Electrophysiological evidence for immature processing capacity and filtering in visuospatial working memory in adolescents

Abstract

The present study investigated the development of visuospatial working memory (VSWM) capacity and the efficiency of filtering in VSWM in adolescence. To this end, a group of IQ-matched adults and adolescents performed a VSWM change detection task with manipulations of WM-load and distraction, while performance and electrophysiological contralateral delay activity (CDA) were measured. The CDA is a lateralized ERP marker of the number of targets and distracters that are selectively encoded/maintained in WM from one hemifield of the memory display. Significantly lower VSWM-capacity (Cowan's K) was found in adolescents than adults, and adolescents' WM performance (in terms of accuracy and speed) also suffered more from the presence of distracters. Distracter-related CDA responses were partly indicative of higher distracter encoding/maintenance in WM in adolescents and were positively correlated with performance measures of distracter interference. This correlation suggests that the higher interference of distracters on WM performance in adolescents was caused by an inability to block distracters from processing and maintenance in WM. The lower visuospatial WM-capacity (K) in adolescents in the high load (3 items) condition was accompanied by a trend (p<.10) towards higher CDA amplitudes in adolescents than adults, whereas CDA amplitudes in the low load (1 item) condition were comparable between adolescents and adults. These findings point to immaturity of frontal-parietal WM-attention networks that support visuospatial WM processing in adolescence.

Figure 1. Example of distracters-present trial (T1D2) for the left hemifield.

Figure 2. Behavioral data from the VSWM change detection task.

Bar graphs of (A) Cowan's K, (B) average reaction times (in ms), and (C) percentage of correct responses for adolescents and adults in T1D0 (one target), T1D2 (one target, two distracters) and T3D0 (three targets) conditions. Error bars indicate 95% confidence intervals.

Figure 3. Average ERP activity during the VSWM change detection task.

(A) HEOG activity ((HEOG left visual field trials*-1+HEOG right visual field trials)/2) and (B) CDA activity (computed by subtracting ipisilateral from contralateral activity), after smoothing with a 6 Hz low-pass filter, time-locked to the memory array and averaged across occipital and posterior parietal electrode sites for adolescents and adults, in conditions T1D0, T1D2 and T3D0. The analyzed window (300–550 ms) is indicated by a grey rectangle.

Figure 4. CDA amplitudes for adults with high and low K-scores.

Mean amplitudes between 300 and 550 ms in T1D0, T1D2 and T3D0 conditions in adults with high (N = 5) and low (N = 12) working memory capacity (WMC), determined by a K-T3D0 score larger or smaller than 2.4. * p = .087, one-tailed, ** p<.001, one-tailed, n.s. = non-significant.

Figure 5. Scatterplots of significant correlations between behavioral and CDA measures.

(A) Correlation between K-T3D0 and Unnecessary Storage (K-T1D0 minus K-T1D2) for adolescents (triangles) and adults (squares). (B&C) Correlation between distracter related parietal-occipital CDA effects (CDA-T1D2 minus CDA-T1D0) and Unnecessary Storage (K-T1D0-K-T1D2; panel B) or RT distracter effects (RT-T1D2 minus RT-T1D0; panel C). Negative CDA effects are observed when T1D2 CDA amplitude is larger than T1D0 CDA amplitude. Larger negative CDA (T1D2-T1D0) values reflect larger CDA increases when distracters are present. Larger positive distracter-related RT values reflect larger RT increases when distracters are present.