Rhythm evokes action: early processing of metric deviances in expressive music by experts and laymen revealed by ERP source imaging

Abstract

To examine how musical expertise tunes the brain to subtle metric anomalies in an ecological musical context, we presented piano compositions ending on standard and deviant cadences (endings) to expert pianists and musical laymen, while high-density EEG was recorded. Temporal expectancies were manipulated by substituting standard "masculine" cadences at metrically strong positions with deviant, metrically unaccented, "feminine" cadences. Experts detected metrically deviant cadences better than laymen. Analyses of event-related potentials demonstrated that an early P3a-like component (~150-300 ms), elicited by musical closure, was significantly enhanced at frontal and parietal electrodes in response to deviant endings in experts, whereas a reduced response to deviance occurred in laymen. Putative neuronal sources contributing to the modulation of this component were localized in a network of brain regions including bilateral supplementary motor areas, middle and posterior cingulate cortex, precuneus, associative visual areas, as well as in the right amygdala and insula. In all these regions, experts showed enhanced responses to metric deviance. Later effects demonstrated enhanced activations within the same brain network, as well as higher processing speed for experts. These results suggest that early brain responses to metric deviance in experts may rely on motor representations mediated by the supplementary motor area and motor cingulate regions, in addition to areas involved in self-referential imagery and relevance detection. Such motor representations could play a role in temporal sensory prediction evolved from musical training and suggests that rhythm evokes action more strongly in highly trained instrumentalists.

Figure 1

Examples of stimuli. Metrically standard and deviant endings for 3 (a–c) out of 30 different polyphonic piano compositions (composer Nicolaas Ravenstijn).

Figure 2

Figure 2. Grand‐average ERP waveforms for an array of 15 electrode sites. ERPs in response to (a) metrically standard endings, (b) metrically deviant endings, and (c) difference ERPs between conditions (deviant minus standard), plotted for Experts (in red) and Laymen (in black). Gray shaded areas show the time interval used for the statistical analysis of the ERPs (150–300 ms). Topographic scalp configurations (d) are shown for each group and each condition (S: standard; D: deviant) and for difference ERPs (D‐S: deviant minus standard), depicting the average voltage over the 150–300 ms time period at all 128 electrode sites. These maps are 2‐D projections of the 3‐D electrode configuration (view from above, nasion on top). (e) Left panel: highlighted head positions of the 15 electrodes within the array depicted in gray amongst all other electrode sites (total n = 128), right panel: enlarged labeled array.

Figure 3

Results of spatiotemporal ERP analyses. (a) A k‐means clustering analysis yielded 7 distinct microstate maps that optimally represent the data of both groups and both conditions (E: Experts, L: Laymen, std: metrically standard ending, dev: metrically deviant ending). The segments under the GFP curves represent the time periods during which each of these microstate maps was most represented in the group data. The segments are marked in black and gray when common to both groups, and in color when unique for one group or one condition at a certain time period. The average onset and offset of each microstate are indicated in ms. (b) Scalp configurations of the microstate maps, framed in corresponding color‐code; positive voltages in red, negative in blue. These maps are 2‐D projections of the 3‐D electrode configuration (view from above, nasion on top). (c) Mean duration of microstate maps resulting from individual subject fitting for both experimental groups and conditions for two consecutive time periods. Vertical bars depict 95% confidence intervals.

Figure 4

Results of statistical source analyses of ERPs; time‐period 143–261 ms. (a) Mean current density values (μA/mm³) for all 50 ROIs (cf. Table I) of both experimental groups and conditions for the 143–261 ms time period resulting from the spatiotemporal ERP analyses. On top of the graphs a division of ROIs in frontal (Front), parietal (Par), temporal (Temp), and occipital (Occ) lobes is provided. Vertical bars depict standard errors. Gray‐shaded areas highlight values for the 8 predefined ROIs. (b) The 8 predefined ROIs are highlighted, superimposed on MNI 152 template brain slides. Talairach coordinates are provided underneath, corresponding to the position of the superimposed white cross. Corresponding Brodmann areas (BA) are provided when available. (c) Below each ROI mean current density values for both groups (Ex: experts, in black; Lay: laymen in gray) and both conditions (S: standard endings, D: deviant endings) are depicted for each cerebral hemisphere (L: left, R: right). Vertical bars depict standard errors. For significant differences, we refer to Table III, top panel (Time period 1: 143–261 ms).

Figure 5

Statistical parametric mapping (SPM) of estimated sources, comparing experts and laymen for metric and harmonic deviance processing over 300 ms from stimulus onset. Results are shown within 50 Regions of interest (ROI, on the y‐axis; Table 1) estimated by a distributed linear inverse solution (WMN). Time course of current density differences between experts and laymen is shown in graduated light‐gray shades for metric deviance and in graduated dark‐gray shades for harmonic deviance (James et al., 2008). The shaded areas indicate when current density significantly differed between the two groups for more than 40 ms; significant p‐values vary from P < 0.005 (darkest shades) up to P < 0.0001 (lightest shades). Vertical bars depict 95% confidence intervals. For metric deviance significant differences occurred in bilateral SMA (ROIs 9 & 10); for harmonic incongruity significant differences occurred in right hippocampal complex, amygdala and insula (respectively ROIs 38, 40 and 12). In both cases the enhanced responses occurred in the expert group. (Adapted with permission from James CE et al. 2008, 42:1597–1608, Elsevier).