Cognitive mechanisms at play: the role of supramodal representations and working memory in melodic perception

Abstract

Music is highly relational and in this manner shares much in common with other human behavior. While this may suggest that the processes used in music perception could be domain general, the characteristics and flexibility of these representations remain less understood. If the underlying representations required for perceiving music are shared with those in other domains it should be possible to map such representations to other domains. Through explicit and implicit learning tasks in Experiment 1 and 2, this hypothesis was partially supported through novel experiments with melodic stimuli in the auditory modality and analogous stimuli (Gabor sequences) in the visual modality. Transfer of representations was defined as successful mapping of categories from one modality to another. Two classes of representations (contour and intervals) were used in an explicit learning paradigm in Experiment 1. Crossmodal mapping was evident in three out of the four conditions, implying domain-generality and flexibility of representational transfer. Experiment 2 extended this to an implicit learning paradigm using both unimodal and crossmodal testing conditions. Under unimodal conditions participants were able to discriminate novel instances based on learned exemplars, however there was considerably less evidence of crossmodal discrimination for the same representations. These results provide novel insights into the differential flexibility of melodic representations under explicit and implicit learning conditions, and how explicit learning may play a role in creating domain-general representations. The final experiment explored the extent to which cognitive resources such as working memory underpin the processing of relational information in melodies. Participants in Experiment 3 listened to a melodic stream while performing concurrent n-back tasks with different gradients of working memory taxation. Following each task, participants were given a recognition test between two stimuli that matched the previously heard melodic stream on either relational or featural properties. A weak trend between memory taxation and rate of relational responses was observed, providing insights for future research. Combined, the research presented here extends knowledge of explicit and implicit learning, and their interactions with domain-general and domain-specific representations, providing a novel way of exploring not only music perception, but also relational processing and the underlying mechanisms.