§ 1

The question with which the last chapter terminated brings us to the heart of the problem I desire to bring before you in this chapter. Let us consider it more in detail.

It has already become plain that in the routine of daily life we find ourselves treating sense experience in two quite different ways, and employing two quite different methods of gaining knowledge through perception.

The first is the way of ordinary common sense in its most ordinary moments. It is instinctive and immediate. It is used by all mankind, and presumably (mutatis mutandis) by the lower animals also. When following this procedure, we accept a world of material things as we perceive it and because we perceive it. Thus do we obtain our main stock of working beliefs about our physical environment. Perceptions thus employed provide subjects for the painter and inspiration to the poet; as most persons suppose, they introduce us to the “real” in its most “solid” and substantial form; and whatever may happen to them in philosophic or scientific crucibles, by mankind, including philosophers and men of science, they will never in practice be abandoned. This way of using them I shall call the direct method.

But there is quite another way of treating perceptual experiences, also known to common sense, but greatly and increasingly developed by science. I shall call it the indirect method. It consists in regarding perceptions, not as bringing us into immediate cognitive relation with a portion of external reality, but only as supplying us with the data from which the character of external reality may (it is thought) be indirectly inferred. From this point of view, seeing, hearing, touching, smelling, and the rest, tell us about what is seen, heard, touched, and smelt, in no other way than a wound tells us about a bullet, or drunkenness tells us about alcohol. They show us effects; we infer causes as we can.

Beliefs reached in this second way are evidently due to experience. So far, therefore, they may claim the title of empirical. But they are based on a very different sort of empiricism from that to which we usually apply the name. Nature on this view is not observed; it is conjectured. The infinite fabric of the physical universe is treated as a hypothetical construction devised to account for the experiences of mortal men. These experiences (e.g. the sight of a blue sky) are no longer held to involve the immediate apprehension of external fact; they are merely experienced events which challenge curiosity—transitory consequences of which we look for a more enduring cause.

§ 2

Now how are we to value the respective claims of these two familiar though utterly different methods of penetrating by the way of experience into the world of external fact? Is it possible, for example, wholly to reject the direct method, placing our sole reliance on the indirect? The direct method, as we shall presently find, is full of difficulties whose edge would be blunted could we put all our trust in its rival. But can we?

Note what such a course involves. It requires us, in the first place, to treat as symbolical the familiar experiences of the external world which colour our whole lives, which are common to all mankind, which provide the historical origin, if not the logical foundation, of all the sciences of Nature. For these inevitable intuitions of an external reality it requires us to substitute, as our sole empirical guide to the world of physical fact, inferences from perceptions regarded as the passing effects of causes that (by hypothesis) are never themselves perceived. On this frail foundation of argument we are expected to construct a scheme of beliefs that shall embrace the whole fabric of the natural world. Surely the shadowy procession in Plato's cave would provide a surer guide than this to truths which lie beyond the limits of immediate vision.

For, observe that, on this hypothesis, we have literally nothing to guide us; no laws except those, if any, which obtain between these transitory experiences themselves; no antecedent probability that truth lies more in one direction than another; no antecedent probability that what we are in search of, namely, the enduring material universe required by science, is to be found in any direction whatever. The whole realm of the imaginable lies, indeed, before us at our absolute disposal. We may believe what we like. We may invent whatever kind of cosmic scaffolding seems best to suit our needs. But wealth like this is indistinguishable from beggary. Who would willingly sit down and speculate in the void on the kind of external world which of all possible external worlds is the one best suited to engender his individual train of sense perceptions? Could a more barren occupation be imagined? Could fancy be put to uses less responsible?

But (you may be tempted to say) where is the difficulty? Take down any series of scientific textbooks, and it will be found that the work just declared to be impossible has already been happily accomplished. In their kindly pages is described the very thing we are in search of—“the external world best suited to engender our individual train of sense perceptions”—and (you will probably add) the value of its claims can in the immense majority of cases be, directly or indirectly, tested by experimental verification.

But how was the “most suitable” external world originally discovered? By unguided explorations through the vacant tracts of the “possible”? Not at all. As I have already noted, the scientific conceptions which seem most remote from the naïve experiences of the plain man are all historically rooted in them, and in the name of “verification” are always appealing to them for support. Science could never have been arrived at, nor can its conclusions now be verified, by the sole use of the indirect method. This, in its purity, is completely impotent. It is (by definition) an argument from effect to cause; and whenever this argument is used it assumes a framework of beliefs within which, and only by the help of which, we can legitimately draw conclusions. We must know, for example, the causal sequences on which our inferences are founded. This can only be known if we have already broken through into that external world where, according to science, its causes are mainly to be sought for. In fact, of course, we always do break through, lawfully or unlawfully. At no stage in the history of the individual, at no stage in the history of the race, have men ever depended for their knowledge of physical reality solely upon conjectures as to the character of the unperceived causes which have produced their individual perceptions. Always there has been, and always there must be, some use made of direct observations after the manner of crude, common-sense realism. Always among the things observed have been independent material objects with which, throughout the period of observation, the observers believed themselves to have direct acquaintance. These were the original occasion of all scientific speculation: they still constitute the main substance of our thoughts when we deal with the external world. To that external world we can never penetrate without their help. Deprived of their aid we should for ever stand, each in a little world of his own, as powerless to reach outside reality as if we were philosophers hopelessly entangled in some representative theory of perception.

§ 3

The indirect method, therefore, taken by itself, carries us nowhere. How far can we travel if we use only the direct method?

Consider a particular case. I get a glimpse, let us say (with apologies to “Kim”), of a red bull on a green field. This seems, on the face of it, to be an immediate apprehension of external fact, a direct intuition of physical reality. I do not arrive at a conclusion—I see, and I believe. But on reflection it becomes obvious that I may believe wrongly. The red object may not be a bull. The green object may not be a field. Though, therefore, there was no element of conscious inference in the act of perceiving, nevertheless the perception, as I have described it, cannot be regarded as supplying us with immediate knowledge wholly uncontaminated by alien ingredients. So much must be conceded. But though as a whole this perceptual deliverance may be thus unsatisfactory, can we not find in it some element which resists this (or any other) process of destructive criticism? I conceive that we can. If I cut down the contents of what I at first regarded as the immediate experience of a bull in a field to much narrower limits; if I admit that what I directly observed was not necessarily a red bull in a green field, but only a red patch on a green patch, my position, it would seem, becomes impregnable. No amount of reflection on my own part, or argument on the part of other people, will persuade me that I did not see what I know that I did see. I may have been drugged or dreaming; but I was not in error. The experience thus reduced may have had small significance; it may even have been what we call “illusory”; but it was mine, and, even if illusory, it was true.1

But granting that it was true, for what purposes can it be deemed sufficient? If this, and only this, kind of knowledge is directly supplied by sense experience, can we even assert with confidence that perception makes us directly acquainted with an independent world at all?2 We may, of course, interpret our perceptions of the red patch and the green patch by the indirect method of which we have already spoken. We may treat them simply as effects. We may investigate their causes, and may come to the conclusion that among them was a red bull in a green field. But it is not the indirect method that we are now discussing; and it seems plain that if we can make no better use of the direct method than the one I have indicated; if its messages when purged and purified are of an import so trifling, it can never furnish our scientific beliefs with anything even distantly resembling an adequate foundation. The deliverances of experience may be indubitable, yet if they are all of this pattern they must surely be quite insufficient.

§ 4

But if, when taken separately, neither the indirect nor the direct method of using sense perception provides us with a satisfactory groundwork for our familiar beliefs in their scientific form, can the two methods be combined? In the ordinary practice of life we constantly employ both. Sometimes perception is treated as the immediate intuition of external fact—this is the direct method; sometimes as an effect from which the external fact may be inferred—this is the indirect method. We use them alternatively without scruple; travelling from one to the other with no very clear consciousness of what we are doing.

But such a hand-to-mouth procedure can hardly be described as using them in intelligent combination. This can only be accomplished (so it seems to me) if the provinces of each are defined on some intelligible principle. I do not recall any serious attempt to do this except by the old and famous theory of the “primary” and “secondary” qualities of matter. In this scheme the direct and the indirect methods were both employed, and their spheres of operation more or less delimited. The primary qualities, such as shape and mass, were known, or were theoretically knowable, by immediate acquaintance. The secondary qualities, such as colour and sound, were supposed to be no more than effects produced by the action of the primary qualities on the perceiver's sensibilities; the real nature of that action being arrived at indirectly and by inference.

Modern philosophers have dealt somewhat roughly with this scheme, of which fragments still lie loosely embedded in much scientific exposition. But they have not seen the need for something of the kind, and have never provided a substitute. Untenable though it be, it had merits which it would be unkind to forget. Its realistic side did something to satisfy common sense; for it left standing at least the bare skeleton of that external world of matter with which we suppose ourselves to be familiar. At the same time it left a place for the indirect interpretation of perceptions, by treating the secondary qualities as effects, and seeking their causes among the energies of the primary. There was thus something for everybody—except perhaps the Neo-realists. Less than might be desired for common sense; still less for critics of common sense like Berkeley and Hume; but a good deal for science as science at that time stood. I am not sure that for a century and a half, or more, after the publication of Newton's “Principia,” physicists had much to say against it, however ill fitted it may have been to survive serious criticisms. But with every step in the progress of physics difficulties have multiplied. So long as science was content to explain the character of matter perceived in bulk by the behaviour of more or less similar matter, divided into molecules or atoms, these difficulties hardly obtruded themselves. Molecular was molar writ small; and to the pictorial imagination, the kind of thing which truly existed did not seem very alien to the kind of thing which appeared to be. We may remember Maxwell comparing atoms to “manufactured articles,” and the ordinary way of visualizing the kinetic theory of gases was to regard an enclosed gas as consisting of very diminutive and perfectly elastic material globules untiringly dashing now against their prison walls, now against each other. This sort of thing corresponds as well as we could expect with a theory of matter which divides its qualities into primary and secondary. Not so, however, do later developments.

In truth, through the progress of scientific knowledge, appearance and reality are now most widely sundered. The external world in its true character recedes more and more into the realm of the imperceptible and the unimaginable. We have no senses wherewith to apprehend it. On the other hand external objects, as we perceive them, are no more than a mirage of transitory effects, having little resemblance to their more enduring causes. And though we must indeed believe that physical realities are objects of knowledge, that their character may be expressed in appropriate concepts, and their movements formulated in suitable equations, it is plain that we shall never extract from the theory of “primary and secondary qualities” any substantial contribution to the philosophy by which these conclusions can be rationally supported.

§ 5

Let me now endeavour to recapitulate, from the point of view of methodological doubt, the results of this discussion on perception, and on the sort of access which perception gives us to the physical world of science and common sense.

In every case we may regard the causal process of perception as beginning, for the purposes of our argument, in the object perceived, and ending, after a complicated series of transformations, in the experience of some percipient. In every case, this end product of the causal process is also the intuitive beginning of the cognitive process. But it may initiate the cognitive process in two quite different ways. It may either provide the percipient with direct knowledge of the object perceived: or it may itself, when treated as an effect, constitute an important part of the material from which the character of the object can be indirectly conjectured.

Now can perception, thus schematically described, give us what we want? Can it, either in its direct form or its indirect, supply us with a body of intuitive truths sufficient to support our familiar beliefs about the world of matter? Let me re-submit concisely, but in a convenient order, the main reasons which suggest an unfavourable answer.

It will be recollected that in opening our enquiry I began with the causal aspect of perception. In these concluding observations I shall reverse the procedure; and starting from the perceptions which the causal process has produced, shall travel back by the cognitive road to the self-revealing world of nature—making a few notes by the way.

1. We cannot flatter ourselves that our theories of perception are in order, until the relation between the direct and the indirect methods of using it is systematically explained. I cannot think that this has as yet been accomplished; and if it be true, as I contend, that neither the direct nor the indirect method suffices by itself, the importance of the task is manifest.

2. The direct method is, I suppose, rarely used in its integrity. Our immediate perceptions of external fact are always, or almost always, contaminated with instinctive interpretations, which modify their primary import.

What we treat as intuitions are thus more and other than what is immediately given us through the senses. But how much more, and how far other, remains, so far as I am aware, still to be determined. It has moreover also to be determined whether, if purified from all extraneous accretions (were such an operation possible), these intuitions would not be so attenuated as to be quite unequal to the calls which science and common sense habitually make on them. I rather think they would. But however this may be, it is evidently most inconvenient, from the point of view of theory, that the direct method on which everything depends should so often in practice have to use perceptions qualified to an unknown degree by alien elements.

3. As regards the indirect method, its practical importance cannot be exaggerated, though it plays a part which of necessity is ancillary and dependent. Alone it is powerless. It cannot work in vacuo. It always assumes an external world partially revealed by the direct method, whose outline it can do no more than correct and expand. If therefore the direct method were wholly to fail us, the indirect method would provide no substitute.

4. So much for the particular premises on which our knowledge of nature depends; the unnumbered rootlets nourishing the growth whose final flower is science. Obviously we should next enquire how these premises can best be used; by what methods of inference we can proceed from perceptions treated as effects to the external causes which produce them; on what assumptions the individual can build on the experiences of the race, and how these assumptions are to be justified. These and other cognate questions are of vital importance to any philosophy of the familiar, but I have touched elsewhere on some of the many difficulties which beset them, and in this chapter they must be passed over with only a passing reminder that they are there, and cannot be ignored.

§ 6

We may now suppose ourselves to have reached the goal of the cognitive process, namely the physical universe as described by science. Let us assume that our double journey has been successfully accomplished—and that we know in broad outline both how our experiences of nature are produced, and how our knowledge of nature is arrived at. What do we make of the result? Do the two processes fit into the same picture? Can they be regarded as related fragments of one consistent whole?

It is hard to think so. When (as it were from the outside) we contemplate the elaborate process of perception-making, what is there to inspire confidence in the veracity of its final message? Why, for instance, should the complex physical and physiological transformations, which precede and produce an act of vision, turn out to be the oracles of truth? If they are, if we really see what really is, can anything short of a miracle explain the marvel?

Perhaps not, it may be replied; but why be troubled? The marvel never happens. What happens is that the oracle delivers itself of a message properly, if unkindly, described by those most competent to speak as crude and naïve; a message which travesties the facts, and gives us a notion of the physical world very different from any revealed by science.

This explanation, we may allow, relieves us of one difficulty, but does it not immediately plunge us in another? On what is science ultimately founded, unless it be on these same crude and naïve deliverances? From these we cannot escape. We cannot regard them as merely symbolical; we cannot treat them merely as effects; we depend upon them in the conduct of life; they are with us through all our explorations of nature; we begin with them; we return to them—yet all the time we know them to be false! Can any conclusion give more food for thought to those who faithfully employ the instrument of methodological doubt?

• 1.

  This argument, of course, assumes the trustworthiness of memory, which is not here in question.

• 2.

  The reader will please observe that I do not describe the perception of the red patch on the green patch as a sensation; nor do I wish to suggest that it is merely subjective. I desire to beg no philosophic question.