Attribute amnesia as a product of experience-dependent encoding.

Attribute amnesia, a phenomenon in which participants fail to report a just-attended attribute in a surprise test, reflects the importance of expectation in determining memory for attended information. To investigate how expectations arise in the context of attribute amnesia, the present study examined whether and how different response histories, independently of task instruction, can shape expectation, thereby driving or eliminating attribute amnesia. Participants were assigned to three groups and completed variations of the attribute amnesia task, where they were initially instructed to encode both target location and identity. Two groups of participants were probed four times on target identity before a critical identity probe, in one case intermittently while in the other case repeatedly during the first few trials. Another group of participants was never probed on identity until the critical trial, which occurred on the 370th trial (after many location probes). The results showed that, in spite of common task instruction, performance on the critical trial depended strongly on response history, with initial identity probes providing some protection against attribute amnesia and intermittent probes completely eliminating it. The findings suggest that the encoding of information into working memory is determined by task experience, independently of task instruction.