Rhythm in the Premature Neonate Brain: Very Early Processing of Auditory Beat and Meter

Abstract

The ability to extract rhythmic structure is important for the development of language, music, and social communication. Although previous studies show infants' brains entrain to the periodicities of auditory rhythms and even different metrical interpretations (e.g., groups of two vs three beats) of ambiguous rhythms, whether the premature brain tracks beat and meter frequencies has not been explored previously. We used high-resolution electroencephalography while premature infants (n = 19, 5 male; mean age, 32 ± 2.59 weeks gestational age) heard two auditory rhythms in the incubators. We observed selective enhancement of the neural response at both beat- and meter-related frequencies. Further, neural oscillations at the beat and duple (groups of 2) meter were phase aligned with the envelope of the auditory rhythmic stimuli. Comparing the relative power at beat and meter frequencies across stimuli and frequency revealed evidence for selective enhancement of duple meter. This suggests that even at this early stage of development, neural mechanisms for processing auditory rhythms beyond simple sensory coding are present. Our results add to a few previous neuroimaging studies demonstrating discriminative auditory abilities of premature neural networks. Specifically, our results demonstrate the early capacities of the immature neural circuits and networks to code both simple beat and beat grouping (i.e., hierarchical meter) regularities of auditory sequences. Considering the importance of rhythm processing for acquiring language and music, our findings indicate that even before birth, the premature brain is already learning this important aspect of the auditory world in a sophisticated and abstract way.SIGNIFICANCE STATEMENT Processing auditory rhythm is of great neurodevelopmental importance. In an electroencephalography experiment in premature newborns, we found converging evidence that when presented with auditory rhythms, the premature brain encodes multiple periodicities corresponding to beat and beat grouping (meter) frequencies, and even selectively enhances the neural response to meter compared with beat, as in human adults. We also found that the phase of low-frequency neural oscillations aligns to the envelope of the auditory rhythms and that this phenomenon becomes less precise at lower frequencies. These findings demonstrate the initial capacities of the developing brain to code auditory rhythm and the importance of special care to the auditory environment of this vulnerable population during a highly dynamic period of neural development.

Keywords: electroencephalography; entrainment; frequency tagging; music; phase coupling; premature human brain.

Figure 1.

A, B, The two rhythmic stimulus patterns used in this study. Both rhythmic patterns consisted of 333-ms-long tones and rests. The dashed lines show the beat and metrical levels. C, D, The frequency spectra of the stimulus sound envelopes.

Figure 2.

A, B, Frequency spectra of the EEG while listening to Duple/Triple Rhythm (A) and Quadruple Rhythm (B). The values are presented as noise subtracted and averaged across all electrodes (electrodes included in the averaging process, after removing the outer ring, are shown in bold). Top, Topographical maps averaged across participants corresponding to the triple meter frequency (1 Hz), duple meter frequency (1.5 Hz), and beat frequency (3 Hz) for Duple/Triple Rhythm (A), and to the quadruple meter frequency (0.75 Hz) and beat frequency (3 Hz) for Quadruple Rhythm (B). C, D, Violin plots depict the distribution of individual responses to the beat and meter frequencies as well as the average noise floor, for Duple/Triple Rhythm (C) and Quadruple Rhythm (D). Lines connect the beat, meter, and unrelated frequency results for each subject. The white dot and the horizontal line indicate the median and mean for each condition, respectively. Paired-samples t tests corrected for multiple comparisons showed that the amplitudes of beat- and meter-related frequencies were significantly above the average noise floor. Specifically, p = 0.0009 for triple meter frequency (1.5 Hz), p = 0.0018 for duple meter frequency (1 Hz), and p = 0.0013 for beat frequency (3 Hz), corresponding to Duple/Triple Rhythm, and p = 0.0009 for quadruple frequency (0.75 Hz),and p = 0.0004 for beat frequency (3 Hz), corresponding to Quadruple Rhythm.

Figure 3.

A, B, Comparison of frequency spectra of stimulus and EEG rhythmic patterns for Duple/Triple Rhythm (A) and Quadruple Rhythm (B). The amplitude values are normalized by the amplitude at beat frequency (3 Hz). A paired-samples t test showed that the amplitude of the spectrum corresponding to Duple/Triple Rhythm at duple meter frequency (1.5 Hz) was significantly larger than the amplitude of the spectrum of the stimulus sound envelope at this frequency (t₍₁₆₎ = 2.17, p = 0.0461, Cohen's d = 0.75). There were no significant differences at any other frequencies. For the Quadruple Rhythm, there was no significant difference between the stimulus and EEG amplitudes at 1.5 Hz or any other frequencies.

Figure 4.

A, B, The topographical maps of averaged coupling strength between neural activity and rhythmic patterns are shown for Duple/Triple Rhythm (A) and Quadruple Rhythm (B). The electrodes for which circular nonuniformity of phase values was confirmed using the Rayleigh test (p < 0.05) are indicated by black dots. The individual coupling phase and strength are shown for sample electrodes (white dots) below the topographical maps. The black vector illustrates the circular group average.