Deficit in feature-based attention following a left thalamic lesion

Abstract

Selective attention enables us to prioritise the processing of relevant over irrelevant information. The model of priority maps with stored attention weights provides a conceptual framework that accounts for the visual prioritisation mechanism of selective attention. According to this model, high attention weights can be assigned to spatial locations, features, or objects. Converging evidence from neuroimaging and neuropsychological studies propose the involvement of thalamic and frontoparietal areas in selective attention. However, it is unclear whether the thalamus is critically involved in generating different types of modulatory signals for attentional selection. The aim of the current study was to investigate feature- and spatial-based selection in stroke survivors with subcortical thalamic and non-thalamic lesions. A single case with a left-hemispheric lesion extending into the thalamus, five cases with right-hemispheric lesions sparing the thalamus and 34 healthy, age-matched controls participated in the study. Participants performed a go/no-go task on task-relevant stimuli, while ignoring simultaneously presented task-irrelevant stimuli. Stimulus relevance was determined by colour or spatial location. The thalamic lesion case was specifically impaired in feature-based selection but not in spatial-based selection, whereas performance of non-thalamic lesion patients was similar to controls' performance in both types of selective attention. In summary, our thalamic lesion case showed difficulties in computing differential attention weights based on features, but not based on spatial locations. The results suggest that different modulatory signals are generated mediating attentional selection for features versus space in the thalamus.

Keywords: Attention weights; Endogenous control; Feature-based attention; Stroke; Thalamus.

Fig. 1

Lesion distribution. (a) Lesion overlay. The colour code indicates in how many individuals of our sample (n = 6) a given voxel was lesioned. (b) Overview of the location and extent of the individual lesions.

Fig. 2

Analysis of thalamic lesion overlap for Case 1. (a) The 7 sub-thalamic regions included in the Oxford Thalamic Connectivity atlas by Behrens and colleagues (Behrens et al., 2003, Johansen-Berg et al., 2005), and the lesion of Case 1 (dotted lines), visualized on axial slices of the brain. (b) We assessed the ratio between the number of lesion voxels in each sub-thalamic region and the total number of voxels in the same region. The analysis was conducted by thresholding the probability maps of sub-thalamic region at different levels, ranging from 10% to 100%.

Fig. 3

SART schematic: Each block started with a 1200 ms cue presentation indicating the relevant stream (i.e. single-lined ring or single arrow in the single condition) or streams (double-lined ring or double arrow in the double condition). The cue remained visible during the whole block, i.e. two random sequence repetitions of the numbers 1 through 9. Each digit pair was presented for 250 ms. Participants were asked to respond to each pair by pressing the space bar on the keyboard with their dominant hand, but to withhold their response, when a “3” was presented in the relevant stream or streams respectively. Each digit pair was followed by a mask, before a new pair was shown. For instance, in the single magenta (or left) condition, the first number pair depictures a go trial with a low salient distractor (no “3” present), the second a no-go trial (magenta or left “3” present) and the third a go trial with a highly salient distractor (cyan or right “3” present). The digit sequence 1 through 9 was randomised in each stream and all nine numbers were shown before the next sequence started. In the feature task, the numbers were partly overlapping, so that attention to location could not be used to separate the objects. Their presentation in the front or the back randomly changed trial-wise.

Fig. 4

Proportion of misses on go trials made by controls (n = 34), the single thalamic-lesion case (n = 1) and non-thalamic lesion patients (n = 5) during a) the feature-based SART and b) the spatial-based SART. The difference between the grey bar (condition 2) and the white bar (condition 1) reflects the competition effect. The difference between the black bar (condition 4) and the white bar (condition 1) reflects the divided attention effect. Error bars represent SEM.

Fig. 5

Proportion of false alarms on no-go trials made by controls (n = 34), the single thalamic-lesion case (n = 1) and non-thalamic lesion patients (n = 5) during a) the feature-based SART and b) the spatial-based SART. The difference between the black bar (condition 5) and the white bar (condition 3) reflects the divided attention effect. Error bars represent SEM.