The proposition which is fundamental to the discussion is that we have knowledge only where we perceive the agreement or disagreement of ideas. Locke then takes up each of his four classes of connection, in order to ascertain the extent of knowledge in it. Our knowledge of "identity and diversity extends as far as our ideas," because we intuitively perceive every idea to be "what it is, and different from any other." Locke afterwards states, however, that all purely identical propositions are "trifling," that is, they contain no instruction; they teach us nothing. Thus the first class of relations cannot be said to be of much avail. If we consider the fourth kind of knowledge, that of real existence, we have an intuitive knowledge of self, a demonstrative knowledge of God, and a sensitive knowledge of other things. But sensitive knowledge, it must be noted, "does not extend beyond the objects _actually present_ to our senses." It can hardly be said, therefore, to assure us of the existence of _objects_ at all. It only tells us what experiences are being at the time undergone. Furthermore, knowledge of all three (God, self, and matter), since of real being, and not of relations between ideas, contradicts his definition of knowledge. But perhaps we shall find knowledge more extended in the other classes. And indeed Locke tells us that knowledge of relations is the "largest field of our knowledge." It includes morals and mathematics; but it is to be noticed that, according to Locke, in both of these branches our demonstrations are not regarding facts, but regarding either "modes" framed by ourselves, or relations that are the creatures of our minds,--"extraneous and superinduced" upon the facts, as he says. He thus anticipates in substance, though not in phraseology, Hume's distinction between "matters of fact" and "connections of ideas," in the latter of which we may have knowledge, but not going beyond the combinations that we ourselves make.