Commentary on “The Causal Theory of Perception” Chapter IV of Perception, Physics, and Reality: An Enquiry Into the Information that Physical Science Can Supply About the Real by Charlie Dunbar Broad. Cambridge University Press. 1914. pp. 186-274.

by Steven R. Bayne
May 19, 2006

I’m going to be commenting on Broad’s dissertation published in 1914. I will focus on Chapter IV as it had an influence on Russell’s treatment of structure and the causal theory of perception in The Analysis of Matter. This work is the key to understanding not only most all of what is of value in Russell’s theory of structure, but stimulated a discussion between Russell and the topologist M. H. A. Newman which created a controversy that remains of some interest. I may stray from Broad from time to time, but my course is set for Russell and my point of departure is Broad 1914. Here is the first of what will be multiple installments.

Common sense begins in naive realism; but naive realism is challenged by the distinction between the way things appear and the way they are, or the way we take them to be in reality. To overcome this challenge, and thereby reinstate appearance as a reliable guide to reality, Common Sense invokes causal principles accounting for the appearances in terms of reality, but in doing so Common Sense, as well as naive realism, come under scrutiny. According to Broad, Common Sense relinquished its acceptance of secondary properties, but could not do the same in the case of primary qualities. The causal theory upon which it relied moved the philosopher towards the Kantian “thing in itself,” but in so doing rendered the very Common Sense claims that motivated the original theory suspiciously uncommonsensical. Causation, then, supplies arguments against the naive realism of Common Sense. These arguments supply the subject matter of Broad’s investigations in this chapter.

Around the time Broad was considering the causal theory of perception it was held by some, e.g. G. E. Moore, that a sense datum could be a patch of colour and that in order that that patch exist it was necessary that it be experienced. This was related to the idealistic position that esse est percipi but differed from it insofar as for Moore there was more to the world than sense data. Still, the relation of perception, according to some, was essential to existence. If a thing did not exist as a relatum in the relation of perceiving, then that thing did not exist. Broad discusses a similar view, based on the argument that if a secondary quality does not occur in relation to some organ, then that quality cannot exist. The argument, itself, derives from Bradley, who stated it as follows:

A thing is coloured, but, except to some eye, it seems not coloured at all. And the eye...relation to which appears somehow to make the quality – does that itself possess colour? Clearly not unless there be another eye that sees it. Nothing therefore is really coloured; colour seems only to belong to what is colorless. And the same result holds for heat and cold. (Appearance and Reality Chap. 1. p. 12).
Broad observes that this argument is very similar to certain arguments for phenomenalism. In particular the argument that says that since nothing is knowable independently of a perceiving mind nothing can justifiably be said to exist independently of perception. In the case of the above argument it is the relation to a sense organ and not a perceiving mind that is at issue. Broad then undertakes to distinguish these two arguments. He begins by noting that while it may be the case that Bradley’s argument gives no reason to suppose that red exists outside some relation to the eye, neither does it provide a reason to believe that when red is unperceived it does not exist in such and such relation to the eye. Having no reason to accept a claim is not the same as having a reason for rejecting it. Perceiving red may imply red’s being in a certain relation to the eye, but the converse does not hold. That is one point. The other, and, and the more important one for understanding the linkage between Bradley’s argument and the argument for phenomenalism is that Bradleys argument does not entail the phenomenalistist’s conclusion because being in such and such relation to the eye does not entail being in the relation of being perceived by some mind. Short of accepting this entailment, we are left with two possibilities. Either in the absence of perception red, nonetheless, stands in such and such relation to the eye, or when red is no longer in such and such relation to the eye it still exists. Perceiving red may entail red’s being in a certain relation to the eye; but the converse does not obtain. Unperceived, red either exists in the absence of such and such relation to the eye; or it continues to be in that relation to the eye.

At the time Broad wrote it had been widely maintained that “Cartesian dualism has been perpetually confirmed by the habits of common sense” (R. B. Perry Present Philosophical Trends. 1912. p. 309). The causal theory of perception was held by Broad to be part of the common sense view of perception. The external object would provide the cause of our conscious awareness of a subjective object, an appearance. The appearance and the object are regarded as distinct. This chapter in Broad’s thinking influenced Russell considerably when he constructed a theory of sense data which possessed all the elements of a conscious and deliberate contrast with the views of Moore on the same subject. Moore it will be recalled also distinguished the objects of perception and the objects of immediate acquaintance. Moore was a realist, but a representational realist. Russell would become a realist without reliance on representations, a feature of his theory of knowledge he came to share, after some struggle, with James. Broad will be among the first philosophers to give careful consideration to the causal theory of perception. This is why it is important for Broad to compare the relation of perception, familiar in discussions of sense data and idealism, to the relation of object to eye which induced Bradley, as we have seen, to question secondary properties. Later we shall have occasion to see that secondary properties introduce issues of enormous importance, but for now, having accepted the common sense distinction between mind and body we must consider whether common sense is equally intent on distinguishing sense data, as representational and subjective entities, and objects which cause them, on the theory under discussion. There is one complication Broad does not immediately address but is of considerable importance.

In the case of Moore, and earlier philosophers, a distinction was drawn between acts of consciousness and the objects of those acts. In the context of a discussion of the causal theory of perception this raises certain questions. Among the most important is this: if the object in the external world causes the act of consciousness, does it also cause the appearance of the object that, also, serves as the object of the act? Which does that object cause? The appearance? The act? Or both. One can see that should sense data be capable of existence independently of the act, that is, if there are unsensed sense data, then the choice is made more difficult. For if we distinguish the cause of the appearance and the cause of the act, which has the appearance as its object, then the object in the external world, while it may cause the act, may not (perhaps cannot) be the cause of the appearance. Rather than either collapse the appearance into the act, as one of its constituents, in some sense, or collapse the appearance into the object, and thereby fall into classical phenomenalism, Russell will accept neutrality of the objects of immediate acquaintance and run together the appearance with the external object. According to Broad, the external object on the causal theory relates the mental act to the external object. But if there are unsensed sense data one might ask “What connects the object of the act, the appearance, and the object. Some, e.g., Stout, had claimed that redness, for example, belonged to the appearance as well as the object; but, then, what accounts for this correspondence, if the causal nexus is between the act component and the external object, rather than the external object and the appearance? Clearly, the structure of the sense organ is correlated with certain features of the appearance; e.g., rods and cones are associated with color. But this will not tell us whether such features belong to the external object or the appearance alone, or whether these features obtain independently of a certain relation to the eye. There is no way of telling whether the structure of the eye explains certain features of the appearance of the object or explains how we perceive real features of the object itself. At this point Broad raises a question having enormous implications.

A few years before Broad wrote, T. P. Nunn and Samuel Alexander had proposed a solution to a problem connected with the so called “relativity of perception.” A penny viewed from different angles will appear sometimes elliptical and sometimes round etc. Which shape is the “real” one? Although this is the way the question is usually framed there is another way, one which I don’t believe was ever discussed. As a single subject moves about the penny, the shape of the penny appears to change from round, for example, to elliptical, and so on. Even though one may argue that only one shape can be real and the others illusory, it can also be argued that the apparent change is what is illusory. It’s not just a question of illusory qualities but illusory change. I will not pursue this at the moment, but since there is no change in the object, one argument for sense data is the fact of change, and not the fact that the apparent change is illusory. Whether there are sense data associated with change and not, merely, the qualities indicative of the change is something we won’t discuss at this point. Before Nunn and Alexander, and owing to the belief, as Broad puts it that “realities are dear and perceptions of appearance cheap” (p. 196), the most commonly stated view was that only one appearance was the real one and the rest were illusory. But Nunn and Alexander, resistant to the idea of representations of perception, held that no such privilege could be extended to one appearance alone. They had held that all the appearances were appearances of the real, and that when a penny looked elliptical from a certain angel that was because it was elliptical from that perspective. This would become a critical notion in Russell’s theory of sense-data, and perhaps more importantly, his theory of space. In other words, a penny had such and such shapes but from a perspective. Later in The Mind and Its Place in Nature (Routledge & Kegan Paul. p. 161) Broad would discuss a very similar theory, the theory of “multiple inherence.” We shall have occasion to discuss this further, but for now what interests us is that Broad will extend the perspective theory of inherence, as we will call it, to include as fundamental not only the geometrical relation of organ and object, but physiological considerations as well. This comes out in his comparison of what an insect sees and what a person sees in regard to the issue of whether what the human sees is in some sense more real.

Here again it might be that both what we perceive and what the insect perceives are equally real, but that only minds furnished with eyes like ours can perceive the reality that we do, and only those furnished with eyes like insects’ can perceive the reality they do. And once more I must insist that until you assume that all these perceptions of very similar objects have a common real cause there is nothing improbable in a word of reals, each of which can only be perceived by one creature from one place. (ibid)
Extending the notion of relativity of perspective from the purely geometrical to include the physiological is a radical move. Evander McGilvary, a champion of perspectival thinking, took this view seriously and developed it in a series of papers, which however neglected today remain models of analysis (see “Realism and Physiological Argument and Two Other Articles on Realism” Journal of Philosophy 1907 and numerous other essays by McGilvary).

Relativity to space and relativity to organ of sense are, then, two important respects in which our awareness of the data is not determined by intrinsic content. This leads Broad to an important distinction between two theories of sense-data. On one theory, our sense organs are instruments of perception. The distinctions we find in objects are actual, but it is only by means of such instruments that they can be brought before the mind. By contrast, there is another theory of the role of such organs, and that is that they supply the conditions of the properties of sense-data (that is, appearances). The contrast, then, is between instruments and conditions. Broad supplies the most concise way of putting it when he remarks:

We have seen that all the facts capable of two interpretations, viz., the Instrumental one which holds that our organs and their detailed structure are the instruments by which the mind perceives real things and their real qualities and characteristics; and the Causal one which holds that our organs and their internal structure are conditions of the perception by the mind of objects and distinctions in them, both of which for aught we can tell are mere appearances. (p. 197).
At this stage of the discussion, economy argues against the Instrumental view inasmuch as it introduces an infinite number of real properties and not just one. The so-far unmentioned down side is that we now have a distinction between the real property and the property of the datum, which may coincide with that of the external object.

Here again it might be that both what we perceive and what the insect perceives are equally real, but that only minds furnished with eyes like ours can perceive the reality that we do, and only those furnished with eyes like insects’ can perceive the reality they do. And once more I must insist that until you assume that all these perceptions of very similar objects have a common real cause there is nothing improbable in a word of reals, each of which can only be perceived by one creature from one place. (ibid). E Extending the notion of relativity of perspective from the purely G geometrical to include the physiological is a radical move. Evander McGilvary, a champion of perspectival thinking, took this view seriously and developed it in a series of papers, which however neglected today remain models of analysis (see “Realism and Physiological Argument and Two Other Articles on Realism” Journal of Philosophy 1907 and numerous other essays by McGilvary).

Relativity to space and relativity to organ of sense are, then, two important respects in which our awareness of the data is not determined by intrinsic content. This leads Broad to an important distinction between two theories of sense-data. On one theory, our sense organs are instruments of perception. The distinctions we find in objects are actual, but it is only by means of such Instruments that they can be brought before the mind. By contrast, there is another theory of the role of such organs, and that is that they supply the conditions of the properties of sense-data (that is, appearances). The contrast, then, is between instruments and conditions. Broad supplies the most concise way of putting it when he remarks:

We have seen that all the facts capable of two interpretations, viz., the Instrumental one which holds that our organs and their detailed structure are the instruments by which the mind perceives real things and their real qualities and characteristics; and the Causal one which holds that our organs and their internal structure are conditions of the perception by the mind of objects and distinctions in them, both of which for aught we can tell are mere appearances. (p. 197).
At this stage of the discussion, economy argues against the Instrumental view inasmuch as it introduces an infinite number of real properties and not just one. The so-far unmentioned down side is that we now have a distinction between the real property and the property of the datum, which may coincide with that of the external object.

I will continue discussing Broad’s view in the next post, again with an eye towards Russell’s views on structure, down the way. Common sense does not offer an immediate answer to the question as to which of these two points of view is correct. One may either view the relation of the structure of the eye, etc., as enabling us to perceive qualities of the real, in which case the organ is more properly regarded as an instrument; or, one may view the eye, etc., as being causally related to the external object in such a way that an appearance is all that we are acquainted with, as long as the causal relation obtains. In this later case, the existence of the objects of awareness, that is, appearances depends on the disposition of the organ, unlike when the organ is regarded as a mere instrument. Three questions arise in connection with considering which alternative is most acceptable.

Is the detailed structure of the organ essential to our awareness of actual properties of objects which otherwise would go unnoticed, or is it a causal condition of the mere appearance of objects, where appearances are transitory and may not reflect the actual properties of objects, assuming they exist?

Is there a proper state or condition of the organ which reveals actual properties, whereas others states or conditions supply erroneous data about the objects?

Can we justify the view that instruments, such as microscopes, provide new data about the world whereas drugs such as santonin produces illusory qualities? Broad sets out to examine the instrumental view in light of these three questions. A closer look at precisely what is “instrumental” about the instrumental view is in order. For this purpose, Broad selects the typewriter as the instrumentalists analogue of the eye. Just as the structure of the typewriter is correlated with the production of certain shapes or letters, similarly, the structure of the eye is said to be correlated with visual properties. But the analogy is not perfect. In the case of the typewriter, what is at issue is the relation between the mind, the typewriter and its structure, and the paper. When I type the mind causes me (in some sense) to strike the keys, resulting in letters on the paper. The paper and the mind are distinct, of course. In the case of the eye what is involved is the mind, the eye, and the external object. The result of the action in the typewriter case is an object, or objects, in the external world, viz. letters. In the case of the eye, the result is the minds awareness of certain properties. In other words, unlike the typewriter case the action has a mental result, and not a result changing the world. Indeed, were it the case that the eye changed the world as the typewriter does, then we would never perceive the world in its actuality. But because on the instrumentalist view the eye’s structure is causally efficacious the distinction between the causal theory and the instrumental theory becomes blurred. There is another potential problem Broad does not grapple with.

The compelling feature of the instrumental view is that there are properties existing independently of the mind and appearance that the mind accesses with the help of the eye as an instrument. Some have thought of vision as a faculty or, using more primitive language, a “power.” In this case, not the power to produce but the power to access features of the world. Faculties are not instruments, however. A tool or instrument may supplement a faculty, but it cannot, without distorting our usual conception, be regarded as a faculty. The notion of a faculty is indifferent to the causal/instrumental distinction; the later suggesting accounts of the faculty’s operation. But while this indifference possesses an element of truth there is a difference: a tool is the tool of an agent, whereas a faculty construed as a set of causal conditions, or dispositions, does not, as commonly understood, require an agent. As we shall see, Broad is not indifferent to such considerations. It is no minor detail that an instrument is something we learn to use, whereas a faculty of the senses is innate, although it can be “shaped” through training to experience things not everyone can experience; still, to be used no learning is required. A faculty is a capacity; in the absence of a capacity a tool is a mere piece of junk. Viewing matters in terms of faculties, although the notion is fraught with problems making it the subject of considerable joking, does allow one advantage over he instrumentalist position. A ypewriter alters the world, not so a faculty. Whether the faculty is that of memory, perception, judgment, reasoning, there is no alteration of the world in its exercise; and, of course, there is no faculty of appearance. But such considerations are not what causes Broad to reject the instrumentalist view.

What causes Broad to reject the instrumentalist view is a problem as old as Plato’s Sophist, and that is the problem of explaining the possibility of error. In the case of illusion, what is it that the instrument selects on the basis of its structure? The instrumentalist may attempt the solution that the object comes into being with the cognitive relation which the organ makes possible; but, then there must be some accounting in instrumentalist terms of the difference between illusory and non-illusory perception. To do so, according to Broad (p. 204) will land him in a situation where his own position cannot be distinguished from the causal approach. There may be a solution to the problem Broad has overlooked, however. It may be argued that the properties brought to mind by means of the organ instrumentally conceived are not the material of the errors of perception. It may be argued, that is, that error is a notion reliant not on the cognitive faculties, alone, but on the faculty of judgment. If conceptual content and cognitive content can be distinguished, and here I do not mean necessarily that they can exist independently of one another but only that they can be distinguished in our reasoning about them, then error might be attributed to interpretation and not to the objects given for interpretation. This relies on the concept of a given, but we have not yet absorbed the more contemporary arguments, such as those to be offered by Sellars, for example, that the given is a faulty notion. If we were unfamiliar with sticks, a stick in water, which otherwise would appear bent, would not create an illusion or even a tendency to accept what we see as illusory. Since on this theory illusion requires learning, the properties presented by means of the sense organs cannot be in themselves illusory.

The instrumental theory, thus, gives way to the causal theory; following recognition that it incorporates essentially causal elements. On the causal approach in perception what we have is an external object causing an appearance in the mind. The object of perception is, then, an appearance. Since the appearance cannot exist independently of the mind the relation of appearance to mind, usually thought of as the relation ‘being immediately aware of’, comes to be thought of as incorporating the object, or appearance. In this case, the appearance becomes adverbial, rather than objectual. There can be little question that future epistemologists, such as Chisholm, came to such an “adverbial” view of perception by taking such facts into consideration. There is more, however, to be learned from the instrumentalist position despite its being reckoned as being absorbed into a causal account. Although Broad does not explicitly raise the matter, it remains the case that the adverbial account will render the act/object distinction unnecessary.

The instrumentalist view conceives of the perceiving subject as an active or agentive subject. This is only natural since the idea of an instrument implies use and use requires an active subject. There is a change in the mind yielding use of the instrument, but there is, also, the effect of the instrument’s use on the mind. In any event, the mind uses the body to use the instrument. The mind’s use of the body introduces volition; not all perceptions are a matter of voluntary action. But some are. As Broad puts it:

The volition enters when we voluntarily place and adjust our organs and when we pay attention. It is only in such cases that the mind’s use of organs of perceptions is strictly analogous to the use of a typewriter. (p. 207)
Shortly afterwards Broad says
...the whole effect of volition is to produce a preliminary adjustment of the organ after which everything proceeds involuntarily. (op cit. p. 208)
In using a typewriter, for example, we are constantly active agents; not so in the case of perception. Thus, the analogy of instrument and organ is said to fail.

Whereas Moore had spoken of “common sense,” Broad spoke of “reflective common sense” (p. 210). But, unlike Moore, who in the early going of analytical philosophy took “common sense” as a body of propositions without philosophical content, themselves, Broad took “reflective common sense” as providing the basis of a philosophical position. Thus “reflective common sense” would have it that the appearance and the cause of appearances must be distinguished. This is owing to the fact that the cause precedes the appearance and, so, cannot be identified with it. Once this distinction of real object as cause and appearance as effect is accepted it becomes an easy matter, according to this theory, to avoid having to attribute different, and contrary properties, to the same object seen from different perspectives. The differences in perspective amount to being differences in the antecedent conditions of perceiving the shape of a penny, for example; and, while the shape of the penny, itself, remains constant the appearances vary. On the common sense view, therefore, there are more causal conditions of perceiving an object than the properties of the object itself; involved in perception are conditions of mind and the properly “adjusted” sense organs. The common sense theory depends on the truth of two synthetic a posteriori propositions: 1) “Certain objects of perception have events in them which are causes of those objects being perceived”; 2) “All objects that are real and perceived have the perceptions of themselves caused by events in them” (p. 211). These propositions can be known and supported but not by direct means. They require the backing of the hypothetical methods of science, making use of lawlike propositions. So the discussion shifts to the scientific view of these matters. The important philosophical point in doing so is that the view of science is the view of common sense under discussion, that is, “reflective common sense.”

The idea of such a relation between science and common sense may eventuate in a distancing of naive realism and common sense, but science will be considered common sense more carefully and systematically elaborated. Gustav Bergmann once, at least, referred to science as the “long arm of common sense,” and this is certainly in the spirit of Broad’s proposal. The scientific view will accept the fundamentality of primary qualities only after causality is introduced mediate the relation of external world object and its appearances. But if we give up on secondary qualities what accounts for our knowledge of objects, at least insofar as it is assumed that the real objects of perception are subatomic events within a certain volume with which we have no acquaintance? How, in other words, can we know real objects, if these real objects are subatomic events, and if we refuse to countenance secondary qualities? Broad’s answer is that primary qualities are connected in a lawlike way with the appearances of the object and that by inferring from appearances in a lawlike way the existence of the object and its properties we can come to know them without acquaintance with their subatomic constituency. The scientific extension of common sense introduces laws of more than one sort.

The scientific common sense view introduces the need for three kinds of laws. First, laws relating the primary qualities to the sense organ (say, the eye); second, laws relating conditions of the eye to those of the brain; and, finally, laws connecting brain states to properties of the appearances contained in the perceptual situation. If Broad has been right in saying that the instrumental view and the causal view are not entirely mutually exclusive from the standpoint of the instrumental theory we are inclined to wonder as to where the science and with it common sense stands n relation to these competing theories.

The scientific perspective on our knowledge of the external world can be argued to have it that primary qualities are known as the instrumentalists suggest; but should primary qualities loss this special status and all qualities are dealt with as the causalists maintain, then nothing is really knowable beyond the appearances. But how shall we construe the position of science in this regard? It is natural to suppose that as far as concerns science what is presented to us is determined by events in the object perceived. The commonly accepted position is that objects consist of primary qualities and the events that enable their effect on the eye and, ultimately, the brain. Even if it could be maintained that what we perceive is a function of primary qualities in the object this will not resolve the dispute between instrumentalists and pure causalists since the object whose primary qualities have this effect may consist in properties which include those manifest in appearance. Thus an object may be red, notwithstanding reliance on primary qualities to bring them to our attention. It is only if the only properties an object has are primary that locating the cause of our perceptions in the object will exclude from realistic consideration secondary qualities, rendering them effects and, therefore, mere matters of appearance. One logical possibility Broad does not examine is that events in the object require that it be red, but that the appearance of the object’s being red is determined by primary qualities indifferent to the color associated with events in the object. But what this would mean would be that the color of the object would be “selected,” instrumentally, by the visual apparatus but there would be no causal connection between the actual color and the apparent color even though they coincided. There would, on this account, be no causal connection at all between the color of the object and the color of which we are aware in its appearance. How, then, might we account for the correspondence? It is probably owing to such considerations that Broad doesn’t entertain such possibilities, but they remain possibilities that may carry implications further along in the discussion.

It is implicit in the scientific view, as Broad understands it, that when we perceive a real object we are caused to do so by events in the very object we perceive. What is the nature of the causal relation that makes this a reality? The answer to this question is not altogether transparent. Says Broad,

When we perceive a definite reality, the event hose causation we are seeking is the establishment of a relation between mind and that reality. Certainly on the ordinary view of causation this event must have a cause; but I know no reason why that cause should be found in the reality which is relatum to this relation (p. 225)
As the discussion proceeds, one important side to the issue will not be examined but it is one which will figure highly. This has to do with the nature and status of relations in epistemology. In particular, what is at issue is the controversy between so called “external” and “internal” relations, a topic thoroughly and brilliantly exploited by Russell; and an issue that vexed Moore in his treatment of necessity and other issues. What it concerns is the question whether the fact that a relation obtains can be accounted for solely in terms of properties of the relata of that relation; in the case of binary relations either one or both of the relata. Perceiving is a relation of object and mind, when it is veridical. Is this relation determined entirely by the properties of the mind and the object perceived? If we deny this, then it at once becomes difficulty to argue that the relation accrues to certain properties of events in the objects perceived, whether primary qualities or not. In other word, the question is whether scientific realism entails the internality of the relation of awareness or perception. It may be argued that there is a middle ground between saying that this relation, or all relations, are external; while it may be true that there are real relations over and above properties of the relata this may not be to say that the properties of the relata do not figure at all in there being a relation between the objects in question. Surely, the properties of an object enter somehow in determining the relations between it and other objects. In particular, if the properties of the objects are not internal to the relation, how do we account for object X being standing in the relation of awareness, say, at one instant, and at the very next some other object?

We have it, then, that on one view of scientific realism there are only primary qualities. But on this theory as stated the argument against secondary qualities is that they are no longer necessary, not that they have in any way been shown to be unreal or impossible in a direct way. Broad, however, raises a certain basis for doubt. The laws that have been culled from scientific observation, in particular those having to do with primary qualities, have never been seen to obtain in the absence of secondary qualities. Without admitting to the reality of at least some qualities besides primary qualities of extension we cannot justify the very laws we use in justifying our disregard for secondary properties. If we take the instrumental theory as possible for extension and relegate secondaries to mere appearance we dismantle the scientific basis in observation of the laws justifying the procedure. The way Broad has framed the scientific argument is this: all properties except for properties of extension fall within the instrumental theory; whereas the causal theory accounts for the secondaries as mere appearances. The fundamental problem here, he maintains, is that for science and common sense, most primary qualities are, also, mere appearances. The question then becomes one of whether their are ideal conditions of the ye and the medium of perception allowing us to make the distinction between those primary qualities which are real and those which are not? The realist with respect to primaries will answer yes.

On the instrumental theory, as originally stated, the reality behind secondary properties were subatomic events which bore no similarity to the secondary properties. But here we are dealing with primaries and the question is whether we select these out using as instruments of doing so the sense organs that bring us the appearances. In the case of shapes, there was arguably a real shape which could be determined by laws of refraction etc. A penny was round, whereas the elliptical shapes associated with it as appearances were the various ways the roundness of the penny presented itself under certain conditions that fell within the purview of physical laws. But when we apply these principles to shapes of objects in three dimensions we encounter a crucial difference calling for emendation of the instrumental view. In the case of a sphere, for example, all we are ever acquainted with are two imensional images associated with it. It is as though, and this is intended only by way of analogy, the only shapes we perceived were elliptical projections from the round penny. But in such a case, if we never acquainted with a sphere how can we formulate laws relating the property of being spherical with what is presented to acquaintance? The instrumental view must be altered to fit these facts.

The instrumental theory must now rely on a)real events ‘corresponding’ to appearances b) real primary qualities (as opposed to merely apparent primary qualities); c) the real relative spatial orientation of the sense organ and d) the state of the mind and the organ involved. What is crucial here is the notion of correspondence. Because we are never acquainted with the property, spherical, in order to “set up” the correspondence we must either maintain a similarity between the reality and the image in the mind; or we must deduce it from what is presented in appearance (p. 232). On the old theory and in particular in the case of the penny it was alleged that one appearance shared a quality with the reality, the penny’s roundness; but here we are compelled either to arrive at the privilege real property by deduction or we assert a correspondence based on similarity, despite its being characteristic of the instrumentalist view that there is no similarity between appearance and reality, even when veridical. It is important to keep in mind, as Broad notes, that even while I cannot be said to perceive the shape of a three dimensional object I can be said to be acquainted with some shape or other in two dimensions. If I can correlate the features of apparent shapes under such and such conditions I may be able to deduce the real shape of the object, even though it is not given to me in any appearance. Broad goes on from this point to consider another view, one which maintains that the various shapes given to an observer as he moves around a penny, say, are real; that the penny actually changes shape with the change in perspective of the subject.

The main difficulty with this notion, according to Broad, is that there may be more than one agent. Suppose one remains stationary while another moves around the penny. But then if the thesis is correct that the penny doesn’t change there arises a contradiction, since the other subject will view the penny with different shapes and if these are real their existence contradicts the first subject’s view. This is an early treatment by Broad, related to what in The Mind and Its Place in Nature would be called the “multiple inherence theory.” This theory has it that the inherence of a quality in an object is a function of the perspective from which it is viewed. The above contradiction is offset by the fact that a penny in fact has one shape from one perspective, but in fact has another shape from yet another perspective. In other words, the exemplification of a quality by an object is triadic, not diadic, as is usually thought! Notice, however, a subtle discrepancy. It is one thing for the appearance of a quality to be relative to perspective; it is another for a change of quality to be relative to perspective. One can imagine a quality’s being exemplified being a function of location, but it is much more difficult to imagine that something changes, and here we are not talking about properties related to the relativity of space-time, but only from a perspective. Change does not inhere in an object as one of its qualities. We do say that it is a property of an object to change from being of one quality to another, but the changes themselves are not qualities. This is not a conclusive argument against Broad’s main point. It may turn out that even a change in a diadic property is merely a conjunction of contradictory predicates taken at different times. Once time is factored in the relativity of time may allow for consistently maintaining that change as well as quality is relative to perspective. I will not attempt a resolution of this issue. In addition to the issue that perspective introduces there is also a sort of relativity to the particular sense or senses involved. Such as when I feel a penny to be round but see it as elliptical.

What Broad has to say in connection with this problem isn’t always clear as he sometimes relies on earlier discussion; but the thrust of his argument seems to be this. The visual perception of the elliptical penny can be explained once laws of perspective are established from our familiarity with the changes accruing to change of position. No such laws of perspective are involved in the cased of felt roundness. There is no contradiction here, given that scientific laws allow for an explanation of the experienced difference consistent with the claim that both the penny as visually experienced and tactually experienced are understood as being of one and the same shape. There is, however, another approach; one suggested by various defenders of the instrumentalist position. This involves regarding the difference between the eye and the hand as instruments as contributing to what property is selected, much as a change in medium will effect changes in the visual property presented of the same object. In other words, physiology is build into the idea of a perspective. In this way, the appearance of contradiction no longer has the force it earlier had. Much the same might be said with respect to Broad’s argument incorporating multiple observers who see the same object differently, a fact that can be brought to bear against the instrumentalist. But, again, it may be averred that different observers constitute a basis for difference in accessible property or aspect of the object perceived, rather than difference in appearances of some identical underlying object as cause. Perhaps the reader has lost sight of the significance of the instrumentalist claim, particularly when it is advocated by scientifically minded epistemologists who wish to retain instrumentalism but only for a few primary properties.

Recall that instrumentalism is an alternative to introducing appearances in addition to the objects perceived. An instrumentalist with respect to certain properties will reject the notion that there are representations standing between himself and the objects of perception. The causal component which gets introduced by Broad as essential to this sort of representationalism is, in fact, inessential to the denial of representationalism advocated by instrumentalists. For the instrumentalists there is direct realism, but one which is not necessarily naive. Why might this be the case. The reason is that the naive realist will affirm our direct acquaintance with ordinary common sense physical objects; but even if we adopt a Russellian view (circa 1914) and claim that ense data are physical and more or less as the instrumentalists maintain, it is not the case that it is ordinary physical objects which are objects of immediate awareness. The idea goes back as far as Leibniz, that of distinguishing a things “aspects” from the thing of which these aspects are constitutive. Samuel Alexander who, along with T. P. Nunn, provides strong arguments for this distinction identifies “aspects” with “objects” and “things” with the of which these objects are constitutive [vide “Things and Objects” in Space, Time and Deity vol. II. Macmillan. 1927 pp. 92-95, same pagination as Dover edition]. An account of the instrumentalist position, including developments subsequent to Broad’s and Alexander’s treatment can be found in H. H. Price’s Perception (Methuen 1973 (1932) pp. 40-53). I’m going to pass by Price’s treatment now as I intend to return to it as this inquiry into Russell’s theory of structure unfolds more completely. Let’s refresh our memory a bit on what issues are under discussion.

Some scientifically minded epistemologists deny the reality of secondary properties, invoking appearances explainable in terms of primary qualities and the states of the organs of perception. But some primary qualities are, themselves, mere appearances as the relativity of perception suggests. Those primary qualities which are taken as real, for example the roundness of a penny in contrast with the elliptical shapes of its other mere appearances, are accessed by the sense organs on the analogy of instrumentalists; so this position is one of attenuated instrumentalism, that is, an instrumentalism limited to certain primary qualities in a particular perceptual situation. But, now, Broad makes a suggestion hoping to catch the attention of the defender of such a view: why not regard even the roundness as appearance, even though it is veridical and as such is susceptible to treatment along instrumentalist lines.

Can you really believe that practically the same mechanism can produce such utterly different results? Again, you have granted that most of the objects that you perceive are appearances. You only see the circle which you believe to be real in one set of positions; but you see the ellipses which you believe to be appearances from an infinite number of sets of positions. Surely it would be more reasonable to hold that it too is an appearance. Look at the advantages that accrue from this slight change. At once you will be able to drop this incredible belief that the same mechanism sometimes brings your mind into relation with a previously existing object and sometimes creates the object in the relation. (p. 240)
If we take the route that is being proposed here and take all presentations as appearances, then we have, says Broad, fallen into idealism; yet, somewhat paradoxically he says that all we can know of the appearances is that they are caused by something about which we know nothing. I say “paradoxically” since, strictly speaking, to admit there is something besides appearances, such as causes of appearances, is consistent with an attenuated realism. Our not having knowledge of things existing independently of perception is, nevertheless, to acknowledge that there is something besides ideas in some mind, barring, of course, the introduction of mental causes of appearances themselves. The “paradox” may be resolved, however, in favor of the realist, as Broad puts it:
Thus it is hardly true to say that a purely causal theory leads to complete agnosticism, for it does allow of a certain amount of analysis of the causes of our perceptions in accordance with the actually observed conditions of their production. (p. 242)
The “conditions of their production” on Broad’s view are not so much Kantian things in themselves but, rather, structures. Thus there is a permanent “structure” to reality which accounts for the possibility of mind’s relation to the world (ibid). Properties given in appearance, such as color possess “for their particularization a small number of characteristics which may be compared to independent variables. The structural component, then, appears in this comparison with variables; whereas content, or particularizations, are values associated with such variables. It is the structural component that affords us a connection to causes. This description seems most consistent, but Broad is not entirely explicit in this, at least at this stage, and his expression lacks the lucidity of his later writings. His emphasis on structure, however, almost certainly is what Russell was getting at in his acknowledgement in The Analysis of Matter of the influence of Broad on the causal theory of perception. Still we are not entirely certain whether on Broad’s view this structure belongs to the mind or the world. This may have been in some measure the animus for Russell’s neutral monism, a possibility we shall explore as we move towards a discussion of Russell’s views. What we are told by Broad is that the possibility of perception “depends on the connexion of the mind with a real structure,” such structures enabling access to general qualities such as colors and sounds. One can hardly restrain the impression that at some level Broad is introducing something like the Kant’s categories, not as elements of content, but “structure,” and doing so in such a way as to avoid the necessity of a Kantian thing in itself. Thus the particularizations of the “variable” content of the general properties encountered in the appearances depend on “states in that permanent structure which we believe to be the condition for the perception of the general quality of which this is one possible particularization” (244). With respect to particularizatons there is no appeal to unknown particular features of a thing in itself, only appeal to states of structures required for the perception of the particular instantiations of the “variable” content of those structures. These particular values may not, as Broad surmises, occur to the visible organ of sense, but they may, and I think this is something which, as we shall discover, Russell believed, be occurrences in brain states, and which correspond to aspects or properties of objects in the external world.

Recall that earlier we spoke of Broad’s introduction of structure and the relation this has to his employment of the notion of variables connected with qualities receiving particularization in individual appearances. This particularizations are tethered to the individual subject’s experience and are said to be incommunicable. It may be argued that on Broad’s theory structure makes communication possible without reliance on such particularizations. Communication does not entail identity of experienced qualities, and neither is it necessary that the reality we share exemplify the differing particularizations associated with filling in the structure of appearance in each perceiver’s individual experience, even when those experiences are veridical.

Just as we cannot possibly know whether A and B, who always use the words ‘red’ and ‘green’ under the same circumstances, really have the same experience, so we cannot and do not need to know whether the real has the peculiar sensuous quality that we experience when we use those terms. (p. 248)

It is an important feature of Broad’s position with respect to the scientific theory that while it relies on primary qualities this is not an absolute necessity, for if it can be shown that our experience of those primary qualities can be explained without postulating their independent existence then all the theory will need are external, albeit, unknown objects, not a special class of appearances. (p. 250) Is it possible to state the scientific theory without assuming the reality of primaries?

In the process of examining whether a scientific theory of the perception of primaries is possible without accepting their reality, Broad draws on the contrasting example of sounds. In the case of sound, it seems we need no reliance on primaries at all, beyond appearances. But the situation is not so clear cut in the case of visual perception. In the visual case there is no question on the scientific theory of perceiving the causes; the causes are beyond the reach of vision. This is in contrast to the case of sound where it is at least possible under certain conditions to see the thing which when it vibrates causes us to be aware of the sound. Furthermore, in the case of explaining the appearance of sound we do not rely on postulating real sounds as the cause of the perceived sound. By contrast, in the visual case on the standard theory, which we are attempting to modify, the perceived primary qualities are explained in terms of the primary qualities of unobserved subatomic events. Finally, in the sound case there is no arguing that what explains the perceived sound resembles the sound understood as the cause of the perceived sound. But there is a further difference making the difficulties of the visual case virtually intractable.

On the scientific theory with which we begin, primary qualities are real, and our task is to formulate a theory in which they are merely appearances; thus simplifying the ontological presuppositions of the causal theory. If we jettison primaries from reality and affirm their appearance, then we must explain their appearance without introducing them as real. We are aware of the appearances; the appearances are caused by subatomic events with which we have no acquaintance. How can the underlying reality of these primaries in appearance be anything but primaries themselves? Only by doing so can we say, for example, that this object is located where it appears to be. In the case of vision, we are at once told that we have a relation to reality, in the form of primaries, at least; but, we are also told that we are only in a cognitive relation to the appearance of those primaries in our own minds in the form of appearances. This is what I take to be the “paradox” Broad describes, although his description is not entirely perspicuous. Broad explores the possibility of avoiding this “paradox” by considering shapes as common sensibles, that is, as properties experienceable by more than one sense.

I may hold a penny in my hand, attending to its roundness. Being able to do so is a function in part of my desire to do so. Whenever I choose to feel the penny I can do so and the experience is roughly unchanged. A persisting quality of this penny is its ability to evoke this experience; whatever quality is involved here is, then, a more or less permanent feature of the object felt. I may open and close my eyes and, thereby, alternately see a shape I associate with the penny. If I look at it from above at ninety degrees I see a shape coincidental with the one I feel, should I decide to hold it in my hand. The cause of this perceived quality, in the visual case, is then not always present. It does not persist as it appears to in the case of the felt shape (p. 253). The possibility, then, appears available to us of accepting the instrumental theory with respect to touch while rejecting it for vision.

If it can be successfully maintained that in the case of touch the primaries given to us are one and the same as those in reality, then the instrumental theory would be established with respect to touch without our having to accept it with respect to vision. Recall that in the case of vision, although the instrumental theory was internally consistent, a paradox arose from the fact that

...precisely the same sort of causal processes in the same bodily organs must be capable of producing two entirely different results, viz., the establishment of a relation between the mind and a reality, and the production of a whole perception consisting of an apparent object very like the one that is supposed to be real + the same sort of relation to the mind as we had before. (p. 254)
By “two different results” Broad means that whereas the instrumentalist needed only one result, the result of “selection” of an aspect, as long as appearances distinguishable from aspects were not required for the analysis, once it was accepted that there are appearances of primary qualities in addition to the real properties, what Russell and Leibniz called “aspects,” then a second set of results was required. In the case of touch, should there be no basis for this sort of distinction, the paradox does not arise with respect to touch; it does with vision but only until we relegate all primaries to appearances, but the question remains, how is this possible?

While illusions associated with touch are a sometime occurrence, they are far less common than in the case of vision. And, even in cases where illusion is alleged, the problem is not one where an object presented to acquaintance is believed to be an appearance, merely. If I gently poke my skin with two pins situated close together, there is the impression of one pin only; but no illusory appearance results, only an imperfect awareness of what is being presented; and this is not owing to an illusory appearance, as in the case of the elliptical look of a penny, but, rather, to the imperfect instrument provided by the sensory capacities of the skin surface. Moreover, it should be noted that in the case of errors involving touch that the culprit is not the occurrence of an illusory appearance that causes difficulty, but faulty judgment based on an imperfect or limited experience of touch. In the case of the pins which on one occasion seem at no distance and on another at some distance, the problem in the former case is misjudgment of the distances separating them. The instrumental theory says nothing about the possibility of misjudgment. It would seem, then, that touch provides the basis for defending the causal theory; for now it can be maintained that, when visual perception “tells us” that the penny is round, this is corroborated by the instrumental nature of touch, and that the “paradox” that ensued upon the need for a dual explanation of the primary qualities in appearance and those in reality is no longer in the offing. All can be viewed as appearances without embracing with respect to visual perception the instrumental theory, thus establishing a simplicity that had been challenged by the argued need for appearances, which were not veridical, in addition to presentations of what we have called aspects, which accurately present, without need of appearances as representations, the primary qualities of objects despite the invisibility of the microscopic events that result in our awareness of them. As Broad says,

For recognition of the reality of the tangible carries with it a solution of the paradoxes mentioned on p. 252. For instance, there is now no difficulty about the real, but imperceptible events in imperceptible little bodies which occupy the same place as that in which the body is perceived to be. For it was granted that touch did not tell us the whole truth about tangible realities. But is was claimed that it told us, so far as we know nothing but the truth about their geometrical qualities. (p. 260)
The core of Broad’s proposals, then, with respect to the causal theory is that tactual experience is fundamental, and that the problems encountered by a causal theory of visual perception can be remedied by accepting this as a fact. Five advantages of such a theory are stated, which I shall recapitulate briefly.

First, tactual experience is rarely deceptive, and when it is this is not owing to there being a number of competing appearances only one of which is the “right” one, such as in the case of the visual presentations of a penny, for example. Second, we avoid need for the unverifiable contention that there is a similarity between the properties of the appearances and those of the reality “behind” the appearances. Instead, in the case of tactual experience there is only need for a one to one correspondence. Third, this simplifies among other things the account to be given of several perceivers seeing the “same” thing from different perspectives, and accounts for the correlation of visual and tactual appearances. Fourth, we are able to provide a theory of visual appearances which takes as its foundation tactual experience, and finally, unlike the instrumental theory even tactual experience does not commit us to accepting every appearance as a presentation of the object; thus, we avoid the problem presented by the possibility of error. One potential problem area is that now that we have provided for real primary qualities we don’t know exactly what to do with secondaries. This, one should recall, was not a problem for the instrumental theory as such.