Control of spatial and feature-based attention in frontoparietal cortex

Abstract

Visual attention selects task-relevant information from scenes to help achieve behavioral goals. Attention can be deployed within multiple domains to select specific spatial locations, features, or objects. Recent evidence has shown that voluntary shifts of attention in multiple domains are consistently associated with transient increases in cortical activity in medial superior parietal lobule, suggesting that this may be the source of a domain-independent control signal that initiates the reconfiguration of attention. To investigate this hypothesis, we used fMRI to measure changes in cortical activation while human subjects shifted attention between spatial locations or between colors at a location. Univariate multiple regression analysis revealed a common, domain-independent transient signal [in posterior parietal cortex (PPC) and prefrontal cortex] time-locked to shifts of attention in both domains. However, multivariate pattern classification conducted on the cortical surface revealed that the spatiotemporal pattern of activity within PPC differed reliably for spatial and feature-based attention shifts. These results suggest that the posterior parietal cortex is a common hub for the control of attention shifts but contains subpopulations of neurons with domain-specific tuning for cognitive control.

Figure 1.

Static examples of moving displays. Each of the two apertures of dot fields measured 2.6° in diameter and was centered 4.5° to the left and right of fixation. In each aperture, one dot field color was moving in one of eight directions while the other was stationary. Once per second, the direction of the moving dots changed, or they became stationary (and the stationary dots began moving). During each run, subjects covertly attended to one location and one color (e.g., green dots on the left side) as they monitored for cues defined by the direction of motion of the currently attended dot field. Subjects responded by button press to three different types of cues: shift color cues directed subjects to shift attention from one color value to the other on the currently attended side of space; shift location cues directed subjects to shift attention from the currently attended location to the same color dots on the other side of fixation; and hold cues directed subjects to continue attending to the currently attended color and location.

Figure 2.

Effects of attention in extrastriate visual cortex: group data. A, Functional map showing mixed effects group analysis resulting from a contrast between epochs of attention to the left stimulus aperture versus attention to the right aperture. B–E, Event-related average time course plots of left (B, C) and right (D, E) extrastriate cortex. Time courses locked to the onset of the shift/hold cue are the average percentage change (across trials) relative to mean BOLD signal in each run. In each plot, activation contralateral to hold cues is sustained at a high level and activation ipsilateral to hold cues is sustained at a low level. Activation contralateral to location shift cues begins high and ends low as the subject shifts attention from the contralateral, preferred visual field to the ipsilateral, nonpreferred visual field, and vice versa for activity ipsilateral to location shift cues. Color shift cues evoked small transient increases (above sustained hold cue levels) both ipsilateral and contralateral to extrastriate cortex regions. Time courses are generated from a nonindependent analysis and are therefore merely illustrative. Shaded error regions represent ±1 SEM across subjects.

Figure 3.

Sources of attentional control in frontoparietal cortex: group data. This functional map is the result of a conjunction of four specific contrasts (see text for full description) that depict common regions (colored yellow) activated by both color shifts and location shifts relative to hold cues. Regions specific to color shift activity are shown in green and regions specific to location shift activity are in red. L, Left; R, right.

Figure 4.

Transient cue-evoked activity in attentional control regions. A, Frontal and parietal regions selected from the conjunction maps in the individual subject leave-one-run-out GLM procedure (see text for details) were used to extract time courses locked to the onset of the three cue types: shift-location, shift-color, and hold. B–F, Mean event-related averages of all trials across subjects. In each region that reached significance in the conjunction, the activation for both types of shift cues is significantly larger than that of hold cues. The two types of shift cues evoked similar activation profiles throughout the frontoparietal attention network. Shaded error regions represent ±1 SEM across subjects. L, Left; R, right.

Figure 5.

Pattern classification results from our group of subjects. A leave-one-run-out procedure (see text for details) was used to generate ROIs (independent of the extracted data) that were, in turn, used to define the features (surface nodes) that were submitted to the SVM classifier. A–D, Event-related classification accuracy (time 0 marks cue onset) for the group (thick black line) and each individual (thin colored lines). The linear classifier performed at chance levels (dotted line) on shift cue discrimination at all time points tested in frontal regions (A, B). However, the classifier reached accuracy levels significantly (p < 0.01) above chance for several time points (indicated by gray shading) in the parietal ROIs (C, D).