From How We Think (1910)
Conceptions are not derived from a multitude of different definite objects by leaving out the qualities in which they differ by retaining those in which they agree. The origins of concepts is sometimes described to be as if a child began with a lot of different particular things, say his particular dogs; his own Fido, his neighbor’s Carlo, his cousin’s Tray. Having all these different objects before him, he analyzes them into a lot of different qualities, say (a) color, (b) size, (c) shape, (d) number of legs, (e) quantity and quality of hair, (f) digestive organs and so on; and them strikes out all the unlike qualities (such as color, size, shape, hair), retaining traits such as quadruped and domesticated, which they all have in general.
As a matter of fact, the child begins with whatever significance he has got out of the one dog and has seen, heard, and handled. He has found that he can carry over from one experience of this object to subsequent experience certain expectations of certain characteristic modes of behavior – may expect these even before they show themselves. He tends to assume this attitude of anticipation whenever any clue or stimulus presents itself’ whenever the object gives him any excuse for it. Thus he might call cats little dogs, or horses big dogs. But finding that other expected traits and modes of behavior are not fulfilled, he is forced to throw out certain traits from the dog-meaning, while he contrasts certain other traits are selected and emphasized. As he further applies the meaning to other dogs, the dog-meaning gets still further defined and refined. He does not begin with a lot of ready-made objects from which he extracts a common meaning; he tries to apply every new experience whatever from is old experience will help him understand it, and as this process of constant assumption and experience is fulfilled and refuted by results, his conceptions get body and clearness.