some philosophers, that there are certain
innate principles of knowledge, it seems now to be generally allowed
that these, apart from the particular facts of experience, are merely
abstractions; and that they only develop into complete knowledge when
they receive some empirical content, which must be supplied either by
present perception or by memory. So that in this case, too, all definite
concrete knowledge would seem to be either presentative cognition,
memory, or, lastly, some mode of inference from these.

A little inquiry into the mental operations which I here include under
the name belief will show, however, that they are by no means uniformly
process of inference. To take the simplest form of such knowledge,
anticipation of some personal experience: this may arise quite apart
from recollection, as a spontaneous projection of a mental image into
the future. A person may feel "intuitively certain" that something is
going to happen to him which does not resemble anything in his past
experience. Not only so; even when the expectation corresponds to a bit
of past experience, this source of the expectation may, under certain
circumstances, be altogether lost to view, and the belief assume a
secondarily automatic or intuitive character. Thus, a man may have first
entertained a belief in the success of some undertaking as the result of
a rough process of inference, but afterwards go on trusting when the
grounds for his confidence are wholly lost sight of.

This much may suffice for the present to show that belief sometimes
approximates to immediate, or self-evident, conviction. How far this is
the case will come out in the course of our inquiry into its different
forms. This being so, it will be needful to include in our present study
the errors connected with the process of belief in so far as they
simulate the immediate instantaneous form of illusion.

What I have here called belief may be roughly distinguished into simple
and compound belief. By a simple belief I mean one which has to do with
a single event or fact. It includes simple modes of expectation, as well
as beliefs in single past facts not guaranteed by memory. A compound
belief, on the other hand, has reference to a number of events or facts.
Thus, our belief in the continued existence of a particular object, as
well as our convictions respecting groups or classes of events, must be
regarded as compound, since they can be shown to include a number of
s