Amodal population clock in the primate medial premotor system for rhythmic tapping

Abstract

Precise timing is a fundamental requisite for a select group of complex actions, such as music appreciation and performance, where subjects extract the regular beat of a rhythmic sequence to generate an internal pulse representation that allows for predictive responses to the beat. The neural substrate for beat extraction and response entrainment to auditory and visual rhythms is still largely unknown. Here we recorded and analyzed the population activity of hundreds of MPC neurons of two rhesus monkeys performing an isochronous tapping task guided by brief flashing stimuli or auditory tones. The animals showed a strong bias towards visual metronomes, with rhythmic tapping that was more precise and accurate than for auditory metronomes. The population dynamics in state space as well as the corresponding neural sequences shared the following properties across modalities: the circular dynamics of the neural trajectories and the neural sequences formed a regenerating loop for every produced interval, producing a relative time representation; the trajectories converged in similar state space at tapping times while the moving bumps restart at this point, resetting the beat-based clock; the tempo of the synchronized tapping was encoded by a combination of amplitude modulation and temporal scaling in the neural trajectories. The latter correlated with a mixture of response duration and recruitment properties in the neural sequences. In addition, the modality induced a displacement in the neural trajectories in auditory and visual subspaces without greatly altering time keeping mechanism. These results suggest that the interaction between the amodal internal representation of pulse within MPC and a modality specific external input generates a neural rhythmic clock whose dynamics define the temporal execution of tapping using auditory and visual metronomes.

Monkeys have rhythm

“…At a theoretical level, it also provides evidence in favor of the 4Cs hypothesis which supposes that many other species may show some ability to sense musical beat, opening the study of musical beat perception to the larger range of high spatial and temporal resolution recording and manipulation techniques possible in animal models. Such techniques are already shedding light on the auditory and premotor cortical dynamics at play during sensory-motor synchronization to metronomes [30, 31, 73], which coincides with evidence from human studies that implicate auditory-motor [27, 74–76] and fronto-parietal networks [77] in audiomotor synchronization.…”

“…Such techniques are already shedding light on the auditory and premotor cortical dynamics at play during sensory-motor synchronization to metronomes [30,31,73], which coincides with evidence from human studies that implicate auditory-motor [27,[74][75][76] and fronto-parietal networks [77] in audiomotor synchronization.…”

“…However, macaques are able to produce human-like predictive and tempo-flexible synchronization to metronome beats when reward was made contingent on the asynchrony between taps and beats [28], indicating that predictive synchronization can be trained in a non vocal-learning species. As a result of establishing the macaque as a model of predictive auditorymotor synchronization to explicit metronome beats, extracellular recordings performed in awake, behaving macaques are beginning to reveal in unprecedented detail the neural activity patterns and dynamics that underlie this ability [29][30][31][32]. Crucially, these findings enable direct comparison with human behavior and neurophysiology during perception and synchronization to explicit metronome beats [27], and thereby begin bridging the neurophysiological and evolutionary void in our understanding of beat perception.…”

Dynamic models for musical rhythm perception and coordination

“…In humans, there is much evidence that motor areas are used for auditory timing ( Schubotz, 2007 ; Iversen et al, 2009 ; Fujioka et al, 2012 ). Evidence from non-human primates synchronizing with periodic stimuli also shows interactions between sensory and motor areas ( Betancourt et al, 2022 ). Thus, fully modeling rhythm processing in the human brain will likely require an architecture with both local areas and bidirectional interactions between areas.…”

Preconfigured cortico-thalamic neural dynamics constrain movement-associated thalamic activity

“…Then, upon aligning the average population response to the onset of each stimulus of the train, we identified virtually identical responses for both ISIs (Figure 2B), suggesting that, as a population, the VL/VM network may not be temporally adapting to the two stimulating conditions. To further characterize this possibility, we applied a geometric approach (Betancourt, 2022; Zhou et al, 2020) to compare the matrices from our two ISI conditions. For the first comparison, called “relative” comparison, 30 bin matrices were constructed for both ISIs, that is, the bin sizes were adjusted so that both ISIs fitted 30 bins (10 / 16.6 ms bins for 300 / 500 ISIs; Figure S1B).…”