Subjective uncertainty, variability of experience, and the infant's response to discrepancies

4-month-old infants were assessed in a habituation/discrepancy paradigm using random shapes to test several propositions deduced from the discrepancy hypothesis. First, an inverted-U distribution of attention was found for discrepancies produced by random shapes which differ in the density of contours or spatial frequencies rather than in terms of specific overt pattern. Second, the distribution of attention was more similar for discrepancies defined by increases and by decreases in contour density for random shapes than for checkerboard stimuli. This result supported the role of the hypothesized intervening variable of subjective uncertainty, because while simple checkerboards are subsets of more complex checkerboards, simple random shapes are not unmodified portions of more complex shapes. Third, greater variability of experience with stimuli about a given level of standard stimulation reduced attention to simple discrepancies but increased attention to larger magnitudes of more complex discrepancies.