Neural bases of rhythmic entrainment in humans: critical transformation between cortical and lower‐level representations of auditory rhythm

Abstract

The spontaneous ability to entrain to meter periodicities is central to music perception and production across cultures. There is increasing evidence that this ability involves selective neural responses to meter-related frequencies. This phenomenon has been observed in the human auditory cortex, yet it could be the product of evolutionarily older lower-level properties of brainstem auditory neurons, as suggested by recent recordings from rodent midbrain. We addressed this question by taking advantage of a new method to simultaneously record human EEG activity originating from cortical and lower-level sources, in the form of slow (< 20 Hz) and fast (> 150 Hz) responses to auditory rhythms. Cortical responses showed increased amplitudes at meter-related frequencies compared to meter-unrelated frequencies, regardless of the prominence of the meter-related frequencies in the modulation spectrum of the rhythmic inputs. In contrast, frequency-following responses showed increased amplitudes at meter-related frequencies only in rhythms with prominent meter-related frequencies in the input but not for a more complex rhythm requiring more endogenous generation of the meter. This interaction with rhythm complexity suggests that the selective enhancement of meter-related frequencies does not fully rely on subcortical auditory properties, but is critically shaped at the cortical level, possibly through functional connections between the auditory cortex and other, movement-related, brain structures. This process of temporal selection would thus enable endogenous and motor entrainment to emerge with substantial flexibility and invariance with respect to the rhythmic input in humans in contrast with non-human animals.

Hysteresis in the selective synchronization of brain activity to musical rhythm

“…Another noteworthy result was the absence of context effects when the EEG responses from individual trials were directly transformed into the frequencydomain instead of first averaging in the time-domain. Averaging in the time-domain has been used across a number of EEG studies, as it increases signal-to-noise ratio of the responses to the input by cancelling out activity that is not time-locked to the input, and thus not aligned across trials (Picton, 2010;Mouraux et al, 2011;Nozaradan et al, 2011Nozaradan et al, , 2018Luck, 2014). An important assumption of this processing step is that the phase of the neural response of interest is constant across trials.…”

Response to perturbations as a built-in feature in a mathematical model for paced finger tapping

“…The number of theoretical and experimental works dedicated to understand this behavior and its neural bases is rapidly growing, especially imaging and electrophysiology studies like EEG, MEG, and fMRI [9][10][11][12][13][14][15][16]. It is very simple to show that there is an error correction mechanism in the brain in charge of keeping average synchrony that operates based on past performance [1,17].…”

Attention affects overall gain but not selective contrast at meter frequencies in the neural processing of rhythm

“…We used two different rhythmic patterns (depicted in Figure 1). These two patterns were selected based on previous evidence that they both induce a perception of musical meter, consistent across individuals, based on nested grouping of the individual event rate (200 ms) by 2 (2 x 200 ms = 400 ms), 2 (2 x 400 ms = 800 ms) and 3 (3 x 800 ms = 2400 ms) (Nozaradan et al, 2012(Nozaradan et al, , 2018Lenc et al, 2018).…”

“…That is, if selective contrast at meter periodicities is observed in the EEG in response to a sequence lacking such prominent periodic cues, this selective contrast at meter periodicities cannot be explained easily by the stimulus structure or low-level processing of the stimulus. Importantly, the decision as to what frequencies would correspond to meter periodicities was informed by previous studies, which used tapping tasks to carefully test the metric pulses most consistently induced by these two rhythmic patterns across listeners (Nozaradan et al, 2012(Nozaradan et al, , 2018Lenc et al, 2018). This ensured that these specific frequencies were relevant for meter perception, in contrast to other frequencies that are also elicited by the rhythms but are irrelevant to the perceived meter.…”

“…To this end, we used biologically plausible models to simulate responses to the rhythmic stimuli in the auditory nerve, as well as inferior colliculus. To complement these simulations, we also directly captured responses presumed to be predominantly driven by brainstem auditory nuclei and primary auditory areas using the same frequency-tagging method as Nozaradan et al (2016cNozaradan et al ( , 2018). These early responses were observed at a faster timescale (> 150 Hz) due to neural tracking of the amplitude-modulated fine structure of the sound input.…”