This research compared free recall and recognition memory for stereotype-consistent and stereotype-inconsistent information as a function of attentional capacity during encoding. Whereas recall was better for consistent information under conditions of limited capacity, recognition accuracy favored inconsistent information in the same conditions. Contrary to previous theorizations, these data demonstrate that stereotype-inconsistent information is encoded more thoroughly and represented more accurately in memory than stereotype-consistent information when resources are depleted. The recall advantage for consistent information appears to be due to retrieval advantages rather than more thorough encoding or representation. Implications of these findings for models of stereotype efficiency are discussed.

According to the encoding flexibility model, stereotypes are efficient because they facilitate, in different ways, the encoding of both stereotype-consistent and stereotype-inconsistent information when capacity is low. Because stereotypical information is conceptually fluent, it may be easily understood, even when resources are scant. As a result, processing resources may shift from stereotypical toward counterstereotypical information, which is difficult to comprehend under such conditions. Thus, whereas inconsistent information receives greater attention (Experiments 1-3) and perceptual encoding (Experiment 4) when resources are depleted, the conceptual meaning of consistent information is extracted to a greater degree under such conditions (Experiment 5). Potential moderating roles of stereotype strength and perceiver motivations are discussed, as are the implications of these results for dual process models of stereotyping.