Abstract

To understand functional roles of the thalamic mediodorsal nucleus (MD) in sensory-to-motor information transformation during spatial working memory performance and compare with those of the dorsolateral prefrontal cortex (DLPFC), we calculated population vectors using a population of MD activities recorded during 2 tasks. In the oculomotor delayed-response (ODR) task, monkeys needed to make a memory-guided saccade to the cue location, whereas in the rotatory oculomotor delayed-response (R-ODR) task, they needed to make a memory-guided saccade 90o clockwise from the cue direction. The directions of population vectors calculated from populations of cue- and response-period activities were similar to the cue and saccade target directions, respectively, which confirmed that population vectors represent information regarding the directions of the visual cue and the saccade target. We then calculated population vectors of delay-period activity using a sliding 250-ms time window. In the ODR task, population vectors were directed toward the cue direction throughout the delay. However, in the R-ODR task, they gradually rotated from the cue direction to the saccade target direction. Based on a comparison with the results obtained from DLPFC neurons, the rotation of population vectors started earlier in the MD than in the DLPFC, suggesting that the motor information regarding forthcoming saccade is provided from the MD.

Introduction

Working memory is a dynamic neural process that includes the temporary maintenance of information as well as information processing (Baddeley 1986; Funahashi and Kubota 1994; Miyake and Shah 1999; Funahashi 2001). The dorsolateral prefrontal cortex (DLPFC) is known to play significant roles in working memory (Goldman-Rakic 1987; Petrides 1994; Fuster 1997). However, the DLPFC is not the only brain area that participates in working memory. Other brain areas, such as the posterior parietal cortex (Gnadt and Andersen 1988; Chafee and Goldman-Rakic 1998) and the inferior temporal cortex (Miyashita and Chang 1988; Fuster 1990), also participate in working memory. The mediodorsal nucleus (MD) of the thalamus also participates in working memory. The MD has strong reciprocal connections with the DLPFC (Kievit and Kuypers 1977; Goldman-Rakic and Porrino 1985; Giguere and Goldman-Rakic 1988; Ray and Price 1993). Delay-period activity has been observed in the MD while monkeys performed delayed-response tasks (Fuster and Alexander 1971, 1973; Tanibuchi and Goldman-Rakic 2003; Watanabe and Funahashi 2004a, 2004b). The characteristics of delay-period activity observed in the MD were very similar to those observed in the DLPFC (Watanabe and Funahashi 2004a, 2004b). These results indicate that the MD plays significant roles in working memory (Watanabe and Funahashi 2004a, 2004b). However, there were also some important differences in working memory–related activity. First, more neurons with delay-period activity represented the direction of the saccade in the MD than in the DLPFC (Watanabe and Funahashi 2004b). Second, more neurons with presaccadic activity were present in the MD than in the DLPFC (Watanabe and Funahashi 2004a). These results suggest that, although both the MD and the DLPFC participate in working memory processes, the MD participates more in motor aspects, whereas the DLPFC participates more in the sensory aspects.

Recently, Takeda and Funahashi (2004) used a population vector analysis to examine temporal changes in the information represented by a population of DLPFC activities. The population vector analysis was originally proposed and advanced by Georgopoulos and his group (Georgopoulos et al. 1983, 1986, 1988, 1989, 1993; Georgopoulos 1988; Kettner et al. 1988; Schwartz et al. 1988; Lurito et al. 1991; Smyrnis et al. 1992). They showed that population vectors calculated from a population of motor cortical activity can predict movement directions during arm-reaching behavior (Georgopoulos et al. 1983, 1986, 1988; Kettner et al. 1988; Schwartz et al. 1988). Takeda and Funahashi (2004) used 2 kinds of oculomotor delayed-response (ODR) task. In the ODR task, monkeys were required to make a memory-guided saccade in the direction of the visual cue. In the rotatory oculomotor delayed-response (R-ODR) task, they were required to make a memory-guided saccade 90o clockwise from the cue direction. In the ODR task, all population vectors calculated in a 250-ms time window were directed toward the cue direction throughout the delay period, whereas in the R-ODR task, the directions of population vectors rotated from the cue direction to the saccade direction during the late phase of the delay period. Thus, these studies showed that spatial information represented by a population of neurons can be visualized as the direction of the population vector and that the temporal change in the information represented by a population of neurons during task performance can be visualized as the temporal change in the direction of the population vector.

The population vector analysis using a population of DLPFC activities revealed that, although the directions of population vectors rotated from the cue direction to the saccade direction in the R-ODR task, the rotation started in the late phase of the delay period. Takeda and Funahashi (2002) showed that delay-period activity of DLPFC neurons represented forthcoming motor information, although the number of these neurons was small, and that delay-period activity representing motor information tended to show gradually increasing activation toward the motor performance. These observations suggest that the DLPFC receives information regarding forthcoming saccade performance from other structures. The MD is one of the possible structures that provide information regarding forthcoming saccade information to the DLPFC because the MD, especially the intermediate portion of the MD, has strong reciprocal connections with the DLPFC and because many MD neurons exhibited delay-period activity representing forthcoming saccade information. If the MD is the source structure that provides information regarding forthcoming saccade information to the DLPFC, the information regarding the saccade direction would develop earlier during the delay period in the MD than the DLPFC. To test this hypothesis, we constructed population vectors using a population of MD activities and compared the present findings with those observed in DLPFC neurons reported by Takeda and Funahashi (2004). Preliminary results have been published in abstract form (Watanabe and Funahashi 2006).

Materials and Methods

Subjects and Apparatus

The 2 rhesus monkeys (monkey P, 4.0 kg; monkey Q, 3.5kg) used in this study were the same as those used in our previous study (Watanabe and Funahashi 2004a, 2004b). The experimental apparatus, surgical procedures, and histological examinations have been described in detail previously (Watanabe and Funahashi 2004a). In brief, during training and recording sessions, the monkey sat quietly in a primate chair in a dark sound-attenuated room, and its head movement was restricted painlessly by a stainless steel rod which was fixed to the skull. The monkey faced a 21-inch color TV monitor (PC-TV471; NEC, Tokyo Japan), on which a fixation point (FP) and visual cues were presented. The TV monitor was placed 30 cm from the monkey's face. The monkey's eye positions were monitored by the magnetic search coil technique (Robinson 1963). Two laboratory computers (PC-486HX; Epson, Suwa, Nagano, Japan) controlled the monkey's behavior, presented visual stimuli on the monitor, recorded neural activity, and monitored eye movements. All experiments were conducted in accordance with the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. This experiment was approved by the Animal Research Committee at the Graduate School of Human and Environmental Studies, Kyoto University.

Behavioral Tasks

In the present experiment, we sought to compare the results obtained from MD neurons with the results in the DLPFC reported by Takeda and Funahashi (2004). Therefore, we used the same ODR tasks (ODR and R-ODR tasks) under the same behavioral conditions as were used by Takeda and Funahashi (2004) in the DLPFC.

In the ODR task, the monkey was required to make a memory-guided saccade to the location where a visual cue had been presented. The temporal sequence of this task is illustrated in Figure 1A (top). After a 5-s intertrial interval, an FP (a white circle, 0.5° in diameter in visual angle) was presented at the center of the TV monitor. If the monkey looked at the FP for 1 s (fixation period), a visual cue (a white circle, 1° in diameter in visual angle) was presented for 0.5 s (cue period) randomly at 1 of 8 predetermined locations around the FP (eccentricity was 17°) (Fig. 1B). The monkey was required to maintain fixation on the FP throughout the 0.5-s cue period and the subsequent 3-s delay period. At the end of the delay period, the FP was extinguished. This was the GO signal for the monkey to make a saccade within 0.4 s (response period) to the location where the visual cue had been presented. If the monkey made a correct saccade, a drop (0.2 ml) of water was given as a reward. To determine whether or not the monkey made a correct saccade, we set a square window (4–7° in visual angle) around the target location and judged that the monkey performed a correct saccade if its eye position fell within this window. If the monkey broke fixation during the cue period or the delay period, or failed to perform a saccade within the 0.4-s response period, or if the saccade did not fall within the correct window, the trial was aborted immediately without a reward and the next trial began.

Figure 1.

Diagrams of 2 types of ODR tasks. (A) Temporal sequences of task events. Top: A standard ODR task. The monkey was required to make a saccade to the location where the visual cue had been presented. Bottom: An R-ODR task. The monkey was required to make a saccade 90° clockwise from the location where the visual cue had been presented. (B) Locations of the visual cues for each task. The eccentricity of cue locations was 17°.

Figure 1.

Diagrams of 2 types of ODR tasks. (A) Temporal sequences of task events. Top: A standard ODR task. The monkey was required to make a saccade to the location where the visual cue had been presented. Bottom: An R-ODR task. The monkey was required to make a saccade 90° clockwise from the location where the visual cue had been presented. (B) Locations of the visual cues for each task. The eccentricity of cue locations was 17°.

In the R-ODR task, the monkey was required to make a saccade 90° clockwise from the location where the visual cue had been presented. Figure 1A (bottom) shows the temporal sequence of this task. After a 5-s intertrial interval, the FP (a white plus [+] sign, 0.5° in visual angle) was presented at the center of the TV monitor. If the monkey looked at the FP for 1 s (fixation period), a visual cue (a white circle, 1° in diameter in visual angle) was presented for 0.5 s (cue period) randomly at 1 of 4 predetermined locations around the FP (eccentricity was 17°). The monkey was required to maintain fixation on the FP throughout the 0.5-s cue period and subsequent 3-s delay period. At the end of the delay period, the FP was extinguished. This was the GO signal for the monkey to make a saccade within 0.4 s (response period) to the direction 90° clockwise from the direction where the visual cue had been presented. If the monkey made a correct saccade, a drop (0.2 ml) of water was given as a reward. We used a square window of the same size around the target location and the same criterion to determine whether or not the monkey made a correct saccade as were used in the ODR task.

Analysis of Single-Neuron Activity

In the present experiment, we recorded the activities of single neurons in the MD. We used an epoxy-coated tungsten microelectrode (25-10-2L; HFC instruments, Bowdoinham, ME) to record single-neuron activity from the MD. After we isolated single-neuron activity, we first made raster and histogram displays of recorded activity aligned at several task events and visually inspected whether or not the recorded neuron exhibited task-related activity in relation to any task events. To confirm our visual observations, we then conducted statistical analyses. First, to obtain the neuron's baseline discharge rate, we calculated the mean discharge rate during the last 500 ms of the fixation period for each cue condition. For activity during the cue period, we calculated the mean discharge rate during the 300-ms period (from 50 to 350 ms after the onset of the visual cue) for each cue condition. If this mean discharge rate differed significantly from the baseline discharge rate by the Mann–Whitney U-test (P < 0.05), we considered that the neuron had cue-period activity. For activity during the delay period, we calculated the mean discharge rate during the 3-s delay period for each cue condition. If the mean discharge rate differed significantly from the baseline discharge rate by the Mann–Whitney U-test (P < 0.05), we considered that the neuron had delay-period activity. Similarly, for activity during the response period, we calculated the mean discharge rate during the 300-ms response period (from −150 to 150 ms from the initiation of the saccade) for each cue condition. If the mean discharge rate during this period differed significantly from the baseline discharge rate by the Mann–Whitney U-test (P < 0.05), we considered that the neuron had response-period activity.

Determination of Preferred Directions

The population vector in a particular cue condition can be obtained by the weighted sum of cell vectors obtained from the individual neuron's activity under the same cue condition. The direction of the cell vector represents the preferred direction of task-related activity, whereas the length of the cell vector represents the magnitude of the activity in a particular cue condition. Therefore, we need to determine each neuron's preferred direction. In the present study, we first calculated the mean discharge rate during a particular period (e.g., cue period) and constructed a vector for each cue condition (cue vector). The direction of the cue vector corresponded to the direction of the visual cue and the length of the cue vector corresponded to the mean discharge rate. We then obtained a single vector (a resultant vector, R0) by summing up all cue vectors (8 cue vectors in the ODR task and 4 cue vectors in the R-ODR task). The direction of the resultant vector corresponds to the direction for which the neuron was maximally excited. We then tested the statistical significance of the obtained directional tuning using a statistical bootstrapping technique (Lurito et al. 1991; Smyrnis et al. 1992; Constantinidis et al. 2001; Kruse et al. 2002). To evaluate whether or not the obtained directional tuning could arise by chance, we constructed 1000 resultant vectors (R1R1000) using the previously described method by randomly assigning the discharge rate of each trial to 1 of the 8 (ODR task) or 4 (R-ODR task) cue directions. We then estimated the probability (P) that the lengths of 1000 randomly generated resultant vectors would exceed the length of the resultant vector R0. If the P value is less than 0.05, we defined that the neuron is spatially tuned and the preferred direction of the activity is the direction of the resultant vector R0.

Calculation of the Population Vector

The population vector in a particular cue condition can be obtained by the weighted sum of all cell vectors in that cue condition. Cue condition means a particular cue direction. Therefore, the weighting function for each neuron's contribution to the population vector is computed for one cue condition only, using each neuron's firing rate when saccades were directed toward that cue direction. The preferred direction for calculating the cell vector corresponds to the direction of the resultant vector R0, which is determined by the procedure described above. The population vector P of population size N under a particular cue condition is calculated as 

graphic
where Ci is the preferred direction of the ith cell and Wi is a weighted value of the ith cell's activity in a particular cue condition. Wi is calculated as 
graphic
where di is a square root–transformed value of the mean discharge rate of the ith cell during a particular period (e.g., the cue period) and ai is a square root–transformed value of the mean discharge rate of the ith cell during the fixation period.

Results

Database

We recorded the activities of 238 neurons one-at-a-time from the MD in 2 monkeys (monkey P, n = 201; monkey Q, n = 37). Because 153 of these 238 neurons were recorded in both tasks, the activities of these 153 neurons were analyzed in detail in the present experiment. In the ODR task, 51 (33%) had directional cue-period activity, 63 (41%) had directional delay-period activity, and 78 (51%) had directional response-period activity. Similarly, in the R-ODR task, 34 (22%) had directional cue-period activity, 47 (31%) had directional delay-period activity, and 70 (46%) had directional response-period activity. The histologically identified locations of recorded neurons within the MD have been reported in our previous study (Watanabe and Funahashi 2004a).

Population Vectors Calculated from a Population of Cue- and Response-Period Activities of MD Neurons

To confirm that population vectors calculated from MD activities correctly represent information regarding the direction of the visual cue or the direction of the saccade, we calculated population vectors using a population of cue-period activity and a population of response-period activity, respectively, recorded during ODR performance.

First, we calculated population vectors using a population of cue-period activity for all cue conditions. In the ODR task, 51 neurons exhibited directional cue-period activity. For this analysis, cue-period activity was defined as the mean discharge rate during the 300-ms period from 50 to 350 ms after the onset of the visual cue. Figure 2A shows the distributions of the preferred directions of these 51 cue-period activities. The preferred directions were distributed evenly around the FP and did not show a statistically significant directional bias (Rayleigh test for uniformity, P > 0.1). Figure 2B shows population vectors (thick lines) and cell vectors (thin arrows) calculated from the directional cue-period activity for each cue condition. The population vector in each cue condition was in a direction similar to the direction of the visual cue. The differences between the directions of population vectors and the directions of the visual cues were distributed between 1.0° and 23.2° (mean = 9.1°). Our previous study showed that cue-period activity observed in MD neurons represents information regarding the visual cue (Watanabe and Funahashi 2004b). Therefore, this result indicates that population vectors calculated from a population of cue-period activities recorded from the MD correctly represent information regarding the direction of the visual cue.

Figure 2.

Distributions of preferred directions and population vectors calculated from a population of cue- and response-period activities. (A) Distribution of preferred directions of 51 directional cue-period activities in the ODR task. No statistically significant directional bias was observed. (B) Cell vectors (thin arrows) and population vectors (thick lines) obtained from cue-period activities for all cue conditions in the ODR task. The directions of population vectors were 7.7° for the 0° condition, 49.2° for the 45° condition, 97.5° for the 90° condition, 146.0° for the 135° condition, 179.0° for the 180° condition, 208.5° for the 225° condition, 246.8° for the 270° condition, and 313.4° for the 315° condition. (C) Distribution of preferred directions of 78 directional response-period activities in the ODR task. No statistically significant directional bias was observed. (D) Cell vectors (thin arrows) and population vectors (thick lines) obtained by response-period activities for each cue condition in the ODR task. The directions of population vectors were 40.7° for the 0° condition, 58.9° for the 45° condition, 81.9° for the 90° condition, 105.3° for the 135° condition, 141.5° for the 180° condition, 209.5° for the 225° condition, 275.1° for the 270° condition and 339.2° for the 315° condition.

Figure 2.

Distributions of preferred directions and population vectors calculated from a population of cue- and response-period activities. (A) Distribution of preferred directions of 51 directional cue-period activities in the ODR task. No statistically significant directional bias was observed. (B) Cell vectors (thin arrows) and population vectors (thick lines) obtained from cue-period activities for all cue conditions in the ODR task. The directions of population vectors were 7.7° for the 0° condition, 49.2° for the 45° condition, 97.5° for the 90° condition, 146.0° for the 135° condition, 179.0° for the 180° condition, 208.5° for the 225° condition, 246.8° for the 270° condition, and 313.4° for the 315° condition. (C) Distribution of preferred directions of 78 directional response-period activities in the ODR task. No statistically significant directional bias was observed. (D) Cell vectors (thin arrows) and population vectors (thick lines) obtained by response-period activities for each cue condition in the ODR task. The directions of population vectors were 40.7° for the 0° condition, 58.9° for the 45° condition, 81.9° for the 90° condition, 105.3° for the 135° condition, 141.5° for the 180° condition, 209.5° for the 225° condition, 275.1° for the 270° condition and 339.2° for the 315° condition.

Second, we calculated population vectors using a population of response-period activity for all cue conditions. In the ODR task, 78 neurons showed directional response-period activity. For this analysis, response-period activity was defined as the mean discharge rate during the 300-ms period from −150 to 150 ms at the initiation of the saccade. Figure 2C shows the distributions of the preferred directions of response-period activity in these 78 neurons. The preferred directions were distributed around the FP. Although more preferred directions were directed upward, we did not observe a significant bias in the distributions of the preferred directions (Rayleigh test for uniformity, P = 0.08). Figure 2D shows population vectors (thick lines) and cell vectors (thin arrows) calculated from a population of directional response-period activity for each cue condition. Although the directions of population vectors showed upward bias, the differences between the directions of population vectors and the directions of the saccade targets were distributed between 5.1° and 40.7° (mean = 22.0°) and did not exceed 45o. Thus, the population vector in each cue condition was in a direction similar to the direction of the saccade target. Our previous study showed that most of the response-period activity in the MD was presaccadic and that most response-period activities represented information regarding saccade directions (Watanabe and Funahashi 2004b). The present results indicate that population vectors calculated from a population of response-period activities again represent information regarding the direction of the saccade target.

Thus, population vectors calculated from a population of cue-period activities and a population of response-period activities represent the directions of the visual cues and the directions of the saccade targets, respectively. These results indicate that the population vector analysis could be a suitable method for visualizing spatial information represented by a population of MD neurons and the temporal change in spatial information represented by MD neurons during the progress of the ODR task.

Overall Preferred Direction for Each Neuron

Among the 153 MD neurons examined in the present experiment, neurons exhibiting task-related activity at 2 or more task events were present. The preferred directions of task-related activities occurred at different task events were not always the same for these neurons. Therefore, to analyze the temporal change in the directions of population vectors during the progress of the trial, we need to determine one overall preferred direction for each MD neuron. We compared the preferred directions of task-related activities for neurons that exhibited significant activity at 2 or more task events in our previous report (Watanabe and Funahashi 2004a). In neurons that exhibited both cue- and delay-period activities (n = 25), the preferred directions of both activities were almost identical (correlation coefficient, r = 0.92, P < 0.01). Therefore, the preferred direction of cue-period activity was used as the overall preferred direction for these neurons. In neurons that exhibited both delay- and response-period activities (n = 57), a positive but nonsignificant correlation (r = 0.12) was observed in the preferred directions between these 2 activities (Watanabe and Funahashi 2004a). This population included neurons that exhibited all cue-, delay-, and response-period activities (n = 23), as well as neurons which exhibited only delay- and response-period activities (n = 34). Because the preferred directions of cue- and delay-period activities were almost identical, the preferred direction of cue-period activity was used as the overall preferred direction for 23 neurons that exhibited all cue-, delay-, and response-period activities. However, in 34 neurons that exhibited only delay- and response-period activity, a significant positive correlation (r = 0.77, P < 0.01) was observed in the preferred directions between these 2 activities. Therefore, the preferred direction of response-period activity was used as an overall preferred direction for these 34 neurons,

Temporal Changes in the Directions of Population Vectors during the Delay Period

ODR Task

To examine the temporal changes in the spatial information represented by a population of MD neurons during the delay period of the ODR task, we obtained population vectors using a population of MD activities during a 250-ms time window that moved from the initiation of the cue period to the end of the delay period in a 50-ms step. As a result, 70 population vectors were obtained for each cue condition. Figure 3A shows an example of this analysis at the 90° cue condition of the ODR task. Most of the population vectors were pointed toward the direction of the visual cue, and this tendency was maintained throughout the delay period. To confirm this observation, we calculated the mean differences between the directions of the population vectors and the direction of the visual cue during the delay period using the data obtained from all cue conditions. As shown in Figure 3B, the mean differences were close to 0° during the delay period. Because the direction of the visual cue and the direction of the saccade were the same in the ODR task, this result indicates that the directional information of either the visual cue or the saccade is maintained by a population of MD neurons during the delay period in the ODR task.

Figure 3.

(A) Temporal change in population vectors in the 90° cue condition of the ODR task. Population vectors were calculated using the activities of 67 neurons. (B) Temporal change in the differences between the directions of population vectors and the directions of the visual cues using the database obtained from all cue conditions of the ODR task. The mean and the standard error are shown for each point. (C) Temporal changes in population vectors in the 90° cue condition of the R-ODR task. Population vectors were calculated using the activities of 67 neurons. (D) Temporal change in the differences between the directions of population vectors and the directions of the visual cues using the database obtained from all cue conditions of the R-ODR task. The mean and the standard error are shown for each point.

Figure 3.

(A) Temporal change in population vectors in the 90° cue condition of the ODR task. Population vectors were calculated using the activities of 67 neurons. (B) Temporal change in the differences between the directions of population vectors and the directions of the visual cues using the database obtained from all cue conditions of the ODR task. The mean and the standard error are shown for each point. (C) Temporal changes in population vectors in the 90° cue condition of the R-ODR task. Population vectors were calculated using the activities of 67 neurons. (D) Temporal change in the differences between the directions of population vectors and the directions of the visual cues using the database obtained from all cue conditions of the R-ODR task. The mean and the standard error are shown for each point.

R-ODR Task

In the R-ODR task, the direction of the saccade is 90° clockwise from the cue direction. The population vector analysis using a population of DLPFC activities revealed that the directions of population vectors rotated from the cue direction to the saccade direction in the R-ODR task (Takeda and Funahashi 2004). However, the rotation started in the late phase of the delay period in the DLPFC. This is because, although delay-period activity of DLPFC neurons represented forthcoming saccade information, the proportion of these DLPFC neurons was small (13%) (Takeda and Funahashi 2002) and because delay-period activity representing saccade information tended to show gradually increasing activation toward the saccade performance (Takeda and Funahashi 2004). On the other hand, a large proportion of MD neurons (41%) having delay-period activity represented forthcoming saccade information and the MD may be one of the source structures that provide information regarding forthcoming saccade information to the DLPFC (Watanabe and Funahashi 2004b). If this is the case, the information regarding the saccade direction would develop earlier during the delay period in the MD than the DLPFC. To test this hypothesis, we constructed population vectors using a population of MD activities during the R-ODR task and compared the present findings with those observed in DLPFC neurons reported by Takeda and Funahashi (2004).

The preferred direction of each neuron was determined from the activity observed in the ODR task. Figure 3C shows an example of the temporal change in population vectors for the 90° cue condition of the R-ODR task. The directions of population vectors gradually rotated from the cue direction toward the saccadic direction during the delay period. This rotation of the vector direction began in the early phase of the delay period (approximately 0.5 s after the start of the delay period). Figure 3D shows the temporal change in the mean differences between the directions of population vectors and the direction of the visual cue during the delay period using the data obtained from all cue conditions of the R-ODR task. The mean differences between 2 directions moved away from 0° in the early phase of the delay period. The direction of the population vector rotated slowly and gradually pointed toward the direction of the saccade as the trial progressed. The deviation of the mean value from 0° began within 0.5 s after the start of the delay period. Thereafter, population vectors were maintained at around 45° from the direction of the visual cue until the last 1 s of the delay period. During the last 1 s of the delay period, population vectors further rotated toward the direction of the saccade. These results indicate that directional information represented by a population of MD neurons began to change from the direction of the visual cue to the direction of the saccade in the early phase of the delay period.

Temporal changes of directions of population vectors during the delay period were compared between MD neurons and DLPFC neurons (Takeda and Funahashi 2004). Figure 4 shows temporal changes in the mean differences between the directions of population vectors and the direction of the visual cue during the delay period of the R-ODR task. The mean values obtained from MD neurons moved away from 0° earlier during the delay period than DLPFC neurons. To determine when the population vector began to rotate, we first calculated the mean and the standard deviation (SD) of the mean differences between the directions of population vectors and the direction of the visual cue during the cue period. Then, if the absolute value of the mean differences exceeded 2SD from the mean at 3 consecutive windows during the delay period, we defined the first window of these 3 consecutive windows as the start timing to rotate. The mean and the SD for both DLPFC and MD populations were –2.3 and 5.5, respectively. For MD neurons, the mean differences exceeded 2SD from the mean at 300–550 ms after the start of the delay period. On the other hand, for DLPFC neurons, the mean differences exceeded 2SD from the mean at 1500–1750 ms after the start of the delay period. These results indicate that the information regarding the saccade direction develops earlier during the delay period in the MD than in the DLPFC. Thus, this result supports the idea that the MD is one of the source structures that provide information regarding forthcoming saccade information to the DLPFC.

Figure 4.

Comparison of the temporal change in the differences between the directions of population vectors and the directions of the visual cues in the R-ODR task between MD neurons and DLPFC neurons. Black line, MD activities; dotted line, DLPFC activity based on Takeda and Funahashi (2004).

Figure 4.

Comparison of the temporal change in the differences between the directions of population vectors and the directions of the visual cues in the R-ODR task between MD neurons and DLPFC neurons. Black line, MD activities; dotted line, DLPFC activity based on Takeda and Funahashi (2004).

Discussion

In the present study, we calculated population vectors using a population of MD activities recorded while monkeys performed 2 types of ODR tasks and tried to understand functional roles of the MD in sensory-to-motor information transformation during spatial working memory performance and compare with those of the DLPFC. To confirm the usefulness of the population vector for the present study, population vectors were first constructed using a population of cue- and response-period activities in both tasks. The directions of these population vectors were similar to the cue directions and the directions of the saccade targets, respectively, indicating that the population vector calculated from a population of MD activities is a useful index to visualize spatial information represented by a population of neurons. We then examined the temporal change in spatial information represented by a population of MD neurons as the trial progressed for the 2 tasks. In the ODR task, population vectors were directed mostly toward the cue direction during the delay period, whereas in the R-ODR task, the direction of the population vector began to rotate in the early phase of the delay period and gradually pointed toward the direction of the saccade as the trial progressed. These results indicate that the transformation from visual information to saccade information occurs during the delay period in the MD. In addition, compared with the results obtained with DLPFC neurons (Takeda and Funahashi 2004), the rotation of population vectors started earlier in the MD than in the DLPFC, indicating that the MD could be the source structure that provides information regarding forthcoming saccade information to the DLPFC.

Population Vectors Calculated from a Population of Cue- and Response-Period Activities

Neurons in the MD that respond to visual stimuli have been reported previously (Fuster and Alexander 1973; Tanibuchi and Goldman-Rakic 2003; Wyder et al. 2003; Sommer and Wurtz 2004a; Watanabe and Funahashi 2004a). We showed that cue-period activity represented information regarding a visual cue by comparing the preferred directions of cue-period activity between the ODR and R-ODR tasks in the same MD neurons (Watanabe and Funahashi 2004b). Based on these observations, we first evaluated how well the directions of population vectors calculated from a population of cue-period activities correctly represented the directions of the visual cues in MD neurons. As shown in Figure 2B, the differences between the directions of population vectors and the directions of the visual cues were distributed between 1.0° and 23.2° (mean = 9.1°). Therefore, we conclude that the directions of population vectors calculated from a population of cue-period activities well represent the directions of the visual cue in MD neurons.

We also evaluated how well the directions of population vectors calculated from a population of response-period activities represented the directions of the saccades in MD neurons. By comparing the preferred directions of response-period activity between the ODR and R-ODR tasks in the same MD neurons, we found that most of these neurons represented the direction of the saccade (Watanabe and Funahashi 2004b). In addition, most of response-period activity observed in MD neurons was presaccadic (Watanabe and Funahashi 2004a). As shown in Figure 2D, the differences between the directions of population vectors and the directions of the saccades were distributed between 5.1° and 40.7° (mean = 22.0°). Although an upward bias was observed in the directions of population vectors, the magnitude of the bias was less than 45o. Therefore, this result indicates that the directions of population vectors calculated from a population of response-period activities well represent the directions of the saccade targets. The direction of the population vector is known to be strongly affected by the distribution of preferred directions. If the preferred directions of response-period activities were evenly distributed, the direction of the population vector would be close to the direction of the saccade target. Because preferred directions of response-period activities are biased toward upward directions in this study, this bias may cause the upward bias in directions of population vectors calculated from a population of response-period activities.

Temporal Changes in the Directions of Population Vectors during the Delay Period

In both the ODR and R-ODR tasks, a visual cue was initially presented and, after the 3-s delay period, a saccade toward the correct direction was needed based on the task rules. In the present study, the direction of a population vector calculated from a population of cue-period activities represented the direction of the visual cue, whereas the direction of a population vector calculated from a population of response-period activities represented the direction of the saccade. This result indicates that the transformation from visual information to motor information occurs in the MD during the delay period. This result also suggests that the temporal change in the directions of population vectors during the delay period can be used to visualize this transformation process. Therefore, to visualize the process of information transformation during the delay period, we calculated population vectors using a population of MD activities during a 250-ms time window that moved from the onset of the cue period to the end of the delay period at a 50-ms step.

Population vectors obtained during the delay period of the ODR task were pointed mostly in the direction of the visual cue because the direction of the visual cue and the direction of the saccade were the same in the ODR task. This result indicates that population vectors calculated from a population of activities during the delay period correctly represent information regarding the direction of the visual cue and/or the direction of the saccade. On the other hand, in the R-ODR task, we found that the direction of the population vector moved away from the direction of the visual cue in the early phase of the delay period and gradually rotated toward the direction of the saccade as the trial progressed. Because the direction of the saccade was 90o clockwise from the direction of the visual cue in the R-ODR task, this result indicates that information represented by a population of MD neurons gradually transformed from visual information to motor information during the delay period. A similar transformation from visual information to motor information could occur during the delay period of the ODR task, although we could not observe such a phenomenon because the direction of the visual cue and the direction of the saccade were the same in the ODR task.

Takeda and Funahashi (2004) examined temporal change in the directions of population vectors calculated from a population of DLPFC activities during the delay period in the R-ODR task. They also reported the rotation of population vectors as the R-ODR trial progressed. We compared the temporal change in population vectors during the delay period between MD neurons obtained in the present study and DLPFC neurons reported by Takeda and Funahashi (2004). Figure 4 shows temporal changes in the mean differences between the directions of population vectors calculated from a population of MD and DLPFC activities and the direction of the visual cue in the R-ODR task. The mean difference obtained from MD neurons moved away from 0° in an earlier phase of the delay period compared with the mean difference obtained from DLPFC neurons. For MD neurons, population vectors began to rotate toward the direction of the saccade within 0.5 s after the start of the delay period, whereas for DLPFC neurons, population vectors began to rotate during the last 1.5 s of the delay period. These results indicate that the transformation from visual information to saccade information begins sooner in the MD than in the DLPFC. Presaccadic activity accounted for most of response-period activity in the MD (Watanabe and Funahashi 2004a), whereas postsaccadic activity comprised most of response-period activity in the DLPFC (Funahashi et al. 1991). In addition, neurons with presaccadic activity were found mainly in the lateral part of the MD (Watanabe and Funahashi 2004a). Sommer and Wurtz (2002, 2004b) indicated that presaccadic activity observed in MD neurons is a corollary discharge that represents the vector of the impending saccade, which is transmitted from the superior colliculus to the frontal eye field through the lateral part of the MD. For MD neurons, 56% of directional delay-period activity represented information regarding the location of the visual cue (retrospective information coding), whereas the remaining 41% represented information regarding the direction of the saccade (prospective information coding) (Watanabe and Funahashi 2004b). On the other hand, in the DLPFC, 86% of directional delay-period activity represented retrospective information, whereas only 13% represented prospective information (Takeda and Funahashi 2002). A similar result that most delay-period activity represents retrospective information in the DLPFC has also been reported in other studies (Niki and Watanabe 1976; Funahashi et al. 1993). These results indicate that the MD is more related to the motor aspect of information processing, whereas the DLPFC is more related to the sensory aspect of information processing and that the DLPFC receives a feedback of motor information from the MD. The present study showed that the transformation from the visual information to the motor information begins earlier in the MD than the DLPFC during the delay period. Although it is not known whether the signals to initiate the transformation from the visual information to the saccade information are generated within the MD or transmitted from other brain structures to the MD, the MD might send information processed within the MD to the DLPFC and contribute to the information processing occurred in the DLPFC.

Funding

The Japanese Ministry of Education, Science, Sports, Culture, and Technology (MEXT; Grant-in-Aid for Scientific Research on Priority Areas––System study on higher order brain functions [18020016 to S.F.] and Grants-in-Aid for Scientific Research [14380367, 17300103 to S.F. and 18730481 to Y.W.); the 21st Century COE Program (D-2 to Kyoto University), MEXT, Japan.

The authors thank Dr N. Matsumoto for valuable advices to analyze neural activity. Conflict of Interest: None declared.

References

Baddeley
A
Working memory
 , 
1986
Oxford
Oxford University Press
Chafee
MV
Goldman-Rakic
PS
Matching patterns of activity in primate prefrontal area 8a and parietal area 7ip neurons during a spatial working memory task
J Neurophysiol
 , 
1998
, vol. 
79
 (pg. 
2919
-
2940
)
Constantinidis
C
Franowicz
MN
Goldman-Rakic
PS
Coding specificity in cortical microcircuits: a multiple-electrode analysis of primate prefrontal cortex
J Neurosci
 , 
2001
, vol. 
21
 (pg. 
3646
-
3655
)
Funahashi
S
Neuronal mechanisms of executive control by the prefrontal cortex
Neurosci Res
 , 
2001
, vol. 
39
 (pg. 
147
-
165
)
Funahashi
S
Bruce
CJ
Goldman-Rakic
PS
Neuronal activity related to saccadic eye movements in the monkey's dorsolateral prefrontal cortex
J Neurophysiol
 , 
1991
, vol. 
65
 (pg. 
1464
-
1483
)
Funahashi
S
Chafee
MV
Goldman-Rakic
PS
Prefrontal neuronal activity in rhesus monkeys performing a delayed anti-saccade task
Nature
 , 
1993
, vol. 
365
 (pg. 
753
-
756
)
Funahashi
S
Kubota
K
Working memory and prefrontal cortex
Neurosci Res
 , 
1994
, vol. 
21
 (pg. 
1
-
11
)
Fuster
JM
Inferotemporal units in selective visual attention and short-term memory
J Neurophysiol
 , 
1990
, vol. 
64
 (pg. 
681
-
697
)
Fuster
JM
The prefrontal cortex
 , 
1997
Philadelphia (PA)
Lippincott-Raven
Fuster
JM
Alexander
GE
Neuron activity related to short-term memory
Science
 , 
1971
, vol. 
173
 (pg. 
652
-
654
)
Fuster
JM
Alexander
GE
Firing changes in cells of the nucleus medialis dorsalis associated with delayed response behavior
Brain Res
 , 
1973
, vol. 
61
 (pg. 
79
-
91
)
Georgopoulos
AP
Neural integration of movement: role of motor cortex in reaching
FASEB J
 , 
1988
, vol. 
2
 (pg. 
2849
-
2857
)
Georgopoulos
AP
Caminiti
R
Kalaska
JF
Massey
JT
Spatial coding of movement: a hypothesis concerning the coding of movement direction by motor cortical populations
Exp Brain Res
 , 
1983
, vol. 
7
 
Suppl
(pg. 
327
-
336
)
Georgopoulos
AP
Kettner
RE
Schwartz
AB
Primate motor cortex and free arm movements to visual targets in three-dimensional space. II. Coding of the direction of movement by a neuronal population
J Neurosci
 , 
1988
, vol. 
8
 (pg. 
2928
-
2937
)
Georgopoulos
AP
Lurito
JT
Petrides
M
Schwartz
AB
Massey
JT
Mental rotation of the neuronal population vector
Science
 , 
1989
, vol. 
243
 (pg. 
234
-
237
)
Georgopoulos
AP
Schwartz
AB
Kettner
RE
Neuronal population coding of movement direction
Science
 , 
1986
, vol. 
233
 (pg. 
1416
-
1419
)
Georgopoulos
AP
Taira
M
Lukashin
A
Cognitive neurophysiology of the motor cortex
Science
 , 
1993
, vol. 
260
 (pg. 
47
-
52
)
Giguere
M
Goldman-Rakic
PS
Mediodorsal nucleus: areal, laminar, and tangential distribution of afferents and efferents in the frontal lobe of rhesus monkeys
J Comp Neurol
 , 
1988
, vol. 
277
 (pg. 
195
-
213
)
Gnadt
JW
Andersen
RA
Memory related motor planning activity in posterior parietal cortex of macaque
Exp Brain Res
 , 
1988
, vol. 
70
 (pg. 
216
-
220
)
Goldman-Rakic
PS
Circuitry of primate prefrontal cortex and regulation of behavior by representational memory
Plum F, editor. Higher functions of the brain, part I, handbook of physiology. Section I: the nervous system
 , 
1987
, vol. 
Vol. V
 
374–417. Bethesda (MD)
American Physiological Society
Goldman-Rakic
PS
Porrino
LJ
The primate mediodorsal (MD) nucleus and its projection to the frontal lobe
J Comp Neurol
 , 
1985
, vol. 
242
 (pg. 
535
-
560
)
Kettner
RE
Schwartz
AB
Georgopoulos
AP
Primate motor cortex and free arm movements to visual targets in three-dimensional space. III. Positional gradients and population coding of movement direction from various movement origins
J Neurosci
 , 
1988
, vol. 
8
 (pg. 
2938
-
2947
)
Kievit
J
Kuypers
HG
Organization of the thalamo-cortical connections to the frontal lobe in the rhesus monkey
Exp Brain Res
 , 
1977
, vol. 
29
 (pg. 
299
-
322
)
Kruse
W
Dannenberg
S
Kleiser
R
Hoffmann
KP
Temporal relation of population activity in visual areas MT/MST and in primary motor cortex during visually guided tracking movements
Cereb Cortex
 , 
2002
, vol. 
12
 (pg. 
466
-
476
)
Lurito
JT
Georgakopoulos
T
Georgopoulos
AP
Cognitive spatial-motor processes. 7. The making of movements at an angle from a stimulus direction: studies of motor cortical activity at the single cell and population levels
Exp Brain Res
 , 
1991
, vol. 
87
 (pg. 
562
-
580
)
Miyake
A
Shah
P
Models of working memory: mechanisms of active Maintenance and executive control
1999
Cambridge (UK)
Cambridge University Press
Miyashita
Y
Chang
HS
Neuronal correlate of pictorial short-term memory in the primate temporal cortex
Nature
 , 
1988
, vol. 
331
 (pg. 
68
-
70
)
Niki
H
Watanabe
M
Prefrontal unit activity and delayed response: relation to cue location versus direction of response
Brain Res
 , 
1976
, vol. 
105
 (pg. 
79
-
88
)
Petrides
M
Boller
F
Spinnler
H
Hendler
JA
Frontal lobes and working memory: evidence from investigation of the effects of cortical excisions in nonhuman primates
Handbook of neuropsychology
 , 
1994
, vol. 
Vol. 9
 
Amsterdam
Elsevier
(pg. 
59
-
82
)
Ray
JP
Price
JL
The organization of projections from the mediodorsal nucleus of the thalamus to orbital and medial prefrontal cortex in macaque monkeys
J Comp Neurol
 , 
1993
, vol. 
337
 (pg. 
1
-
31
)
Robinson
DA
A method of measuring eye movement using a scleral search coil in a magnetic field
IEEE Trans Biomed Eng
 , 
1963
, vol. 
10
 (pg. 
137
-
145
)
Schwartz
AB
Kettner
RE
Georgopoulos
AP
Primate motor cortex and free arm movements to visual targets in three-dimensional space. I. Relations between single cell discharge and direction of movement
J Neurosci
 , 
1988
, vol. 
8
 (pg. 
2913
-
2927
)
Smyrnis
N
Taira
M
Ashe
J
Georgopoulos
AP
Motor cortical activity in a memorized delay task
Exp Brain Res
 , 
1992
, vol. 
92
 (pg. 
139
-
151
)
Sommer
MA
Wurtz
RH
A pathway in primate brain for internal monitoring of movements
Science
 , 
2002
, vol. 
296
 (pg. 
1480
-
1482
)
Sommer
MA
Wurtz
RH
What the brainstem tells the frontal cortex. I. Oculomotor signals sent from superior colliculus to frontal eye field via mediodorsal thalamus
J Neurophysiol
 , 
2004
, vol. 
91
 (pg. 
1381
-
1402
)
Sommer
MA
Wurtz
RH
What the brain stem tells the frontal cortex. II. Role of the SC-MD-FEF pathway in corollary discharge
J Neurophysiol
 , 
2004
, vol. 
91
 (pg. 
1403
-
1423
)
Takeda
K
Funahashi
S
Prefrontal task-related activity representing visual cue location or saccade direction in spatial working memory tasks
J Neurophysiol
 , 
2002
, vol. 
87
 (pg. 
567
-
588
)
Takeda
K
Funahashi
S
Population vector analysis of primate prefrontal activity during spatial working memory
Cereb Cortex
 , 
2004
, vol. 
14
 (pg. 
1328
-
1339
)
Tanibuchi
I
Goldman-Rakic
PS
Dissociation of spatial-, object-, and sound-coding neurons in the mediodorsal nucleus of the primate thalamus
J Neurophysiol
 , 
2003
, vol. 
89
 (pg. 
1067
-
1077
)
Watanabe
Y
Funahashi
S
Neuronal activity throughout primate mediodorsal nucleus of the thalamus during oculomotor delayed-response performances. I. Activity encoding visual versus motor signal
J Neurophysiol
 , 
2004
, vol. 
92
 (pg. 
1738
-
1755
)
Watanabe
Y
Funahashi
S
Neuronal activity throughout primate mediodorsal nucleus of the thalamus during oculomotor delayed-response performances. II. Task-related activity represents either visual or motor information
J Neurophysiol
 , 
2004
, vol. 
92
 (pg. 
1756
-
1769
)
Watanabe
Y
Funahashi
S
Population vector analysis of primate mediodorsal thalamic activity during oculomotor delayed-response performances
 , 
2006
 
Atlanta (GA) Program No. 138.15 Neurosci Meeting Planner, Soc Neurosci CD-ROM
Wyder
MT
Massoglia
DP
Stanford
TR
Quantitative assessment of the timing and tuning of visual-related, saccade-related, and delay period activity in primate central thalamus
J Neurophysiol
 , 
2003
, vol. 
90
 (pg. 
2029
-
2052
)