The dimensionality of recognition memory: A state-trace analysis of the effects of dividing attention

Theorizing in psychology about latent processes often relies on inferences drawn from empirical interactions. Dual-process models of recognition memory are popular in part because of the straightforward explanations they provide for such empirical interactions, including those between memory for items and memory for contexts in which they occur. Attention is one variable that reveals such a dissociation: When attention is divided at encoding, memory for context appears to be more severely impaired than item memory. However, empirical dissociations, even double dissociations, can be found even when only one latent process underlies performance. State-trace analysis offers an alternative means of assessing the number of latent variables underlying recognition memory. Unlike the dissociation approach, it has the desirable characteristic of treating the dependent variables as having ordinal, not interval, measurement characteristics. In three experiments, memory for items and context was measured under full and divided attention across four levels of study time. Across all three experiments, state-trace analyses revealed that a single, common latent variable underlies performance in all conditions and so indicates the sufficiency of a single memory process in explaining the underlying dissociations. (PsycInfo Database Record (c) 2026 APA, all rights reserved).