Preprint: “A silent disco: Persistent entrainment of low-frequency neural oscillations underlies beat-based, but not pattern-based temporal expectations”

Is something special about beat-based expectations in rhythm? In this paper, we try to find out! The preprint is now fully updated with all the latest analyses, and online on bioRxiv:

Bouwer, F.L., Fahrenfort, J.J., Millard, S.K., Kloosterman, N.A., Slagter, H.A. (preprint). A silent disco: Persistent entrainment of low-frequency neural oscillations underlies beat-based, but not pattern-based temporal expectations. bioRxiv, 2020.01.08.899278; doi:

Temporal expectations (e.g., predicting “when”) facilitate sensory processing, and are suggested to rely on entrainment of low frequency neural oscillations to regular rhythmic input, such as a beat in music (“beat-based” expectations). However, temporal expectations can also be based on predictable repeated patterns (“pattern-based” expectations). These two types of expectations often occur simultaneously, but whether they are subserved by the same neural mechanisms is currently a topic of active debate. Here, we addressed this outstanding issue by examining EEG activity and behavioral responses during silent periods following rhythmic auditory sequences designed to elicit only beat-based or pattern-based expectations, or with random timing. In Experiment 1 (N = 32), participants rated how well probe tones at various time points fitted the previous rhythm. Beat-based expectations affected fitness ratings for at least two beat-cycles, while the effects of pattern-based expectations subsided after the first expected time point in the silence window. In Experiment 2 (N = 27), using EEG, we found a Contingent Negative Variation (CNV) following the final tones of pattern-based, but not beat-based sequences. Moreover, we found enhanced power in the EEG signal at the beat frequency for beat-based sequences both during listening and the silence, while for pattern-based sequences, enhanced power at a pattern-specific frequency was only present during listening, not during the silence. Finally, we show how multivariate pattern decoding and multi scale entropy – measures sensitive to non-oscillatory components of the signal – can be used to probe temporal expectations. Taken together, we show that beat-based and pattern-based expectations affect behavior differentially, with beat-based expectations exerting longer-lasting effects than pattern-based expectations. At a neural level, climbing activity may specifically reflect pattern-based expectations, while persistent low frequency oscillations may be specific to beat-based expectations. When studying responses to complex rhythmic stimuli, like in music and language, both types of expectations should therefore be considered.