Abstract
What limits capacity in visual working memory (VWM)? Alvarez & Cavanagh (2004) reported that the greater the visual detail within an object, the fewer the objects stored. However, we found that VWM capacity for faces (Jackson & Raymond, 2004) is similar to that found for simple polygons. Perhaps, VWM capacity is not determined by ‘physical’ complexity but, rather, by ‘perceived’ complexity. We propose that, when people gain experience with a physically complex stimulus, it is perceived and encoded more simply, and thereby gains capacity. To test this we measured VWM capacity for Chinese characters (‘Hanzi’) in a group of fluent Chinese readers and a group of participants naïve to Hanzi. By using the same stimuli in both groups, physical complexity remained constant while only perceived complexity differed. To minimise input from verbal WM we used a verbal suppression task and ‘traditional’ Hanzi as stimuli. (Chinese readers, fluent only with modern, ‘simple’ Hanzi, were able to recognise stroke groupings but could not pronounce traditional Hanzi.) On each trial, stimuli were presented in a memory array (set sizes 1–8), followed 900 ms later by a test array. Participants reported whether one item had changed between arrays. To ensure sufficient encoding time, arrays were displayed for durations derived from each observer's mean RT in a Hanzi visual search pre-task. Chinese readers had superior change detection performance, p = .01, and capacity (K = 2.48), p < .01, compared to non readers of Chinese (K = 1.64). Furthermore, inverting the Hanzi significantly reduced Chinese readers' performance, p < .01, and capacity (K = 2.09), p = .03, whereas inversion had non-significant effects for non readers (K = 1.79). We conclude that experience enables the integration of detailed object information to form a perceptually simplified representation. The degree of within-object element integration, modulated by experience, therefore appears to determine capacity.