Essays in Experimental Logic

by John Dewey

[58] I am indebted to Dr. Bush's article on "Knowledge and Perception," Journal of Philosophy, Psychology, and Scientific Methods, Vol. VI, p. 393, and to Professor Woodbridge's article on "Perception and Epistemology" in the James Memorial Volume, as well as to his paper on "Sensations," read at the 1910 meeting of the American Philosophical Association. Since my point of departure and aim are somewhat different, I make this general acknowledgment in lieu of more specific references.

[59] Plato's use of shadows, of reflections in the water, and other "images" or "imitations" to prove the presence in nature of non-being was, considering the state of physical science in his day, a much more sensible conclusion than the modern use of certain images as proof that the object in perception is a psychical content. Hobbes expressly treats all images as physical, as on the same plane as reflections in the water and echoes; the comparison is his.

[60] It is impossible, in this brief treatment, to forestall every misapprehension and objection. Yet to many the use of the term "seen" will appear to be an admission that a case of knowledge is involved. But is smelling a case of knowledge? Or (if the superstition persists as to smell) is gnawing or poking a case of knowledge? My point, of course, is that "seen" involves a relation to organic activity, not to a knower, or mind.

[61] This is the phase of the matter, of course, which the rationalistic or objective realist, the realist of the type of T. H. Green, emphasizes. Put in terms of systems, the difficulty is that in escaping the subjectivism latent in treating perception as a case of knowledge, the realist runs into the waiting arms of the objective idealist.

[62] Professor Perry says (The New Realism, p. 115): "Professor Dewey is mistaken in supposing that realism assumes 'the ubiquity of the knowledge-relation.' Realism does not argue from the 'ego-centric predicament,' i.e., from the bare presence of the knowledge-relation in all cases of knowledge." If the text has not made my point clear, it is probably too much to expect that a footnote will do so. But I have not accused the realist of arguing from the ego-centric predicament. I have said that if any realist holds that the sole and exclusive relation of the one who is knower to things is that of being their knower, then the realist cannot escape the impact of the predicament. But if the one who knows things also stands in other connections with them, then it is possible to make an intelligible contrast between things as known and things as loved or hated or appreciated, or seen or heard or whatever. The argument, it should be noted, stands in connection with that of the last section as to whether hearing a sound and seeing a color are of themselves (apart from the use made of them in inference) cases of knowledge. It is significant that Perry holds (New Realism, p. 150) that "sensing" is per se a case of knowing. Hence it must be in relation to a knower; it must fall within the "predicament," for "it makes the mind aware of a characteristic of the environment." That it is used (or may be used) to make us aware of some characteristic of the environment, I of course hold. To say that it is an awareness by the mind of a characteristic of the environment is at once to involve a philosopher immediately in the discussion of whether red qualities, or only certain vibrations, are "really" characteristics of the environment. Then, when the authority of physics is invoked in behalf of the latter proposition, the epistemologist (however realistic in his intention) is forced to consider color as a misapprehension of the environment, a case of error or illusion, while the idealist triumphantly flourishes it as a case of the transformative or constitutive efficacy of "mind" in knowing. But if the color is simply a natural event, and if "mind" does not enter except when color is made the basis of inference to some characteristic of the environment, then there is no predicament; and there is no problem of error save as a false inference is made. Moreover, since errors in inference are an undoubted fact, the principle that entities are not to be multiplied beyond need gives a prima facie superiority to any theory which connects all error with inference till adequate evidence to the contrary is produced.

[63] I shall pass over the terms "our own" so far as specific reference is concerned, but the method employed applies equally to them. Who are the "we," and what does "own" mean, and how is ownership established?

[64] Contrast the statement: "When I speak of a fact, I do not mean one of the simple things of the world, I mean that a certain thing has a certain quality, or that certain things have a certain relation" (p. 51).

[65] In view of the assumption, shared by Mr. Russell, that there is such a thing as non-inferential knowledge, the conception that a thing offers evidence for itself needs analysis. Self-evidence is merely a convenient term for disguising the difference between the indubitably given and the believed in. Hypotheses, for example, are self-evident sometimes, that is, obviously present for just what they are, but they are still hypotheses, and to offer their self-evident character as "evidence" would expose one to ridicule. Meanings may be self-evident (the Cartesian "clear and distinct") and truth dubious.

[66] "Really known" is an ambiguous term. It may signify understood, or it may signify known to be there or given. Either meaning implies reference beyond.

[67] The reply implies that the exhaustive, all-at-once perception of the entire universe assumed by some idealistic writers does not involve any external world. I do not make this remark for the sake of identifying myself with this school of thinkers, but to suggest that the limited character of empirical data is what occasions inference. But it is a fallacy to suppose that the nature of the limitations is psychologically given. On the contrary, they have to be determined by descriptive identifications which involve reference to the more extensive world. Hence no matter how "self-evident" the existence of the data may be, it is never self-evident that they are rightly delimited with respect to the specific inference in process of making.