Article Figures & Data
Figures
- Figure 1.
Experimental setup, task, and behavioral results. A, On separate blocks, periodic stimuli were presented either auditorily (acoustic “flutter”), visually (“flicker” lights), or by vibrotactile stimulation of the fingertips (tactile flutter). All stimuli were presented bilaterally. B, Delayed-match-to-sample frequency discrimination task. For each modality, the to-be-maintained stimulus frequency (f1) was varied between 14 and 34 Hz, for later discrimination against a comparison stimulus (f2) of the same modality. C, Average behavioral discrimination performance (left) and response times (right) for the different task conditions. Clipart courtesy of Microsoft.
- Figure 2.
Temporal evolution of spectral activity, expressed as percentage change relative to prestimulus baseline. A, Alpha activity (8–12 Hz) over parieto-occipital channels increases throughout the WM task in each of the modality conditions. B, Comparison of parieto-occipital alpha power changes between conditions. Colored ribbons indicate significance of pairwise comparisons. The alpha increase in the visual condition is significantly reduced during early retention (red shading). C, 3D source reconstructions. Top, Conjunction analysis of alpha source activity increases common to all three modality conditions, during late retention (blue shading in B). Bottom, Difference in source power between the visual and the remaining conditions, during early retention (red shading in B).
- Figure 3.
Modality-specific alpha topographies. A, Statistical topographical maps of modality-specific alpha activity. In each row, alpha activity in one modality (blue, tactile; green, auditory; red, visual) was contrasted against the two remaining modalities. Modality-specific patterns were most pronounced during stimulation (f1, f2) and late retention (2–3 s). B, 3D source reconstructions of the modality-specific alpha contrast decreases shown in A. C, Correlations of modality-specific contrast decreases over somatosensory (left) and visual (right) areas during retention (1–3 s) with individual task-performance levels. Different bar colors represent correlations with performance in the different tasks (normalized to subject mean). Representative channels used for computation are outlined in A. By tendency, correlations with performance in the contrasts' target conditions were always positive, while any correlations with performance in the contrasts' non-target conditions were negative. For the A-TV contrast decrease over temporal channels, none of the correlations reached significance (all p values >0.30).
- Figure 4.
Parametric modulations of oscillatory activity by the stimulus frequency maintained in working memory. A, Left, Time-frequency statistical parametric map indicating the significance of overall linear modulations of oscillatory activity by f1 frequency, collapsed across modality conditions, for representative channel F2. Maximum at 26 Hz, 2800 ms (p < 0.05, FWE). Right, Changes of prefrontal upper beta amplitudes as a function of f1. Amplitudes were zero-centered relative to overall mean. B, Changes of prefrontal upper beta amplitudes as a function of f1 frequency, for the different modality conditions. Amplitudes were zero-centered relative to modality condition mean. In all modalities, significant linear modulations were found (indicated by solid colored lines). In the visual condition (right), the fastest flicker frequencies (>30 Hz) may have been partially “fused” (see Discussion) and were excluded from subsequent analyses (light red indicates modulation across the remaining visual f1s, 14–30 Hz). C, Topographical statistical maps (color-scaling as in A) and 3D source reconstructions of the above parametric modulations. Source renderings appear larger for more superficial sources and may not reflect actual differences in activity strength between modality conditions. D, Conjunction analysis of parametric contrast activity common to all three modality conditions. E, Overall parametric modulation by f1 (cf. A, computed from correct trials (dark brown) compared with incorrect trials (light brown). On incorrect trials, the parametric modulation was significantly reduced (z = 3.06, p < 0.005; resampling statistics based on 1000 iterations).
Tables
14 Hz 18 Hz 22 Hz 26 Hz 30 Hz 34 Hz Linear slope Accuracy (%) Tactile 76.7 78.4 77.7 77.5 71.3 68.5 −0.45* Auditory 82.5 81.3 80.0 76.2 72.1 72.3 −0.59* Visual 81.6 82.2 77.8 68.8 65.0 61.4 −1.16* RT (ms) Tactile 547 550 539 530 555 518 −0.99 Auditory 474 488 506 524 524 516 2.40 Visual 514 527 542 542 549 579 2.79* Top, Mean f1–f2 frequency discrimination accuracy (% correct) for each f1 frequency, and linear regression slopes for the different modality conditions. Asterisks indicate significance of a linear trend across f1 frequencies (p < 0.05, Bonferroni corrected). Bottom, Same as top, for RT (milliseconds).
