Past experience influences object representation in working memory Brandon M. Wagar * , Mike J. Dixon Department of Psychology, University of Waterloo, Waterloo, Ont., Canada N2L 3G1 Accepted 12 August 2004 Abstract The nature of object representation in working memory is vital to establishing the capacity of working memory, which in turn shapes the limits of visual cognition and awareness. Although current theories discuss whether representations in working memory are feature-based or object-based, no theory has considered the role of past experience. However, work with humans and non-hu- man primates suggests that once participants learn which features are important for category membership, these diagnostic features become more salient than non-diagnostic features in long-term memory and object recognition. Critically, the brain areas involved in this diagnosticity effect are also recruited during working memory tasks. We report two experiments testing whether a diagnos- ticity effect exists in working memory; and whether it is present when visual information is encoded into working memory, or if it is the result of maintenance within working memory. Results showed a diagnosticity effect which was present at encoding. Mainte- nance did not influence the nature of object representation in working memory. These findings show that the meaning we glean from our past experience has a profound influence on the nature of object representation in working memory. Ó 2004 Elsevier Inc. All rights reserved. 1. Introduction Working memory (WM) consists of a set of capacity limited short-term information stores and an attentional component responsible for the maintenance and manip- ulation of information within awareness (Baddeley, 1986; Cowan, 1995). There is general agreement that the capacity for visual information stored in WM is around four ‘‘items’’ (Cowan, 2001; Vogel, Woodman, & Luck, 2001; Wheeler & Treisman, 2002). What re- mains to be determined is the nature of object represen- tation in WM. Current theories propose that visual information stored in WM is feature-based (the number of features in the display limits capacity, see Frick, 1988), object-based (the number of objects in the display limits capacity, see Vogel et al., 2001), or based on a combination of features and objects (the number of fea- tures in the display limits capacity, but feature conjunc- tion can group features into higher order chunks, see Wheeler & Treisman, 2002). Implicit in the feature- based versus object-based discussion is the assumption that all features (whether individual or fused into ob- jects) have an equal likelihood of being represented in WM. To our knowledge, no theory has considered the role of past experience when attempting to discern the nature of object representation in WM. Research in object recognition has shown that cate- gorizing objects influence how they are perceived and represented in long-term memory (Olivia & Schyns, 1997; Schyns & Rodet, 1997). If certain features are more important for categorizing objects, then these diagnostic features exhibit increased salience relative to non-diagnostic features in long-term memory. Although little is known about the neural basis of the categorical information that gives visual features mean- ing (e.g., diagnosticity effect), the neural correlate of such information is considered to reside in the brain areas that process visual form, such as the inferior tem- poral (IT) and prefrontal cortex (PFC). In humans, cat- egorization involves the IT cortex, and damage to this area leads to impairments in categorization (see Farah, www.elsevier.com/locate/b&c Brain and Cognition 57 (2005) 248–256 0278-2626/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.bandc.2004.08.054 * Corresponding author. Fax: +1 519 746 8631. E-mail address: bmwagar@watarts.uwaterloo.ca (B.M. Wagar).