Distinct neural mechanisms of distractor suppression in the frontal and parietal lobe

Abstract

The posterior parietal cortex and the prefrontal cortex are associated with eye movements and visual attention, but their specific contributions are poorly understood. We compared the dorsolateral prefrontal cortex (dlPFC) and the lateral intraparietal area (LIP) in monkeys using a memory saccade task in which a salient distractor flashed at a variable timing and location during the memory delay. We found that the two areas had similar responses to target selection, but made distinct contributions to distractor suppression. Distractor responses were more strongly suppressed and more closely correlated with performance in the dlPFC relative to LIP. Moreover, reversible inactivation of the dlPFC produced much larger increases in distractibility than inactivation of LIP. These findings suggest that LIP and dlPFC mediate different aspects of selective attention. Although both areas can contribute to the perceptual selection of salient information, the dlPFC has a decisive influence on whether and how attended stimulus is linked with actions.

Figure 1

A. Task Task stages are shown with time running from left to right. An array of 8 placeholders remained continuously on the screen, and a trial began with a variable period of central fixation. This was followed by a 100 ms flash indicating the target location. After a variable target distractor onset asynchrony (TDOA) a distractor flashed after target presentation. The distractor was identical to the target in appearance and duration, but appeared either at a near-target location (angular separation of 45°) or at far locations (separations 135° or 180°). One third of trials (randomly interleaved) were no-distractor trials. After an additional delay (bringing the total delay period to 1600 ms) the fixation point disappeared (“Go”) and monkeys were rewarded for making a saccade to the target location. B. Performance for each monkey as a function of distractor distance and TDOA (mean and standard error across all recording sessions, n = 89 sessions in monkey S, 47 sessions in monkey M).

Figure 2. Neural responses in LIP and dlPFC

Average normalized firing rates in each area aligned on the onset of the target and “Go” signal (times 0 and 1600 ms). The gray traces show trials in which the target (T) was in the RF and no distractor appeared. The colored traces show trials in which a distractor (D) appeared in the RF, at 100, 200, 300 or 900 ms TDOA (red, green, blue and orange arrows and traces). Raw firing rates are smoothed by convolving with a Gaussian kernel (15 ms SD). Firing rates were normalized by dividing each neurons’ activity by its peak target response (T in RF, no-distractor) and the normalized traces were averaged to obtain the population response. Shading shows SEM. As shown in the cartoons, distractor trials were sorted according to whether the target had appeared near the RF (top row) or far from the RF (bottom row).

Figure 3. Correspondence between distractor responses and error rates

A The peak normalized response to the distractor (mean and SEM) as a function of distance and TDOA. Stars indicate significant difference between near and far distractors (paired t-test, p<0.001). B The correlation between distractor responses and error rates. Each point shows the fraction of errors and distractor response (mean and SEM) for a different monkey, distance and TDOA.

Figure 4. Analysis of error trials

Population responses preceding correct saccades (teal) and error saccades (purple) on trials with a near distractor at 100 ms TDOA. After an initial visual response, neural activity became stronger whenever the saccade was directed to the RF center. When a distractor was in the RF (top panels), activity was stronger for an error relative to a correct saccade (top panels). When a target was in the RF (bottom panels), activity was stronger for a correct relative to an error trial. The red trace and axes (log-scale) show the p-values resulting from a sliding window t-test comparing correct and error trials (window width, 1 ms, step size, 1 ms). The dashed line shows the p = 0.05 significance level and the black arrows on the x-axis show the time of consistent discrimination (when the p-values remained consistently below 0.05).

Figure 5. Anticipatory and visual suppression

A, Pre-distractor responses in LIP and dlPFC on trials when the target appeared opposite the RF and a distractor appeared at 900 ms TDOA (distractor responses not shown). Histograms show unsmoothed firing rates measured in 2 ms time bins. Black traces show the exponential fit of firing in the 100–500 ms interval indicated by shading. Stars denote 100-ms non-overlapping bins where firing rates differed significantly between LIP and dlPFC (p<0.05). B. Pre-distractor responses for near separations are unrelated to RF size Each point shows the average pre-distractor activity of one neuron for near distractors (800–900 ms after target onset, 900 ms TDOA) as a function of the neuron’s RF size. RF size is defined as the ratio of the magnitude of response to the target presented 45 degree away from the center of RF and in the RF center. Therefore, values close to 1 indicate a wide RF with equivalent responses at the center and adjacent location, while values close to 0 indicate a smaller RF. The lines are best-fit linear regressions. While RF sizes were larger in the dlPFC relative to LIP (p < 0.05) they were not consistently related to the pre-distractor response in either area. C. Visual distractor responses (mean and SEM across all neurons), computed as shown in the left panel. For each distance and TDOA, the additional distractor response (ΔR_d) in dlPFC was significantly smaller than in LIP (all p<0.001).

Figure 6. Effects of reversible inactivation of LIP and dlPFC

As shown in the cartoon at the top left, in the trials shown here the target was in the hemifield contralateral to the inactivation site (shading) and was accompanied either by a near distractor (shown in red) or by a far distractor (blue). In the data panels each circle shows the difference in error rate between a single inactivation and control session. The bars show the average and SEM. In both monkeys for all TDOAs and distances, dlPFC inactivation induced more errors than LIP inactivation. The deficits were larger on trials that did, relative to those that did not contain a distractor (no-D, gray bars).