objection, I describe Russell’s project in the period in question, and explain why the issue of the visual field arises for this project. In section two I explain Russell’s account of the visual field, and a particular difficulty arises for it. In sections three and four I depict Wittgenstein’s alternative picture.

Wittgenstein’s alternative hinges on the idea of analysing the visual field in terms of facts rather than of objects, and results in a picture of the field which is, in a sense I will endeavour to explain, depsychologised. But, I will argue, that it is the first point that makes Wittgenstein’s analysis crucially different from Russell’s.

§ 1. Russell’s project

In order to understand Russell’s interest in the visual field in the Theory of Knowledge manuscript, it is necessary to enter into some discussion of his broader project during the period when it was written. He himself characterised this project as demonstrating how matter was to be constructed from sense data. In writings of 1914, especially the article “The relation of sense data to physics” and the Boston lecture series published as Our Knowledge of the External World as a Field for Scientific Method in Philosophy, Russell enters into some detail as to how the construction of matter is supposed to proceed. The following sketch is by no means a full account of the construction: it is intended merely to show what role the visual field plays in this area of Russell’s thought.

A major concern of Russell’s writings around 1913 and 1914 was to provide an account of the relation between the world of sense experience and the world as described by science. Russell’s motivation is in large part epistemological: he wishes to explain how it is that scientific knowledge is possible. But this aim in Russell’s hands is not a merely conservative one, of showing that we indeed know what we had already taken ourselves to know.1 Instead, he is prepared to demarcate real from merely apparent knowledge, and also to reinterpret what we know so that the content of scientific knowledge looks different from what we had taken it to be.

In his The Problems of Philosophy, written during 1911, Russell had treated our knowledge of the external world as obtained by causal inference from the data of sense experience.2 But he soon rejected this early assumption, and by early 1914 could write that “the supreme maxim of scientific philosophising is this: Wherever possible, logical constructions are to be substituted for inferred entities.” Characteristically, Russell’s justification of this maxim is an epistemological one: constructions are “less doubtful” than inferences.3

In the period with which we are concerned, then, Russell was committed to the view that matter can be constructed logically either from sense data (the immediate objects of experience), or sensibilia (objects of the sort which are apt to be immediate objects of experience, but which may not in fact be such).4 Given his reformist ambitions in epistemology, Russell is not committed to treating all common sense objects in this way, but he is committed to vindicating by logical construction those entities posited by confirmed science.

It is well known, and was often repeated by Russell himself, that his rejection of idealism was a seminal moment in the development of his thought. The nature of this rejection is, however, easy to misunderstand. Russell’s anti-idealism does not consist in the view that there is in the world not only sense data but also matter, so that the question of whether we can know whether there is matter is the question of whether idealism is true. On the contrary, Russell denies that sense data are themselves mental entities.5 He conceives of the mental as that which is the subject of awareness: that is, takes up the subject position in awareness relations. Sense data, as he conceives them, are rather objects of awareness, and thus not mental.

Russell is not, in the writings with which we are concerned, primarily focused on the question of the truth or falsity of idealism. Rather, the construction of matter out of sense data is, if successful, a vindication of at least some of our scientific knowledge. Further, the sciences thus vindicated include psychology as well as physics. Since matter is constructed out of sense data “[t]he discrepancy between the world of physics and the world of sense … will be found to be more apparent than real”.6 The fundamental difference between physics and psychology is that they examine the same phenomena from different points of view: psychology relates those sense data which are objects of awareness for the same subject; physics relates the momentary experiences of different subjects.7

The aim of Russell’s writings in this period is an account of the construction, from sense data, of entities of the sort studied both by physics and by psychology. The perceiving subject is a construct just as a table is. Implicit in this view is a further thesis: that the distinction between appearances and the things that appear is not a fundamental one. Sense data are themselves neither one nor the other. The construction of the self and of the external world from those sense data will however encompass a distinction between classes of sense data that we may call things and that we may call appearances. Physics studies the first and psychology the second.

Two distinct sorts of space are constructed: Russell calls them “private space” and “perspective space”.8 Each private space constitutes a perspective. The perspective space is the space occupied by many such perspectives, and it is at the level of perspective space that external objects are constructed. Thus the nature of perspective space is the particular concern of physics.

Our concern is, however, with the construction of private spaces themselves. Once again, this proceeds in stages. Russell remarks that “[p]eople who have never read any psychology seldom realise how much mental labour has gone into the construction of the one all-embracing space into which all sensible objects are supposed to fit”, speaking here not of a shared public space but one which is specific to a given subject.9

First, there is the construction of multiple sensory spaces, each corresponding to a sensory modality. An entire visual scene is constructed from visual sense data. Second, there is the correlation of these spaces during infancy through a process of learning. One learns, for example, to correlate the sight of a table with the tactile sensation of hardness. Thus the construction of the visual fields, and of the other sensory fields, forms a stage in the construction of the external world.

§ 2. A Russellian difficulty

The notion of the visual field is, from the point of view of common sense, ambiguous. It can mean that portion of the environment which is visible to a given perceiver at a given time. Or it can instead mean a psychological entity, something like the environment as it appears to a perceiver at a time.

If we assume the common sense distinction between appearances and things that appear, we will find that the visual field qua psychological entity is more finely grained than the visual field qua portion of the environment. The same environmental object can appear in more than one way: thus to each inhabitant of the environment can correspond more than one appearance. The familiar Necker cube, for example, is a single environmental object. But it can be seen in two distinct ways. Thus a description of the contents of the visual field qua psychological entity requires a distinction which a description of the visual field qua portion of the environment does not.

As we saw in section one above, Russell cannot take for granted the distinction between appearances and things that appear, so he cannot appeal to that distinction to understand the visual field. Let us now see how he does understand it.

Russell discusses the visual field at some length in the abandoned Theory of Knowledge manuscript, composed during 1913. Indeed, Russell’s views on judgement, the main topic of the manuscript, are closely bound up with his views on perception. It seems reasonable to surmise that Wittgenstein’s remarks on the visual field in the Tractatus may themselves be partially a reaction to Russell’s views, since it is known that his views on judgement arose from such a reaction. In any case, the contrast between the two accounts is a revealing one.

The visual field, for Russell, is not itself given in perceptual experience but rather is inhabited by the objects which are so given.10 The objects that are first given are the objects of selective attention. Our first consciousness is of objects as objects. Only by reflection on such objects do we come to conceive of the visual field itself. “The ‘sensible continuum’ of psychologists endeavouring to undo the work of thought is itself a late discovery of thought. All primitive consciousness is selective…” For all that, the visual field is a discovery and not an invention of thought. It is the domain from which the objects of which we are conscious are selected. Russell conceives of the investigation of the phenomena of consciousness and attention as the investigation of the visual field.

The conception of the visual field as inhabited by visible objects is important to Russell, and has a bearing on his theory of judgement. Russell in this period advocated a multiple relation theory of judgement, according to which judgement consists in a relation between a judging subject and several other objects. The theory is well-known, but a brief synopsis of it, insofar as it bears directly on our topic, may be useful. Suppose that I judge that some object a stands in a relation R to b. Russell was concerned to avoid supposing that I thereby bear a dual relation to some unitary object that could be characterised by the proposition aRb, largely because it seemed to him that unless a does indeed stand in R to b, there could be no such object. Thus a dual relation theory would make it impossible to judge a falsehood.

Instead, Russell proposed in 1910, the judgement consists in a multiple relation involving, at least, the judging subject, the objects a and b, and the relation R. The story of the development of Russell’s thought on judgement after 1910 is in part the story of the attempt to find a stable list of relata which could satisfactorily account for the facts of judgement.

The connection between this theory of judgement and the issue of perceptible complexes is as follows. Suppose that I judge that the knife is to the left of the fork.11 According to the simplest version of the multiple relation theory, which Russell still held in 1910, this judgement consists in a relation between me, the knife, the fork and the to-the-left-of relation. The judgement is fallible because, although this relation could not occur unless all these relata existed, it could occur were the relata differently disposed towards each other. An example would be if the fork were to the left of the knife. Perception, however, is infallible. If I see the knife to the left of the fork, then they are indeed disposed towards each other in that way. According to Russell, we have here a dual relation between a perceiving subject and a unitary, perceptible complex object: knife-to-left-of-fork. Since there are no false perceptions, the problem that arose for dual relation theories of judgement does not arise here.

In analysing a perceptible complex, we come to a criterion for the truth of certain judgements. My judgement that the knife is to the left of the fork is true if and only if there is such a complex as the knife-to-left-of-fork. If I arrive at a false judgement through perception, this shows that I have misanalysed a perceived complex, and thus wrongly identified its constituents or the way that they are disposed towards each other.

As we have seen, in analysing perceptible complexes we are investigating the contents of the visual field. In order to understand how Russell conceives of this investigation as proceeding, we need to consider some aspects of his overall project in this period, and in particular how he conceives of the procedure of “analysis”.

Russell took himself to be engaged in an epistemological task; Wittgenstein’s concerns in his early period are resolutely non-epistemological. But there is a danger of overstating this contrast. Because of certain specific Russellian doctrines, what Russell took to be the major task of a theory of knowledge to a great extent falls into what we might understand as philosophy of language, or, perhaps better, a theory of understanding.

We can put the point, first, in terms of knowledge of objects, rather than knowledge of facts. That is, one thing that a theory of knowledge aims to do is to account for our having knowledge of, as well as about, certain objects. Russell takes the distinction between knowledge by acquaintance and by description to be key to the construction of a theory of knowledge. Acquaintance is a primitive knowledge relation: the bare knowledge of something we gain by being acquainted with it is not in need of explanation. What calls for a theory of knowledge is rather our apparent knowledge of things with which we are not acquainted.

Such problems are to be solved by correctly stating the relation between objects of acquaintance and the other things in the world. Thus London is not, for Russell, an object of my acquaintance, but my knowledge of London, if genuine, depends on my knowledge of some object of my acquaintance.12 Suppose, for instance, that the reason one cannot be acquainted with London is that it is too big (this is not precisely what Russell thought, but will serve for clarificatory purposes). In that case, we might suppose that an object of acquaintance, in virtue of which I count as knowing London, is some part of London. Let us suppose, then, that while one cannot be acquainted with London as such, one can be acquainted with Mornington Crescent.

At this point there remains the problem of showing how my knowledge by acquaintance of Mornington Crescent is apt to yield any knowledge at all of London. That Mornington Crescent is part of London may play a role in the explanation; so might properties of Mornington Crescent in virtue of which it is recognisable as part of London, and so on. But, for Russell, a correct statement of the relation between objects of acquaintance and other objects of knowledge will also be a statement of the relation between the names of objects of acquaintance and expressions for the other things.

The point may most easily be put in terms of sentences. Sentences about, say, material objects turn out on analysis to have a form quite different from their apparent form. No singular reference to a chair occurs in an a fully analysed sentence about the chair. The result, for Russell, is that what had previously seemed impossible demands on our knowledge come on analysis to seem capable of being satisfied. In a precisely parallel way, our understanding of certain sentences had appeared mysterious, but now seems intelligible.

A new theory of how to understand sentences about the material world just is a theory of how we come to know it. This is already determined by Russell’s view that only possible objects of acquaintance can be named, and that the apparent names for other possible objects of knowledge will reveal their real form on analysis. Again, given Russell’s thought that objects of acquaintance are transparent to the subject, problems of knowledge do not arise for such objects. It follows that a correct assessment of our capacity for knowledge will be achieved by understanding the relation between such objects and other objects of knowledge.

The theory of knowledge is thus advanced by the analysis of sentences. It is important to be clear on what it is that such analysis, successfully completed, reveals. It reveals which objects are complexes and, for those complexes, what their constituents are, and how those constituents are disposed to each other. Thus Russell says: “Analysis may be defined as the discovery of the constituents and the manner of combination of a given complex.”13

As we shall see in more detail shortly, we can be acquainted with complexes as well as simples. Thus, from the point of view of epistemological explanation, analysis can work in either of two directions. It can show how I can have knowledge of a complex in virtue of having knowledge of its constituents. But it can also show how I can have knowledge of constituents in virtue of having knowledge of a complex.

The latter point turns out to have particular importance for Russell’s ideas about the visual field. For, in many if not all ordinary perceptual cases, we precisely are acquainted with complex objects. We may take Russell’s own example:14


Call this token of the letter T (that you can see on paper or on screen) λ. Call the vertical stroke a and the horizontal stroke b. Now, it is the case that a, b and λ are all possible objects of your acquaintance. But a and b are constituents of λ: it is their being disposed to each other as they are that constitutes it.

Because of this, a peculiar problem arises for Russell. As I have emphasised, acquaintance is a simple cognitive relation. Once the object of acquaintance has been identified, the relation itself is specified. Now, suppose that I am acquainted with some complex, by virtue of being confronted with it, as we are in perception. Suppose also that I have arrived, through a process of analysis, at knowledge of the constituents of that same complex. Russell’s problem is then this: how do I know that the complex object of my acquaintance is the product of the analysed constituents?

This problem may not at first sight seem very compelling. But I think Russell is right to raise it, and that understanding why will help us to see some persistent problems in his account. Suppose some object λ consists of two further constituents a and b. I am acquainted with λ. I am also acquainted with a and b. All three acquaintance relations are transparent: in specifying the object we also specify what I know. Now, the acquaintance relation is in itself sufficient to yield knowledge of each of its objects individually. Thus, I have knowledge of λ, a, and b. Nevertheless, certain specifiable knowledge, required to yield knowledge of how λ is constituted, is lacking. Here are two ways of characterising the knowledge that I lack. The point is that, on Russell’s assumptions, these two ways are equivalent. First, my knowledge of a and b is not sufficient to yield knowledge of how a and b are related. Second, my knowledge of a and b is not sufficient to yield knowledge that a and b together constitute λ .

Insofar as I lack such knowledge, I do not count as knowing that λ is constituted by a and b, at least insofar as it is understood that λ, a and b are known by acquaintance. We might say that I do not have de re knowledge, of λ, that it is constituted by a and b.

Our first formulation of my ignorance already suggest what needs to be added to make up the lack. I must be acquainted with, as Russell puts it, the relatedness of a and b.15 Russell is clear that this relatedness is distinct from the “abstract” relation, which even in 1910 he was content to include among the constituents of complexes. The point of distinguishing the two formulations should now be clear: being acquainted with this relatedness of a and b just is knowing that λ consists of a and b.

Here we come to a crucially important point. Let us call whatever relation a stands in to b R. Thus, if a is on top of b, R is the relation on-top-of; if a is in love with b, R is the relation in-love-with. Russell thinks of such relations as themselves possible objects of acquaintance.

Allowing all this, we come to two Russellian theses:

(i) Acquaintance with R is necessary for knowledge of the relatedness of a and b.

(ii) Acquaintance with a, b, and R is not sufficient for such knowledge.

Thesis (i) depends on the general Russellian view that in order to understand a proposition I must be acquainted with its constituents. The upshot is that the relation is itself taken to be a further object, along with a and b, in the complex λ . Thesis (ii) arises from the thought that the relatedness of a and b is not such an object. Even if the relation is a constituent of the complex, the relatedness is not.

This may already remind us of a characteristic Wittgensteinian criticism of Russell’s theory of judgement, and thus his philosophy of mind and language. The criticism stems from Wittgenstein’s insistence that the relation between a and b is not a further object but rather a fact. We will return to these criticisms presently, but it is important to see that there is a tension even within Russell’s own position. Criticisms in Wittgenstein’s style are not merely external to Russell’s own view.

Russell is faced, on his own terms, with a dilemma. Is our knowledge of the relatedness of a and b a further example of object knowledge? If he supposes that it is, we risk losing the insight expressed in thesis (ii). Even allowing that the relation R is both a constituent of the complex and an object of acquaintance there is a difference in category between it and the relatedness of a and b.

But what, then, is the alternative? We could suppose that the relatedness of a and b is not an object but a fact: the fact that a and b are related. But if we do so we lose the idea that knowledge of the inhabitants of the visual field is a simple cognitive relation between the perceiving subject and objects of acquaintance.

In fact, Russell opts for a third possibility.16 The relatedness of a and b is indeed an object. However, it is not a further constituent of the complex λ. Rather, it just is that complex itself. “When there are two strokes in a certain spatial relation, there is a T; the T consists of these two strokes so related; the T is these two strokes so related.” But this is an unstable compromise. As we saw above, acquaintance with λ is a simple cognitive relation. It is one that holds when λ is given to me in perceptual experience. It does not require attention to the constituents of λ, nor does it require me to analyse, much less to analyse correctly, what those constituents are.

Russell has fallen into a view which takes the inhabitants of the visual field to manifest to us their own structure. If this meant merely that the constituents of objects in the visual field were themselves visible, this might not be objectionable. But if this was what is meant, all that could legitimately be concluded is that acquaintance with the complex allows for a knowledge of the relatedness of the parts, not that it is such knowledge.

Russell himself acknowledges that we can make a distinction between attention to a complex and attention to its parts, and thus that a complex can be given without its parts being so. But he has argued himself into a position which is directly in tension with this, according to which complexes in the visual field display their own structure, in such a way that acquaintance with them just is knowledge of that structure.

A non-Russellian, but apt, way to put the point would be to say that Russell has begun with a conception of the visual field as a domain of visible complex objects. It is this conception that is required for his account of the truth of perceptual judgements. But he has ended with a conception of the visual as a domain both of objects and of facts, such that knowledge of λ is equated with knowledge that a and b are related.

§ 3. Wittgenstein contra Russell

Perhaps the best known reference to the visual field in Wittgenstein’s early work occurs at Tractatus 5.633-5.6331. The main purpose of this reference is to make a point, by analogy, about the world. The relation between “metaphysical” subject and world is compared to the relation between eye and visual field. Just as an exhaustive description of the visual field will not include reference to an eye, an exhaustive description of the world will not include reference to a subject. And “nothing in the visual field allows you to infer that it is seen by an eye.”

Wittgenstein is making a distinction between being a limit and being a component. The eye is not a component of a complex that constitutes the visual field; nor is it a relatum in a relation that constitutes it. We will return to this passage after a brief consideration of (what I will interpret as) Wittgenstein’s remarks on what the visual field is.

At Tractatus 5.5423 Wittgenstein gives an account of the perception of complex objects, according to which “to perceive a complex means to perceive that its constituents are related to one another in such and such a way”. The ambiguity of theNecker cube is explained in terms of the availability to perception of different facts. There are two possible ways of seeing the cube because there are two different facts to be seen. In drawing out the consequences of this latter passage, we can shed some light also by what is meant by the first, which I will contend is a substantive remark about the visual field and not merely inserted for the purposes of an analogy.

As 5.5423 shows, Wittgenstein, against Russell, sees the visual field as consisting outright in visible facts. This means that a specification of a certain range of facts will give a full account of the contents of the visual field. It will also, according to Wittgenstein, yield the perceptible complexes. It is no longer correct to say that the visual field is inhabited by perceptible complex objects. We should say, instead, that the perceptibility of complexes is constituted by the visual field being as it is.

Russell takes the alternative to a multiple relation theory of judgement to be a dual relation view, whereby judgement consists in a relation between a judging subject on the one hand and a judgeable content on the other. In the case of perception, on the other hand, there is no need for the relatum to be itself judgeable. The relatum is just a complex object, not something capable of being evaluated as true or false.

The contrast with Russell might be seen as follows. How might Russell deal with the case of the cube? Here we have a perceptible complex, and thus apparently an object of acquaintance, which can be viewed in two ways. The two ways are naturally described in terms of seeing-as. We see the cube as extending towards the viewer, or as receding from the viewer.

This is just the sort of case to bring trouble for Russell over how the visual field is to be characterised. For if the way the visual field is determines not only that the cube is seen but in which way it is seen, then the visual field consists not in objects but in something more finely grained than objects. A natural way to interpret this is to say that it consists in possible interpretations of objects. But we are then placed in the position of taking the visual field to consist in self-interpreting objects. Facts are indeed more finely grained than complex objects. There are more facts about how the constituents of complexes are disposed towards each other than there are complexes. Turning from object-perception to fact-perception thus in a sense solves Russell’s problem.

It may help here to summarise both the comparison and contrast between Russell and Wittgenstein. Wittgenstein has retained Russell’s idea of analysis: that understanding what is meant by an adequate ascription of a perceptual experience to a subject yields a specification of the contents of the visual field. But he has modified Russell’s conception of how analysis proceeds. Instead of ending with a specification of the way in which simple objects are combined into complex, we have a specification of truth-functionally related facts.

It is this conception of the visual field that allows Wittgenstein to apply the notion of logical impossibility to the visual field directly:

6.375: Just as the only necessity that exists is logical necessity, so too the only impossibility

that exists is logical impossibility.

6.7351: For example, the simultaneous presence of two colours at the same place in the

visual field is impossible, in fact logically impossible, since it is ruled out by the logical

structure of colour.

If the visual field consisted of objects, one could at best say that two descriptions of the visual field were logically incompatible if they ascribed two colours to the same part of the field. Logical incompatibility does not exist as between objects. But then the question would arise of what it was about the visual field that made these descriptions incompatible, and one would have to posit that different colours excluded each other necessarily without that necessity being a logical one.

If this is the right way to understand the passage, it appears at first sight to leave us with two ways to read the notion of the visual field in Wittgenstein, corresponding to the ambiguity in the common sense conception of the visual field mentioned at the beginning of section two above.

(i) We might suppose that, for Wittgenstein, the visual field is a psychological entity. But, for Wittgenstein, the perceptibility of a complex is just constituted by the state of this psychological entity. The world external to the viewing subject falls away as irrelevant, and we have arrived at a sort of idealism.

(ii) Alternatively, we might suppose that Wittgenstein has tacked to the other extreme. He is thinking rather of the visual field as a portion of the environment: that is, Wittgenstein’s visual field is composed of the facts about the environment in virtue of which the viewing subject has the perceptual experience she has.

The visual field, on this view, is a portion of the external world, individuated relative to the viewer. But since the world is now conceived of as consisting of finely-grained facts, rather than coarsely-grained objects, the conception holds out the prospect of avoiding the introduction of a further level of proto-propositional content.

What is striking is that nothing in the passage itself determines a choice between these opposite alternatives; it is not quite obvious that anything in the context of the passage does, either. But, I will now argue, the immediate context, understood as contrasting with Russell’s view, shows that neither of them is correct.

§ 4. Wittgenstein’s alternative

The context in which the Tractatus remarks on the cube occur is Wittgenstein’s own account of judgement. His ontology of facts effectively rules out a multiple relation theory of judgement like Russell’s. It is precisely unitary facts which propositions picture. This may seem to lead us directly back to a dual relation theory. But Wittgenstein makes a quite different move. Judgement is not a relation between subject and fact, because the subject drops out of the relation: once judgement is analysed, the subject no longer appears as a term in the analysis.

At 5.541-5.5421, in a passage denying that propositions occur in each other in any way which is not truth-functional, Wittgenstein turns to propositions like ‘A believes that p’, ‘A has the thought that p’ and so on. Prima facie, these are cases in which p is not a truth-functional component of the complex proposition in which it appears.

if these are considered superficially, it looks as if the proposition p stood in some kind of relation to an object A. …

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.

The discussion of the cube appears immediately afterwards, and in a subordinate remark. It is clear that the point is meant to carry over from judgement to perception. Just as the judging subject drops out of an analysis of judgement, the perceiving subject drops out of an analysis of the visual field.

This is reflected in the famous diagram of 5.6331. The eye is not included in a representation of the visual field: it is a limit of the field, not a part. Thus a complete characterisation of the visual field does not mention the eye. The eye’s not appearing in the representation is analogous to the subject dropping out of the analysis.

An interpretive difficulty arises at this point, however, since at 6.4311 Wittgenstein writes “[o]ur life has no end in just the way in which our visual field has no limits [wie unser Gesichtsfeld grenzenlos ist].” The passage suggests that I have just overplayed the analogy between world and visual field: the subject may be the limit of the world, as Wittgenstein writes at 5.632, but the visual field has no limits.

By pressing the analogy, however, we come to an alternative interpretation of 6.4311. To say that the subject is the limit of the world is not to say that the world has this in it, but not that. As Wittgenstein writes at 5.61, I cannot draw a boundary between what is in my world and what is not. Analogously, I cannot see the bounds of my visual field. From my perspective the visual field as it were goes on forever.17

Lucy O’Brien has, similarly, argued that in these passages of the Tractatus perceptual representation acts as a model for representation generally.18 Just as the perceiving subject does not perceive itself, the representing subject does not, quite generally, represent itself. The exclusion of the perceiving subject from the visual field is thus of a piece with the exclusion of the self from its world.

It is far from clear just what implications this view has for our conception of the visual field itself. I will suggest that these implications include a rejection of options (i) and (ii) of § 3 above.

To suppose that the perception of facts consisted in a relation between subject and facts would be to suppose that a characterisation of the visual field requires reference to a perceiving subject, and to suppose that would be to suppose, as it were, that the eye is part of the visual field rather than its limit. But what does it mean to deny these suppositions? I will suggest that we an draw two conclusions, and contend that they rule out options (ii) and (i) respectively.

According to option (i), the visual field is part of the subject’s psychological economy. Take for example, looks ascriptions, which specify how an object looks to a subject, and thus how that subject sees the world. They are essentially subject-relative; they specify how a subject represents the world as being. If the subject drops out of the picture, then so does the distinction between how an object is and how it looks. Wittgenstein’s visual field, insofar as it is not a visual field for any subject, is by the same token not a psychological entity. Option (i) is therefore ruled out.

What of option (ii), according to which the visual field is to be identified with a portion of the environment? Such a portion is precisely not part of the subject’s psychological economy. It is part of the external world, a part which is appealed to in explanation of why the world looks as it does to a subject. Does such externality make it a suitable candidate to be identified with Wittgenstein’s visual field?

Further problems arise if we take this option. In the terms of the Tractatus, the field qua portion of the environment can only be some range of that totality of facts which constitutes the external world. But how is this range itself identified? It includes those facts which stand in an appropriate causal relation with the experiences of the subject.

The problem here is not merely that the causal relation by which the stimulus field is individuated has the subject as a term. The real problem is that the very notion of such a causal relation has been undermined. The relation is supposed to play an explanatory role, but this explanatory role has already been undercut, because reference to the subject and thus the subject’s point of view has been dropped.

Both of these conclusions can be seen as tending towards the depsychologisation of the visual field. But the second is considerably more radical than the first. The first leads to an anti-psychologism about the visual field in roughly the sense that Frege was anti-psychologistic about semantic content. It denies that the visual field belongs to a private mental domain of any subject. The second is anti-psychologistic in a more radical sense. The perceptible facts in the world are not to be identified by any relation, causal or otherwise, that they stand in to a private mental domain. Taken together, these two theses entail that Wittgenstein’s visual field does not fit either of our two options.

A more general moral may be drawn. For the subject to drop out of an analysis of the visual field is for the distinction between appearance and that which appears to lapse. The distinction, that is, is external to a characterisation of the visual field itself. In this respect Wittgenstein is only making explicit a view that is already present in Russell’s treatment, since for Russell differences in how things appear always mark differences in the objects of perception.19 Russellian objects of acquaintance are only objects for a given subject, and are transparent to that subject.

Where Wittgenstein does go beyond Russell is in extending the point from perception to judgement. Not only the visual field, but the world itself as the domain of judgement, is characterised independently of the subject. This ‘solipsistic’ doctrine of Tractatus 5.6 and the following propositions is not one that can be adequately treated here, but a few brief remarks are possible.

Considered on its own, the doctrine of the non-subject-relativity of the visual field is not foreign to some contemporary views on perception. Take the question of whether in a given perceptual experience things are as they appear to be. One currently popular view, associated with disjunctivism, is that nothing in a perceptual experience itself settles this question, or even allows it to be raised. To see a stick in a glass of water is neither to see the stick as it is or as it is not. On the contrary, questions of correctness or incorrectness arise only at the level of judgement. At that level, we do distinguish between the way the world is and the way it appears to a subject.

From this perspective, the radical element of Wittgenstein’s idea, the element that justifies his partial self-identification as a solipsist, is his extension of the point to judgement. For this entails that even at the level of judgement we cannot speak of world-views, or ways subjects take things to be. But an examination of this issue would take us beyond the confines of this paper.


Wittgensteinian solipsism, while a striking doctrine, does not really mark a crucial difference from Russell’s theory. Russell’s objects of acquaintance are only ever objects of acquaintance for a given subject. The solipsistic consequences of the theory are merely made explicit by Wittgenstein.

Indeed, we may conclude that the root of the solipsistic conclusion lies in the premise that Russell and Wittgenstein share: that the visual field can be specified by an adequate analysis of the terms in which a perceptual experience is described, by why we might nowadays call an ascription of perceptual content.

The crucial difference lies rather in the substitution of facts for objects as constituting this content, and thus the visual field itself. It is this move that determined the substance of Wittgenstein’s analysis.20

Flat 13

180 Marlborough Road

London N19 4NL


McGinn, M. (2006). Elucidating the Tractatus: Wittgenstein’s early philosophy of logic and language. Oxford: Clarendon.

O’Brien, L. (1996). “Solipsism and self-reference”. European Journal of Philosophy. Vol. 4, pp. 175-194.

Russell, B. (1910/2009). “On the nature of truth and falsehood”. In Philosophical Essays. London: Routledge, pp. 140-152.

Russell, B. (1912/1998). The Problems of Philosophy. Oxford: Oxford University Press.

Russell, B. (1914/1926). Our Knowledge of the External World as a Field for Scientific Method in Philosophy. London: Allen and Unwin.

Russell, B. (1953). Mysticism and Logic. London: Penguin.

Russell, B. (1992). Theory of Knowledge: the 1913 Manuscript. London: Routledge.

Sainsbury, M. (1979) Russell. London: Routledge.

Wittgenstein, L. (1921/1961). Tractatus Logico-Philosophicus, trans. D. Pears and B. McGuinness. London: Routledge.

1For the distinction between conservative and reformist ontology see Sainsbury (1979: 236-7).

2Russell (1912/1998: 3-4).

3Russell (1953: 148-9).

4 Russell occasionally uses the term ‘sensibile’ to overcome the verbal objection that since an entity only counts as a sense datum if actually present to sense experience, it cannot be the case that there are unsensed sense data. Sensibilia are those entities which, if sensed, count as sense data. See in particular Russell (1953: 142ff).

5Russell (1953: 153ff).

6Russell (1914/1926: 72-3).

7Russell (1914/1926: 100).

8Russell (1953: 151-4; 1914/1926: 94-8).

9Russell (1914/1926: 118).

10Russell (1992: 123-4).

11Russell (1910/2009: 150).

12Russell (1953: 205).

13Russell (1992: 119).

14Russell (1992: 124).

15Russell (1992: 126-7).

16Russell (1992: 126).

17On the analogy between world and visual field see, further, McGinn (2006: 263-4).

18O’Brien (1996).

19See chapter one of Russell (1912/1998).

20An early version of this paper was presented to the Wittgenstein Seminar at the University of East Anglia in March 2013: thanks to participants for helpful comments. Thanks also to an anonymous reviewer of a previous draft.