The role of virtual standards in visual discrimination

Abstract

Psychometric functions were measured in various visual discrimination tasks involving test stimuli whose values spanned a standard stimulus. In two-interval trial blocks, the standard was either always in the first or always in the second interval, or appeared randomly in either interval. In one-interval blocks, the standard stimulus was never presented. Fitting the data with cumulative Gaussian functions revealed that discriminability was highest on one interval trials, where the observer had to rely on an implicit standard. On two-interval trials, discriminability was higher when the standard was in the first rather than the second interval, regardless of whether those two types of trials were intermixed or not, also possibly implicating the operation of an implicit standard in two-interval trials as well. A time-order error occurs on two-interval trials: in effect the value of the stimulus presented in the first interval is underestimated relative to that in the second interval. An analogous error occurs in one-interval trials, as if there were an implicit standard whose value is underestimated.