On the neural origin of pseudoneglect: EEG-correlates of shifts in line bisection performance with manipulation of line length

Abstract

Healthy participants tend to show systematic biases in spatial attention, usually to the left. However, these biases can shift rightward as a result of a number of experimental manipulations. Using electroencephalography (EEG) and a computerized line bisection task, here we investigated for the first time the neural correlates of changes in spatial attention bias induced by line-length (the so-called line-length effect). In accordance with previous studies, an overall systematic left bias (pseudoneglect) was present during long line but not during short line bisection performance. This effect of line-length on behavioral bias was associated with stronger right parieto-occipital responses to long as compared to short lines in an early time window (100-200ms) post-stimulus onset. This early differential activation to long as compared to short lines was task-independent (present even in a non-spatial control task not requiring line bisection), suggesting that it reflects a reflexive attentional response to long lines. This was corroborated by further analyses source-localizing the line-length effect to the right temporo-parietal junction (TPJ) and revealing a positive correlation between the strength of this effect and the magnitude by which long lines (relative to short lines) drive a behavioral left bias across individuals. Therefore, stimulus-driven left bisection bias was associated with increased right hemispheric engagement of areas of the ventral attention network. This further substantiates that this network plays a key role in the genesis of spatial bias, and suggests that post-stimulus TPJ-activity at early information processing stages (around the latency of the N1 component) contributes to the left bias.

Keywords: Attention; EEG; Event-related potentials; Landmark task; Pseudoneglect; Spatial bias.

Fig. 1

Experimental paradigm and sequence of events during each trial. Each trial was initiated by the appearance of a fixation cross for 1000 ms followed by presentation of the line stimulus for 150 ms followed by the fixation cross, which remained on the screen until the end of the trial. Participants were requested to delay their manual response for 1000 ms following the presentation of the stimulus in order to obviate for motor artifacts in the EEG signal. The onset of the response period was indicated by an auditory beep (100 Hz). In landmark task blocks, participants were asked to judge which end of the pre-transected line appeared shortest. The long line displayed is veridically transected but lines could be transected at any 1 of 29 points ranging symmetrically from ± 4.36% of absolute line length to veridical center. Long (15.3° × .39°) and short (1° × .39°) lines were presented in separate blocks. In control task blocks, 25% of the presented lines were not transected (plain white lines) and participants were asked to indicate whether the line was transected or not.

Fig. 2

Behavioral bias data. Group-averaged (N = 17) point of subjective equality (± 1 SE) for both long (gray bar) and short (white bar) landmark task performance (in % of absolute line length relative to veridical center). Negative values indicate leftward bias. Note the typical systematic leftward error (pseudoneglect) is stronger for long than short lines.

Fig. 3

Group-averaged (N = 17) voltage waveforms (62-channel butterfly plot) and topographic maps at selected time points corresponding roughly to the traditional P1, N1, P2 and P3 components of the ERP. Data are shown separately for (a) long line landmark task, (b) short line landmark task, (c) long line control task and (d) short line control task performance. L: Left, R: right, P: posterior.

Fig. 4

Line-length EEG-effects. (A) Global field power (GFP) over time for each experimental condition (upper panel) and mass univariate analysis results of the line-length effect (lower panel). Note in the GFP the early grouping of conditions according to line length (red & blue solid lines vs. red & blue dashed lines: i.e. long vs. short lines). The corresponding mass univariate analysis revealed these differences to be significant, peaking at 140 ms post-stimulus onset. (B) Topographical t-map (long minus short) across the scalp at 140 ms (left panel) and source estimate p-value maps of the effect (right panel, only p-values reaching a significance level of p < 0.01 are displayed, p-value coded by voxel size and color). Note that the line length effect peaked at electrode PO4 (electrode marked in white) and localized to the temporo-parietal junction of the RH (max. significant voxel: Talairach coordinate: 65, − 39, 20, peak t-value = 4.59, p < 0.001). (C) Hemispheric asymmetry data for electrodes PO3/PO4. Long lines were associated with a hemispheric asymmetry (RH > LH), not present in short lines. (D) Relationship between the line-length effect in ERPs (long–short lines) at PO4 and the line-length effect in behavioral bias (long–short lines) across individuals (left panel), and histograms of the corresponding Pearson and Spearman bootstrapped correlation values (right panels, red bars = 95% confidence intervals). The correlation proved significant by both correlation methods (p < 0.05) and the bootstrapped 95% confidence intervals for both did not include 0. The positive relationship suggests that the level to which the RH is engaged by “long” line processing during the early time period influences the direction and magnitude of lateralized behavioral bias.

Fig. 5

Line-bisection EEG-effect. (A) Global field power (GFP) over time for each experimental condition (upper panel) and mass univariate analysis results of the line bisection effect (lower panel). Note in the GFP the late grouping of conditions according to task (red lines vs. blue lines: i.e. bisection vs. control task). The corresponding mass univariate analysis revealed these differences to be significant, peaking at 280 ms and 378 ms post-stimulus onset. (B) Topographical t-maps (control minus landmark task) across the scalp at 280 ms and 378 ms (left panel) and source estimate p-value maps of the effect (right panel). Note that the line bisection effect peaked at electrodes CP6 (280 ms) and CP4 (378 ms), shown in white, and localized largely to the right superior parietal cortex (max. significant voxel: Talairach coordinate: 35, − 61, 43 (peak t-value = − 3.3, p < 0.01)). (C) Hemispheric asymmetry data for electrodes CP5/CP6. Landmark task performance was associated with a hemispheric asymmetry, not present during the control task. (D) Relationship between the line bisection effect in ERPs (control–landmark tasks) at CP6 and the length effect in behavioral bias (long–short lines) across individuals (left panel) and histograms of the corresponding Pearson and Spearman bootstrapped correlation (right panel, red bars = 95% confidence intervals). The correlation was not significant for either correlation method (p > 0.05) and the bootstrapped 95% confidence intervals for both included 0.