The evolutionary neuroscience of musical beat perception: the Action Simulation for Auditory Prediction (ASAP) hypothesis

Abstract

a perceived periodic pulse that structures the perception of musical rhythm and which serves as a framework for synchronized movement to music. What are the neural mechanisms of musical beat perception, and how did they evolve? One view, which dates back to Darwin and implicitly informs some current models of beat perception, is that the relevant neural mechanisms are relatively general and are widespread among animal species. On the basis of recent neural and cross-species data on musical beat processing, this paper argues for a different view. Here we argue that beat perception is a complex brain function involving temporally-precise communication between auditory regions and motor planning regions of the cortex (even in the absence of overt movement). More specifically, we propose that simulation of periodic movement in motor planning regions provides a neural signal that helps the auditory system predict the timing of upcoming beats. This "action simulation for auditory prediction" (ASAP) hypothesis leads to testable predictions. We further suggest that ASAP relies on dorsal auditory pathway connections between auditory regions and motor planning regions via the parietal cortex, and suggest that these connections may be stronger in humans than in non-human primates due to the evolution of vocal learning in our lineage. This suggestion motivates cross-species research to determine which species are capable of human-like beat perception, i.e., beat perception that involves accurate temporal prediction of beat times across a fairly broad range of tempi.

Keywords: brain; comparative psychology; evolution; music cognition; rhythm perception.

Figure 1

Top: Spectrogram of an excerpt of jazz music (“Stompin at the Savoy,” by Benny Goodman; for corresponding audio, see supplementary sound file 1). Inverted arrows above the spectrogam show times of double bass and snare drum onsets, respectively. Bottom: time at which 9 human subjects (S1–9) tapped when instructed to “tap to the beat you hear.” Each tap is indicated by a vertical red bar. See text for details.

Figure 2

Sound evoked responses are modulated by beat perception, from Iversen et al. (2009). (A) Illustration of two conditions in the study of Iversen et al. (2009). A simple two-note repeating pattern is heard by listeners. On some trials listeners imagine the beat is on the first tone (condition IB1); on other trials, they imagine the beat is on the 2nd tone (condition IB2). The accents indicate the imagined beat and do not correspond to any physical differences in the stimulus, which were identical in the two conditions. (B) Normalized evoked neural responses for the two imagined beat conditions, measured with MEG. Thin gray vertical lines at 0 and 200 ms indicate onset of the two tones (each 45 ms long and 1 KHz in frequency). Solid blue line: evoked response when beat was imagined on tone 1; dashed red line: evoked response when beat was imagined on tone 2. Grand averages are shown for three frequency bands: Event-related field ERF (1–10 Hz), beta (20–30 Hz), and gamma (30–50 Hz). For beta and gamma frequencies, the mean power envelopes were averaged across individuals after first normalizing each individual's peak power across both conditions to one. Statistically significant effects of imagined beat location occurred only in the beta frequency response, where the response to both tones 1 and 2 was larger when that tone was imagined to be on the beat (arrows).

Figure 3

Patterns of induced beta-band neural activity as a function of imagined beat location in a syncopated rhythm, adapted from Iversen et al. (2009). (A) Constant rhythmic pattern where individual notes are indicated by black rectangles. (B) Music notation of the same rhythm pattern. (C) On different trials, participants mentally organized the perceived beat structure of a syncopated rhythm so that all beats fell on sounded tones (condition IB0) or some beats fell on silent positions just before (IB−) or after (IB+) sounded tones, (IB = Imagined Beat). The horizontal line with vertical tick marks indicates the timing of the imagined beats, and the rectangles above indicate the repeating rhythmic unit for each beat organization. (D) Music notation of three beat structures and associated rhythms shown in (C). Black notes show sounded tones, while notes with dotted note heads show imagined beats. The sounded-tone pattern is physically identical in all three conditions, but is psychologically distinct depending on where one feels the beat. In both (C,D) black, red and blue squares indicate the analysis window for MEG data, which is centered on the same tone. (E) Patterns of induced beta-band neural activity for beats on actual tones (marked by the vertical black line at 0 ms) vs. beats on silent positions just before (red) or after (blue) sounded tones. For the beat-at-silent-position conditions, the vertical dashed lines show the location of the imagined beats, relative to the sounded tone at time 0, and the grand mean normalized fluctuation of induced beta-band activity shows a peak of power that reflects the timing of the imagined beat (arrows), not the auditory input. Note how in all three conditions, the power of the induced beta-band signal rises in anticipation of the time of the beat, and sharply decreases around the time of the beat. Beta-band fluctuation was computed by subtracting the mean over the entire interval (−300 to 300 ms).

Figure 4

Model of dual-stream auditory processing in the primate brain, from Rauschecker (2011). Dorsal (red) and ventral (green) auditory pathways are shown in the macaque brain (A) and the human brain (B). Solid arrows indicate ascending projections from auditory cortex, while dashed arrows indicate reciprocal projections back to the auditory cortex. AC, auditory cortex; AL/CL, anterolateral/caudolateral superior temporal gyrus; CS, central sulcus; DLPFC, dorsolateral prefrontal cortex; IFC, inferior frontal cortex; IPL, inferior parietal lobule; IPS, inferior parietal sulcus; PFC, prefrontal cortex; PMC, premotor cortex; STS, superior temporal sulcus; VLPFC, ventrolateral prefrontal cortex.

Figure 5

Details of some of the fiber tracts thought to be involved in the dorsal auditory processing stream in humans, from Gierhan (2013). 44, Brodman Area 44; AG, angular gyrus; dPMC, dorsal premotor cortex; pSTG/MTG, posterior superior temporal gyrus/middle temporal gyrus; PTL, posterior temporal lobe; SMG, supramarginal gyrus; vPMC, ventral premotor cortex.

Figure 6

Circular histograms of relative phase values for human (A) vs. chimpanzee (B) taps to an auditory metronome with a period of 600 ms. In these plots asynchronies between taps and tones are expressed as relative phase values: 0° indicates taps perfectly aligned with tones, 180° indicates taps midway between tones, negative values (e.g., −15°) indicate taps before tones, and positive values (e.g., 15°) indicate taps after tones. Human data (Left) from Iversen et al., in press. Chimpanzee data (Right) redrawn from Hattori et al. (2013). In both graphs, the mean relative phase angle is shown with an inset arrow.