Abstract

Humans are capable of learning sequential information from multiple stimulus dimensions simultaneously, but the factors influencing the selection of specific sequences for learning remains unclear. In order to investigate the role of cross-dimensional distinctiveness in sequence learning, in a serial reaction time task, sequences were presented on two dimensions that were very low in distinctiveness. Repeating sequences were presented in two visuospatial dimensions nested hierarchically within each other-a local spatial dimension that specified the correct response on each trial and a global spatial dimension that indicated the general display region of the stimulus. In the phase-repeat condition, the two sequences were consistently matched in phase, allowing an integrated representation of the two sequences to be formed. In the phase-change condition, the two sequences differed in length and were not correlated. In the phase-repeat condition, integrative learning was found for the cross-dimensional pattern, but individual sequence learning was not found. In the phase-change condition only the global sequence was learned, but not the local sequence. Thus, integrative and individual sequence learning did not occur simultaneously, neither was individual sequence learning complete. Furthermore, learning for individual sequences and the cross-dimensional information had an over-additive influence on performance. This pattern of results contrasts with previous research that showed simultaneous learning for both cross-dimensional and individual sequence information presented in highly distinct dimensions. The current results are attributed to the combination of the two closely related dimensions and suggest that individual and integrative sequence learning can be constrained due to interactions between indistinctive dimensions at encoding or working memory activation stages of processing.