Posterior Intraparietal Sulcus Mediates Detection of Salient Stimuli Outside the Endogenous Focus of Attention

Abstract

Visual consciousness is shaped by the interplay between endogenous selection and exogenous capture. If stimulus saliency is aligned with a subject's attentional priorities, endogenous selection will be facilitated. In case of a misalignment, endogenous selection may be compromised as attentional capture is a strong and automatic process. We manipulated task-congruent versus -incongruent saliency in a functional magnetic resonance imaging change-detection task and analyzed brain activity patterns in the cortex surrounding the intraparietal sulcus (IPS) within the Julich-Brain probabilistic cytoarchitectonic mapping reference frame. We predicted that exogenous effects would be seen mainly in the posterior regions of the IPS (hIP4-hIP7-hIP8), whereas a conflict between endogenous and exogenous orienting would elicit activity from more anterior cytoarchitectonic areas (hIP1-hIP2-hIP3). Contrary to our hypothesis, a conflict between endogenous and exogenous orienting had an effect early in the IPS (mainly in hIP7 and hIP8). This is strong evidence for an endogenous component in hIP7/8 responses to salient stimuli beyond effects of attentional bottom-up sweep. Our results suggest that hIP7 and hIP8 are implicated in the individuation of attended locations based on saliency as well as endogenous instructions.

Keywords: cytoarchitectonics; endogenous selection; exogenous capture; intraparietal sulcus; visual attention.

Figure 1

(A) The 3 possible saliency conditions in the experimental design. On the leftmost panel, the 2 letter targets are of a different color than the distracters and saliency is congruent with task-relevance. In the middle panel, in the saliency-incongruent condition, the 2 target pairs are in red alongside 4 other distracters, whereas the 2 opposite distracters to the target pairs are in the other color. Hence, saliency is incongruent with task-relevance. In the righmost panel, in the low-saliency condition, all stimuli share the same color. (B) The 6 squares depict the 6 possible spatial configurations for the 2 targets, presented here for the saliency-congruent condition. (C) Example trial of the paradigm. In this trial, the focus of exogenous and endogenous attention coincided, saliency is thus congruent with task-relevance.

Figure 2

Behavioral experiment. (A) Mean accuracy in the behavioral experiment over all subjects across the 3 saliency-condition. (B) Reaction time kernel density distribution for the 3 saliency conditions. The 3 conditions did not statistically differ in terms of reaction time. (C) Evolution of the duration of the stimulus array during the staircase procedure. The duration of the stimulus array has been averaged over all subjects. The x-axis shows the trial order from the first to the last (144 trials total). The y-axis shows the duration of the stimulus array, starting from 300 ms to the minimum duration reached by the subjects. fMRI experiment: (D) Mean accuracy scores in the fMRI experiment over all subjects across the 3 saliency-condition and the 3 target axis configurations. SC = saliency-congruent, SI = saliency-incongruent, and NS = low-saliency. (E) Reaction time kernel density distribution and for the 3 saliency conditions.

Figure 3

Comparisons of the difference of each saliency-condition to the mean of the BOLD response within each ROI. On the x-axis are the 12 cytoarchitectonic regions ordered based on their location in the landmark anatomical regions (inferior parietal lobule, IPS, superior parietal lobule). On the y-axis is the difference between the mean area under the curve for a given condition to the mean area under the curve of all-conditions within a specific ROI. Saliency-congruent (SC) is in green, saliency-incongruent (SI) in red, and low-saliency (LS) in black.

Figure 4

Overview of the univariate and multivariate analysis in the different hIP areas. Seven areas are highlighted here. hIP8, hIP4, hIP3, and 7P were significantly more active in the saliency-incongruent trials compared with the saliency-congruent or low-saliency trials. HIP7, hIP8, and hIP4 could distinguish between saliency-incongruent and saliency-congruent or low-saliency trials. PGa and PGp were able to discriminate between saliency-congruent and low-saliency trials.

Figure 5

Results of the searchlight classification analysis between saliency-congruent and saliency-incongruent trials, with the relative difference in stimulus array duration and relative difference in RTs controlled as covariates. (A) Results of the searchlight classification analysis between saliency-congruent and saliency-incongruent. (B) Results of the searchlight classification analysis between saliency-incongruent and low-saliency trials. (C) Results of the searchlight classification analysis between saliency-congruent and low-saliency trials. Results are FWE-corrected at P < 0.05 at the cluster level. Color bars represent T-values of the univariate analysis.

Figure 6

Univariate results for the saliency-incongruent > saliency-congruent contrast. Results are FWE-corrected at P < 0.05 at the cluster level. Color bars represent T-values of the univariate analysis.

Figure 7

Lesions on the IPS and superior parietal lobule of patient H.H. from Gillebert et al. (2012). In red is the left focal ischemic lesion. The 3 other regions overlayed are the maximum probability maps of hIP8 (orange), 7A (light blue), and 7P (green).