James Sully, answered from the texts and cited to the page.
There is a common prejudice that illusion is a mere defect, an accident of a mind gone momentarily wrong, of no interest to the student of its sound working. I hold the contrary: that illusion is one of the surest windows into the normal mind. The illusions of sense, of memory, of belief, of introspection itself are not produced by some special faculty of error; they are produced by the very same processes — the associations, the expectations, the swift unnoticed inferences — that in ordinary circumstances give us truth.12
The eye that is deceived by the bent stick in water is the eye that ordinarily sees truly, following the same habits of interpretation; the memory that misremembers is the memory that ordinarily recalls, betrayed by the same reconstructive tendency. To study where and how the mind goes wrong is therefore to catch its machinery working in the open, its ordinary operations thrown into relief by their failure.
Error is continuous with knowledge, and the psychology of illusion is the psychology of the normal mind read by the light of its mistakes.345
It is evident, indeed, that the very conception of an illusion implies a criterion of certainty: to call a thing illusory, is to judge it by reference to some accepted standard of truth. The mental processes involved in detecting, resisting, and overcoming illusion, are a very interesting subject for the psychologist, though we have not space here to investigate them fully.Illusions A Psychological Study, pp. 351–352
And illusion, in all its forms, is bad grouping or carelessly performed synthesis. This holds good even of the simplest kinds of error in which a presentative element is wrongly classed; and it holds good of those more conspicuous errors of perception, memory, expectation, and compound belief, in which representations connect themselves in an order not perfectly answering to the objective order.Illusions A Psychological Study, pp. 334–335
So again, to take a narrower group of illusions, we find that in the recalling of distant events the proportion of error is vastly greater than in the recalling of near events. So much as to the extent of illusion as brought to light by our preceding study. Let us now glance at the conclusions obtained respecting its nature and its causes. Causes of Illusion.Illusions A Psychological Study, pp. 331–332
In this way they closely resemble those slight errors of perception which are due to erroneous classing of senseimpressions.Illusions A Psychological Study, pp. 336–337
They are sometimes called deceptions of the senses; but this is a somewhat loose expression, suggesting that we can be deceived as to sensation itself, though, as we shall see later OD, this is only true in a very restricted meaning of the phrase. To speak correctly, senseillusions must be said to arise by a simulation of the form of just and accurate perceptions.Illusions A Psychological Study, pp. 19–20
This is Sully's most original and most durable idea. The strategy of studying normal mental function through its systematic breakdowns and errors became one of psychology's most productive methods: perceptual illusions as evidence of the visual system's inferential machinery; the cognitive-bias program of Kahneman and Tversky, which reads the mind's judgment heuristics off its systematic errors; neuropsychology's use of deficits to map normal function. The modern account of illusions as products of the perceptual system's inferential, expectation-driven (Bayesian) processing is precisely Sully's thesis — that the same machinery which usually yields truth yields the illusion under revealing conditions — given a rigorous mechanism. Here Sully anticipated a permanent feature of the science.