Parallel Acquisition of Uncorrelated Sequences is Hard to Find

The distinction between implicit and explicit learning is fundamental to theories of human learning. However, it remains an open question whether implicit and explicit learning are based on a single, unitary learning meachnism, or multiple learning mechanisms or systems. The dual-systems model (Keele, Ivry, Mayr, Hazeltine, & Heuer, 2003) posits that, in addition to a multidimensional learning system that is accessible to consciousness, implicit learning may also proceed in a unidimensional learning. Learning in this system is considered to proceed within a set of dimension-specific, encapsulated processing modules, where strong separation between different kinds of information protects against interference. Parallel acquisition of multiple uncorrelated sequences has been considered crucial evidence in favor of encapsulated processing. However, recent work has demonstrated that parallel learning is also well accounted for by a single learning mechanism, and can therefore not provide convincing evidence for encapsulated processing. In a series of three experiments, we therefore aimed at testing the assumption of encapsulated processing more stringently. To this aim, we orthogonally manipulated sequences of stimulus and response features, and tested for selective influence of these sequences on the cognitive processes that are involved in SRTT performance. In Experiments 1 and 2, stimulus colors and response locations followed independent six-item probabilistic sequences. In Experiment 3, stimulus colors and stimulus locations followed such independent sequences. Replicating previous findings, we found learning of sequences of response locations. By contrast, we consistently found evidence against learning of sequences of stimulus colors or stimulus locations. First implications for the debate on encapsulated processing, and the organization of sequence learning systems, are discussed.