Two experiments are described which investigate recall of hierarchically organized melodic sequences. An adaptation of the technique of melodic dictation was employed, in which musically trained listeners notated each sequence after it was presented. All sequences were 12 tones in length. Pronounced effects of sequence structure were demonstrated. For those sequences that could be parsimoniously encoded in hierarchical fashion, near‐perfect recall was obtained. Performance was substantially inferior for sequences that could not be parsimoniously encoded. The effect of temporal segmentation was also investigated, and this was found to have a powerful effect on perceived structure, virtually overriding that of tonal structure. Temporal groupings tended to be retained or lost in an all‐or‐none fashion. These findings are related to the theoretical assumption that we encode tonal materials by inferring sequence structures and alphabets at different hierarchical levels, together with rules of combination. [Work supported by NIMH.]

This content is only available via PDF.