Functional MRI of Macaque Monkeys Performing a Cognitive Set-Shifting Task

See allHide authors and affiliations

Science  22 Feb 2002:
Vol. 295, Issue 5559, pp. 1532-1536
DOI: 10.1126/science.1067653


Functional brain organization of macaque monkeys and humans was directly compared by functional magnetic resonance imaging. Subjects of both species performed a modified Wisconsin Card Sorting Test that required behavioral flexibility in the form of cognitive set shifting. Equivalent visual stimuli and task sequence were used for the two species. We found transient activation related to cognitive set shifting in focal regions of prefrontal cortex in both monkeys and humans. These functional homologs were located in cytoarchitectonically equivalent regions in the posterior part of ventrolateral prefrontal cortex. This comparative imaging provides insights into the evolution of cognition in primates.

The prefrontal cortex (PFC) is evolutionarily most developed in primates and supports higher cognitive functions. Macaque monkeys have been widely used for investigating the PFC, mainly in anatomical, electrophysiological, and lesion studies (1–5), whereas the human PFC has been mainly investigated in neuroimaging and neuropsychological studies (6–10). We thus wanted to compare directly the functional organization of the PFC between the two species using a common physiological technique (11, 12). Recently, functional magnetic resonance imaging (fMRI) studies have been performed in both anesthetized and awake macaque monkeys (13–18). In the present study, we applied event-related fMRI to macaque monkeys and humans performing a modified version of the Wisconsin Card Sorting Test (WCST). The WCST (8) has been used to probe behavioral flexibility, i.e., the ability to shift from one response tendency (cognitive set) based on previous experiences, to another that is suitable for the current circumstances. Performance in this test is characteristically impaired by lesions of the PFC in both humans (8, 9) and monkeys (4, 5). In our modified WCST, equivalent task sequence and visual stimuli were used in subjects of the two species for the direct comparison (19, 20) (Fig. 1, A and B).

Figure 1

The behavioral task and strategy for event-related fMRI analysis. (A) Visual stimuli. A cue and three choice stimuli were presented on a screen. Among the choice stimuli, one is matched to the cue stimulus in the color dimension, another in the shape dimension, and the other in neither dimension. Subjects were required to match the cue stimulus to one of choice stimuli based on currently relevant dimension by pressing one of the three buttons. (20). (B) Task sequence. In each trial, first a cue stimulus is presented, and then three choice stimuli. Visual feedback notifies subjects whether a response is correct or not. After several successive correct matches, the relevant dimension was changed (dimensional change, arrow). ITI, intertrial interval (19). (C) An event-related fMRI analysis. FMRI signal for each set-shifting event and error trial were modeled independently (25).

Two macaque monkeys (monkeys Y and Z) were trained to perform the modified WCST (21). To confirm that the monkeys behaved according to the rule of the task, behavioral data in fMRI sessions were assessed in three different ways. First, we assessed how efficiently the monkeys could complete each set shifting. For both monkeys, more than 80% of shifts were completed within three trials (monkey Y, 327 out of 395 shifts; monkey Z, 343 out of 394 shifts) (22). The mean number of trials required to complete a shift was 2.3 ± 1.9 in monkey Y and 1.7 ± 1.6 in monkey Z (mean ± SD). Second, we assessed types of errors to analyze the monkeys' strategy after dimensional changes. Two types of errors are possible: perseverative errors and nonperseverative errors. Perseverative errors were the errors that resulted from continuing to make responses according to the previously relevant dimension. Nonperseverative errors were the errors that resulted from making responses that did not match in any dimension (Fig. 1A). If monkeys tend to make random selections (i.e., neglecting dimensions) expecting chance hits, the percentages of perseverative errors will approach 50%. Among errors at the time of dimensional changes, excluding inevitable errors immediately after dimensional changes, the perseverative errors accounted for 83.4 ± 4.9% in monkey Y and 86.4 ± 5.8% in monkey Z, significantly above 50% (P < 10−7; binomial test). Finally, once the monkeys completed each instance of set shifting and established a current cognitive set, they had to maintain it until the next dimensional change. We examined performance of trials between each completion of shifting and the next dimensional change. The mean correct response was 93.8 ± 2.7% in monkey Y and 92.3 ± 2.0% in monkey Z (chance level, 33.3%, P < 10−7; binomial test). These results confirmed that the monkeys' performances were proficient and that their behaviors were in good accordance with the rule of the WCST that requires both shifting and maintenance of cognitive set.

Functional images of the two monkeys were obtained by echo-planer imaging (EPI) with a voxel size of 2 × 2 × 2.5 mm (23). All of the images were spatially normalized to a template that we made from three-dimensional (3D) anatomical images of one monkey's whole brain (24). This procedure enabled us to perform a group analysis of the monkeys using SPM 99, and to introduce a stereotaxic coordinate arranged in bicommissural space in which the origin was placed at the anterior commissure (16). The EPI images overlapped faithfully with the corresponding reference images (23). The predicted hemodynamic responses to each dimensional change were modeled based on the general linear model implemented in SPM 99 (Fig. 1C). Such an event-related modulation of the fMRI signals reflects brain activation that was locked to the time of occurrence of cognitive set shifting (25).

The group analysis revealed transient activation associated with set shifting in the monkeys' bilateral inferior prefrontal convexity (Fig. 2, A and B) (26). The peak of the activation was located in the rostral bank at the ventral end of the inferior ramus of the arcuate sulcus. FMRI signals from these activation foci were transiently increased (0.5 to 0.6%) time-locked to the set-shifting event (Fig. 2B). Transient activation was also found in the rostral bank of the left intraparietal sulcus, the bilateral posterior cingulate cortex, the precuneus, and the insula (26). To confirm the reproducibility of this activation across the monkeys, the data from each monkey were analyzed independently. Bilateral prefrontal activation foci were observed in the same cortical regions across the monkeys (Fig. 2C), and their coordinates showed overlap between the subjects (26). These results support intersubject reproducibility of activated regions in the PFC. We further performed another type of group analysis, a conjunction analysis that is suggested to be adequate to infer typical characteristics of a population from a relatively small number of subjects (27). The resultant activations were reproducibly found in the rostral bank of the arcuate sulcus or the caudalmost part of the inferior prefrontal convexity, the left intraparietal sulcus, the posterior cingulate, and the precuneus (Fig. 3A, upper).

Figure 2

Shift-related activation in monkeys. (A) Set-shifting related activation in monkeys revealed by a group analysis. Areas that showed shift-related activation were superimposed on transverse sections of normalized anatomical images. In Fig. 2, A and C, statistical threshold was P < 0.05 corrected for multiple comparisons across the brain volume examined. Color scales indicate t values. (B) Event-related, averaged time courses with the fitted models of hemodynamic response function. The activities are taken from the peak voxels in the left (left panel: x = –20, y = 13,z = 9) and the right (right panel: x = 18, y = 10, z = 11) inferior prefrontal convexity. Lateral views of 3D-rendering of a monkey brain are presented in insets, and arrows point to activated regions in the inferior prefrontal cortex. (C) Intersubject reproducibility of the prefrontal activation. Shift-related activation of each monkey was estimated independently and was superimposed on anatomical images of each monkey. (D) Comparison of EPI images (right in each panel) and the corresponding reference images (left in each panel). The slices at the level of the inferior prefrontal cortex and the principle sulcus were shown. Arrowheads indicate the ventral end of the inferior ramus of the arcuate sulcus. Individual grid elements are 14 mm square.

Figure 3

Comparisons of shift-related activation in the PFC between monkeys and humans. (A) (upper) Shift-related activation in monkeys estimated by a conjunction analysis. A threshold of P < 0.001 (uncorrected) was used for display purpose. (lower) Shift-related activation in humans estimated by a random-effect model (P < 0.001, uncorrected). (B) Lateral views of 3D-rendered brain image on which the activation shown in (A) was superimposed (upper left, monkeys; upper right, humans). In the human data, prominent activation focus in the posterior part of the inferior frontal sulcus was shown. Other activations in the precentral gyrus and in the anterior insula were also observed. For references, cytoarchitectonic map of macaque monkeys by Walker (lower left) (34) and that of humans by Brodmann (lower right) (35) were presented. These maps correspond approximately to areas in the white squares in the upper panels. Green arrowhead, the principal sulcus; blue arrowhead, the inferior ramus of the arcuate sulcus; yellow arrowhead, the inferior frontal sulcus. Scale bar, ∼30 mm. (C) The same activation as in (B) were displayed on an inflated surface reconstruction of a monkey brain (left) and that of a human brain (right) (12, 39). Symbols are denoted as in (B).

We further proceeded to fMRI sessions with human subjects using the same task sequence and visual stimuli as in the monkey experiments (28). Functional images from 10 human participants were subjected to an event-related analysis as in the monkey experiments (25), and a random effect model was used to estimate the data (P < 0.001, uncorrected; 19 or more contiguous voxels) (29). The most prominent shift-related activation of the PFC was found in the posterior part of the bilateral inferior frontal sulcus (Fig. 3A, lower) (26). This focus conforms to the one reported in our previous studies of set shifting [see table 2 in (29)]. This activation was presumably related to the inhibition of the previously relevant response (9). Activated regions were also found in the extrastriate cortex, the inferior parietal lobule, and the anterior insula (26). In the present task condition the parietal activation was weaker than that observed in the previous studies; nevertheless, the global activation pattern including the main activation focus in the PFC was consistent with the preceding studies (29–33).

These fMRI experiments enabled us to compare directly the functional organization of the PFC in monkeys and humans. In the monkeys, the two types of group analysis consistently revealed prominent activation foci at the ventral end of the inferior ramus of the arcuate sulcus or caudalmost part of the inferior convexity [Figs. 2, A and B, and 3A (upper)]. On the other hand, in the humans, prominent activation foci were found in the posterior part of the inferior frontal sulcus (Fig. 3A, lower). Thus the main shift-related activation was commonly found in the posterior part of the ventrolateral PFC across the two species. We hypothesize that these activated regions in the PFC could be functionally homologous regions across the two species with regard to cognitive set-shifting function (Fig. 3, B and C). Estimated cortical areas of these regions corresponded to area 45/posterior12 by Walker (34) or area 45B/45A/44 by Petrides and Pandya (3) in monkeys and Brodmann's area (BA) 44/45 in humans (35) (Fig. 3B). Thus, in our set-shifting paradigm, the functionally homologous regions are largely equivalent in cytoarchitecture across the two species. However, this consistency is not always the case. The frontal eye field, a major region responsible for control of the saccadic eye-movement in the frontal cortex, is assigned to BA 8 in monkeys by electrophysiological and microstimulation studies (36), whereas it is assigned to BA 6 in humans by neuroimaging studies (37). This diversity emphasizes the importance of direct functional comparison of the PFC between species.

  • * To whom correspondence should be addressed. E-mail: yasushi_miyashita{at}


Stay Connected to Science

Navigate This Article