Introduction

The prefrontal cortex is known to participate in a variety of cognitive functions including thinking, reasoning, and decision-making, and is an important brain area for understanding the neural mechanisms of these functions (Goldman-Rakic 1987; Fuster 1997; Miller and Cohen 2001). The prefrontal cortex consists of three main sectors; the lateral sector, the medial sector, and the orbital sector. The lateral sector is the multimodal association area which participates in higher cognitive functions (e.g., executive functions), whereas the medial and orbital sectors are considered to be the limbic (or paralimbic) area which participates in emotional and motivational functions (Mesulam 2000). The lateral part of the prefrontal cortex, and especially the dorsolateral prefrontal cortex (DLPFC), has been extensively examined in relation to a variety of cognitive functions. Among these, the neural mechanisms of spatial working memory have been extensively examined neurophysiologically using an oculomotor version of the delayed-response task (ODR task) (Funahashi et al. 1989, 1990, 1991; Sawaguchi 1998, 2001; Chafee and Goldman-Rakic 2000; Constantinidis et al. 2001; Takeda and Funahashi 2002, 2004; Compte et al. 2003; Tsujimoto and Sawaguchi 2004; Roesch and Olson 2005a). It has been shown that many DLPFC neurons exhibit directionally selective delay-period activity (Funahashi et al. 1989), which has been considered to be a neural correlate of the mechanism for the temporary maintenance of spatial information (Funahashi and Kubota 1994; Funahashi and Takeda 2002). Two groups of delay-period activity have been identified; one represents retrospective information (e.g., the location of the visual cue) and the other represents prospective information (e.g., the direction of a forthcoming movement) (Niki and Watanabe 1976; Funahashi et al. 1993; Takeda and Funahashi 2002). Takeda and Funahashi (2004) showed that these two groups of delay-period activity play a significant role in information processing such as visuo-motor information transformation. Thus, neural correlates of spatial working memory have been identified in the DLPFC and their functional characteristics have been examined neurophysiologically using the ODR task.

The orbitofrontal cortex (OFC) has strong reciprocal connections to the DLPFC (Barbas and Pandya 1989; Carmichael and Price 1996; Cavada et al. 2000), which suggests that the OFC also significantly contributes to cognitive functions such as working memory. OFC functions have been examined mainly with regard to emotional and motivational aspects as well as decision-making (Critchley and Rolls 1996; Bechara et al. 1998; Rolls 1999; Tremblay and Schultz 1999, 2000; Hikosaka and Watanabe 2000, 2004; Wallis and Miller 2003; Roesch and Olson 2004, 2005b; Padoa-Schioppa and Assad 2006). Tremblay and Schultz (1999) examined how OFC neurons processed reward information during spatial delayed response performance and showed that the differential activity observed in most OFC neurons represented the monkey’s relative preference among the available rewards rather than any physical properties of the rewards. They concluded that OFC neurons process the motivational value of rewarding outcomes of voluntary action. Roesch and Olson (2005b) examined the effect of delay length (shorter delay or longer delay) on reward-related activity in OFC neurons while monkeys performed a memory-guided saccade task. They showed that OFC neurons tended to fire more strongly in response to a cue that predicted a shorter delay, and concluded that OFC activity represents the time-discounted value of the expected reward. Further, Padoa-Schioppa and Assad (2006) examined how and where the economic values of reward are represented in the brain and showed that OFC neurons encode the value of offered and chosen goods independently of visuospatial factors and motor responses. Thus, these results indicate that OFC neurons encode the motivational values of the expected reward and that these values could be used for an assignment that underlies economic choice.

Although the motivational aspects of OFC functions have been examined extensively, it is not well understood how the OFC participates in cognitive functions such as working memory. To fully understand the functions of the OFC as part of the prefrontal cortex, it is necessary to understand the cognitive aspects of OFC functions in addition to their motivational aspects and to examine whether or not the OFC and DLPFC contribute equally to cognitive functions. A comparison of the characteristics of task-related activities in these two areas recorded under the same behavioral tasks could provide important clues to address these issues. Therefore, in the present experiment, we selected the spatial working memory process as a cognitive process and used the ODR task as a behavioral task, since the characteristics of task-related activities observed during ODR performance and their functional significance have been extensively examined in the DLPFC (Funahashi et al. 1989, 1990, 1991; Takeda and Funahashi 2002, 2004). In the present study, we analyzed the characteristics of task-related activities recorded from the DLPFC and the OFC while a monkey performed the ODR task and compared the characteristics of task-related activities between these two areas to understand how the OFC contributes to spatial working memory processes.

Methods

Subject and apparatus

One female rhesus monkey (monkey U, 6.0 kg) was used as a subject. This experiment was conducted according to the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health, USA. This experiment was approved by the Animal Research Committee at the Graduate School of Human and Environmental Studies, Kyoto University. The monkey was housed individually in a home cage. Food was always available in the home cage. Although water intake was restricted in the home cage, the daily requirement of water was given as a reward in the laboratory. To ensure that the monkey remained healthy, body weight and water intake were measured daily.

The monkey sat in a primate chair in a dark, sound-attenuated room during training and recording sessions. The monkey’s head movements were restricted by a head-holder attached to the skull. The monkey faced a 21-in. TV monitor (F520; Nanao, Japan), on which a fixation point and visual cues were presented. The monitor was placed 40 cm from the monkey’s face. The monkey’s eye positions were monitored by the magnetic search coil technique (Robinson 1963). A program called TEMPO (Reflective Computing, USA) was used to control behavioral tasks, sample neural and behavioral data, and to display these data on the monitor on-line.

Behavioral tasks

We used an oculomotor delayed-response (ODR) task, since this task has been used in our previous studies in the DLPFC (Funahashi et al. 1989, 1990, 1991; Takeda and Funahashi 2002, 2004) and since the aim of the present experiment was to analyze the characteristics of task-related activity in OFC neurons and to compare these characteristics with those of DLPFC neurons. Figure 1a shows a schematic representation of the ODR task. In this task, the monkey was required to make a memory-guided saccade after a 3-s delay to the location where a visual cue had been presented. After a 5-s intertrial interval (ITI), a fixation point (FP; a white circle, visual angle of 0.5°) was presented at the center of the TV monitor. If the monkey maintained fixation at the FP for 1 s (fixation period), a visual cue (white circle, visual angle of 1°) was presented for 0.5 s (cue period) randomly at one of four predetermined locations around the FP (Fig. 1b). The monkey was required to maintain fixation at 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 location where the visual cue had been presented. If the monkey performed a correct saccade, a drop (about 0.2 ml) of liquid reward was given. To determine whether the monkey made a correct saccade or not, a square-shaped window (4°–6° in visual angle) was set around the target location and the TEMPO program examined whether or not the end of the saccade fell within this window (correct saccade). If the monkey broke fixation during the cue period or the delay period, if the monkey failed to perform a saccade within the 0.4-s response period, or if the end of the saccade did not fall within the correct window, the trial was aborted immediately without a reward and the next trial began.

Fig. 1
figure 1

Diagram of an oculomotor delayed-response (ODR) task. a Temporal sequences of task events in the ODR task. b Spatial arrangement of the visual cue. The eccentricity was 17°

Full size image

Surgical procedures

Before the start of the experiment, MRI photographs of the monkey’s brain were taken at the National Institute for Physiological Sciences, Okazaki, Japan, to determine the recording area. At the beginning of the experiment, we performed surgery to place an eye coil and a stainless steel head-holder under aseptic conditions. The monkey was anesthetized by an intravenous injection of pentobarbital sodium (25 mg/kg). Using the technique described by Judge et al. (1980), the eye coil was placed under the conjunctiva of the left eye to monitor eye movements. In addition, a head-holder made of stainless steel was placed on the skull to restrict the monkey’s head movements during the experiment. To secure the head-holder, several stainless steel bolts were implanted in the skull. The connector for the eye coil, the head-holder, and the stainless steel bolts were fixed to the skull with dental acrylic.

After the monkey’s training was completed, we performed a second surgery to place a stainless steel cylinder on the skull to record neural activity from the DLPFC and the OFC. The monkey was anesthetized by an intravenous injection of pentobarbital sodium (25 mg/kg). We placed the monkey’s head in the stereotaxic apparatus. We made a small hole (20 mm in diameter) in the skull over the center position of the recording cylinder using a trephine and attached a stainless steel cylinder. The center position of the recording cylinder was determined based on the MRI photographs of the monkey’s brain. The stereotaxic coordinates of the center of the recording cylinder were 35.0 mm anterior from the interaural plane and 12.0 mm lateral from the midline for the right hemisphere, and 33.0 mm anterior from the interaural plane and 12.0 mm lateral from the midline for the left hemisphere. Several stainless steel bolts were also implanted in the skull and fixed to the cylinder by dental acrylic. After each surgery, the monkey was treated with antibiotics (Cephalosporin, Astellas Pharma, Japan) for several days to prevent infection. The monkey was given a full amount of food and water until it fully recovered from surgery.

Recording single-neuron activity

After the monkey fully recovered from the second surgery, we started recording single-neuron activity from the DLPFC (mainly area 46) and the OFC (mainly areas 11, 12, 13, and 45). Epoxy-coated tungsten microelectrodes (0.2–2.0 MΩ at 1 kHz, FHC Instruments, USA) were used to record single-neuron activity. The electrode was advanced vertically using an oil-driven micromanipulator (MO-95, Narishige Instrument, Japan). We estimated whether the tip of the electrode was in the DLPFC, the OFC or the white matter by observing spontaneous discharge patterns. We determined the depth of the orbital surface at 80 penetration sites by measuring the depth at which the tip of the electrode touched the cranial bone. Raw neural activity was amplified using an amplifier (DAM80, World Precision Instruments, USA) and monitored by an oscilloscope (SS-7802, Iwatsu Electronics, Japan). At the same time, single-neuron activity was isolated by a window discriminator (DIS-1, BAK Electronics, USA), monitored with raw activity by an oscilloscope, and input to a computer. Neural data were displayed on-line on the monitor and stored together with event signals and behavioral data on magnetic media using TEMPO.

Data analysis

To examine whether or not the recorded single-neuron activity was task-related, we made rastergrams and histograms of recorded activity aligned at several task events (e.g., onset of the cue period, onset of the delay period, onset of reward delivery) for each cue condition. We first visually inspected whether or not recorded single-neuron activity was related to one or more task events. We identified the following task-related activities by visual inspection; fixation-related activity, cue-period activity, delay-period activity, response-period activity, and reward-period activity.

We then conducted statistical analyses to confirm our visual inspection. We first obtained the neuron’s baseline discharge rate by calculating 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 100 to 400 ms after the onset of the visual cue) 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 exhibited cue-period activity. For the 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 during this period differed significantly from the baseline discharge rate by the Mann–Whitney U test (P < 0.05), we considered that the neuron exhibited delay-period activity. For activity during the response period, we first determined the cue condition for which the maximum excitatory or inhibitory activity was observed in the histogram displays aligned at the initiation of the saccade, and determined when the maximum activity was observed. We then calculated the mean discharge rate during the 300-ms of the response period (150 ms before and 150 ms after the time when the maximum activity was observed) 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 exhibited response-period activity. In previous studies (Funahashi et al. 1991; Takeda and Funahashi 2002), response-period activity in DLPFC neurons was further classified into pre- or post-saccadic activity. Therefore, we also classified response-period activity into pre- or post-saccadic activity based on whether or not the initiation of response-period activity preceded the initiation of saccadic eye movements. The reward period was defined as the 800-ms period after reward delivery. If the mean discharge rate during the reward period was significantly different from the baseline discharge rate by the Mann–Whitney U test (P < 0.05), we considered that the neuron had reward-period activity. Most of the neurons that exhibited reward-period activity were also tested with regard to whether or not they responded to unexpected delivery of the reward (free-reward). Free-reward (0.2 ml liquid reward) was delivered at a random timing during the delay period at randomly selected trials of the ODR task.

Directional selectivity of task-related activity

Most prefrontal task-related activities have been shown to exhibit directional selectivity (Funahashi et al. 1989, 1990, 1991; Takeda and Funahashi 2002). To determine whether or not task-related activity exhibited directional selectivity, we first examined the differences in the mean discharge rates across all cue conditions for each task-related activity using one-way ANOVA. If the difference was significant (P < 0.05), we considered that this task-related activity showed directional selectivity. If significant task-related activity was observed in all cue conditions and if mean discharge rates were not significantly different across cue conditions by one-way ANOVA, we considered that this task-related activity was omni-directional.

The directional selectivity of task-related activity has been quantitatively characterized by constructing a tuning curve of this activity (e.g., Funahashi et al. 1989, 1990, 1991). To compare the directional characteristics of task-related activities in DLPFC and OFC neurons quantitatively, we constructed tuning curves of task-related activities using a Gaussian function (Bruce and Goldberg 1985; Funahashi et al. 1989, 1990, 1991). The tuning curve was created from the mean discharge rate of task-related activity under each cue condition by its best fit to the Gaussian function

$$ f(d) = B + R \times {\rm exp} \;{\left( { - 0.5\;[(d - D)/{\rm Td}]^{2} } \right)} $$

where f (d) is the discharge rate as a function of the visual cue location d. The constants can be interpreted as follows: B is the baseline discharge rate, R is the discharge rate above the baseline discharge rate at the best direction, D is the best direction where the maximum task-related activity was observed, and Td is an index of the tuning width. We obtained D and Td for each directional task-related activity by constructing tuning curves. In some cases, discharge rates were not well fit to the obtained tuning curve. Therefore, for further analysis, we only selected tuning curves in which r 2 values were greater than 0.95.

Histological examinations of the recorded areas

At the end of the experiment, we made electrolytic lesions at several selected locations within each hemisphere by passing positive current through elgiloy microelectrodes to identify recording sites during histological examinations. The monkey was deeply anesthetized by an intravenous injection of an overdose of pentobarbital sodium (45–50 mg/kg). The brain was perfused first by phosphate buffer solution and then by 10% formalin solution containing 2% potassium ferrocyanide. The brain was removed from the skull and stored in 10% formalin solution containing 30% sucrose. The brain was cut into 50-μm-thick coronal sections, and each section was stained with cresyl violet.

Results

Neural database

While the monkey performed the ODR task, the activities of 205 OFC neurons and 204 DLPFC neurons were recorded. Among these, 121 OFC (59%) and 167 DLPFC (82%) neurons exhibited task-related activity in relation to at least one task event of the ODR task. Numbers and proportions of task-related neurons are summarized in Table 1 and Fig. 2. The proportion of task-related neurons was significantly smaller in the OFC than in the DLPFC (χ 2 = 25.60, P < 0.001). The proportions of OFC neurons that exhibited cue-, delay-, or response-period activity were significantly smaller than those of DLPFC neurons (cue, χ 2 = 12.64, P < 0.001; delay, χ 2 = 23.27, P < 0.001; pre-saccadic, χ 2 = 9.91, P < 0.01; post-saccadic, χ 2 = 23.59, P < 0.001). However, the proportion of OFC neurons with reward-period activity was significantly greater than that of OFC neurons with other task-related activities (cue vs. reward, χ 2 = 22.44, P < 0.001; delay vs. reward, χ 2 = 37.50, P < 0.001; response vs. reward, χ 2 = 29.95, P < 0.001). In contrast, the proportion of DLPFC neurons with reward-period activity was significantly smaller than that of DLPFC neurons with other task-related activities (cue vs. reward, χ 2 = 17.61, P < 0.001; delay vs. reward, χ 2 = 14.44, P < 0.001; response vs. reward, χ 2 = 34.03, P < 0.001). Among 121 task-related OFC neurons, 95 (79%) exhibited task-related activity in relation to only one task event (1 at the fixation period, 8 at the cue period, 7 at the delay period, 13 at the response period, and 66 at the reward period), while the remaining 26 (21%) exhibited task-related activity in relation to two or more task events. On the other hand, among 167 task-related DLPFC neurons, 80 (48%) exhibited task-related activity in relation to only one task event (9 at the cue period, 21 at the delay period, 28 at the response period, and 22 at the reward period), while the remaining 87 (52%) exhibited task-related activity in relation to two or more task events.

Table 1 Number of neurons exhibiting task-related activity
Full size table
Fig. 2
figure 2

a Comparison of the proportion of task-related activity between the orbitofrontal cortex (OFC) and dorsolateral prefrontal cortex (DLPFC). b Comparison of the proportion of combinations of task-related activity between the OFC and DLPFC. C, D, R and RW indicate cue-period activity, delay-period activity, response-period activity, and reward-period activity, respectively

Full size image

Thus, all of the task-related activities observed in the DLPFC were also observed in the OFC. However, the proportions of neurons with each task-related activity and the proportions of neurons exhibiting task-related activity in relation to two or more task events were different between the two areas. Reward-period activity was the most frequently observed task-related activity in the OFC, whereas cue-, delay- and response-period activities were observed more frequently in the DLPFC.

Delay-period activity

Among 121 task-related OFC neurons, 18 (15%) showed excitatory delay-period activity. Similarly, among 167 task-related DLPFC neurons, 69 (41%) showed delay-period activity (68 excitatory and one inhibitory). Excitatory delay-period activity was classified into three types based on its temporal pattern (tonic, Fig. 3a; gradually increasing, Fig. 3b; gradually decreasing, Fig. 3c). Among 18 OFC neurons, 4 (22%) showed a tonic pattern, 8 (45%) showed a gradually increasing pattern, and 6 (33%) showed a gradually decreasing pattern (Fig. 3d). On the other hand, among 69 DLPFC neurons, 50 (72%) showed a tonic pattern, 17 (25%) showed a gradually increasing pattern, and 2 (3%) showed a gradually decreasing pattern.

Fig. 3
figure 3

Temporal patterns of delay-period activity recorded from the OFC in the ODR task. a Tonic sustained excitation. b Gradually increasing activity. c Gradually decreasing activity. C, D, and R indicate the cue period, delay period, and response period, respectively. The delay length was 3 s. The histogram bin width was 20 ms. The activity was aligned at the start of the delay period. d The proportions of OFC and DLPFC neurons with different temporal patterns of delay-period activity

Full size image

Both OFC and DLPFC neurons with delay-period activity showed directional selectivity. Figure 4a shows an example of directional delay-period activity in the OFC. The strongest excitatory delay-period activity was observed when the visual cue was presented at the 90° position (one-way ANOVA, F(3,41) = 21.81, P < 0.001; 90°, Mann–Whitney U test, z = 3.42, P < 0.001). Figure 4b is an example of omni-directional delay-period activity observed in the OFC. The mean discharge rate of delay-period activity was not significantly different across all cue conditions by one-way ANOVA [F(3,47) = 0.70, P > 0.05], although significant delay-period activity was observed in all cue conditions (Mann–Whitney U test, P < 0.05). Among 18 OFC neurons with delay-period activity, 8 (44%) showed directional selectivity and 10 (56%) showed omni-directional selectivity (Fig. 5, top left). On the other hand, among 69 DLPFC neurons with delay-period activity, 61 (88%) showed directional selectivity and 8 (12%) showed omni-directional activity (Fig. 5, top right). Thus, although directional delay-period activity was observed in the OFC, the proportion of omni-directional delay-period activity in the OFC was significantly greater than that in the DLPFC (χ 2 = 16.81, P < 0.001).

Fig. 4
figure 4

a An example of directional delay-period activity recorded from the OFC in the ODR task. The strongest excitatory delay-period activity was observed when the visual cue was presented at the 90° position. The central diagram shows a polar plot of this delay-period activity. The average discharge rate during the delay period for each cue location is depicted as the radial eccentricity of the plot in that direction. The baseline discharge rate is shown by dotted lines. Asterisks indicate cue directions with statistically significant delay-period activity. The delay length was 3 s. The histogram bin width was 20 ms and was aligned at the start of the delay period. b An example of omni-directional delay-period activity recorded from the OFC. This neuron exhibited significant delay-period activity in all cue conditions and the mean discharge rate of delay-period activity was not significantly different across all cue conditions. The delay length was 3 s. The histogram bin width was 20 ms and was aligned at the start of the delay period

Full size image
Fig. 5
figure 5

Comparisons of the ratio of neurons with directionally selective task-related activity in the OFC and DLPFC

Full size image

To quantitatively compare the directional characteristics of delay-period activity between two areas, we constructed tuning curves of this activity by finding the best fit to the Gaussian function for 6 OFC and 54 DLPFC neurons that showed directional delay-period activity. More DLPFC neurons (78%) showed directional delay-period activity and a good fit than OFC neurons (33%) (Table 2). This difference was significant (χ 2 = 13.46, P < 0.001). Figure 6a shows an example of a tuning curve for the delay-period activity of an OFC neuron. The best direction (D) was 57° and the tuning index (Td) was 26°. Figure 6b shows polar plots of the best directions of delay-period activity for 6 OFC neurons (left) and 54 DLPFC neurons (right). Among the six OFC neurons, two had best directions toward the contralateral visual field, two had best directions toward the ipsilateral visual field, and two had best directions along the vertical meridian. On the other hand, among 54 DLPFC neurons, 37 (69%) had best directions toward the contralateral visual field, 14 (26%) had best directions toward the ipsilateral visual field, and the remaining 3 had best directions along the vertical meridian. A statistically significant contralateral bias was only observed in DLPFC neurons (χ 2 = 10.37, P < 0.01). Tuning indices of delay-period activity ranged from 11° to 77° (median 38°; mean and SD 39 ± 26°) for OFC neurons and from 4° to 109° (median 25°; mean and SD 37 ± 23°) for DLPFC neurons. A significant difference in tuning indices was not observed between OFC and DLPFC neurons (t test, t = 0.15, df = 58; P > 0.05).

Table 2 Characteristics of directional selectivity
Full size table
Fig. 6
figure 6

a A tuning curve of delay-period activity of an OFC neuron during the ODR task (the same neuron as in Fig. 4a). This excitatory delay-period activity had a best direction at 57° and Td = 26°. The horizontal line in this figure indicates the neuron’s baseline discharge rate. b Polar plots of the best directions of 6 OFC and 54 DLPFC neurons with delay-period activity. A significant contralateral bias was not observed in OFC neurons, but was observed in DLPFC neurons. In these figures, the best directions of neurons recorded from the right hemisphere were transformed into mirror-image directions, as if all neurons were recorded from the left hemisphere

Full size image

Cue-period activity

Cue-period activity was observed in 29 OFC (24%) and 74 DLPFC (44%) neurons (Table 1). All OFC neurons showed excitatory cue-period activity. However, 68 DLPFC neurons showed excitatory cue-period activity and 6 showed inhibitory cue-period activity. The latency of cue-period activity was measured by making a cumulative histogram aligned at the onset of the visual cue in the cue condition in which the maximum cue-period activity was observed. Latencies ranged from 50 to 270 ms (mean 131.3 ± 57.0 ms; median 110 ms) in OFC neurons and from 10 to 310 ms (mean 110.0 ± 45.7 ms; median 110 ms) in DLPFC neurons (Fig. 7a). The mean latency in DLPFC neurons was significantly earlier than that in OFC neurons (t test, t = 1.99, df = 101, P < 0.05).

Fig. 7
figure 7

a Distribution of onset latencies of cue-period activity in OFC neurons (upward) and DLPFC neurons (downward). The mean latency was 131 ms in OFC neurons and 110 ms in DLPFC neurons. b An example of directional cue-period activity recorded from the OFC in the ODR task. This neuron had significant cue-period activity when visual cues were presented at the 90° locations. The central diagram shows a polar plot of this activity. The histogram bin width was 20 ms and was aligned at the start of the delay period

Full size image

Among 29 OFC neurons, 24 (83%) showed directional cue-period activity and 5 (17%) showed omni-directional activity (Fig. 5, middle left). Figure 7b shows an example of directional cue-period activity observed in an OFC neuron. Significant cue-period activity was observed when the visual cue was presented at the 90° location (one-way ANOVA, F(3,37) = 28.46, P < 0.001; 90°, Mann–Whitney U test, z = 3.61, P < 0.001). Figure 8a shows an example of a tuning curve of cue-period activity observed in this OFC neuron. The best direction (D) was 94° and the tuning index (Td) was 43°. Tuning curves were constructed using 19 OFC neurons. Many neurons (66%) showed good fitting (Table 2). Figure 8b (left) shows polar plots of the best directions for these OFC neurons. Among these, 8 had best directions toward the contralateral visual field, 9 had best directions toward the ipsilateral visual field, and 2 had best directions along the vertical meridian. No contralateral bias was observed (χ 2 = 0.06, P > 0.05). Tuning indices ranged from 5° to 77° (median 23°; mean and SD 33 ± 21°).

Fig. 8
figure 8

a A tuning curve of cue-period activity of an OFC neuron during the ODR task. This excitatory cue-period activity had a best direction at 94° and Td = 43° (the same neuron as in Fig. 7b). b Polar plots of the best directions of 19 OFC (left) and 64 DLPFC (right) neurons with cue-period activity. A significant contralateral bias was not observed in OFC neurons, but was observed in DLPFC neurons

Full size image

On the other hand, among 74 DLPFC neurons, 69 (93%) showed directional cue-period activity and 5 (7%) showed omni-directional activity (Fig. 5, middle right). Tuning curves were constructed using 64 DLPFC neurons. Most neurons (86%) showed good fitting (Table 2). The proportion of DLPFC neurons that showed good fitting was significantly greater than that of OFC neurons (χ 2 = 5.86, P < 0.05). Figure 8b (right) shows polar plots of the best directions for these DLPFC neurons. Among these, 48 (75%) had best directions toward the contralateral visual field, 12 (19%) had best directions toward the ipsilateral visual field, and 4 had best directions along the vertical meridian. A significant contralateral bias was observed (χ 2 = 21.60, P < 0.001). Tuning indices ranged from 5° to 115° (median 25°; mean and SD 38 ± 22°).

Response-period activity

Response-period activity was observed in 23 OFC (19%) and 97 DLPFC (58%) neurons (Table 1). All neurons showed excitatory response-period activity. The latency of response-period activity was calculated by constructing a cumulative histogram of response-period activity aligned at the initiation of eye movement. This latency was defined as the time from the initiation of saccadic eye movement to the start of response-period activity. Figure 9a shows the distribution of the latencies of response-period activity. Latencies of response-period activity ranged from −110 to 210 ms (mean and SD 72.6 ± 113.3 ms; median 110 ms) for OFC neurons and from −150 to 370 ms (mean and SD 50.8 ± 119.3 ms; median 30 ms) for DLPFC neurons. There was no significant difference in mean latency between OFC and DLPFC neurons (t test, t = 0.80, df = 118, P > 0.05), although the mean latency of DLPFC neurons tended to be shorter than that of OFC neurons. Based on this latency, response-period activity could be classified into two groups (pre- and post-saccadic activity). Among 23 OFC neurons, 9 (39%) had pre-saccadic activity and 14 (61%) had only post-saccadic activity, and among 97 DLPFC neurons, 35 (36%) had pre-saccadic activity and 62 (64%) had only post-saccadic activity.

Fig. 9
figure 9

a Distribution of latencies of response-period activity in OFC neurons (upward) and DLPFC neurons (downward). The mean latency was 72.6 ms in OFC neurons and 50.8 ms in DLPFC neurons. b An example of omni-directional pre-saccadic activity recorded from the OFC. This neuron exhibited significant pre-saccadic activity for all directions. The histogram bin width was 20 ms and was aligned at the initiation of the saccade

Full size image

Among the 9 OFC and 35 DLPFC neurons with pre-saccadic activity, 5 OFC (56%) and 34 DLPFC (97%) neurons showed directional selectivity and the others exhibited omni-directional selectivity (Fig. 5). Figure 9b shows an example of omni-directional pre-saccadic activity observed in an OFC neuron. Saccade-related activity was initiated 90 ms before initiation of the saccades. Interestingly, all of the OFC neurons with omni-directional pre-saccadic activity also showed omni-directional delay-period activity, as shown in Fig. 4b. Of the neurons with pre-saccadic activity, 7 OFC and 27 DLPFC neurons had only pre-saccadic activity, and the remaining 2 OFC and 8 DLPFC had both pre- and post-saccadic activity. Post-saccadic activity was observed when the monkey made saccades toward the direction that was approximately opposite the direction for which pre-saccadic activity was observed.

Post-saccadic activity was observed in 14 OFC and 62 DLPFC neurons. Among these, 9 OFC (64%) and 55 DLPFC (89%) neurons showed directional post-saccadic activity and the remaining neurons showed omni-directional activity (Fig. 5). Although directional post-saccadic activity was observed in the OFC, the proportion of this activity in the DLPFC was significantly lower than that in the OFC (χ 2 = 5.12, P < 0.05).

To quantitatively compare the directional characteristics of response-period activity between the two areas, tuning curves were constructed using this activity of 9 OFC (3 pre-saccadic and 6 post-saccadic) and 83 DLPFC neurons (31 pre-saccadic and 52 post-saccadic). Figure 10a shows an example of a tuning curve of post-saccadic activity observed in an OFC neuron. The best direction was −40° and the tuning index was 25°. Figure 10b shows polar plots of the best directions for neurons with directional response-period activity. Among 3 OFC neurons with pre-saccadic activity, one had a best direction toward the contralateral visual field, whereas two had best directions toward the ipsilateral visual field. Among six OFC neurons with post-saccadic activity, two had best directions toward the contralateral visual field and four had best directions toward the ipsilateral visual field. On the other hand, among 31 DLPFC neurons with pre-saccadic activity, 23 (74%) had best directions toward the contralateral visual field, 6 (19%) had best directions toward the ipsilateral visual field, and 2 had best directions along the vertical meridian. A significant contralateral bias was observed (χ 2 = 9.97, P < 0.01). Among 52 DLPFC neurons with post-saccadic activity, 25 (48%) had best directions toward the contralateral visual field, 22 (42%) had best directions toward the ipsilateral visual field, and 5 had best directions along the vertical meridian. No significant contralateral bias was observed (χ 2 = 0.19, P > 0.05).

Fig. 10
figure 10

a A tuning curve of post-saccadic activity of an OFC neuron during the ODR task. This excitatory post-saccadic activity had a best direction at −40° and Td = 25°. b Polar plots of the best directions of 9 OFC (left) and 83 DLPFC (right) neurons with response-period activity. Among these, 3 OFC and 31 DLPFC neurons showed pre-saccadic activity at their best directions. A significant contralateral bias was not observed in OFC neurons, but was observed in DLPFC neurons. On the other hand, 6 OFC and 52 DLPFC neurons showed post-saccadic activity at their best directions. No significant contralateral bias was observed in either area

Full size image

Tuning indices of pre-saccadic activity ranged from 13° to 78° (median 24°; mean and SD 39° ± 35°) for OFC neurons and from 8° to 74° (median 29°; mean and SD 37° ± 19°) for DLPFC neurons. Tuning indices of post-saccadic activity ranged from 22° to 25° (median 22°; mean and SD 23° ± 1°) for OFC neurons and from 8° to 110° (median 26°; mean and SD 39° ± 23°) for DLPFC neurons. There was no significant difference in tuning indices between OFC and DLPFC neurons (t test, pre-saccadic, t = 0.13, df = 32, P > 0.05; post-saccadic, t = −1.73, df = 56, P > 0.05). In addition, there was no significant difference in tuning indices between pre- and post-saccadic activity (t test, OFC, t = 1.2, df = 7, P > 0.05; DLPFC, t = −0.47, df = 81, P > 0.05).

Reward-period activity

Reward-period activity was observed in 78 OFC neurons (64% of task-related neurons) and 31 DLPFC neurons (19% of task-related neurons). Figure 11b shows an example of reward-period activity observed in an OFC neuron. Onset times of reward-period activity were compared between the two areas. The onset time of reward-period activity was determined as the time when the magnitude of activity reached two standard deviations above the baseline activity after reward delivery. Onset times ranged from 30 to 450 ms (mean and SD 196.0 ± 109.9 ms; median 190 ms) for OFC neurons and from 50 to 450 ms (mean and SD 223.6 ± 105.8 ms; median 210 ms) for DLPFC neurons (Fig. 11a). Although the onset time of reward-period activity in OFC neurons tended to be earlier than that in DLPFC neurons, this difference was not significant (t test, t = −1.19, df = 107, P > 0.05).

Fig. 11
figure 11

a Distribution of onset time of reward-period activity in OFC neurons (upward) and DLPFC neurons (downward). The mean latency was 196.0 ms in OFC neurons and 223.6 ms in DLPFC neurons. b An example of reward-period activity recorded from the OFC. This neuron exhibited significant reward-period activity for all cue directions. The histogram bin width was 20 ms and was aligned at reward delivery. c The same neuron’s responses to free-reward delivery. The response latency was 60 ms. Free-reward was delivered at randomly selected times during the delay period in randomly selected trials of the ODR task. The histogram bin width was 20 ms and was aligned at reward delivery. d Proportion of neurons that responded to free-reward delivery

Full size image

Among neurons with reward-period activity, 60 OFC and 26 DLPFC neurons were tested with regard to whether or not they also responded to free-reward delivery. Free-reward (0.2 ml liquid reward) was delivered during the delay period in randomly selected trials of the ODR task. Figure 11c shows an example of the response to free-reward delivery. The response latency to free-reward delivery was 60 ms. Among 60 OFC neurons, 46 (77%) showed excitatory responses, 2 (3%) showed inhibitory responses, and 12 (20%) did not respond to free-reward delivery. Among 26 DLPFC neurons, 13 (50%) showed excitatory responses and 13 (50%) did not respond to free-reward delivery (Fig. 11d). The proportion of neurons that responded to free-reward delivery was significantly higher in OFC neurons than in DLPFC neurons (χ 2 = 7.92, P < 0.01).

Histological examination of recorded neurons

Figure 12 shows cortical areas where we examined single-neuron activity (gray areas). This figure also depicts histologically identified locations where task-related activity was recorded. Most task-related activities were recorded from the lateral part of the OFC (area 12 and 45). Some activities were also recorded from areas 11 and 13. Neurons with cue-, delay- or response-period activity tended to be located in the lateral part of the OFC and the inferior convexity of the lateral prefrontal cortex. Neurons with reward-period activity were mostly located in the lateral part of area 12 and the inferior convexity of the lateral prefrontal cortex. In the DLPFC, neurons with cue- or response-period activity were distributed in both the dorsal and ventral banks of the principal sulcus. However, neurons with delay-period activity tended to be distributed in the dorsal as well as the ventral surface around the principal sulcus.

Fig. 12
figure 12

Cortical distribution of task-related neurons in the OFC and DLPFC. Coronal sections at level a to level e are shown. Black dots indicate cortical locations of neurons that showed cue-, delay-, response-, or reward-period activity. Gray areas indicate cortical areas where electrode penetrations were made. AS arcuate sulcus, PS principal sulcus

Full size image

Discussion

The DLPFC participates in spatial working memory processes. The neural mechanisms of spatial working memory processes in the DLPFC have been extensively examined using the ODR task (Funahashi et al. 1989, 1990, 1991; Chafee and Goldman-Rakic 2000; Constantinidis et al. 2001; Takeda and Funahashi 2002, 2004). The OFC has strong reciprocal connections to the DLPFC (Barbas and Pandya 1989; Carmichael and Price 1996; Cavada et al. 2000). Although it is well know that the OFC participates in motivational processes, little is known about the cognitive aspects of OFC functions. Therefore, in this experiment, we examined how the OFC contributes to spatial working memory processes. To achieve this goal, we collected single-neuron activity from the same monkey’s OFC and DLPFC while the monkey performed the ODR task, and compared the characteristics of task-related activity in the two areas. All task-related activities observed in the DLPFC were also observed in the OFC. However, the proportions of neurons with each task-related activity were significantly smaller in the OFC than in the DLPFC. In the DLPFC, most neurons with delay-period activity were directionally selective and showed tonic sustained activation, while in the OFC, most neurons with delay-period activity were omni-directional and showed either gradually increasing or decreasing activity. In addition, neurons with reward-period activity were predominant in the OFC. These results indicate that, although both the OFC and DLPFC participate in spatial working memory processes, the OFC is related more to motivational processes such as the expectation and detection of reward delivery, whereas the DLPFC is related more to cognitive processes such as the temporary maintenance of spatial information and its processing for spatial working memory.

Delay-period activity

Delay-period activity observed in the delayed-response task has been considered to be a neural correlate of the mechanism for the temporary active maintenance of spatial information (Niki 1974; Goldman-Rakic 1987; Funahashi et al. 1989; Funahashi and Kubota 1994; Fuster 1997). In the present study, we found a significant difference in the proportion of neurons with delay-period activity between the OFC and DLPFC. Only 15% of task-related OFC neurons exhibited delay-period activity, whereas 41% of task-related DLPFC neurons exhibited delay-period activity. Previous studies using the ODR task have reported that 51% (Funahashi et al. 1989) and 47% (Takeda and Funahashi 2002) of task-related DLPFC neurons exhibited delay-period activity. The proportion of DLPFC neurons with delay-period activity in the present study is similar to that in these previous studies. Therefore, the present result suggests that the proportion of neurons with delay-period activity in the OFC is significantly smaller than that in the DLPFC. In addition, we found a significant difference in the proportion of neurons with directional delay-period activity between the two areas. Among OFC neurons, 44% showed directional delay-period activity and the remaining 56% showed omni-directional delay-period activity. On the other hand, among DLPFC neurons, 88% showed directional delay-period activity and only 12% showed omni-directional delay-period activity. Previous studies have shown that 79% (Funahashi et al. 1989) and 93% (Takeda and Funahashi 2002) of DLPFC neurons with delay-period activity show directional selectivity. The present result obtained in DLPFC neurons agrees with these previous results. Therefore, the present results indicate that more OFC neurons exhibit omni-directional delay-period activity, whereas most DLPFC neurons exhibit directional delay-period activity. Further, we found a difference in the temporal pattern of delay-period activity between the OFC and DLPFC. In the OFC, delay-period activity mostly showed either gradually increasing or decreasing activity toward the end of the delay period. However, in the DLPFC, most delay-period activity showed tonic sustained activation. Previous studies have shown that most DLPFC neurons (Funahashi et al. 1989) and posterior parietal neurons (Chafee and Goldman-Rakic 1998) exhibited a tonic sustained pattern of delay-period activity while monkeys performed the ODR task. Therefore, the temporal pattern of delay-period activity observed in the OFC is different from those observed in the DLPFC and the posterior parietal cortex.

Thus, we found several differences in the characteristics of delay-period activity between the OFC and DLPFC. In the OFC, the proportion of neurons with delay-period activity was small, most delay-period activities were omni-directional, and the temporal pattern of delay-period activity either gradually increased or decreased. Previous studies have shown that the magnitude of OFC activity observed during the delay period was affected by the expectation or the quality or quantity of the reward (Tremblay and Schultz 1999, 2000; Hikosaka and Watanabe 2000, 2004; Roesch and Olson 2005b). Therefore, the present results suggest that, although directional delay-period activities observed in OFC neurons represent spatial information during ODR performance, most delay-period activity is strongly affected by the expectation of reward delivery.

Directionally selective delay-period activity observed in the DLPFC has also been observed in many other brain areas including the frontal eye field (Funahashi et al. 1989; Roesch and Olson 2005a), the posterior parietal cortex (Chafee and Goldman-Rakic 1998), and the caudate nucleus (Hikosaka et al. 1989; Kawagoe et al. 1998). However, as revealed in the present study, most neurons in the OFC showed omni-directional delay-period activity. Similarly, delay-period activity was mostly omni-directional in some brain areas, including the posterior cingulate cortex (Carlson et al. 1997) and the medial part of the mediodorsal nucleus of the thalamus (Watanabe and Funahashi 2004). The posterior cingulate cortex has reciprocal connections to the OFC (Vogt and Pandya 1987; Morecraft et al. 1992; Carmichael and Price 1995). The medial part of the mediodorsal nucleus of the thalamus also has strong reciprocal connections to the OFC (Cavada et al. 1995, 2000). The similarity of the directional selectivity in delay-period activity among these brain areas might be caused by these anatomical connections. Since all of these areas participate in emotional and motivational processes, omni-directional delay-period activity might be related to motivational processes such as the expectation of reward delivery, rather than to spatial working memory processes. The fact that the magnitude of most omni-directional delay-period activity in the OFC gradually increased toward the end of the delay period supports this view.

Cue-period activity

Cue-period activity has been observed in DLPFC neurons while monkeys performed the ODR task (Funahashi et al. 1990; Takeda and Funahashi 2002). Cue-period activity is usually phasic excitation and is considered to be in response to presentation of the visual cue, since DLPFC neurons respond to even simple visual stimuli (e.g., spot or bar stimulus) and have visual receptive fields (Mikami et al. 1982; Suzuki and Azuma 1983) and because similar activity with similar response characteristics was also observed in the visual probe task (Funahashi et al. 1990). In the present study, 24% of task-related OFC neurons exhibited cue-period activity, whereas 44% of task-related DLPFC neurons exhibited this activity in the ODR task. In previous studies, Funahashi et al. (1990) observed this activity in 28% of task-related DLPFC neurons and Takeda and Funahashi (2002) observed this activity in 38% of task-related DLPFC neurons. The proportion of OFC neurons with cue-period activity seems to be comparable to that of DLPFC neurons. Regarding the directional selectivity of cue-period activity, 83% of neurons with this activity in the OFC exhibited directional selectivity, whereas 93% in the DLPFC exhibited directional selectivity. Previous studies performed in the DLPFC have also shown that a great majority of cue-period activity showed directional selectivity (96%, Funahashi et al. 1990; 100%, Takeda and Funahashi 2002). Although a significant contralateral bias was observed in the distribution of the best directions in DLPFC neurons in both the present and previous studies (Funahashi et al. 1990; Takeda and Funahashi 2002), a significant contralateral bias was not observed in the OFC. These results indicate that cue-period activity observed in OFC neurons could be a visual response to presentation of the visual cue, similar to cue-period activity observed in DLPFC neurons.

The present study showed that the mean latency of cue-period activity in OFC neurons was significantly later than that in DLPFC neurons. The OFC has strong reciprocal connections to the DLPFC (Cavada et al. 2000) and the DLPFC has strong reciprocal connections to the posterior parietal cortex and the inferior temporal cortex (Ungerleider and Mishkin 1982; Petrides and Pandya 1984). Therefore, cue-period activity observed in the OFC might be transmitted from the DLPFC. In addition, since area 7a of the posterior parietal cortex is anatomically connected to the OFC (Cavada and Goldman-Rakic 1989), OFC neurons might also receive visual information directly from the posterior parietal cortex.

Response-period activity

In the present study, 19% of task-related neurons in the OFC and 58% of those in the DLPFC exhibited response-period activity. The proportion of response-period activity was significantly higher in the DLPFC than in the OFC (χ 2 = 44.08, P < 0.001). Among OFC neurons with response-period activity, 39% were pre-saccadic and 61% were post-saccadic. Among OFC neurons with pre-saccadic activity, 44% were omni-directional. The proportion of omni-directional pre-saccadic activity was significantly higher in the OFC than in the DLPFC. Interestingly, all OFC neurons with omni-directional pre-saccadic activity also showed gradually increasing delay-period activity. These results suggest that pre-saccadic activity observed in the OFC is not directly related to the trigger signal for saccadic eye movement, but is related to the expectation of reward delivery.

As with pre-saccadic activity, the proportion of post-saccadic activity was significantly higher in the DLPFC than in the OFC (χ 2 = 23.59, P < 0.01). Post-saccadic activity has been considered to be a feedback signal from oculomotor centers and could modulate task-related activity, especially delay-period activity (Goldman-Rakic et al. 1990; Funahashi and Kubota 1994; Funahashi and Takeda 2002). The present results suggest that the OFC also receives feedback signals from oculomotor centers and that these signals might play an important role in modulating delay-period activity in the OFC.

Reward-period activity

Among 205 OFC neurons recorded, 64% showed reward-period activity. The proportion of OFC neurons with this activity was significantly high compared to neurons with either cue-, delay-, or response-period activity. In addition, the proportion of OFC neurons with reward-period activity was also significantly high compared to DLPFC neurons with this activity (19%). The response of OFC neurons to reward delivery has been reported previously (Niki et al. 1972; Rosenkilde et al. 1981; Thorpe et al. 1983; Tremblay and Schultz 2000; Ichihara-Takeda and Funahashi 2006). In the present study, most OFC neurons with reward-period activity responded not only to the task reward but also to free-reward delivery. OFC neurons with reward-period activity were found in the lateral part of area 12 and the inferior convexity. Animals with lesions in the lateral part of area 12 and the inferior convexity showed deficits in the object reversal task (Butter 1969; Rosenkilde 1979). These results suggest that reward-period activity observed in the OFC represents the expectation or detection of reward delivery. Reward-period activity might contribute to behavioral control based on the reward outcome.

Dorsolateral prefrontal cortex neurons also showed reward-period activity. However, half of DLPFC neurons with reward-period activity did not respond to free-reward delivery, suggesting that reward-period activity is not related to the detection of reward delivery but rather may represent the signal indicating the end of the trial (Watanabe 1989; Ichihara-Takeda and Funahashi 2006). The remaining DLPFC neurons also responded to free-reward delivery. The onset times of reward-period activity in the OFC were earlier than those in the DLPFC. The OFC area where reward-period activity was mainly recorded has reciprocal connections with the DLPFC (Zald and Kim 2001). Therefore, these results suggest that reward-related information is conveyed from the OFC to the DLPFC. This notion agrees with the results in a previous study (Wallis and Miller 2003).

Functional contribution of the OFC to cognitive processes

The OFC has strong reciprocal connections to the DLPFC (Barbas and Pandya 1989; Carmichael and Price 1996; Cavada et al. 2000), suggesting that the OFC, together with the DLPFC, participates in higher cognitive functions, such as spatial working memory processes. In fact, all of the task-related activities observed in the DLPFC were found in the OFC while the monkey performed the ODR task. However, the proportion, directional selectivity, and temporal pattern of task-related activity were different between these two areas. In the OFC, reward-period activity was predominant among task-related activities. Most delay-period activity was omni-directional and gradually increased toward the end of the delay period. These results indicate that, although the OFC participates in spatial working memory processes, it participates more in the expectation and detection of reward outcome.

Bechara et al. (1998) reported that patients with orbitofrontal damage were likely to choose cards associated with high risk and eventually tended to lose their money when they performed the Iowa gambling task (Damasio 1994; Bechara et al. 1998, 2000). Thus, the patients were not affected by the negative consequences of their actions. In addition, several studies using reversal tasks have reported that OFC-damaged patients and animals continued to select the same stimulus even after this stimulus was no longer associated with the reward (Butter 1969; McEnaney and Butter 1969; Rolls et al. 1994; Fuster 1997). Lesions in the lateral part of the OFC and the inferior convexity, where most reward-period activity was observed in the present experiment, produced short-term memory deficits and deficits in reversal tasks in animals (Butter 1969; Passingham 1975; Rosenkilde 1979; Kowalska et al. 1991). Thus, OFC functions associated with the expectation and detection of reward delivery contribute significantly to the flexible modulation of the stimulus-reward contingency. These functions are important for the execution of working memory processes.

Ichihara-Takeda and Funahashi (2006) recently showed that the magnitude of reward-period activity was affected by the reward schedule, such that the magnitude of reward-period activity gradually increased depending on the proximity to the reward trial, and indicated that reward-period activity observed in the OFC was strongly affected by the animal’s motivational state. Modulation of the reward schedule would be a good method for examining the motivational influence on neural activity and for comparing the motivational effects on task-related activities between the OFC and the DLPFC.