Essays in Experimental Logic

by John Dewey

III

For simplicity's sake, I have written as if my main problem were to show how, in the face of a supposed difficulty, a strictly realistic theory of the perceptual event may be maintained. But my interest is primarily in the facts, and in the theory only because of the facts it formulates. The significance of the facts of the case may, perhaps, be indicated by a consideration which has thus far been ignored. In regarding a perception as a case of knowledge, the presentative realist does more than shove into it a relation to mind which then, naturally and inevitably, becomes the explanation of any differences that exist between its subject-matter and some causal object with which it contrasts. In many cases—very important cases, too, in the physical sciences—the contrasting "real object" becomes known by a logical process, by inference—as the contemporary position of the star is determined by calculations from data, not by perception. This, then, is the situation of the presentative realist: If perception is knowledge of its cause, it stands in unfavorable contrast with another indirect mode of knowledge; its object is less valid than the object of inference. I do not adduce these considerations as showing that the case is hopeless for the presentative realist;[61] I am willing to concede he can find a satisfactory way out. But the difficulty exists; and in existing it calls emphatic attention to a case which is certainly and indisputably a case of knowledge—namely, propositions arrived at through inference, judgments as logical assertions.

With relation to the unquestionable case of knowledge, the logical or inferential case, perceptions occupy a unique status, one which readily accounts for their being regarded as cases of knowledge, although in themselves they are natural events. (1) They are the sole ultimate data, the sole media, of inference to all natural objects and processes. While we do not, in any intelligible or verifiable sense, know them, we know all things that we do know with or by them. They furnish the only ultimate evidence of the existence and nature of the objects which we infer, and they are the sole ultimate checks and tests of the inferences. The visible light is a necessary part of the evidence on the basis of which we infer the existence, place, and structure of the astronomical star, and some other perception is the verifying check on the value of the inference. Because of this characteristic use of perceptions, the perceptions themselves acquire, by "second intention," a knowledge status. They become objects of minute, accurate, and experimental scrutiny. Since the body of propositions that forms natural science hangs upon them, for scientific purposes their nature as evidence, as signs, entirely overshadows their natural status, that of being simply natural events. The scientific man, as scientific, cares for perceptions not in themselves, but as they throw light upon the nature of some object reached by evidence. And since every such inference tries to terminate in a further perception (as its test of validity), the value of inferential knowing depends on perception. (2) Independently of science, daily life uses perceptions as signs of other perceptions. When a perception of a certain kind frequently recurs and is constantly used as evidence of some other impending perceptual event, the function of habit (a natural function, be it noted, not a psychical or epistemological function) often brings it about that the perception loses its original quality in acquiring a sign-value. Language is, of course, the typical case. Noises, in themselves mere natural events, through habitual use as signs of other natural events become integrated with what they mean. What they stand for is telescoped, as it were, into what they are. This happens also with other natural events, colors, tastes, etc. Thus, for practical purposes, many perceptual events are cases of knowledge; that is, they have been used as such so often that the habit of so using them is established or automatic.

In this brief reference to facts that are perfectly familiar, I have tried to suggest three points of crucial importance for a naïve realism: first, that inferential or evidential knowledge (that involving logical relation) is in the field as an obvious and undisputed case of knowledge; second, that this function, although embodying the logical relation, is itself a natural and specifically detectable process among natural things—it is not a non-natural or epistemological relation; third, that the use, practical and scientific, of perceptual events in the evidential or inferential function is such as to make them become objects of inquiry and limits of knowledge, and to such a degree that this acquired characteristic quite overshadows, in many cases, their primary nature.

If we add to what has been said the fact that, like every natural function, the inferential function turns out better in some cases and worse in others, we get a naturalistic or naïvely realistic conception of the "problem of knowledge": Control of the conditions of inference—the only type of knowledge detectable in direct existence—so as to guide it toward better conclusions.