A word must be added about Hume's division of perceptions into impressions and ideas. The point left out of sight in this analysis is that impressions of sense habitually find their reflexes not in revived sensations, but in expressions, in motor reactions which, with human beings, mostly take the form of words uttered or thought. These, no doubt, are associated to some small extent with revived sensations; but they are more commonly grouped with other words, with movements of the limbs, and with actions on the material or human environment of the percipient. Such expressions are incomparably easier to revive in memory, imagination, or expectation than the impressions that originally excited them; and, indeed, it is in connection with them that such revivals of sensation {81} as we actually experience take place. And it is probable that to this active side of our consciousness that we may trace those associative processes which Hume studies next in his analysis of human knowledge. Putting aside principles of doubtful or secondary value, the relations between states of consciousness that first offer themselves to view are, according to Hume, Co-existence and Succession (united under the name of Contiguity), Resemblance, and Causation. It is with the account he gives of this last category that his name is inseparably associated, for from it all subsequent speculation has taken rise. Yet primarily he seems to have had no other object in view than to simplify the laws of knowledge by resolving one of them into a particular case of another, and thus reducing his three categories to two. The relation of cause and effect, he tells us, is no more than a certain relation between antecedent and consequent in time where the sequence is so habitual as to establish in our minds a custom of expecting the one whenever the other occurs. The sequence is not necessary, for one can think, without any self-contradiction, of a change which has not been preceded by another change; nor is it, like the truths of geometry, something that can be known _à priori_. Without experience no one could tell that bread will nourish a man and not nourish a lion, nor even predict how a billiard-ball will behave when another ball strikes it. Should it be objected that the _à priori_ knowledge of a general principle need not involve an equal knowledge of nature's operations in particular cases, Hume would doubtless reply by saying that there is no abstract idea of causation apart from its concrete exemplifications.