## Abstract

In order to investigate whether and how medial prefrontal cortex (mPFC) of the rat is involved in processing of information related to fear conditioning, we recorded from single units in the prelimbic and infralimbic cortex of fear-conditioned rats in response to an explicit conditional stimulus (CS; an auditory tone) or contextual cues (conditioning box). The majority of units changed their activities significantly in response to the CS in a delay or trace conditioning paradigm. Both transient and tonic activity changes, including delay cell activity, were observed as in other behavioral tasks. When exposed to the context without CS delivery, most units changed their activities as well. These results show that both tone and contextual information are processed in the rat mPFC in expectation of the delivery of an aversive stimulus (electric foot shock). Interestingly, fast spiking cells (putative inhibitory interneurons) and regular spiking cells (putative projection neurons) showed different patterns of responses. Fast spiking cells tended to show transient responses and increased their firing rates following CS presentation, whereas a complementary pattern was observed in the regular spiking cells. Our results enhance our understanding of the neural mechanisms underlying prediction of an aversive stimulus in the mPFC.

## Introduction

Emotion is an important component of the function of prefrontal cortex (PFC). Patients with prefrontal cortical lesions manifest affective and emotional abnormalities such as apathy, depression, panic disorder and obsessive-compulsive disorder (Baxter et al., 1989, 1990; Gorman et al., 1989; Drevets et al., 1992; Godefroy and Rousseaux, 1997). In animals, lesions or stimulation of various divisions of the PFC lead to alterations in emotional functions such as aggression, stress, anxiety and autonomic control (Franzen and Myers, 1973; Siegel et al., 1975; Maskati and Zbrozyna, 1989; Zbrozyna and Westwood, 1991; Diorio et al., 1993; Frysztak and Neafsey, 1994; Jinks and McGregor, 1997). Physiological studies also have shown that neurons in the dorsolateral PFC of the monkey (Ono et al., 1984; Inout et al., 1985) and medial PFC (mPFC) of the rabbit (Maxwell et al., 1994) change their activities in association with emotional aspects of a task. Anatomically, the PFC receives profuse projections from the thalamus, hypothalamus, amygdala, hippocampus and cingulate area (Arikuni and Ban, 1978; Swanson, 1981; Musil and Olson, 1988; McDonald, 1991; Ray and Price, 1993; Gigg et al., 1994; Bacon et al., 1996), which all play important roles in emotion.

Recent investigations of neural mechanisms underlying emotion focused on fear conditioning. Much progress has been made owing to the use of this behaviorally well-defined paradigm (LeDoux, 2000). Concerning the role of the PFC in fear conditioning, however, previous behavioral studies have reported conflicting results. In rats, lesions in the mPFC led to an increase, a decrease or no change in fear reactivity (Mason and Fibiger, 1979; Holson, 1986; Rosen et al., 1992; Morgan et al., 1993; Morgan and LeDoux, 1995; Gewirtz et al., 1997; Joel et al., 1997). The absence of any effect of a PFC lesion on the behavioral reaction to a conditional stimulus (CS) does not necessarily indicate that the PFC has no role in fear conditioning, however. For example, animals with bilateral hippocampal lesions show intact conditioning in a delay-conditioning paradigm, i.e. when there is no interval between a CS and an unconditional stimulus (US). Impairment is observed when the interval between a CS and a US is long or in the context conditioning paradigm, however (Selden et al., 1991; Kim and Fanselow, 1992; Phillips and LeDoux, 1992; McEchron et al., 1998). In addition, some hippocampal neurons show activity changes that correlate with different phases of classical conditioning (Berger et al., 1980; McEchron and Disterhoft, 1997), suggesting that relevant information is processed in the hippocampus and utilized when necessary.

Considering the anatomical organization, it is highly likely that information related to fear conditioning is actively processed in the rat mPFC and influences behavior under appropriate conditions. Sensory information about a stimulus is conveyed to the mPFC via sensory cortical projections to the mPFC (Condé et al., 1995). The amygdala, which plays an essential role in fear conditioning (LeDoux, 1996), sends direct as well as indirect projections via the mediodorsal thalamus to the mPFC. Stimulation and inactivation studies have shown that amygdaloid projections strongly influence neuronal activities in the mPFC (Pérez-Jaranay and Vives, 1991; Garcia et al., 1999). In addition, the mPFC has direct access to contextual information by way of direct hippocampal projections to it (Swanson, 1981; Ferino et al., 1987; Jay and Witter, 1991). Thus, the mPFC of the rat is in a position to receive sensory, hippocampal and amygdaloid inputs, which are major components of fear conditioning. Also, other projections to the mPFC, such as those from the hypothalamus (Arikuni and Ban, 1978), could carry information about visceral changes associated with fear conditioning.

The role of the mPFC in fear conditioning can be conjectured by examining changes in mPFC neural activity during fear conditioning. In addition, insights about the way the mPFC processes fear-related information can be obtained by physiological studies. So far, few single unit studies have examined PFC unit activity in the context of negative emotion. Single unit studies in rabbits (Gibbs and Powell, 1991; Maxwell et al., 1994) have shown that neuronal activities in the mPFC show changes that correlated with heart rate conditioning. In the present study, we examined single unit activities in the prelimbic and infralimbic cortex of the rat during fear conditioning. We examined the effects of both an explicit CS (tone) and context. Our results show that an explicit CS, as well as contextual cues, induces changes in mPFC unit activity, and that putative projection neurons and putative inhibitory neurons play different roles in this process.

## Materials and Methods

### Subjects

Thirteen adult male Sprague–Dawley rats, ~10–12 weeks old, weighing 280–350 g, were used. All subjects were maintained on a 12 h light:dark cycle and allowed free access to food and water. The experimental protocol was approved by the Ethics Review Committees for Animal Experimentation of Ajou University School of Medicine.

### Electrode Implantation

Rats were deeply anesthetized with Nembutal (50 mg/kg) and two microelectrode drivers (McNaughton et al., 1989) were installed on opposite sides of the skull, both directed at the medial wall of the PFC (2.5–3.0 mm A and 0.6–1.3 mm L to bregma) at an angle 0–10° toward the midline. The recording electrodes (tetrode) (Recce and O'Keefe, 1989; Wilson and McNaughton, 1993) consisted of bundles of four polyimide insulated nichrome wires (H.P. Reid Co., Palm Coast, FL) twisted together and heated gently to fuse the insulation without short-circuiting the wires (final overall diameter: 40 μm). The electrode tips were cut and gold-plated to reduce their impedance to 0.2–1.0 MΩ measured at 1 kHz. The reference electrode was a stainless steel wire that was fixed to six skull screws. Another stainless steel wire was soldered onto one screw to serve as an earth lead. The entire implant was encased in dental acrylic.

### Unit Recording

Unit signals were recorded via an FET source-follower headstage mounted on the animal's head. Output signals from the headstage were filtered between 0.6 and 6 kHz, digitized at 32 kHz and stored on a SUN 4u workstation for future offline analysis. Data acquisition was performed using the Cheetah system (Neuralynx, Tucson, AZ). Units were isolated by projecting the four-channel relative amplitude data two-dimensionally, and applying boundaries to each subjectively apparent unit cluster (McNaughton et al., 1983). Care was taken to apply the same criteria to all cells in the population.

#### Apparatus

The conditioning chamber was 27 × 25 × 40 cm in dimension and had 18 metal grids (4 mm diameter, 1.5 cm spaced) on the floor that were connected to a shock scrambler. The box was washed thoroughly with 70% alcohol and completely dried before conditioning and testing. The chamber was placed on a speaker (diameter = 26 cm) that converted animal movements into electrical signals. The movement signals were digitized at 2.5 kHz and stored on the workstation.

#### Fear Conditioning

The behavioral task was an aversive classical conditioning task. After ~1 week of recovery from surgery, rats were adapted to transportation and handling for 5 min over three consecutive days. On the initial day of training, animals were placed in the conditioning box and allowed to habituate for 1 min. They were then presented with 20 incidences of an auditory tone (CS; 5 s, 1 kHz) that were paired with electric foot shock (US; 1 mA, 0.5 s) at 50% chance in a quasi-random manner with 50–70 s inter-trial intervals. Animals were removed from the conditioning box 1 min after the last shock and returned to their home cages. The partial punishment schedule was lowered progressively over the next 2 days from a quasi-random 50% level to 15–20% and the number of trials was raised to 40. The punishment rate and number of trials were maintained at 15–20% and 40 trials, respectively, throughout the subsequent recording sessions unless otherwise noted. Recordings began on day 4 whenever well-isolated units were found. A search of units and electrode advancement were made while the animals were sitting on a pedestal outside the conditioning box. Following recording of baseline unit activity on the pedestal, each animal was placed in the conditioning box and recordings were made in 10 out of 40 trials in which the electric foot shock was not delivered. Care was taken not to allow the animal to predict the delivery of the foot shock. The room light was turned off before training and recording. Four and five rats were used for standard (delay) and trace conditioning, respectively. In the standard conditioning task, the US was delivered during the last 0.5 s of the CS so that the CS and US co-terminated. In the trace conditioning task, the US was delivered 2 s after the offset of the CS (Fig. 1A).

#### Testing Effects of Context

To examine the effects of contextual cues on mPFC unit activity, two rats that were trained in the standard conditioning paradigm were used. Once well-isolated and stable units were found, baseline recordings were made for 5 min while the animal was sitting quiet on the pedestal outside the conditioning box. The animal was then placed in the recording chamber and unit activities were recorded for 5 min without delivery of an auditory tone to test whether exposure to contextual cues induces changes in unit activity.

#### Control Experiment

Recordings were made from four naive (i.e. unconditioned) rats as the control for the effects of the CS (two rats) or contextual cues (two rats) on mPFC unit activity. Unit recordings were made as in the standard conditioning (i.e. the CS was delivered) or contextual cue experiments (the CS was not delivered) except that the rats never received the electric foot shock. The results were compared with those from the conditioned rats.

#### Extinction Procedure

Two rats from the trace conditioning group were used. When well-isolated units were found, the CS was presented repeatedly without the US delivery until freezing behavior disappeared (40–60 trials), while unit recordings were made. The animals were then retrained with CS–US pairing (20 trials, 30% punishment rate) and unit activity was measured without US delivery until freezing behavior disappeared again (40–80 trials). Thus, the rats went through two episodes of extinction in a given day. Because the animals did not go through CS–US pairing following the second episode of extinction, and therefore exhibited a lower degree of freezing during the first episode of extinction on the next day, only data from the second episodes of extinction were considered in this study. Once the rats experienced extinction, they were not used again for further trace conditioning experiments.

### Analysis

#### Unit Classification

Deep layer units in the rat mPFC were classified into regular spiking (RS) and fast spiking (FS) cells in our previous study (Jung et al., 1998). RS cells fire at relatively low rates and typically have wide spike waveforms with small after-hyperpolarization. On the other hand, FS cells fire at relatively high rates and have narrow spike waveforms with more pronounced after-hyperpolarization. Overall, as shown in Figure 2, the units recorded in this study showed the same characteristics. Units were therefore classified as RS and FS cells, as in our previous study (Jung et al., 1998). Unit classification was based on a non-hierarchical clustering method using mean firing rate. Using the ratio between the peak and valley amplitudes of the spike waveform (peak–valley ratio) as an additional parameter for clustering yielded the same results. Units had only a few different, stepwise values of the spike width (duration between the peak and the valley of a spike waveform) due to a limited resolution of spike waveform sampling (32 kHz); hence spike width was not used as a parameter for clustering.

#### Determination of Responsive Units to CS

To determine the units that were responsive versus unresponsive to the CS, the task was divided into pre-CS (5 s) and CS (5 s) periods for the standard conditioning and pre-CS, CS, delay (2 s) and US (1 s) periods for the trace conditioning task. CS, delay and US periods were divided into 1 s bins and unit discharge in each bin was compared with that during the pre-CS period over 10 trials using paired t-test (α = 0.05). Those units that contained at least one bin that was significantly different from the pre-CS period were regarded as responsive units.

#### Classification of Response Patterns to CS

To classify activity patterns of responsive units, peri-stimulus time histograms (PSTH) were first constructed with a bin size of 500 ms by aligning responses to the CS onset. The mean firing rate and standard deviation of each PSTH (no. of bins = 20 and 25 for the standard and trace conditioning, respectively; the first 10 bins correspond to the pre-CS period) were calculated, and then the firing rate in each bin was converted to a Z-score using the mean ± SD. A hierarchical clustering algorithm using the squared Euclidean method was applied to the normalized histograms.

#### Behavior

To assess the degree of freezing, animal movement was quantified by adding up peak amplitudes of oscillatory movement signals within a given time period and then dividing this value by the total duration. Animal movement was expressed in arbitrary units per second.

#### Histology

When recordings were completed, an electrolytic current (50 μA cathodal, 30 s) was applied through one of the recording electrodes. The animal was deeply anesthetized and perfused with 0.9% saline followed by buffered 10% formal saline while the electrode remained in situ. The brain was then removed, left in 10% formal saline for 3–5 days and then transferred to a 30% sucrose solution for 2–3 days until it sank to the bottom. Coronal sections (40 μm thick) were cut on a sliding microtome and alternate sections were stained with cresyl violet. Tracks and lesion sites were located under a light microscope.

## Results

### Recording Locations

Recording was made from 459 well-isolated units in the mPFC of 13 rats. These units were located in the superficial and deep layers of the prelimbic and infralimbic cortex (Fig. 1B). Table 1 summarizes the numbers of animals and recorded units in each task. There was no clear relationship between recording location and response patterns.

### Freezing Behavior

The rats also showed significant freezing following context conditioning in which the CS was not delivered. The amounts of animal movement were 93.8 ± 24.7 and 406.1 ± 87.3 (measured for 5 min periods) for the context and control animals, respectively, which were significantly different (t-test P < 0.01).

### Relationship between Unit Response and Freezing

Monitoring spatiotemporal activity patterns of the mPFC unit population following CS presentation would be necessary to fully understand how information related to fear conditioning is processed by the mPFC neural circuitry. On the other hand, certain units might play particularly important roles in this process so that significant relationships between their activities and freezing behavior can be detected. We tested this possibility by examining the relationship between mPFC unit response and freezing behavior. For this, units in the standard and trace conditioning tasks were combined according to each response type (types 1–3; type 4 of the trace conditioning task was excluded).

$magnitude\ of\ unit\ response\ {=}\ |\mathit{CSu}\ {\mbox{--}}\ \mathit{pre-CSu}|/\mathit{pre-CSu}$
where pre-CSu and CSu respectively indicate mean firing rate of each unit during the pre-CS (5 s before the CS onset) and the CS period (5 s) across all trials.

The degree of freezing was quantified as the following:

$degree\ of\ freezing\ {=}\ (\mathit{pre-CSm}\ {\mbox{--}}\ \mathit{CSm})/\mathit{pre-CSm}$

where pre-CSm and CSm respectively indicate animal movements during the pre-CS and the CS period across all trials in each recording session. Then correlation between the magnitude of unit response and the degree of freezing behavior was examined for each response type. No significant relationship was found between any of the three response types and freezing behavior, indicating that there exists no simple relationship between freezing behavior and responses of mPFC units.

### Effects of Context

Forty-nine units were tested for their responses to the contextual cues without CS delivery. Unit activity outside the conditioning box on the pedestal was compared with that inside the conditioning box. Twenty-six units increased and 23 decreased their average firing rates in the conditioning box compared with those outside the box. These units were recorded only once inside and outside the box, thus statistical comparison between the two conditions for a given unit was not possible. Instead, firing rate changes recorded from the conditioned rats were compared with those from naive rats (control). Nineteen units were recorded from two naive rats. Because we were interested in whether PFC units change their activities significantly to the contextual cues, regardless of the direction of change, absolute changes in unit activity were calculated and normalized to the baseline activity (firing rate outside the conditioning box) so as to express unit activity change as a percentage of the baseline activity. Whereas the cells recorded from conditioned animals changed their activities by 165 ± 43% relative to the baseline, those from the naive rats changed by only 42 ± 8% (the difference was statistically significant, Mann–Whitney U-test, P < 0.05). Figure 7 shows responses, recorded from a conditioned rat,  of two units to the contextual cues.

## Discussion

Although recent investigations of fear conditioning revealed much about underlying neural circuit mechanisms, results of behavioral studies that have examined the PFC role in fear conditioning do not agree well, and single unit recordings in the PFC of fear conditioned animals are scarce. Our study was undertaken to obtain information regarding (i) the role of the rat mPFC in fear conditioning and (ii) neural mechanisms underlying information processing related to fear conditioning. Several conclusions could be drawn from the present study. First, the rat mPFC actively processes information related to fear conditioning. Secondly, unit responses could be classified into several categories, and both transient and tonic response patterns, including delay cell activities, were observed as in previous studies. Thirdly, mPFC neurons respond to both explicit CS and contextual cues that predict an aversive stimulus. Fourthly, putative projection neurons and inhibitory interneurons behave differently during fear conditioning.

### Role of mPFC in Fear Conditioning

Previous behavioral studies have reported conflicting results concerning the role of the rat mPFC in fear conditioning. Lesions in the mPFC led to an increase, a decrease or no change in fear reactivity (Mason and Fibiger, 1979; Holson, 1986; Rosen et al., 1992; Morgan et al., 1993; Morgan and LeDoux, 1995; Gewirtz et al., 1997; Joel et al., 1997). Our study showed that the majority of mPFC units respond to the CS in fear-conditioned rats. Of a total of 260 units, 196 (75%) responded to the CS in a significant manner. In control experiments, in which units were recorded from naive rats, only 25% were classified as responsive units and the magnitude of the response was much smaller. These findings indicate that an innocuous auditory stimulus that only slightly influences the activity of mPFC neurons elicits widespread changes in mPFC unit activity after it is temporally paired with an aversive US. Furthermore, the results from the extinction experiments show that unit responses to the CS were correlated to fear reactivity. During extinction, unit responses to the CS diminished gradually as the degree of freezing became smaller. These results suggest strongly that information related to fear conditioning is actively processed in the rat mPFC. Presumably, parallel pathways exist in the brain that mediate a CS to a conditional response during fear conditioning. Expression of some behaviors may not require the intact PFC whereas other behaviors, or the same behaviors but in different context, require the intact PFC to be expressed when a CS is presented. In this regard, Maxwell et al. (Maxwell et al., 1994) reported previously that 68% of units recorded in the rabbit mPFC responded significantly to the CS after being paired with a US, and that 32% changed their activities when an innocuous sound was applied. The proportions agree well with those in the present study. These results suggest strongly that the mPFC of the rat and rabbit participates in processing of external stimuli that predict an aversive stimulus.

### Types of Responses

Fuster (Fuster, 1973) initially described several different unit response patterns in the dorsolateral PFC of monkeys. Komatsu (Komatsu, 1982) later classified PFC unit responses into the following three basic types: a brief increase of activity after event onset, tonic activity between different events and gradual activity change preceding event onset. These response patterns have been repeatedly observed across different behavioral tasks. Similar response patterns were also observed in the present study. Responses were classified, by a hierarchical clustering algorithm that does not presume the number of clusters a priori, into transient elevation of activity after CS onset (type 3 of the standard and trace conditioning task), tonic decrease following CS onset (type 1 of the standard and trace conditioning task) and gradual activity change preceding expected delivery of the US (type 2 of the standard and types 2 and 4 of the trace conditioning task). Responses of the units that showed a transient rise in their activities (type 3) returned slowly to the baseline over the course of the CS presentation. Of the units, 13 showed brief responses that were terminated within 1 s following the CS onset (bottom example of Fig. 4A), which would correspond to ‘brief elevation of activity after event onset’ (Komatsu, 1982). These observations suggest that common patterns of activity seem to exist in different divisions of the PFC across different tasks that involve temporal discontiguity between a sensory stimulus and a behavioral response (Fuster, 1997). The transient reaction (brief elevation after event onset) is probably related to directing attention to the CS, and gradual reactivity may participate in some way in the anticipation of the US. It is conceivable that transient responses are induced by an external sensory input (CS) to the PFC. More difficult questions are how gradually changing activity is generated and whether the PFC interacts with other brain structures in this process; these are subjects of future investigations.

### Delay Cells

Previous studies have shown that a substantial portion of units in the monkey dorsolateral PFC are ‘delay cells’, which increase their activities during the delay period of a delayed response task (Fuster and Alexander, 1971; Kubota and Niki, 1971; Funahashi et al., 1989). In the present study, 73 cells showed a significant rise in their activities during the delay period of the trace conditioning task. Of these cells, 17 changed their firing rates in association with the delay period only. These cells increased their firing rates at the end of the CS until expected delivery of the US (Fig. 6, top). One interesting aspect of our data is that activities of these delay cells returned to the baseline by the ‘expected’ delivery of the US. This indicates that an external sensory cue is not necessary for shaping delay cell activity in the rat mPFC. The rats were trained following a partial reinforcement schedule; thus they learned that the US is either delivered or not delivered 2 s after the offset of the CS. Because delay cell activities terminated at the time of the expected US delivery without an external sensory input, there must be an internal timing mechanism, which is generated by learning. One possibility is that modification of neural circuitry within the mPFC during conditioning, which could be accomplished by synaptic weight changes (Hirsch and Crepel, 1990), is responsible for the observed delay cell activities. On the other hand, no cell showed delay activity that was in perfect register with the delay period. Most delay cells maintained elevated activities 1–2 s past the expected delivery of the US, which may reflect how precisely the mPFC neural circuit can predict the time of US delivery.

### Effect of Context

The hippocampus plays an important role in context learning (Hirsh, 1974). Previous studies have shown that, following bilateral hippocampal lesions in the rat, tone-elicited freezing remained intact whereas freezing induced by context (exposure to conditioning box) was severely impaired (Selden et al., 1991; Kim and Fanslow, 1992; Phillips and LeDoux, 1992; Anagnostaras et al., 1999). Considering these results, it is likely that contextual information is conveyed from the hippocampus to the mPFC and induces changes in firing rates of mPFC neurons when the animals are exposed to the conditioning box. There are several different pathways through which the hippocampus can influence mPFC neural activities. First, it sends direct projections to the mPFC. In rats, pyramidal neurons of the ventral CA1 and subiculum send monosynaptic projections to the mPFC including prelimbic and infralimbic cortex, where recordings were made in the present study (Swanson, 1981; Ferino et al., 1987; Jay et al., 1991). The ventral hippocampus also projects to the amygdala, which in turn projects to the mPFC (Musil and Olson, 1988; Ino et al., 1990; Van Groen and Wyss, 1990; McDonald, 1991; Ray and Price, 1993; Gigg et al., 1994; Bacon et al, 1996; McDonald and Mascagni, 1997). Rats with bilateral lesions in the fornix, but not in the entorhinal cortex, were impaired in context-induced freezing (Phillips and LeDoux, 1995). These results suggest that hippocampal projections to subcortical areas convey contextual information that is sufficient to induce freezing behavior. Because amygdala lesion also impairs context-induced fear conditioning (Lee and Kim, 1998; Sacchetti et al., 1999), contextual information must be conveyed to the amygdala. Presumably both direct hippocampal projections and indirect projections via amygdala to the mPFC contribute to context-induced neural activity changes in the mPFC. The amygdala appears to be a ‘quick and dirty’ system that informs that something dangerous is out there or expected without providing detailed information about the source of danger (LeDoux, 1996). It is then likely that the amygdala sends a warning signal to the mPFC when a conditioned animal is exposed to the context that is associated with an aversive stimulus, whereas detailed information about the context is provided by direct hippocampal projections to the mPFC. It is interesting that direct hippocampal projections to the mPFC support NMDA receptor-dependent long-term potentiation (Jay et al., 1995). This will enable association between hippocampal contextual information and the US in the mPFC and enhance hippocampal influences over mPFC neurons. This projection alone may exert a significant influence over the mPFC after fear conditioning. Examination of mPFC unit activities following inactivation of the amygdala of a conditioned animal may provide an answer to this issue.

We cannot rule out potential involvement of other indirect pathways, such as those via the entorhinal cortex (EC), in context-induced changes in mPFC unit activities. Although EC lesion had no effect on context-induced freezing behavior (Phillips and LeDoux, 1995), this does not exclude the possibility that hippocampal information is conveyed to the mPFC via the EC. Also possible is that part of the contextual information reaches the mPFC through a pathway that does not include the hippocampus. For example, information about certain visual features of the conditioning box can be transmitted to the mPFC through projections from secondary visual cortices (Condé et al., 1995). This information may have contributed to the observed changes in mPFC unit firing when exposed to the conditioning box. Perhaps sensory information through sensory cortical projections, contextual information through hippocampal projections, warning signals through amygdaloid projections and other indirect projections all contribute, albeit to different degrees, to context-induced alterations in mPFC unit activities.

### Inhibitory versus Projection Neurons

Units recorded in the present study were classified into RS and FS cells, as in our previous study (Jung et al., 1998). RS and FS cells most likely represent pyramidal cells and inhibitory interneurons, respectively (McCormick et al., 1985). RS and FS cells showed complementary activity patterns. Whereas the majority (19/23) of FS cells increased their firing rates from the baseline after CS onset, RS cells decreased their firing rates following CS onset in most cases (105/173). In addition, FS cells tended to show transiently elevated responses (type 3; Figs 4 and 5), while a different pattern was observed for the RS cells. This is consistent with known physiological characteristics of the cortex. Inhibitory interneurons have a lower activation threshold than pyramidal neurons, and stimulation of afferent fibers induces powerful feedfoward and feedback inhibition (Fox and Ranck, 1981; Buzsaki and Eidelberg, 1982; Douglas et al., 1989). A significant afferent input, such as a CS-induced volley of sensory input, to the mPFC would initially activate inhibitory interneurons (FS cells), leading to initial elevation responses. In contrast, pyramidal neurons (RS cells) would initially be suppressed. Following termination of inhibitory postsynaptic potentials, the mPFC neural circuit would go through a sequence of changes, during which pyramidal cells and inhibitory neurons would show several types of responses such as those observed in the present study, for estimation of the time of US delivery and behavioral preparation. In the dorsolateral PFC of the monkey, responses of putative interneurons and pyramidal neurons were similar (Rao et al., 1999) or inverted (Wilson et al., 1994), depending on the distance between two neurons (Rao et al., 1999). In these studies, comparisons were focused on tuning profiles of RS and FS cells in different phases of delayed response tasks. In the present study, relative proportions of different response patterns across time were compared for RS and FS cells, thus a direct comparison of the present results with the previous studies is not possible.

### Relationship with Amygdala

It is well known that the amygdala plays a key role in fear conditioning. The amygdala sends direct as well as indirect projections, through the mediodorsal nucleus of the thalamus, to the PFC. The amygdala can also influence PFC neural activity by way of indirect projections to various modulatory centers, such as the basal forebrain cholinergic system (Pare and Smith, 1994). Stimulation of the basolateral nucleus of the amygdala inhibited the majority of units in the rat mPFC (Pérez-Jaranay and Vives, 1991). A recent inactivation study (Garcia et al., 1999) also has shown that amygdaloid projections reduce mPFC neural activity of the rat during fear conditioning. In the present study, the majority of FS cells (putative inhibitory interneurons) increased their firing rates whereas most RS cells (putative pyramidal cells) decreased their firing rates following CS onset. Interestingly, the proportion of inhibited and unresponsive mPFC units to amygdala stimulation is similar to the proportion of responsive and unresponsive mPFC units in this and a previous study (Maxwell et al., 1994). These similarities raise the possibility that the CS effects on mPFC unit activity are in large part mediated by amygdaloid projections to the mPFC. On the other hand, auditory CS information can reach the mPFC directly from the auditory cortex (Condé et al., 1995). Because mPFC neural circuits are plastic (Hirsch and Crepel, 1990), it is quite possible that CS information through this pathway, after conditioning, contributes to the observed changes in mPFC neuronal activities. Our recent results also indicate that sensory cortical projections to the mPFC support long-term potentiation that is dependent upon activation of NMDA receptors (Kim and Jung, 1999), suggesting the possibility of CS–US association in the mPFC. It is likely, as in the case of context-induced conditioning, that both amgdaloid and auditory cortical projections to the mPFC contribute to tone-induced changes in mPFC neuronal activities. It is difficult to conjecture relative contributions from the two pathways. Again, inactivation studies after conditioning probably help resolve this issue.

## Notes

This research was supported by the Korea Ministry of Science and Technology under the Brain Science Research Program, the Korea Science and Engineering Foundation grant through the Brain Disease Research Center at Ajou University, and the Korea Ministry of Science and Technology grant 97-N3-01–01-A-04 to M.W.J.

Address correspondence to Dr Min Whan Jung, Neuroscience Laboratory, Institute for Medical Sciences, Ajou University, Suwon 442-721, Korea. Email: min@madang.ajou.ac.kr.

Table 1

The numbers of animals and recorded units in each task

No. of animals No. of units
Standard conditioning  97
Trace conditioning 163
Effect of context  49
Control for CS effect  95
Control for context effect  19
Extinction  36
No. of animals No. of units
Standard conditioning  97
Trace conditioning 163
Effect of context  49
Control for CS effect  95
Control for context effect  19
Extinction  36
Figure 1.

Behavioral tasks and recording locations. (A) Behavioral tasks. The conditional stimulus (CS, an auditory tone) was delivered for 5 s and the unconditional stimulus (US, electric foot shock) was delivered for 0.5 s either during the last 0.5 s of the CS (standard conditioning) or 2 s after offset of the CS (trace conditioning). (B) Recording locations. Recordings were made in the superficial and deep layers of the prelimbic cortex (PL) and infralimbic cortex (IL) of the mPFC as shown in the coronal sectional views of the brain (front: 2.7 mm, back: 3.2 mm A to bregma).

Figure 1.

Behavioral tasks and recording locations. (A) Behavioral tasks. The conditional stimulus (CS, an auditory tone) was delivered for 5 s and the unconditional stimulus (US, electric foot shock) was delivered for 0.5 s either during the last 0.5 s of the CS (standard conditioning) or 2 s after offset of the CS (trace conditioning). (B) Recording locations. Recordings were made in the superficial and deep layers of the prelimbic cortex (PL) and infralimbic cortex (IL) of the mPFC as shown in the coronal sectional views of the brain (front: 2.7 mm, back: 3.2 mm A to bregma).

Figure 2.

Unit classification. (A) An average spike waveform, auto-correlogram and interspike interval histogram are shown for a typical regular spiking (RS) and fast spiking (FS) cells. Scale bar: 0.25 ms and 0.2 mV. (B) The relationship between average firing rate and peak–valley ratio. A significant negative correlation was observed. Cells were classified based on a clustering algorithm. The solid and open circles indicate FS and RS cells, respectively.

Figure 2.

Unit classification. (A) An average spike waveform, auto-correlogram and interspike interval histogram are shown for a typical regular spiking (RS) and fast spiking (FS) cells. Scale bar: 0.25 ms and 0.2 mV. (B) The relationship between average firing rate and peak–valley ratio. A significant negative correlation was observed. Cells were classified based on a clustering algorithm. The solid and open circles indicate FS and RS cells, respectively.

Figure 3.

Freezing behavior. The graph shows the amounts of animal movement during 10 s periods before and after CS onset of the control, standard conditioning and trace conditioning tasks in steps of 400 ms. The large shade indicates delivery of the CS; the small shade on the right indicates delivery of the US in the trace conditioning task. Data are mean ± SEM.

Figure 3.

Freezing behavior. The graph shows the amounts of animal movement during 10 s periods before and after CS onset of the control, standard conditioning and trace conditioning tasks in steps of 400 ms. The large shade indicates delivery of the CS; the small shade on the right indicates delivery of the US in the trace conditioning task. Data are mean ± SEM.

Figure 4.

Types of responses in the standard conditioning task. (A) Units recorded in the standard conditioning task were classified into three groups by a hierarchical clustering algorithm. The mean response profile (mean PSTH) for each group is shown on the right. The ordinate indicates the normalized firing rate (Z score). Time 0 indicates the onset of the CS. On the left, a spike raster of a typical example is shown for each group. Each tick mark indicates an incidence of unit discharge and each line represents one trial. Numbers (%) indicate the proportion of each response type. (B) Control task. The mean response profile (right) and an example raster plot (left) of a unit recorded in the control task are shown.

Figure 4.

Types of responses in the standard conditioning task. (A) Units recorded in the standard conditioning task were classified into three groups by a hierarchical clustering algorithm. The mean response profile (mean PSTH) for each group is shown on the right. The ordinate indicates the normalized firing rate (Z score). Time 0 indicates the onset of the CS. On the left, a spike raster of a typical example is shown for each group. Each tick mark indicates an incidence of unit discharge and each line represents one trial. Numbers (%) indicate the proportion of each response type. (B) Control task. The mean response profile (right) and an example raster plot (left) of a unit recorded in the control task are shown.

Figure 6.

Delay cells. Two examples are shown for the units that elevated their activities during the delay period of the trace conditioning task.

Figure 6.

Delay cells. Two examples are shown for the units that elevated their activities during the delay period of the trace conditioning task.

Figure 7.

Unit response to the contextual cues. Two examples of the units that were recorded in the context experiments are shown. The ordinate denotes firing rate of the units in 20 s steps. Units were initially recorded outside the conditioning box, inside the conditioning box (contextual cues) without CS delivery, then outside the conditioning box again.

Figure 7.

Unit response to the contextual cues. Two examples of the units that were recorded in the context experiments are shown. The ordinate denotes firing rate of the units in 20 s steps. Units were initially recorded outside the conditioning box, inside the conditioning box (contextual cues) without CS delivery, then outside the conditioning box again.

Figure 8.

Changes in unit response and freezing behavior during extinction. (A) An example. This unit was initially recorded from an animal that went through 60 trials of extinction. After six trials of CS–US pairing (arrow; 30% punishment rate of 20 trials), unit response to the CS changed markedly and the animal movement decreased. As the CS was presented repeatedly without pairing with the US (extinction), unit response to the CS decreased gradually and the animal movement (shown in 20 s steps) slowly increased. (B,C) Grouped data (n = 20) showing changes in unit responses and freezing behavior during extiction. The magnitudes of unit response (B) and animal movement (C) during 5 s periods before (pre-CS) and after (CS) the CS onset are shown separately. Unit responses in this figure represent absolute differences in mean firing rate between a given trial and the last trial. Each data point is a block of 4–8 trials. Data are mean ± SEM. Exp US: expected US.

Figure 8.

Changes in unit response and freezing behavior during extinction. (A) An example. This unit was initially recorded from an animal that went through 60 trials of extinction. After six trials of CS–US pairing (arrow; 30% punishment rate of 20 trials), unit response to the CS changed markedly and the animal movement decreased. As the CS was presented repeatedly without pairing with the US (extinction), unit response to the CS decreased gradually and the animal movement (shown in 20 s steps) slowly increased. (B,C) Grouped data (n = 20) showing changes in unit responses and freezing behavior during extiction. The magnitudes of unit response (B) and animal movement (C) during 5 s periods before (pre-CS) and after (CS) the CS onset are shown separately. Unit responses in this figure represent absolute differences in mean firing rate between a given trial and the last trial. Each data point is a block of 4–8 trials. Data are mean ± SEM. Exp US: expected US.

Figure 9.

Relationship between the type of units and response patterns. (A) Proportions of RS and FS cells that increased and decreased firing rates in response to the CS. (B) Relative distributions of RS and FS cells over three response patterns to the CS. Type I neurons sustained decreased firing rates during the entire 5 s of the CS delivery, type II neurons gradually increased firing rate during the CS delivery and type III neurons showed a transient elevation of discharge rate at the time of CS onset (see Figs 4 and 5). Type 4 of the trace conditioning task was excluded when pooling the responses of the standard and trace conditioning task.

Relationship between the type of units and response patterns. (A) Proportions of RS and FS cells that increased and decreased firing rates in response to the CS. (B) Relative distributions of RS and FS cells over three response patterns to the CS. Type I neurons sustained decreased firing rates during the entire 5 s of the CS delivery, type II neurons gradually increased firing rate during the CS delivery and type III neurons showed a transient elevation of discharge rate at the time of CS onset (see Figs 4 and 5). Type 4 of the trace conditioning task was excluded when pooling the responses of the standard and trace conditioning task.

## References

Anagnostaras SG, Maren S, Fanselow MS (
1999
) Temporally graded retrograde amnesia of contextual fear after hippocampus damage in rats: within-subjects examination.
J Neurosci

19
:
1106
–1114.
Arikuni T, Ban T Jr (
1978
) Subcortical afferents to the prefrontal cortex in rabbits.
Exp Brain Res

32
:
69
–75.
1996
) Amygdala input to medial prefrontal cortex (mPFC) in the rat: a light and electron microscopy study.
Brain Res

720
:
211
–219.
Baxter LR, Schwartz JM, Phelps ME, Mazziotta JC, Guze BH, Selin CE, Gerner RH, Sumida RM (
1989
) Reduction of prefrontal cortex glucose metabolism common to three types of depression.
Arch Gen Psychiat

46
:
243
–250.
Baxter LR, Schwartz JM, Guze BH, Bergman K, Szuba MP (
1990
) PET imaging in obsessive compulsive disorder with and without depression.
J Clin Psychiat

51
:
61
–69.
Berger TW, Laham RI, Thompson RF (
1980
) Hippocampal unit–behavior correlations during classical conditioning.
Brain Res

193
:
229
–248.
Buzsaki G, Eidelberg E (
1982
) Direct afferent excitation and long-term potentiation of hippocampal interneurons.
J Neurophysiol

48
:
597
–607.
Condé F, Maire-Lepoivre E, Audinat E, Crépel F (
1995
) Afferent connections of the medial frontal cortex of the rat. II. Cortical and subcortical afferents.
J Comp Neurol

352
:
567
–593.
Diorio D, Viau V, Meaney MJ (
1993
) The role of the medial prefrontal cortex (cingulate gyrus) in the regulation of hypothalamic–pituitary– adrenal responses to stress.
J Neurosci

13
:
3839
–3847.
Douglas RJ, Martin KAC, Witteridge D (
1989
) A canonical microcircuit for neocortex.
Neural Comp

1
:
480
–488.
Drevets WC, Videen TO, Price JL, Preskorn SH, Carmichael ST, Raichle ME (
1992
) A functional anatomical study of unipolar depression.
J Neurosci

12
:
3628
–3641.
Ferino F, Thierry AM, Glownski J (
1987
) Anatomical and eletrophysiological evidence for a direct projection from Ammon's horn to the medial prefrontal cortex in the rat.
Exp Brain Res

65
:
421
–426.
Fox SE, Ranck JB Jr (
1981
) Electrophysiological characteristics of hippocampal complex–spike cells and theta cells.
Exp Brain Res

41
:
399
–410.
Franzen EA, Myers RE (
1973
) Neural control of social behavior: prefrontal and anterior temporal cortex.
Neuropsychologia

11
:
141
–157.
Frysztak RJ, Neafsey EJ (
1994
) The effect of medial frontal cortex lesions on cardiovascular conditioned emotional responses in the rat.
Brain Res

643
:
181
–193.
Funahashi S, Bruce CJ, Goldman-Rakic PS (
1989
) Mnemonic coding of visual space in the monkey's dorsolateral prefrontal cortex.
J Neurophysiol

61
:
331
–349.
Fuster JM (
1973
) Unit activity in prefrontal cortex during delayed response performance: neuronal correlates of transient memory.
J Neurophysiol

36
:
61
–78.
Fuster JM (1997) The prefrontal cortex. New York: Lippincott-Raven.
Fuster JM, Alexander GE (
1971
) Neuron activity related to short-term memory.
Science

173
:
652
–654.
Garcia R, Vouimba RM, Baudry M, Thompson RF (
1999
) The amygdala modulates prefrontal cortex activity relative to conditioned fear.
Nature

402
:
294
–296.
Gewirtz JC, Falls WA, Davis M (
1997
) Normal conditioned inhibition and extinction of freezing and fear-potentiated startle following electrolytic lesions of medial prefrontal cortex in rats.
Behav Neurosci

111
:
712
–726.
Gibbs CM, Powell DA (
1991
) Single-unit activity in the dorsomedial prefrontal cortex during the expression of discriminative bradycardia in rabbits.
Behav Brain Res

43
:
79
–92.
Gigg J, Tan AM, Finch DM (
1994
) Glutamatergic hippocampal formation projection to prefrontal cortex in the rat are regulated by GABAergic inhibition and show convergence with glutamatergic projections from the limbic thalamus.
Hippocampus

4
:
189
–198.
Godefroy O, Rousseaux M (
1997
) Novel decision making in patients with prefrontal or posterior brain damage.
Neurology

49
:
695
–701.
Gorman JM, Liebowitz MR, Fyer AJ, Stein J (
1989
) A neuroanatomical hypothesis for panic disorder.
Am J Psychiat

146
:
148
–161.
Hirsch JC, Crepel F (
1990
) Use-dependent changes in synaptic efficacy in rat prefrontal neurons in vitro.
J Physiol (Lond)

427
:
31
–49.
Hirsh R (
1974
) The hippocampus and contextual retrieval of information from memory: a theory.
Behav Biol

12
:
421
–444.
Holson RR (
1986
) Medial prefrontal cortical lesions and timidity in rats. I. Reactivity to aversive stimuli.
Physiol Behav

37
:
221
–230.
Ino T, Matsuzaki S, Shinonaga Y, Ohishi H, Ogawa-Meguro R, Mizuno N (
1990
) Direct projections of non-pyramidal neurons of Ammon's horn to the amygdala and the entorhinal cortex.
Neurosci Lett

115
:
161
–166.
Inout M, Oomura Y, Aou S, Nishino H, Sikdar SK (
1985
) Reward related neuronal activity in monkey dorsolateral prefrontal cortex during feeding behavior.
Brain Res

326
:
307
–312.
Jay TM, Witter MP (
1991
) Distribution of hippocampal CA1 and subicular efferents in the prefrontal cortex of the rat studied by means of anterograde transport of Phaseolus vulgaris-leucoagglutin.
J Comp Neurol

313
:
574
–586.
Jay TM, Burette F, Laroche S (
1995
) NMDA receptor-dependent long-term potentiation in the hippocampal afferent fibre system to the prefrontal cortex in the rat.
Eur J Neurosci

7
:
247
–250.
Jinks AL, McGregor IS (
1997
) Modulation of anxiety-related behaviours following lesions of the prelimbic or infralimbic cortex in the rat.
Brain Res

772
:
181
–190.
Joel D, Tarrasch R, Feldon J, Weiner I (
1997
) Effects of electrolytic lesions of the medial prefrontal cortex or its subfields on 4-arm baited, 8-arm radial maze, two-way active avoidance and conditioned fear tasks in the rat.
Brain Res

765
:
37
–50.
Jung MW, Qin Y, McNaughton BL, Barnes CA (
1998
) Firing characteristics of deep layer neurons in prefrontal cortex in rats performing spatial working memory tasks.
Cereb Cortex

8
:
437
–450.
Kim JJ, Fanselow MS (
1992
) Modality-specific retrograde amnesia of fear.
Science

256
:
675
–677.
Kim M, Jung MW (
1999
) Long-term potentiation in rat prefrontal cortex induced by stimulation of occipital cortex and hippocampus.
Soc Neurosci Abs

25
:
1999
.
Komatsu H (
1982
) Prefrontal unit activity during a color discrimination task with GO and NO-GO responses in the monkey.
Brain Res

244
:
269
–277.
Kubota K, Niki H (
1971
) Prefrontal cortical unit activity and delayed alternation performance in monkeys.
J Neurophysiol

34
:
337
–347.
LeDoux JE (1996) The emotional brain. New York: Simon & Schuster.
LeDoux JE (
2000
) Emotion circuits in the brain.
Annu Rev Neurosci

23
:
155
–84.
Lee H, Kim JJ (
1998
) Amygdalar NMDA receptors are critical for new fear learning in previously fear-conditioned rats.
J Neurosci

18
:
8444
–8454.
1989
) Stimulation in prefrontal cortex area inhibits cardiovascular and motor components of the defence reaction in rats.
J Auton Nerv Syst

28
:
117
–126.
Mason ST, Fibiger HC (
1979
) Noradrenaline and avoidance learning in the rat.
Brain Res

161
:
321
–333.
Maxwell B, Powell DA, Buchanan SL (
1994
) Multiple- and single-unit activity in area 32 (prelimbic region) of the medial prefrontal cortex during Pavlovian heart rate conditioning in rabbits.
Cereb Cortex

4
:
230
–246.
McCormick DA, Connors BW, Lighthall JW, Prince DA (
1985
) Comparative electrophysiology of pyramidal and sparsely spiny stellate neurons of the neocortex.
J Neurophysiol

54
:
782
–806.
McDonald AJ (
1991
) Organization of amygdaloid projections to the prefrontal cortex and associated striatum in the rat.
Neuroscience

44
:
1
–14.
McDonald AJ, Mascagni F (
1997
) Projections of the lateral entorhinal cortex to the amygdala: a Phaseolus vulgaris leucoagglutinin study in the rat.
Neuroscience

77
:
445
–459.
McEchron MD, Disterhoft JF (
1997
) Sequence of single neuron changes in CA1 hippocampus of rabbits during acquisition of trace eyeblink conditioned responses.
J Neurophysiol

78
:
1030
–1044.
McEchron MD, Bouwmeester H, Tseng W, Weiss C, Disterhoft JF (
1998
) Hippocampectomy disrupts auditory trace fear conditioning and contextual fear conditioning in the rat.
Hippocampus

8
:
638
–646.
McNaughton BL, O'Keefe J, Barnes CA (
1983
) The stereotrode: a new technique for simultaneous isolation of several single units in the central nervous system from multiple unit records.
J Neurosci Methods

8
:
391
–397.
McNaughton BL, Barnes CA, Meltzer J, Sutherland RJ (
1989
) Hippocampal granule cells are necessary for spatial learning but not for spatially selective pyramidal cell discharge.
Exp Brain Res

76
:
485
–496.
Morgan MA, LeDoux JE (
1995
) Differential contribution of dorsal and ventral medial prefrontal cortex to the acquisition and extinction of conditioned fear in rats.
Behav Neurosci

109
:
681
–688.
Morgan MA, Romanski LM, LeDoux JE (
1993
) Extinction of emotional learning: contribution of medial prefrontal cortex.
Neurosci Lett

163
:
109
–113.
Musil SY, Olson CR (
1988
) Organization of cortical and subcortical projections to medial prefrontal cortex in the cat.
J Comp Neurol

272
:
219
–241.
Ono T, Nishino H, Fukuda M, Sasaki K, Nishijo H (
1984
) Single neuron activity in dorsolateral prefrontal cortex of monkey during operant behavior sustained by food reward.
Brain Res

311
:
323
–332.
Pare D. Smith Y (
1994
) GABAergic projection from the intercalated cell masses of the amygdala to the basal forebrain in cats.
J Comp Neurol

344
:
33
–49.
Pérez-Jaranay JM, Vives F (
1991
) Electrophysiological study of the response of medial prefrontal cortex neurons to stimulation of the basolateral nucleus of the amygdala in the rat.
Brain Res

564
:
97
–101.
Phillips RG, LeDoux JE (
1992
) Differential contribution of amygdala and hippocampus to cued and contextual fear conditioning.
Behav Neurosci

106
:
274
–285.
Phillips RG, LeDoux JE (
1995
) Lesions of the fornix but not the entorhinal or perirhinal cortex interfere with contextual fear conditioning.
J Neurosci

15
:
5308
–5315.
Rao SG, Williams GV, Goldman-Rakic PS (
1999
) Isodirectional tuning of adjacent interneurons and pyramidal cells during working memory: evidence for microcolumnar organization in PFC.
J Neurophysiol

81
:
1903
–1916.
Ray JP, Price JL (
1993
) The organization of projections from the mediodorsal nucleus of the thalamus to orbital and medial prefrontal cortex in macaque monkeys.
J Comp Neurol

337
:
1
–31.
Recce ML, O'Keefe J (
1989
) The tetrode: an improved technique for multi-unit extracellular recording.
Soc Neurosci Abs

15
:
1250
.
Rosen JB, Hitchcock JM, Miserendino MJD, Falls WA, Campeau S, Davis M (
1992
) Lesions of the perirhinal cortex but not of the frontal, medial prefrontal, visual, or insular cortex block fear-potentiated startle using a visual conditioned stimulus.
J Neurosci

12
:
4624
–4633.
Sacchetti B, Lorenzini CA, Baldi E, Tassoni G, Bucherelli C (
1999
) Auditory thalamus, dorsal hippocampus, basolateral amygdala, and perirhinal cortex role in the consolidation of conditioned freezing to context and to acoustic conditioning stimulus in the rat.
J Neurosci

19
:
9570
–9578.
Selden NR, Everitt BJ, Jarrard LE, Robbins TW (
1991
)
Complementary roles for the amygdala and hippocampus in aversive conditioning to explicit and contextual cues Neuroscience

42
:
335
–350.
Siegel A, Edinger H, Dotto M (
1975
) Effects of electrical stimulation of the lateral aspect of the prefrontal cortex upon attack behavior in cats.
Brain Res

93
:
473
–484.
Swanson LW (
1981
) A direct projection from Ammon's horn to prefrontal cortex in the rat.
Brain Res

217
:
150
–154.
Van Groen T, Wyss JM (
1990
) Extrinsic projections from area CA1 of the rat hippocampus: olfactory, cortical, subcortical, and bilateral hippocampal formation projection.
J Comp Neurol

302
:
515
–528.
Wilson MA, McNaughton BL (
1993
) Dynamics of the hippocampal ensemble code for space.
Science

261
:
1055
–1058.
Wilson FA, O'Scalaidhe SP, Goldman-Rakic PS (
1994
) Functional synergism between putative gamma-aminobutyrate containing neurons and pyramidal neurons in prefrontal cortex.

91
:
4009
–4913.
Zbrozyna AW, Westwood DM (
1991
) Stimulation in prefrontal cortex inhibits conditioned increase in blood pressure and avoidance bar pressing in rats.
Physiol Behav

49
:
705
–708.