To the beat of your own drum: Cortical regularization of non-integer ratio rhythms toward metrical patterns

Abstract

Humans perceive a wide range of temporal patterns, including those rhythms that occur in music, speech, and movement; however, there are constraints on the rhythmic patterns that we can represent. Past research has shown that sequences in which sounds occur regularly at non-metrical locations in a repeating beat period (non-integer ratio subdivisions of the beat, e.g. sounds at 430ms in a 1000ms beat) are represented less accurately than sequences with metrical relationships, where events occur at even subdivisions of the beat (integer ratios, e.g. sounds at 500ms in a 1000ms beat). Why do non-integer ratio rhythms present cognitive challenges? An emerging theory is that non-integer ratio sequences are represented incorrectly, “regularized” in the direction of the nearest metrical pattern, and the present study sought evidence of such perceptual regularization toward integer ratio relationships. Participants listened to metrical and non-metrical rhythmic auditory sequences during electroencephalogram recording, and sounds were pseudorandomly omitted from the stimulus sequence. Cortical responses to these omissions (omission elicited potentials; OEPs) were used to estimate the timing of expectations for omitted sounds in integer ratio and non-integer ratio locations. OEP amplitude and onset latency measures indicated that expectations for non-integer ratio sequences are distorted toward the nearest metrical location in the rhythmic period. These top-down effects demonstrate metrical regularization in a purely perceptual context, and provide support for dynamical accounts of rhythm perception.