Top down attentional deficits in macaques with lesions of lateral prefrontal cortex

Abstract

Brain imaging, electrical stimulation, and neurophysiological studies have all implicated the prefrontal cortex (PFC) in the top-down control of attention. Specifically, feedback from PFC has been proposed to bias activity in visual cortex in favor of attended stimuli over irrelevant distracters. To identify which attentional functions are critically dependent on PFC, we removed PFC unilaterally in combination with transection of the corpus callosum and anterior commissure in two macaques. In such a preparation, the ipsilesional hemisphere is deprived of top-down feedback from PFC to visual cortex, and the contralesional hemisphere can serve as an intact normal control. Monkeys were trained to fixate a central cue and discriminate the orientation of a colored target grating presented among colored distracter gratings in either the hemifield affected by the PFC lesion or the normal control hemifield. Locations of the targets and distracters were varied, and the color of the central cue specified the color of the target on each trial. The behavioral response was a bar release, and thus attentional impairments could be distinguished from impaired oculomotor control. When the cue was held constant for many trials, task performance in the affected hemifield was nearly normal. However, the monkeys were severely impaired when the cue was switched frequently across trials. The monkeys were unimpaired in a pop-out task with changing targets that did not require top-down attentional control. The PFC thus appears to play a critical role in the ability to flexibly reallocate attention on the basis of changing task demands.

Figure 1.

Methods and procedure. a, Combination of unilateral PFC lesion and split brain. Top, The lateral view of the right hemisphere showing a lesion of the lateral surface of the right PFC (gray shading). Bottom, The medial surface of the left hemisphere shows the transection, in gray, of the corpus callosum and anterior commissure. b, A postsurgical coronal section obtained with MRI from monkey M1 (for details, see Materials and Methods). This section, 25 mm anterior to the interaural stereotaxic landmark, shows a lesion of right lateral PFC defined medially by the superior arcuate sulcus and laterally by the inferior arcuate sulcus. In addition, the transection of the corpus callosum and anterior commissure is indicated by the top and bottom arrows, respectively. c, The combined lesion and split brain resulted in the contralesional visual hemifield, shown as gray, being processed without PFC and the ipsilesional visual hemifield serving as an experimental control. The effect of the lesion was assessed by comparing visual performance in the two hemifields. d, The temporal sequence of stimulus presentation in the color cueing task. The monkey fixated centrally and discriminated the orientation of the peripheral grating that was cued by the color of the fixation spot. The relative positions of the colored gratings were randomly assigned each trial. The frequency at which the color cue changed was varied to examine the effect of increasing or decreasing the “top–down load” of the task.

Figure 2.

Effects of cue repetition on grating orientation discrimination in a color cueing task. a, The average orientation threshold is plotted as a function of cue repetition for each monkey. Performance in the control hemifield is shown in white, and in the lesion-affected hemifield in black. Error bars represent the SEM. Each bar represents the average of between 40 and 60 thresholds for that condition. b, The average reaction times for release trials as a function of cue repetition for monkey M1. Monkey M2 showed the same pattern of results.

Figure 3.

A comparison of three task conditions to explore the effects of target positional uncertainty on performance of the color cueing task. The bar graphs represent performance thresholds obtained with no distracters (left), randomly assigned target position (right; same as experiment 1), and fixed target position (middle). The black and white bars correspond to performance of the task in the contralesional and ipsilesional hemifields, respectively. Data shown are for M1. Error bars represent the SEM.

Figure 4.

Effects of target-distracter repetition on grating orientation discrimination in a color pop-out task. a, Color pop-out display. The temporal sequence of stimulus presentation was identical with that used in the color cueing task. The monkey fixated centrally and discriminated the orientation of the target grating as defined by color pop-out. The relative positions gratings were randomly assigned each trial. The frequency at which the target and distracter colors changed was varied to examine the effect of changes in target identity on task performance. b, The average orientation threshold is plotted as function of cue repetition for each monkey (top and bottom plots). Performance in the control hemifield is shown in white, and in the lesion-affected hemifield in black. Error bars represent the SEM. Each bar represents the average of between 50 and 60 thresholds for that condition.

Figure 5.

Effect of distracter contrast on grating orientation discrimination. a, The target grating was surrounded by three disk distracters of uniform luminance. The monkey fixated centrally and discriminated the orientation of the target grating. The distracters were randomly assigned each trial. We systematically varied distracter contrast, relative to the background. Values of distracter contrast were 0 (i.e., no distracters), 8, 16, 32, and 64%. The target grating contrast was fixed at 50%. b, The average orientation threshold is plotted as function of distracter contrast for each monkey. Performance in the control hemifield is shown in white, and in the lesion-affected hemifield in black. Error bars represent the SEM. Each bar represents the average of between 40 and 50 thresholds for that condition.