All such previsions are supposed to be formed by a process of inference from personal experience, including the trustworthiness of testimony.  Even allowing, however, that this was so in the first stages of the belief, it is plain that, by dint of frequent renewal, the expectation would soon cease to be a process of inference, and acquire an apparently self-evident character.  This being so, if the expectation is not adequately grounded to start with, it is very likely to develop into an illusion.  And it is to be added that these permanent anticipations may have their origin much more in our own wishes or emotional promptings than in fact and experience.  The mind undisciplined by scientific training is wont to entertain numerous beliefs of this sort respecting what is now going on in unvisited parts of the world, or what will happen hereafter in the distant future.  The remote, and therefore obscure, in space and in time has always been the favourite region for the projection of pleasant fancies.

Once more, besides these oblique kinds of expectation, I may form other seemingly simple beliefs, to which the term expectation seems less clearly applicable.  Thus, on waking in the morning and finding the ground covered with snow, my imagination moves backwards, as in the process of memory, and realizes the spectacle of the softly falling snow-flakes in the hours of the night.  The oral communication of others’ experience, including the traditions of the race, enables me to set out from any present point of time, and reconstruct complex chains of experience of vast length lying beyond the bounds of my own personal recollection.

I need not here discuss what the exact nature of such beliefs is.  J.S.  Mill identifies them with expectations.  Thus, according to him, my belief in the nocturnal snowstorm is the assurance that I should have seen it had I waited up during the night.  So my belief in Cicero’s oratory resolves itself into the conviction that I should have heard Cicero under certain conditions of time and place, which is identical with my expectation that I shall hear a certain speaker to-morrow if I go to the House of Commons.[140] However this be, the thing to note is that such retrospective beliefs, when once formed, tend to approximate in character to recollections.  This is true even of new beliefs in recent events directly made known by present objective consequences or signs, as the snowstorm.  For in this case there is commonly no conscious comparison of the present signs with previously known signs, but merely a direct quasi-mnemonic passage of mind from the present fact to its antecedent.  And it is still more true of long-entertained retrospective beliefs.  When, for example, the original grounds of an historical hypothesis are lost sight of, and after the belief has hardened and solidified by time, it comes to look much more like a recollection than an expectation.  As a matter of fact, we have seen, when studying the illusions of memory, that our personal experience does become confused with that of others.  And one may say that all long-cherished retrospective beliefs tend to become assimilated to recollections.