Purchase this article with an account.
Johannes Haushofer, Chris I. Baker, Nancy Kanwisher; Frequency-based categorization of complex visual objects. Journal of Vision 2007;7(9):210. doi: https://doi.org/10.1167/7.9.210.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
How do we come to group visual objects into discrete categories? Prior evidence suggests a role for stimulus frequency (Rosenthal et al., 2001). We tested whether stimulus frequency could drive category formation for high-level visual objects, and what the underlying mechanisms are. We generated a two-dimensional continuum of complex visual stimuli which resembled eight-armed starfishes. Stimuli were presented following a bimodal frequency distribution: two regions of the stimulus space were shown with high frequency (“peaks”), while intermediate stimuli were shown less often. Subjects categorized the stimuli into “male” and “female”. Importantly, no feedback was given; this enabled us to assess which criteria were used in categorization. We found that subjects closely followed the frequency distribution: one frequency peak was categorized as “male” and the other as “female”, and the category boundary ran along the frequency minimum between the peaks. Since the category boundary was oriented obliquely to the physical dimensions, this suggests that frequency can influence categorization independently of physical stimulus dimensions. Categorization was most stable, and reaction times were shortest, at the extremes of the frequency spectrum, consistent with the formation of a frequency-based category axis. Remarkably, frequency-based categorization emerged within less than 50 trials, suggesting high sensitivity to frequency information. In a second experiment, we asked whether passive exposure to the frequency distribution is sufficient to influence categorization. Subjects passively viewed the same frequency distribution and subsequently guessed the category membership of each stimulus. Again judgments closely followed the frequency distribution. These results suggest that i. stimulus frequency can strongly influence the categorization of complex visual objects, ii. this influence is independent of physical dimensions and may be based on the establishment of a frequency-based category axis, iii. learning of frequency information is extremely fast, iv. passive exposure is sufficient for frequency to influence categorization.
This PDF is available to Subscribers Only