Abstract

To examine how delay-period activity participates in the decision of a saccade direction, we analyzed prefrontal activity while monkeys performed 2 tasks: oculomotor delayed-response (ODR) and self-selection ODR (S-ODR) tasks. In the ODR task, monkeys were required to make a memory-guided saccade to the cue location after a 3-s delay. In the S-ODR task, 4 identical visual cues were presented simultaneously during the cue period and monkeys were required to make a saccade in any one direction after the delay. Delay-period activity was observed in both tasks in the same neuron with similar directional preferences. Neurons with delay-period activity were classified into several groups based on the temporal pattern of the activity itself and of the strength of the directional selectivity. Among these, neurons with an increasing type of delay-period activity with persistent directional selectivity throughout the delay period in the ODR task also showed directional delay-period activity in the S-ODR task. These results indicate that an increasing type of delay-period activity, which is thought to represent motor information, plays an important role in generating and enhancing directional bias in the S-ODR task and therefore contributes significantly to the decision process of the saccade direction in the S-ODR task.

Introduction

Decision making can be described as a process of response selection in which animals choose a particular action among several alternatives. The ability to make rational decisions in complex environments is a crucial higher brain function. A number of electrophysiological and brain imaging studies have shed light on the role of the dorsolateral prefrontal cortex (DLPFC) in decision making (Kim and Shadlen 1999; Barraclough et al. 2004; Lau et al. 2004). Using an oculomotor delayed-response (ODR) paradigm, we have studied neural components that may reflect the decision process of the saccade direction in the DLPFC by comparing task-related activity between ODR and S-ODR tasks (Watanabe et al. 2006). In the self-selection ODR (S-ODR) task, 4 identical visual cues were presented simultaneously during the cue period and monkeys were required to choose 1 out of 4 directions for the saccade after a 3-s delay period. We found that cue-period activity observed in the ODR task lost its directional selectivity in the S-ODR task. In addition, presaccadic activity showed very similar directional selectivity in the ODR and S-ODR tasks. Temporal coupling between the onset timings of presaccadic activity and saccadic latencies was observed in both tasks. Based on these results, we concluded that cue-period activity and presaccadic activity are not directly related to the decision process of the saccade direction in the S-ODR task (Watanabe et al. 2006). On the other hand, delay-period activity observed in the ODR task showed differential activation in the S-ODR task depending on the directions of forthcoming saccades. The strength of the directional selectivity gradually increased during the delay period in the S-ODR task. These results suggest that neurons showing directional delay-period activity in the ODR task are the principal neuronal component of the decision process of the saccade direction in the S-ODR task.

In the present study, we further examined the contribution of delay-period activity to the decision process of the saccade direction in the S-ODR task. Previous studies have indicated that delay-period activity observed in the DLPFC is divided into several types, such as activity representing the sensory nature of the stimulus or the motor aspect of the required response (Niki and Watanabe 1976; Quintana et al. 1988; Quintana and Fuster 1992, 1999; Funahashi et al. 1993; Hasegawa et al. 1998; Takeda and Funahashi 2002). Moreover, there seems to be correspondence between the temporal pattern of delay-period activity and its functional role, such that an increasing type of delay-period activity tends to represent motor or prospective information, whereas a decreasing type of delay-period activity tends to represent sensory or retrospective information (Kojima and Goldman-Rakic 1982; Quintana and Fuster 1992, 1999; Takeda and Funahashi 2002). Based on these results, we hypothesized that, among neurons showing directional delay-period activity in the ODR task, groups of neurons showing particular characteristics of delay-period activity are the source of the gradual increase of the strength of the directional selectivity during the delay period in the S-ODR task, thus playing a crucial role in the decision process of the saccade direction in the S-ODR task. In this study, we classified directional delay-period activity observed in the ODR task based on 2 criteria: the temporal pattern of the strength of the directional selectivity and the temporal pattern of the delay-period activity itself. Then, we examined which group of neurons also exhibited directional delay-period activity in the S-ODR task. Preliminary results have been published in abstract form (Watanabe and Funahashi 2005).

Materials and Methods

Subjects and Apparatus

Two male macaque monkeys (monkey R, ∼9 kg; monkey Z, ∼8 kg) used in this study were the same as those reported previously (Watanabe et al. 2006). The experimental apparatus and the surgical method have also been described in detail (Watanabe et al. 2006). Briefly, during training and recording sessions, the monkey was seated in a primate chair in a dark sound-attenuated room. The monkey's head movement was restricted by a head holder. Visual stimuli were presented on a 21-inch CRT monitor (Eizo Flex Scan, Nanao, Japan) placed in front of the monkey. Eye movements were monitored by a magnetic search coil technique. All experiments were controlled by 3 laboratory computers using TEMPO software (Reflective Computing, Olympia, WA). 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

We used 2 ODR tasks in this study: an ordinary ODR task and the free-choice version of the ODR task (S-ODR task). In the ODR task, monkeys were required to make a memory-guided saccade after a 3-s delay to the location where a visual cue had been presented. The temporal order of task events is illustrated in Figure 1A (left). Each trial began with the appearance of a fixation point (FP; a small white circle, 0.5° in visual angle) at the center of the monitor. After the monkey looked at the FP for 1 s, a visual cue (white circle, 1° in visual angle) appeared for 500 ms (cue period) at 1 of 4 predetermined peripheral locations (0°, 90°, 180°, or 270°; 17° eccentricity, see Fig. 1B). The location of the visual cue varied randomly. Occasionally we used 8 peripheral locations for cue presentation by adding 4 diagonal locations (45°, 135°, 225°, or 315°). Monkeys were required to maintain fixation at the FP until the end of the 3-s delay period. At the end of the delay period, the FP was extinguished and monkeys were required to make a saccade within 400 ms (response period) to the location where the visual cue had been presented. A drop of juice was given as a reward for a correct saccade. Each recording session always started with a block of 50–100 ODR trials to examine whether a neuron showed task-related activity for any task event and was followed by a block of 50–100 S-ODR trials.

Figure 1.

(A) Schematic diagram of the ODR task (left) and the S-ODR task (right). Solid arrows depict rewarded saccade directions in both tasks. (B) Arrangements of visual cue locations presented in the ODR and S-ODR tasks.

Figure 1.

(A) Schematic diagram of the ODR task (left) and the S-ODR task (right). Solid arrows depict rewarded saccade directions in both tasks. (B) Arrangements of visual cue locations presented in the ODR and S-ODR tasks.

In the S-ODR task, monkeys were required to choose the saccade direction by themselves. The temporal sequence of the S-ODR task (Fig. 1A, right) is basically the same as in the ODR task, except that 4 identical visual cues (white circle, 1° in diameter) were presented simultaneously around the FP (0°, 90°, 180°, and 270°; upright version) during the cue period. Occasionally, depending on the preferred direction of a recorded activity, we used a diagonal version of the visual cue locations (45°, 135°, 225°, and 315°, see Fig. 1B). Either the upright or the diagonal version of the visual cue locations was used in a single block of the S-ODR task. Monkeys received the same amount of juice as a reward as that in the ODR task, whichever direction they chose as a saccade target. In this task, monkeys often chose one particular direction repeatedly. To avoid repetition of saccades in the same direction, we stopped giving rewards after the monkey chose the same saccade direction 4 times (i.e., repetitive saccades in the same direction were allowed up to 4 times). This forced the monkey to select another saccade direction.

Data Collection

We recorded single-neuron activity from the cortex within and surrounding the principal sulcus. The area of the recording in the DLPFC was initially determined by the magnetic resonance imaging (MRI) pictures of the brains of each monkey taken at the National Institute of Physiological Sciences, Japan. We used glass-coated Elgiloy microelectrodes (0.5–2.0 MΩ at 1 kHz) for recording single-neuron activity. An electrode was advanced with a hydraulic microdrive (MO-95, Narishige, Tokyo). Raw neuron activity was amplified using an amplifier (DAM80, WPI, Sarasota, FL) and monitored on an oscilloscope (SS-7802, IWATSU, Tokyo) and an audio monitor. During experiments, we isolated single-neuron activity from raw activity using a window discriminator (DIS-1, BAK Electronics, Mount Airy, MD) and monitored the isolated single-neuron activity together with raw activity using an oscilloscope (SS-7802, IWATSU). Single-neuron activity and task events were stored as a data file in a laboratory computer. We first monitored and recorded single-neuron activity during the ODR task. If task-related activity was observed for any task event of the ODR task, we recorded the activity of the same neuron during the S-ODR task.

Analysis of Directional Selectivity of Delay-Period Activity

Among the various task-related activities observed in the ODR task, we focused on the directional delay-period activity. A statistical analysis was conducted to determine whether directional selectivity was present or absent in delay-period activity. First, to obtain the baseline discharge rate of a neuron, the mean discharge rate was calculated during the last 800 ms of the fixation period for each cue condition. We then calculated the mean discharge rate during the 3-s delay period for each cue condition. If the mean discharge rate during the delay period significantly differed from the baseline discharge rate (Mann–Whitney U-test, P < 0.05), the neuron was considered to have delay-period activity. For neurons with delay-period activity, we examined whether delay-period activity was directionally selective by comparing mean discharge rates during the delay period across all cue conditions by 1-way analysis of variance (ANOVA). If the difference was significant (P < 0.05), delay-period activity was considered to show directional selectivity. In this study, we referred to the activity recorded during the ODR task to determine whether a neuron had directional delay-period activity. The maximum and minimum response directions of delay-period activity were also determined on the basis of the activity during the ODR task, unless mentioned otherwise. The maximum and minimum response directions refer to the cue directions for which the highest and lowest delay-period activities, respectively, were observed.

Calculation of the Best Direction of Delay-Period Activity

A tuning curve was constructed to estimate the best direction of delay-period activity quantitatively. The tuning curve was obtained by fitting the mean spike rates of the delay-period activity across all cue directions for the ODR task or all saccade directions for the S-ODR task to the following Gaussian function: 

graphic
where f(d) is the discharge rate of delay-period activity as a function of the cue location d (ODR task) or of the saccade direction d (S-ODR task). The constant B is the baseline discharge rate, R is the discharge rate above the baseline discharge rate for the best direction, D is the best direction where the discharge rate of delay-period activity would be highest according to the fitted Gaussian curve, and Td is an index of the tuning width.

ROC Analysis

A receiver operating characteristic (ROC) analysis was performed to examine how directional information was generated and enhanced during the delay period. ROC analysis has often been used to analyze the temporal pattern of the response selectivity of neuron activity (Britten et al. 1992; Shadlen and Newsome 1996; Thompson et al. 1996). By calculating ROC values using sliding time windows of a short period, we were able to examine the temporal change of the strength of the directional selectivity of the delay-period activity. A time window of 240 ms was established to calculate ROC values, with movement of the window in 20-ms steps from the start of the cue period to the end of the response period. For each time window, distributions of discharge rates were obtained for the maximum and minimum response directions of neurons by calculating trial-by-trial mean discharge rates of the activity. Based on the separation of these 2 distributions, we constructed the ROC curve and calculated the area under the ROC curve in each time window. In constructing the ROC curve, 100 criterion discharge rates were specified evenly between the maximum and minimum discharge rates among all the trials pooled across the maximum and minimum response directions. For discharge rate distributions of both maximum and minimum response directions of a neuron, we calculated the ratio of trials in which the mean discharge rates were smaller than each criterion discharge rate. The ROC curve was obtained by plotting the resultant ratios. The area under the ROC curve (ROC value) was calculated as the sum of the areas of 100 trapezoids.

Results

Database

We recorded 544 single-neuron activities from 2 monkeys (monkey R, n = 245; monkey Z, n = 299) from the cortex within and surrounding the principal sulcus while monkeys performed the ODR and S-ODR tasks. Among the 544 activities, 433 were recorded during the performance of both tasks. For these 433 activities, we statistically tested whether task-related activity was observed (Mann–Whitney U-test, P < 0.05) and whether task-related activity exhibited directional selectivity (ANOVA, P < 0.05). As a result, 176 neurons exhibited directional task-related activity during at least one task epoch in the ODR task (cue-period activity, n = 81; delay-period activity, n = 65; response-period activity, n = 109).

Contributions of Directional Delay-Period Activity to the Decision Process of the Saccade Direction in the S-ODR Task

Of 65 neurons with directional delay-period activity in the ODR task, 26 (40%) also showed directional delay-period activity in the S-ODR task. Figure 2 shows an example of directional delay-period activity in both tasks. In the ODR task (Fig. 2A), significant delay-period activity was observed when the visual cue was presented in the 90° direction (Mann–Whitney U-test; 90°, z = −3.59, P < 0.001). Delay-period activity was directionally selective (ANOVA, F = 21.55, degrees of freedom [df] = [3, 37], P < 0.001). The maximum and minimum response directions were 90° and 270°, respectively, and the best direction obtained from the tuning curve was 96.3°. In the S-ODR task (Fig. 2B), significant delay-period activity was observed in all saccade directions (Mann–Whitney U-test; 0°, z = −3.95, P < 0.001; 90°, z = −3.95, P < 0.001; 180°, z = −2.38, P < 0.05; 270°, z = −2.62, P < 0.01). Significant directional selectivity was also observed (ANOVA, F = 7.28, df = [3, 81], P < 0.001). The best direction was 95.3°, which was almost identical to that in the ODR task. Of 26 neurons with directional delay-period activity in both the ODR and S-ODR tasks, the difference between the best directions in the 2 tasks was less than 45° in 21 neurons (81%). For these 21 neurons, the mean difference between the best directions in the tasks was 24.4 ± 12.2° (mean ± standard deviation) and the median was 27.3°. These results indicate that neurons with directional delay-period activity in the ODR task also exhibit directional delay-period activity in the S-ODR task with similar directional preferences. This suggests that neurons with directional delay-period activity participate in the generation of information regarding the forthcoming saccade direction and that the decision process of the saccade direction operates during the delay period in the S-ODR task.

Figure 2.

(A) An example of directional delay-period activity in the ODR task. Data from correct trials are sorted on the basis of the direction of the visual cue and aligned at the start of the delay period. The central diagram is a polar plot showing the magnitude of delay-period activity in each cue direction. The mean discharge rate of the delay-period activity is depicted as the radial eccentricity in each direction. The histogram bin width is 20 ms. (B) Activity of the same neuron in the S-ODR task. Data from rewarded trials are sorted on the basis of the saccade direction and aligned at the start of the delay period. The central diagram is a polar plot showing the magnitude of delay-period activity in each saccade direction. The histogram bin width is 20 ms. C, cue period; D, delay period; R, response period.

Figure 2.

(A) An example of directional delay-period activity in the ODR task. Data from correct trials are sorted on the basis of the direction of the visual cue and aligned at the start of the delay period. The central diagram is a polar plot showing the magnitude of delay-period activity in each cue direction. The mean discharge rate of the delay-period activity is depicted as the radial eccentricity in each direction. The histogram bin width is 20 ms. (B) Activity of the same neuron in the S-ODR task. Data from rewarded trials are sorted on the basis of the saccade direction and aligned at the start of the delay period. The central diagram is a polar plot showing the magnitude of delay-period activity in each saccade direction. The histogram bin width is 20 ms. C, cue period; D, delay period; R, response period.

A comparison of the histograms constructed from the population of delay-period activities during the 2 tasks and the result of ROC analysis further support this notion. Figure 3A shows population histograms of delay-period activity in the maximum and minimum response directions in the ODR task (n = 65), and Figure 3B shows population histograms of delay-period activity of the same neurons in the S-ODR task in the maximum and minimum response directions obtained in the ODR task. The persistence of directional delay-period activity is evident in the S-ODR task (Fig. 3B), although the magnitude of activity is weaker in the S-ODR task than in the ODR task (Fig. 3A). Figure 3C shows the temporal change of the strength of the directional selectivity (ROC values) in the ODR and S-ODR tasks. In the ODR task, the ROC values began to increase during the cue period and persisted throughout the delay period. In contrast, in the S-ODR task, the increase of ROC values was not prominent until the beginning of the delay period. The ROC values then gradually increased toward the timing of the saccade initiation. Therefore, the gradual increase of ROC values observed during the S-ODR task may reflect the buildup process of the directional bias for the forthcoming saccade, and thus, neurons showing directional delay-period activity may be important participants in the decision process of the saccade direction in the S-ODR task.

Figure 3.

Population histograms and ROC values calculated using 65 neurons that exhibited directional delay-period activity in the ODR task. (A) Population histograms of neural activity in the maximum response direction (solid line) and the minimum response direction (dashed line) in the ODR task. The histogram bin width is 20 ms. (B) Population histograms of neural activity in the maximum response direction (solid line) and the minimum response direction (dashed line) in the S-ODR task. The maximum and minimum response directions were determined based on the activity in the ODR task. The histogram bin width is 20 ms. (C) Comparison of the temporal change of the population ROC values between the ODR (solid line) and S-ODR tasks (dashed line). All data are aligned at the start of the delay period.

Figure 3.

Population histograms and ROC values calculated using 65 neurons that exhibited directional delay-period activity in the ODR task. (A) Population histograms of neural activity in the maximum response direction (solid line) and the minimum response direction (dashed line) in the ODR task. The histogram bin width is 20 ms. (B) Population histograms of neural activity in the maximum response direction (solid line) and the minimum response direction (dashed line) in the S-ODR task. The maximum and minimum response directions were determined based on the activity in the ODR task. The histogram bin width is 20 ms. (C) Comparison of the temporal change of the population ROC values between the ODR (solid line) and S-ODR tasks (dashed line). All data are aligned at the start of the delay period.

Classification of Directional Delay-Period Activity in the ODR Task

As shown in Figure 4, the delay-period activity exhibited various temporal patterns in the maximum response direction in the ODR task. For example, some neurons showed increasing activity (Fig. 4A-1, “neuron r29301”), whereas others showed temporally convex (Fig. 4A-2, “neuron z11702”) or decreasing activity (Fig. 4A-3, “neuron z10101”). Moreover, in the ODR task, delay-period activity also showed variations in the temporal pattern of the strength of the directional selectivity. For example, in “neuron r29301” (Fig. 4B-1), delay-period activity showed significant directional selectivity throughout the 3-s delay period. However, in “neuron z11702” (Fig. 4B-2), delay-period activity exhibited significant directional selectivity only during the last 2 s of the delay period, whereas in “neuron z10101” (Fig. 4B-3) delay-period activity showed significant directional selectivity only during the first 2 s of the delay period. Thus, to identify which characteristics of delay-period activity contributed most significantly to the decision process of the saccade direction in the S-ODR task, we classified directional delay-period activity in the ODR task based on 2 characteristics: the temporal pattern of the strength of the directional selectivity and the temporal pattern of the delay-period activity itself in the maximum response direction.

Figure 4.

(A) Examples of delay-period activity in the ODR task. (A-1) Gradually increasing type of delay-period activity in the maximum response direction (0°). (A-2) Convex type of delay-period activity in the maximum response direction (0°). (A-3) Gradually decreasing type of delay-period activity in the maximum response direction (180°). The histogram bin width is 20 ms. (B) Polar plots of mean discharge rates during three 1-s delay epochs in the 3-s delay period (D1, first 1 s, solid black line; D2, second 1 s, solid gray line; and D3, third 1 s, dashed gray line). (B-1) Significant directional selectivity was observed in all time epochs (ANOVA, P < 0.05/3). (B-2) Significant directional selectivity was observed only in the D2 and D3 epochs. (B-3) Significant directional selectivity was observed only in the D1 and D2 epochs.

Figure 4.

(A) Examples of delay-period activity in the ODR task. (A-1) Gradually increasing type of delay-period activity in the maximum response direction (0°). (A-2) Convex type of delay-period activity in the maximum response direction (0°). (A-3) Gradually decreasing type of delay-period activity in the maximum response direction (180°). The histogram bin width is 20 ms. (B) Polar plots of mean discharge rates during three 1-s delay epochs in the 3-s delay period (D1, first 1 s, solid black line; D2, second 1 s, solid gray line; and D3, third 1 s, dashed gray line). (B-1) Significant directional selectivity was observed in all time epochs (ANOVA, P < 0.05/3). (B-2) Significant directional selectivity was observed only in the D2 and D3 epochs. (B-3) Significant directional selectivity was observed only in the D1 and D2 epochs.

Before classifying delay-period activity, we first selected neurons that showed directional delay-period activity for at least a fraction of the 3-s delay period in the ODR task. For this purpose, we divided the 3-s delay period into three 1-s delay epochs (D1, D2, and D3 epochs). Then, for each delay epoch, we separately examined whether delay-period activity exhibited directional selectivity by ANOVA (P < 0.05/3, corrected for multiple comparisons). Neurons showing significant directional selectivity during at least one delay epoch were considered to represent specific directional information in the ODR task and were included in our database for further analysis. Among 433 neurons recorded during both the ODR and S-ODR tasks, 124 neurons were selected for this analysis.

Temporal Pattern of the Strength of the Directional Selectivity

Delay-period activity was classified based on the temporal pattern of the strength of the directional selectivity in the ODR task. For each of 124 neurons, the number “0” or “1” was given to each 1-s delay epoch (D1, D2, and D3 epochs) to indicate the presence or absence of directional selectivity, respectively. For example, the neuron shown in Figure 4B-1 was assigned a (1, 1, 1) pattern because directional delay-period activity was observed in all time epochs. In contrast, the neuron in Figure 4B-2 was assigned a (0, 1, 1) pattern because directional delay-period activity was only observed in the D2 and D3 epochs. Thus, the 124 neurons were classified into 7 groups based on the temporal pattern of the strength of the directional selectivity during the delay period in the ODR task. The numbers of neurons classified into each group was as follows: (1, 1, 1), n = 16; (1, 0, 1), n = 5; (0, 1, 1), n = 11; (0, 0, 1), n = 34; (0, 1, 0), n = 21; (1, 0, 0), n = 29; (1, 1, 0), n = 8.

We then examined whether the neurons in each group exhibited directional selectivity during at least one delay epoch in the S-ODR task by ANOVA (P < 0.05/3). If this was the case, we further compared the mean best directions (MDODR and MDS-ODR) in the 2 tasks to determine whether activity exhibited a similar directional preference in the 2 tasks. Mean best direction was calculated as follows. For neurons with directional selectivity during at least one delay epoch, we determined the best direction separately for each delay epoch. The obtained best directions were averaged, and the resultant direction was defined as the mean best direction of a neuron in the ODR (MDODR) or S-ODR task (MDS-ODR). For example, if a neuron showed directional selectivity during the D2 and D3 epochs in the ODR task, we determined the best directions separately for the D2 and D3 epochs. Then, MDODR was calculated as the mean of the best directions in the D2 and D3 epochs. If this neuron showed directional selectivity during the D3 epoch in the S-ODR task, the best direction during the D3 epoch was regarded as MDS-ODR. In computing MDODR or MDS-ODR, delay epochs not showing directional selectivity were ignored because we wanted to extract only effective directional information represented by the delay-period activity. If the difference between MDODR and MDS-ODR was less than 45°, the neuron was considered to exhibit a similar directional selectivity between the 2 tasks.

As shown in Table 1, a high proportion (12/16, 75%) of neurons classified with a (1, 1, 1) directional pattern in the ODR task showed directional selectivity during at least one delay epoch in the S-ODR task. In all 12 neurons, the differences between MDODR and MDS-ODR were less than 45°. The mean difference between MDODR and MDS-ODR was 18.1°, and the median was 12.5°. On the other hand, among 29 neurons classified with a (1, 0, 0) directional pattern, only 5 (17%) showed directional selectivity during at least one delay epoch in the S-ODR task. In one neuron, the difference between MDODR and MDS-ODR was 44°, whereas in the remaining 4 neurons, the differences between MDODR and MDS-ODR were 48°, 93°, 137°, and 173°, respectively. Among 8 neurons classified with a (1, 1, 0) directional pattern, only one (13%) showed directional selectivity in the S-ODR task and the difference between MDODR and MDS-ODR was 153°. These results indicate that neurons classified with a (1, 1, 1) directional pattern also tended to show directional selectivity in the S-ODR task with a directional preference similar to that in the ODR task. Therefore, these results suggest that neurons classified with a (1, 1, 1) directional pattern are the strongest source of directional bias during the delay period in the S-ODR task. In contrast, neurons classified with (1, 0, 0) and (1, 1, 0) directional patterns may not produce effective directional information during the delay period in the S-ODR task.

Table 1

Number of neurons that exhibited directional selectivity during at least one delay epoch in the S-ODR task, among neurons showing 7 patterns of directional selectivity in the ODR task

ODR task S-ODR task Mean difference |MDODR−MDS-ODR| |MDODR−MDS-ODR| < 45° 
Patterns Number of neurons Directionally selective 
(1, 1, 1) 16 12 (75%) 18.1° 12 (75%) 
(1, 0, 1) 1 (20%) 11.5° 1 (20%) 
(0, 1, 1) 11 3 (27%) 30.4° 2 (18%) 
(0, 0, 1) 34 8 (24%) 48.4° 5 (15%) 
(0, 1, 0) 21 4 (19%) 107.3° 1 (5%) 
(1, 0, 0) 29 5 (17%) 99.0° 1 (3%) 
(1, 1, 0) 1 (12%) 152.9° 0 (0%) 
ODR task S-ODR task Mean difference |MDODR−MDS-ODR| |MDODR−MDS-ODR| < 45° 
Patterns Number of neurons Directionally selective 
(1, 1, 1) 16 12 (75%) 18.1° 12 (75%) 
(1, 0, 1) 1 (20%) 11.5° 1 (20%) 
(0, 1, 1) 11 3 (27%) 30.4° 2 (18%) 
(0, 0, 1) 34 8 (24%) 48.4° 5 (15%) 
(0, 1, 0) 21 4 (19%) 107.3° 1 (5%) 
(1, 0, 0) 29 5 (17%) 99.0° 1 (3%) 
(1, 1, 0) 1 (12%) 152.9° 0 (0%) 

Note: The first and second columns show 7 patterns of directional selectivity in the ODR task and the number of neurons that exhibited corresponding patterns of directional selectivity in the ODR task. The third and fourth columns show the number of neurons that exhibited directional selectivity during at least one delay epoch in the S-ODR task and the comparison of directional preferences between 2 tasks (|MDODR − MDS-ODR|). The fifth column shows the number of neurons that exhibited a similar directional preference between 2 tasks (|MDODR − MDS-ODR| < 45°).

Temporal Pattern of Delay-Period Activity in the Maximum Response Direction

Delay-period activity in the ODR task can also be classified based on the temporal pattern of the activity itself in the maximum response direction. For the preselected 124 neurons, we first calculated the mean discharge rate in the maximum response direction separately for the D1, D2, and D3 epochs. Then, we compared these mean discharge rates across delay epochs and classified the delay-period activity into 4 types based on the temporal pattern of activity (Fig. 5A). Types A (n = 38) and C (n =32) refer to activity showing increasing (D1 < D2 < D3) and decreasing (D1 > D2 > D3) patterns across delay epochs, respectively. Types B (n = 27) and D (n = 27) refer to activity showing temporally convex (D1 < D2 > D3) and concave patterns (D1 > D2 < D3) across delay epochs, respectively.

Figure 5.

Differences in the temporal pattern of delay-period activity. (A) Four types of delay-period activity were classified by comparing mean discharge rates among the D1, D2, and D3 epochs in the ODR task. Type A activity, D1 < D2 < D3; type B activity, D1 < D2 ≈ D3 < D2; type C activity, D1 > D2 > D3; and type D activity, D1 > D2 ≈ D3 > D2. (B) Population histograms of each type of delay-period activity (type A activity, n = 38; type B activity, n = 27; type C activity, n = 32; type D activity, n = 27) in the maximum (solid line) and minimum (dashed line) response directions in the ODR task. (C) Population histograms of each type of delay-period activity in the maximum (solid line) and minimum (dashed line) response directions in the S-ODR task. The maximum and minimum response directions were determined based on the activity in the ODR task. The histogram bin width is 20 ms.

Figure 5.

Differences in the temporal pattern of delay-period activity. (A) Four types of delay-period activity were classified by comparing mean discharge rates among the D1, D2, and D3 epochs in the ODR task. Type A activity, D1 < D2 < D3; type B activity, D1 < D2 ≈ D3 < D2; type C activity, D1 > D2 > D3; and type D activity, D1 > D2 ≈ D3 > D2. (B) Population histograms of each type of delay-period activity (type A activity, n = 38; type B activity, n = 27; type C activity, n = 32; type D activity, n = 27) in the maximum (solid line) and minimum (dashed line) response directions in the ODR task. (C) Population histograms of each type of delay-period activity in the maximum (solid line) and minimum (dashed line) response directions in the S-ODR task. The maximum and minimum response directions were determined based on the activity in the ODR task. The histogram bin width is 20 ms.

Population histograms of delay-period activity were constructed for each activity type for the maximum and minimum response directions in the ODR task (Fig. 5B). Similar population histograms were also constructed for each type of delay-period activity in the S-ODR task (Fig. 5C) for the maximum and minimum response directions obtained in the ODR task. A comparison of population histograms between the ODR and S-ODR tasks revealed that only neurons showing type A activity exhibited differential activation between the maximum and minimum response directions during the delay period in the S-ODR task. Prominent delay-period activity observed in the maximum response direction during the ODR task disappeared in the S-ODR task for the remaining types of activity. These results indicate that only neurons showing type A delay-period activity in the ODR task exhibited differential activation between the maximum and minimum response directions in the S-ODR task.

To further examine whether neurons showing type A delay-period activity in the ODR task maintained the same type of delay-period activity in the S-ODR task, we compared the temporal pattern of delay-period activity in the maximum response direction in the 2 tasks. As seen in Table 2, a majority of neurons exhibited the same temporal pattern of delay-period activity in both tasks. A comparison among neurons with the 4 types of delay-period activity showed that a large proportion of neurons with type A delay-period activity in the ODR task (21/38, 55%) also showed type A activity in the S-ODR task. These results further support the hypothesis that neurons with type A delay-period activity in the ODR task play an important role in differential activation between the maximum and minimum response directions in the S-ODR task.

Table 2

Comparison of temporal patterns of delay-period activity in the maximum response direction between the ODR and S-ODR tasks

Types in ODR task Types in S-ODR task 
  
38 (16) 21 (15) 4 (0) 5 (1) 8 (1) 
27 (4) 8 (4) 12 (2) 2 (0) 5 (0) 
32 (18) 3 (1) 9 (0) 11 (6) 9 (2) 
27 (7) 5 (1) 3 (0) 7 (0) 12 (4) 
Types in ODR task Types in S-ODR task 
  
38 (16) 21 (15) 4 (0) 5 (1) 8 (1) 
27 (4) 8 (4) 12 (2) 2 (0) 5 (0) 
32 (18) 3 (1) 9 (0) 11 (6) 9 (2) 
27 (7) 5 (1) 3 (0) 7 (0) 12 (4) 

Note: The number of neurons that exhibited each type of delay-period activity is shown outside the parentheses. The numbers in parentheses indicate the number of neurons that exhibited significant differences in mean discharge rates across the D1, D2, and D3 epochs (ANOVA, P < 0.05) at the maximum response direction in the ODR or S-ODR tasks. Maximum response directions were determined on the basis of the activity in the ODR task.

Figure 6 shows the comparison of the distribution of mean discharge rates for the maximum and minimum response directions between the ODR and S-ODR tasks for each of the 4 types of delay-period activity. Although the discharge rates differed from neuron to neuron, a significant difference in discharge rates was observed between the maximum and minimum response directions for all types of delay-period activity during all delay epochs in the ODR task (Wilcoxon signed-rank test, P ≪ 0.001 for all conditions). In the S-ODR task, the number of neurons exhibiting significant directional selectivity decreased for all types of delay-period activity. However, in neurons with type A delay-period activity (Fig. 6A), the distribution bias toward more neurons exhibiting higher discharge rates in the maximum response direction, compared with the minimum response direction, was statistically significant in all delay epochs of the S-ODR task (Wilcoxon signed-rank test, D1, P < 0.05; D2, P < 0.05; D3, P < 0.001; n = 38). In addition, distributions of discharge rates began to slant toward the maximum response direction with progress of the delay period in the S-ODR task, indicating a gradual increase of the magnitude of delay-period activity in the maximum response direction. In neurons with type B delay-period activity (Fig. 6B), the distribution bias toward more neurons exhibiting higher discharge rates in the maximum response direction, compared with the minimum response direction, was statistically significant in the D2 and D3 epochs of the S-ODR task (Wilcoxon signed-rank test, D2, P < 0.01; D3, P < 0.05, n = 27), but the magnitude of delay-period activities in the maximum response direction was attenuated in the S-ODR task, compared with those in the ODR task. In contrast, in neurons with delay-period activity of types C and D (Fig. 6C,D), the distribution bias was not significant in all delay epochs of the S-ODR task (Wilcoxon signed-rank test; type C activity, n = 32; D1, P = 0.07; D2, P = 0.15; D3, P = 0.28; type D activity, n = 27; D1, P = 0.57; D2, P = 0.97; D3, P = 0.336). Thus, these results indicate that neurons with type A delay-period activity in the ODR task conveyed directional information during the delay period in the S-ODR task.

Figure 6.

Scatter diagrams showing variability and bias in the distribution of mean discharge rates in the D1, D2, and D3 delay epochs in the ODR (open circle) and S-ODR tasks (filled circle). Mean discharge rates for the maximum response direction in the D1, D2, and D3 epochs are plotted against those for the minimum response direction. The maximum and minimum response directions were determined based on the activity in the ODR task. The horizontal and vertical axes represent mean discharge rates for the maximum and minimum response directions, respectively. (A) Neurons with type A delay-period activity in the ODR task (n = 38). (B) Neurons with type B delay-period activity in the ODR task (n = 27). (C) Neurons with type C delay-period activity in the ODR task (n = 32). (D) Neurons with type D delay-period activity in the ODR task (n = 27).

Figure 6.

Scatter diagrams showing variability and bias in the distribution of mean discharge rates in the D1, D2, and D3 delay epochs in the ODR (open circle) and S-ODR tasks (filled circle). Mean discharge rates for the maximum response direction in the D1, D2, and D3 epochs are plotted against those for the minimum response direction. The maximum and minimum response directions were determined based on the activity in the ODR task. The horizontal and vertical axes represent mean discharge rates for the maximum and minimum response directions, respectively. (A) Neurons with type A delay-period activity in the ODR task (n = 38). (B) Neurons with type B delay-period activity in the ODR task (n = 27). (C) Neurons with type C delay-period activity in the ODR task (n = 32). (D) Neurons with type D delay-period activity in the ODR task (n = 27).

Relationships between the Temporal Pattern of the Directional Selectivity and the Temporal Pattern of the Delay-Period Activity

To understand what types of delay-period activity mainly contribute to the decision process of the saccade direction in the S-ODR task, we examined the relationships between the temporal pattern of the strength of the directional selectivity and the temporal pattern of the delay-period activity itself in the maximum response direction.

Figure 7 shows an example of a neuron that exhibited type A delay-period activity coupled with a (1, 1, 1) directional pattern in the ODR task. In the ODR task, this neuron showed directional selectivity across all delay epochs (ANOVA, all P < 0.05/3; best directions in the D1, D2, and D3 epochs, 319°, 318°, and 317°, respectively; MDODR = 318°) with type A activity in the maximum response direction (0°). In the S-ODR task, this neuron also showed directional selectivity across all delay epochs (ANOVA, all P < 0.05/3; best directions in the D1, D2, and D3 epochs, 319°, 319°, and 318°; MDS-ODR = 318.5°) and also showed type A activity in the maximum response direction. Thus, this neuron showed persistent directional delay-period activity in both tasks with very similar directional preferences in the 2 tasks (|MDODR − MDS-ODR| = 0.5°).

Figure 7.

(A) An example of a neuron with type A activity coupled with a (1, 1, 1) directional pattern of delay-period activity in the ODR task. The polar plot shows the mean discharge rates during the D1 (solid black line), D2 (solid gray line), and D3 (dashed gray line) delay epochs. Mean discharge rates in the D1, D2, and D3 epochs for the maximum response direction (0°) were 14.2, 19.7, and 23.8 spikes/s. Significant directional selectivity was observed in all epochs (ANOVA, all P < 0.001). (B) The activity of the same neuron in the S-ODR task. Significant directional selectivity was observed in all epochs (ANOVA, D1, P < 0.01; D2, P < 0.005; and D3, P < 0.001). The histogram bin width is 20 ms.

Figure 7.

(A) An example of a neuron with type A activity coupled with a (1, 1, 1) directional pattern of delay-period activity in the ODR task. The polar plot shows the mean discharge rates during the D1 (solid black line), D2 (solid gray line), and D3 (dashed gray line) delay epochs. Mean discharge rates in the D1, D2, and D3 epochs for the maximum response direction (0°) were 14.2, 19.7, and 23.8 spikes/s. Significant directional selectivity was observed in all epochs (ANOVA, all P < 0.001). (B) The activity of the same neuron in the S-ODR task. Significant directional selectivity was observed in all epochs (ANOVA, D1, P < 0.01; D2, P < 0.005; and D3, P < 0.001). The histogram bin width is 20 ms.

Table 3 summarizes these analyses and shows several important observations. First, type A delay-period activity tended to be coupled with (1, 1, 1), (0, 1, 1), and (0, 0, 1) directional patterns in the ODR task. This type of delay-period activity tended to exhibit directional selectivity during at least one delay epoch in the S-ODR task with similar directional preferences in the 2 tasks. In contrast, type C delay-period activity tended to be coupled with (1, 1, 0) and (1, 0, 0) directional patterns in the ODR task, and this type of delay-period activity scarcely exhibited directional selectivity in the S-ODR task. Type B delay-period activity was coupled with most of the directional selectivity patterns in the ODR task. All 4 neurons with type B activity coupled with a (1, 1, 1) directional pattern in the ODR task showed directional selectivity during at least one delay epoch in the S-ODR task with similar directional preferences in the 2 tasks.

Table 3

Relationships between the temporal pattern of directional selectivity and the temporal pattern of delay-period activity in the ODR and S-ODR tasks

Activity pattern Directional selectivity pattern 
Sustaining (1, 1, 1) Increasing (0, 1, 1) and (0, 0, 1) Decreasing (1, 1, 0) and (1, 0, 0) Other (0, 1, 0) and (1, 0, 1) 
Type A 8 (9) 6 (21) 0 (4) 0 (4) 
Type B 4 (4) 0 (9) 0 (3) 1 (11) 
Type C 0 (2) 0 (3) 1 (20) 1 (7) 
Type D 0 (1) 1 (12) 0 (10) 0 (4) 
Activity pattern Directional selectivity pattern 
Sustaining (1, 1, 1) Increasing (0, 1, 1) and (0, 0, 1) Decreasing (1, 1, 0) and (1, 0, 0) Other (0, 1, 0) and (1, 0, 1) 
Type A 8 (9) 6 (21) 0 (4) 0 (4) 
Type B 4 (4) 0 (9) 0 (3) 1 (11) 
Type C 0 (2) 0 (3) 1 (20) 1 (7) 
Type D 0 (1) 1 (12) 0 (10) 0 (4) 

Note: Delay-period activity in the ODR task was classified into 16 subgroups on the basis of 4 temporal patterns of directional selectivity (sustaining, increasing, decreasing, and other) and 4 temporal patterns of activity (types A, B, C, and D). The numbers in parentheses indicate the number of neurons corresponding to each subgroup of delay-period activity in the ODR task. The numbers outside parentheses indicate the number of neurons exhibiting directional selectivity during at least one delay epoch (D1, D2, and/or D3 epoch) in the S-ODR task with a similar directional preference in the 2 tasks.

To elucidate how directional bias for the forthcoming saccade direction was generated and enhanced during the delay period of the S-ODR task, we constructed population histograms and calculated ROC values using activities of neurons that exhibited directional delay-period activity in both tasks with similar directional preferences (type A activity coupled with a [1, 1, 1] directional pattern, n = 8; type A activity coupled with [0, 1, 1] and [0, 0, 1] directional patterns, n = 6; type B activity coupled with a [1, 1, 1] directional pattern, n = 4). Figure 8A,B shows population histograms for each group of activities in the ODR and S-ODR tasks calculated from activities in the maximum and minimum response directions. Figure 8C shows the temporal changes of ROC values for each group in the 2 tasks. Directional biases during the delay period of the S-ODR task gradually increased toward the timing of execution of the saccade in all these groups. Neurons exhibiting type A delay-period activity coupled with a (1, 1, 1) directional pattern and type B delay-period activity coupled with a (1, 1, 1) directional pattern in the ODR task began to show directional bias during the cue period in the S-ODR task. Neurons exhibiting type A delay-period activity coupled with (0, 1, 1) and (0, 0, 1) directional patterns in the ODR task followed the trend of the other groups during the delay period in the S-ODR task. Thus, these results indicate that neurons exhibiting type A activity coupled with a (1, 1, 1) directional pattern and type B activity coupled with a (1, 1, 1) directional pattern in the ODR task play an essential role in generating and enhancing information regarding the forthcoming saccade direction in the S-ODR task.

Figure 8.

Population histograms (A, B) and ROC values (C) constructed from each group of delay-period activities. (A) Population histograms of delay-period activity in the maximum (solid line) and minimum (dashed line) response directions in the ODR task. The histogram bin width is 50 ms. (B) Population histograms of delay-period activity in the maximum (solid line) and minimum (dashed line) response directions in the S-ODR task. The maximum and minimum response directions were determined on the basis of the activity in the S-ODR task, rather than in the ODR task. The histogram bin width is 50 ms. (C) Comparison of the temporal change in population ROC values between the ODR (solid line) and S-ODR tasks (dashed line). In the S-ODR task, the maximum and minimum response directions were determined based on the activity in the S-ODR task. All data are aligned at the start of the delay period.

Figure 8.

Population histograms (A, B) and ROC values (C) constructed from each group of delay-period activities. (A) Population histograms of delay-period activity in the maximum (solid line) and minimum (dashed line) response directions in the ODR task. The histogram bin width is 50 ms. (B) Population histograms of delay-period activity in the maximum (solid line) and minimum (dashed line) response directions in the S-ODR task. The maximum and minimum response directions were determined on the basis of the activity in the S-ODR task, rather than in the ODR task. The histogram bin width is 50 ms. (C) Comparison of the temporal change in population ROC values between the ODR (solid line) and S-ODR tasks (dashed line). In the S-ODR task, the maximum and minimum response directions were determined based on the activity in the S-ODR task. All data are aligned at the start of the delay period.

Discussion

In the present study, we aimed to identify the types of directional delay-period activity that contribute most significantly to the decision of the saccade direction in the S-ODR task. Directional delay-period activity observed in the ODR task was classified on the basis of 2 criteria: the temporal pattern of the strength of the directional selectivity and the temporal pattern of the delay-period activity itself in the maximum response direction. Comparison of the characteristics of delay-period activity in the same neuron between the ODR and S-ODR tasks revealed that increasing or temporally convex types of delay-period activity with persistent directional selectivity throughout the delay period in the ODR task play an essential role in generating and enhancing directional information regarding the forthcoming saccade in the S-ODR task. These results indicate the importance of the prefrontal cortex in the decision process of the saccade direction in the S-ODR task. The results also indicate the importance of neurons with directional delay-period activity as essential neural substrates in the decision-making process.

Types of Directional Delay-Period Activity Contributing to the Decision Process of the Saccade Direction in the S-ODR Task

In our previous report (Watanabe et al. 2006), we suggested that neurons with directional delay-period activity in the ODR task showed differential activation depending on the direction of the saccade that the monkeys freely chose in the S-ODR task. This result suggests that neurons with directional delay-period activity in the ODR task participate in the decision process of the saccade direction in the S-ODR task. However, the temporal pattern and the directional tuning of the delay-period activity are different from neuron to neuron in the DLPFC (Funahashi et al. 1989). Therefore, in the present study, we aimed to identify which types of delay-period activity play the most significant role in the decision process of the saccade direction in the S-ODR task.

Prefrontal neurons showed a variety of temporal patterns of delay-period activity. For example, the delay-period activity of some neurons showed a gradual increase toward the initiation of the saccade, whereas that of other neurons showed a gradual decrease in the best direction. It has been suggested that the difference in temporal pattern observed in delay-period activity reflects a difference in functional roles, such that gradually decreasing types of delay-period activity tend to represent information regarding the sensory stimulus presented during the cue period (cue-coding activity), whereas gradually increasing types of activity tend to represent information regarding the forthcoming motor response (response-coding activity) (Quintana and Fuster 1992, 1999; Constantinidis et al. 2001a; Takeda and Funahashi 2002, 2004; Fukushima et al. 2004). Moreover, it has been shown that prefrontal delay-period activity exhibits variation in directional selectivity during ODR performance (Funahashi et al. 1989; Sawaguchi and Yamane 1999). For example, some neurons showed broadly tuned directional selectivity, whereas others showed sharply tuned directional selectivity (Funahashi et al. 1989). Therefore, in the present study, to identify which types of delay-period activity contributed most significantly to the decision process of the saccade direction in the S-ODR task, we classified the delay-period activity observed in the ODR task based on the temporal pattern of the strength of the directional selectivity and the temporal pattern of the delay-period activity itself in the maximum response direction. We then examined which types of delay-period activity in the ODR task also showed directional selectivity in the S-ODR task.

A high proportion of neurons with increasing or sustained delay-period activity with temporally persistent directional selectivity in the ODR task (types A and B activities coupled with a [1, 1, 1] directional pattern) also held effective directional information in the S-ODR task. On the contrary, neurons exhibiting decreasing delay-period activity with directional selectivity only present during the early phase of the delay period in the ODR task (type C activity coupled with [1, 1, 0] and [1, 0, 0] directional patterns) did not exhibit directional selectivity in the S-ODR task. In the ODR task, both neurons exhibiting cue-coding activity and neurons exhibiting response-coding activity could be activated, because the visual cue indicated the correct saccade direction, and because the saccade planning took place before the end of the delay period. However, in the S-ODR task, only response-coding activity persists, because saccade direction was not externally indicated by a visual stimulus during the cue period, and because information regarding the forthcoming saccade direction needed to be generated during the delay period.

The decreasing types of delay-period activity that exhibited directional selectivity only during the early phase of the delay period in the ODR task might be classified as cue-coding activity, and these types of delay-period activity did not exhibit directional selectivity in the S-ODR task. In the S-ODR task, the disappearance of directional selectivity among neurons with cue-coding activity may be caused by the task, in which the 4 visual stimuli did not indicate a specific saccade direction. It is also possible that simultaneous presentation of visual stimuli within and outside the receptive field of a neuron induces inhibitory interactions among groups of neurons favoring different preferred directions (Rao et al. 1999; Inoue and Funahashi 2002). Although we cannot specify the underlying neuronal mechanisms, our results indicate that this activity did not convey directional information during the delay period in the S-ODR task and suggest that this activity is not directly related to the decision of the saccade direction in the S-ODR task.

In contrast, the increasing types of delay-period activity that exhibited directional selectivity that persisted throughout or during the late phase of the delay period in the ODR task might be classified as response-coding activity. These types of delay-period activity persistently showed directional selectivity in both tasks with very similar directional preferences. Therefore, it could be concluded that this activity represents information regarding the forthcoming saccade direction in both tasks and thus plays an important role in the decision of the saccade direction.

Whether the increasing directional delay-period activity observed in the S-ODR task directly represents the decision process of the saccade direction or merely reflects the outcome of the decision is not certain based on the present results. In the S-ODR task, saccade planning and execution processes need to be activated, and in our previous study (Watanabe et al. 2006), we showed that presaccadic activity exhibited directional selectivity in both tasks with very similar directional preferences. Directional presaccadic activity could contribute to the saccade execution process. Neurons with type A delay-period activity coupled with (1, 1, 1), (0, 1, 1), and (0, 0, 1) directional patterns in the ODR task also showed increasing directional delay-period activity when the monkeys made a saccade toward the preferred direction of the neurons in the S-ODR task. Thus, type A delay-period activity may represent enhancement of directional information about the forthcoming saccade in the S-ODR task. As seen in Figures 3B and 5C, neuronal activity was enhanced after the visual cue presentation, even in trials in which the monkey made a saccade toward the minimum response direction of the neurons. This trend is evident when comparing the activity between the maximum and minimum response directions in the ODR task (see Figs 3A and 5B). However, this initial activation was not followed by a prominent increase of activity when the monkey made a saccade toward the minimum response direction of the neurons. These observations suggest that neuronal signals favoring different saccade directions compete with each other during the early phase of the delay period in the S-ODR task and that only winning directional information is further enhanced in the course of the delay period. Thus, it is suggested that the activity during the early delay period corresponds to the ongoing decision process of the saccade direction and that the activity during the latter phase of the delay period corresponds to the planning or preparation of a particular saccade.

Other Sources of Directional Information during Performance of the S-ODR Task

Other populations of neurons may also contribute to the generation and enhancement of directional information during the delay period of the S-ODR task. One example would be a group of neurons that exhibited directional delay-period activity only in the S-ODR task. Among 433 neurons recorded during the ODR and S-ODR tasks, 19 exhibited directional selectivity in the D1, D2, and/or D3 epochs only in the S-ODR task. All these neurons showed directional selectivity in one delay epoch (D1, n = 6; D2, n = 5; D3, n = 8) of the S-ODR task. It has been shown that prefrontal neurons often exhibit task-specific activity (Hoshi et al. 1998; White and Wise 1999; Asaad et al. 2000; Wallis et al. 2001). For example, Hoshi et al. (1998) showed that a subset of movement-related neurons in the DLPFC exhibited task-specific activation depending on whether the location or the shape of the visual cue guided the direction of the reaching behavior of the monkey. White and Wise (1999) also showed that the magnitude of task-related activity was modulated by the rules of the task. Takeda and Funahashi (2002) observed that a number of DLPFC neurons exhibited task-related activity only when monkeys performed either the ODR task or the R-ODR tasks. Directionally selective delay-period activity observed only in the S-ODR task in 19 neurons is task-specific activation. Because these 19 neurons exhibited directionally selective delay-period activity only in the S-ODR task, the activity of these 19 neurons may participate in the generation and enhancement of directional information regarding the forthcoming saccade in the S-ODR task.

Neural Mechanisms Underlying the Decision Process of the Saccade Direction in the S-ODR Task

Figure 9 shows temporal changes of ROC values in the S-ODR task for 4 groups of delay-period activity that showed directional selectivity during at least one delay epoch in the S-ODR task (type A activity with a [1, 1, 1] directional pattern, type B activity with a [1, 1, 1] directional pattern, type A activity with a [0, 1, 1] directional pattern, and type A activity with a [0, 0, 1] directional pattern). For neurons with type A activity with a (1, 1, 1) directional pattern and neurons with type B activity with a (1, 1, 1) directional pattern, ROC values began to increase during the cue period in the S-ODR task. For neurons with type A activity with a (0, 1, 1) directional pattern and neurons with type A activity with a (0, 0, 1) directional pattern, ROC values began to increase about 1 s after the start of the delay period. The difference in the timing of the increase in ROC values suggests that, in the S-ODR task, both neurons with type A activity with a (1, 1, 1) directional pattern and neurons with type B activity with a (1, 1, 1) directional pattern start to lead the directional trend in the neuronal population in the early phase of the delay period and have a directional influence on neurons with type A activity with (0, 1, 1) and (0, 0, 1) directional patterns. As a result, delay-period activity of a large number of neurons would exhibit substantial directional bias toward a specific saccade direction at the end of the delay period. Previous studies (Funahashi and Inoue 2000; Constantinidis et al. 2001b) have shown that functional interactions of prefrontal neuron pairs with similar directional preferences were observed in various task periods in the ODR task, including the delay period. Therefore, in the S-ODR task, neurons with delay-period activity with a similar directional preference would interact with each other and generate and enhance directional bias in a population level to prepare for the forthcoming saccade.

Figure 9.

Comparison of ROC values among 4 groups of neurons in the S-ODR task. The maximum and minimum response directions were determined on the basis of the activity in the S-ODR task. The width of the time window is 300 ms, and the window was moved every 100 ms from the cue presentation until the end of the response period. Blue solid line, type A activity with a (1, 1, 1) directional pattern (n = 8); blue dashed line, type B activity with a (1, 1, 1) directional pattern (n = 4); red solid line, type A activity with a (0, 1, 1) directional pattern (n = 2); red dashed line, type A activity with a (0, 0, 1) directional pattern (n = 4).

Figure 9.

Comparison of ROC values among 4 groups of neurons in the S-ODR task. The maximum and minimum response directions were determined on the basis of the activity in the S-ODR task. The width of the time window is 300 ms, and the window was moved every 100 ms from the cue presentation until the end of the response period. Blue solid line, type A activity with a (1, 1, 1) directional pattern (n = 8); blue dashed line, type B activity with a (1, 1, 1) directional pattern (n = 4); red solid line, type A activity with a (0, 1, 1) directional pattern (n = 2); red dashed line, type A activity with a (0, 0, 1) directional pattern (n = 4).

Although neurons with particular types of delay-period activity (response-coding activity) play an important role in deciding the direction of the forthcoming saccade, it is not clear how one particular direction of the saccade is selected among other possible directions, how functional interactions among neurons exhibiting similar or different directional preferences bring about one particular direction of the saccade, and when the direction of the saccade is finally decided during the delay period. Answers to these important questions in future studies will increase the understanding of how the prefrontal cortex participates in the decision of the saccade direction in the S-ODR task.

Funding

Japanese Ministry of Education, Science, Sports, Culture, and Technology (No. 18020016 and 17300103 to SF), Japan Society for the Promotion of Science (No. 18050592 to KW), and by the 21st Century Center of Excellence program (D-10 to Kyoto University), Ministry of Education, Science, Sports, Culture, and Technology, Japan.

The authors thank Prof. H. Komatsu for making arrangements to take MRI pictures of monkey brains at the National Institute of Physiological Sciences, Okazaki, Japan. Conflict of Interest: None declared.

References

Asaad
WF
Rainer
G
Miller
EK
Task-specific neural activity in the primate prefrontal cortex
J Neurophysiol
 , 
2000
, vol. 
84
 (pg. 
451
-
459
)
Barraclough
DJ
Conroy
ML
Lee
D
Prefrontal cortex and decision making in a mixed-strategy game
Nat Neurosci
 , 
2004
, vol. 
7
 (pg. 
404
-
410
)
Britten
KH
Shadlen
MN
Newsome
WT
Movshon
JA
The analysis of visual motion: a comparison of neuronal and psychophysical performance
J Neurosci
 , 
1992
, vol. 
12
 (pg. 
4745
-
4765
)
Constantinidis
C
Franowicz
MN
Goldman-Rakic
PS
The sensory nature of mnemonic representation in the primate prefrontal cortex
Nat Neurosci
 , 
2001
, vol. 
4
 (pg. 
311
-
316
)
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
)
Fukushima
T
Hasegawa
I
Miyashita
Y
Prefrontal neuronal activity encodes spatial target representations sequentially updated after nonspatial target-shift cues
J Neurophysiol
 , 
2004
, vol. 
91
 (pg. 
1367
-
1380
)
Funahashi
S
Bruce
CJ
Goldman-Rakic
PS
Mnemonic coding of visual space in the monkey's dorsolateral prefrontal cortex
J Neurophysiol
 , 
1989
, vol. 
61
 (pg. 
331
-
349
)
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
Inoue
M
Neuronal interactions related to working memory processes in the primate prefrontal cortex revealed by cross-correlation analysis
Cereb Cortex
 , 
2000
, vol. 
10
 (pg. 
535
-
551
)
Hasegawa
R
Sawaguchi
T
Kubota
K
Monkey prefrontal neuronal activity coding the forthcoming saccade in an oculomotor delayed matching-to-sample task
J Neurophysiol
 , 
1998
, vol. 
79
 (pg. 
322
-
333
)
Hoshi
E
Shima
K
Tanji
J
Task-dependent selectivity of movement-related neuronal activity in the primate prefrontal cortex
J Neurophysiol
 , 
1998
, vol. 
80
 (pg. 
3392
-
3397
)
Inoue
M
Funahashi
S
Prefrontal delay-period activity is affected by visual cues presented outside the memory field
Neuroreport
 , 
2002
, vol. 
13
 (pg. 
2097
-
2101
)
Kim
JN
Shadlen
MN
Neural correlates of a decision in the dorsolateral prefrontal cortex of the macaque
Nat Neurosci
 , 
1999
, vol. 
2
 (pg. 
176
-
185
)
Kojima
S
Goldman-Rakic
PS
Delay-period activity of prefrontal neurons in rhesus monkeys performing delayed response
Brain Res
 , 
1982
, vol. 
248
 (pg. 
43
-
49
)
Lau
HC
Rogers
RD
Ramnani
N
Passingham
RE
Willed action and attention to the selection of action
Neuroimage
 , 
2004
, vol. 
21
 (pg. 
1407
-
1415
)
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
)
Quintana
J
Yajeya
J
Fuster
JM
Prefrontal representation of stimulus attributes during delay tasks. I. Unit activity in cross-temporal integration of sensory and sensory-motor information
Brain Res
 , 
1988
, vol. 
474
 (pg. 
211
-
221
)
Quintana
J
Fuster
JM
Mnemonic and predictive functions of cortical neurons in a memory task
Neuroreport
 , 
1992
, vol. 
3
 (pg. 
721
-
724
)
Quintana
J
Fuster
JM
From perception to action: temporal integrative functions of prefrontal and parietal neurons
Cereb Cortex
 , 
1999
, vol. 
9
 (pg. 
213
-
221
)
Rao
SG
Williams
GV
Goldman-Rakic
PS
Isodirectional tuning of adjacent interneurons and pyramidal cells during working memory: evidence for microcolumnar organization in PFC
J Neurophysiol
 , 
1999
, vol. 
81
 (pg. 
1903
-
1916
)
Sawaguchi
T
Yamane
Properties of delay-period neuronal activity in the monkey dorsolateral prefrontal cortex during a spatial delayed matching-to-sample task
J Neurophysiol
 , 
1999
, vol. 
82
 (pg. 
2070
-
2080
)
Shadlen
MN
Newsome
WT
Motion perception: seeing and deciding
Proc Natl Acad Sci USA
 , 
1996
, vol. 
93
 (pg. 
628
-
633
)
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
)
Thompson
KG
Hanes
DP
Bichot
NP
Schall
JD
Perceptual and motor processing stages identified in the activity of macaque frontal eye field neurons during visual search
J Neurophysiol
 , 
1996
, vol. 
76
 (pg. 
4040
-
4050
)
Wallis
JD
Anderson
KC
Miller
EK
Single neurons in prefrontal cortex encode abstract rules
Nature
 , 
2001
, vol. 
411
 (pg. 
953
-
956
)
Watanabe
K
Funahashi
S
Prefrontal delay-period activity contributes to the decision of saccade directions in the free-choice task. Program No. 412.4. 2005 Abstract Viewer/Itinerary Planner [Internet]
 , 
2005
Washington (DC)
Society for Neuroscience
 
Online
Watanabe
K
Igaki
S
Funahashi
S
Contributions of prefrontal cue-, delay-, and response-period activity to the decision process of saccade direction in a free-choice ODR task
Neural Netw
 , 
2006
, vol. 
19
 (pg. 
1203
-
1222
)
White
IM
Wise
SP
Rule-dependent neuronal activity in the prefrontal cortex
Exp Brain Res
 , 
1999
, vol. 
126
 (pg. 
315
-
335
)