Abstract

The medial temporal lobe (MTL) plays a critical role in transforming complex stimuli into permanent memory traces, yet little is known on how the activity of neurons in the human brain mediates this process. Recording from single neurons in the human MTL during visual encoding and retrieval of faces and objects, we found that in the hippocampus faces evoked predominantly suppression of neuronal firing below prestimulus baseline (‘inhibitory responses’). These responses were also prevalent in the entorhinal cortex but were absent in the amygdala during the first second of stimulus encoding when all responses to faces were ‘excitatory’ (neuronal firing increased above the prestimulus baseline). Inhibitory responses were more prevalent during recognition than encoding and were also present during processing of objects, albeit less frequently than during processing of faces. Despite the prevalence in the hippocampus of cells with inhibitory responses and their relative specificity to faces, it was mainly the activity of the cells with excitatory responses that was selective for stimulus features such as gender and emotional expression of faces. These findings suggest that a large population of cells with inhibitory responses is engaged in the hippocampal memory network, but primarily cells with excitatory responses process feature-specific information.

Introduction

The human hippocampus and neighboring structures in the medial temporal lobe (MTL) are at the pinnacle of the visual information processing hierarchy (Felleman and Van Essen, 1991). These areas play a critical role in transforming complex stimuli into permanent memory traces (Scoville and Milner, 1957; Eichenbaum and Otto, 1992; Squire, 1992), yet little is known about how neurons in the different MTL subdivisions process and represent information during the various operations which constitute memory (Milner et al., 1998). Although much insight has been gained from rodent and primate electrophysiological studies, from human and animal lesion data (Scoville and Milner, 1957; Zola-Morgan and Squire, 1990; Eichenbaum and Otto, 1992; Squire, 1992; Milner et al., 1998), and from functional neuroimaging studies (Brewer et al., 1998; Wagner et al., 1998; Dolan and Fletcher, 1999; Fernández et al., 1999; Schachter and Wagner, 1999), the differential role of these regions in encoding versus recognition of visual information remains controversial (Suzuki, 1999). Functional MRI studies have shown responses of these regions during both encoding and recognition. Based on these studies, suggestions that different MTL structures are functionally specialized for the memory processes of encoding and retrieval have been recently put forward (Stern et al., 1996; Gabrieli et al., 1997; Lepage et al., 1998; Schacter and Wagner, 1999; Aggleton and Brown, 1999; Brown and Aggleton, 2001). However, activation patterns in the hippocampus and environs have often been difficult to interpret (Stern and Hasselmo, 1999). At the same time, single unit studies in animals that show changes in neuronal firing during encoding and retrieval have been limited as they cannot examine conscious recollection.

In this study we examined the activity of neurons in the human hippocampus, amygdala and entorhinal cortex during encoding and retrieval of complex visual stimuli. We utilized a rare opportunity to perform single unit recordings in awake and conscious subjects who were able to declare their memories. We asked the following questions: Are there differences in the firing characteristics of single neurons during encoding compared with recognition and are there differences among the hippo-campus, amygdala and entorhinal cortex with respect to the neuronal activity during these memory processes? In this study we show that an important clue can be found in the distribution of excitatory and inhibitory responses of single neurons in the various regions of the MTL.

Materials and Methods

Subjects

We conducted single unit recording in 20 patients with intractable epilepsy who underwent implantation of intracranial electrodes in order to identify the seizure focus for possible surgical resection. All patients except one were right-handed. Twelve patients underwent the sodium amobarbital test (Wada and Rasmussen, 1960). Two were found to have right-sided dominance for language (including the left-handed patient), nine had left-hemisphere dominance for language and one had bilateral representation of language function. All eight patients who did not have the Wada test were right-handed, and for comparison of neurons in the dominant versus nondominant hemisphere it was assumed that in these patients the left hemisphere was dominant for language function.

Electrode Placement

The technique of electrode placement and single unit recordings in these patients has been described in detail in previous publications (Fried et al., 1997, 1999; Kreiman et al., 2000a,b). The sites of electrode implantation were selected based exclusively on clinical criteria. Electrodes were placed stereotactically, guided by magnetic resonance imaging (MRI) and cerebral angiography. Targets included the hippocampus, entorhinal cortex and the basolateral nuclear group of the amygdala (Fig. 1). Each electrode consisted of a flexible polyurethane probe containing nine 40 μm platinum–iridium microwires protruding ~4 mm into the tissue beyond the tip of the electrode. Following electrode placement the patient was monitored on the ward for a period of 1–3 weeks until a sufficient number of spontaneous seizures had been recorded. MRI confirmed the locations of the electrodes (Fig. 1).

Tasks

Subjects underwent two sessions of memory testing (Fig. 2). In the first session (‘encoding’) faces and objects were presented to be remembered. The faces were those of eight actors each depicting seven different emotional expressions (Eckmann and Friesen, 1976). Objects were single common household items. Fifty-six faces and 42 objects were presented in an intermixed fashion in a single list that was presented once (in four patients) or twice (in 16 patients). Each stimulus was presented for a duration of 1 s, with an interstimulus delay of 4–5 s. One to 12 h later subjects underwent a ‘recognition’ session where they were shown a series of 165 stimuli (94 faces and 71 objects), which included previously presented stimuli as well as novel faces and objects. Each stimulus was presented for 1 s before a question mark appeared below the picture, signaling the subject to respond with a key press (‘yes’ vs ‘no’) whether they had seen the face or object during the encoding session (Fried et al., 1997). The research was conducted according to protocol approved by an Institutional Review Board, with subject informed consent.

Recording

Data from each of the recorded microwires were amplified and high-pass filtered (300 Hz to 10 kHz), A/D converted and stored for spike sorting using Experiment Workbench data acquisition software (Datawave, Denver, CO, USA). Spike sorting was carried out off-line as previously described (Fried et al., 1997, 1999). Spikes from single neurons were discriminated from the extracellular recordings based on action potential amplitude, duration, slope and other parameters of waveform morphology, using a manual cluster cutting method implemented in Datawave. Nevertheless, it should be pointed out that some units could possibly have had signals from more than one cell.

It is important to note that the recognition task was carried out 1–12 h after the encoding task. It is therefore likely that some cells recorded during encoding were ‘lost’, and other cells that were not detected during encoding became active during the recognition task. We therefore made no assumption of recording from any neuron during both encoding and recognition. Also, our technique picks up only stimulus-related changes and not task-specific changes, which would require direct comparison of encoding versus recognition in the same neuron. Such direct comparison requires different paradigms (Cameron et al., 2001).

Data Analysis

For data analysis, spike counts were obtained for 100 ms bins and collapsed across 1 s episodes: the second preceding stimulus onset (T0), the second of stimulus presentation (T1), and the following second (T2). The choice of 1 s is based on long-term potentiation (LTP) and episode formation timescale as discussed by Trevis and Rolls (Trevis and Rolls, 1994). We have previously shown that stimulus-selective responses of medial temporal neurons can be discerned beyond the duration of stimulus presentation (Fried et al., 1997). The response of each cell to a stimulus during a specific task condition was evaluated by a two-tailed t-test comparing the spike counts during T1 or T2 against the preceding baseline (T0). A criterion level of P = 0.05 was used. The response was termed ‘excitatory’ when it was significantly higher than the prestimulus baseline, and ‘inhibitory’ when it was lower.

A neuron was considered ‘selective’ for a particular stimulus category (i.e. faces or objects) if (i) the activity during T1 or T2 was significantly different from baseline activity, (ii) the neuron's activity in response to the other category was not significantly different from baseline (in the same direction), and (iii) the response to the preferred category was significantly different from the response to the other stimulus category (by a two-tailed t-test).

For each unit, the responses to faces (discharge rate during T1 or T2, corrected for the preceding 1 s baseline) were subjected to analysis of variance (ANOVA) to evaluate the relationship between the response and stimulus features, i.e. emotional expression and gender of faces. We considered a unit discriminant if one of the factors (i.e. gender or expression) or their interaction was significant (P < 0.05) during T1 and/or T2. The inclusion of T2 data was based on the observation that MTL neurons respond to specific stimulus features well beyond the first 1000 ms of stimulus presentation (Fried et al., 1997).

To evaluate the relationship between discharge rate and recognition performance, the data for each stimulus were segregated into one of two bins: correct recognition and incorrect recognition. For neuronal activity during retrieval this simply meant looking at the activity of the neuron during correct or incorrect recognition. For activity during encoding, the responses had to be classified into one of two bins: responses to stimuli that were later successfully recognized (during retrieval) and responses to those stimuli that were not subsequently recognized. A separate repeated measure ANOVA was carried out to evaluate the relationship between discharge rate (during encoding or recognition) and the performance during recognition (i.e. correct vs incorrect key responses). The relationship between firing rate and performance was considered significant if performance or its interaction with stimulus type (faces or objects) and/or time was significant. A neuron was considered performance-selective during recognition if (i) the response (during T1 and/or T2) of the neuron during correct performance was different from the response during incorrect performance and (ii) the response of the neuron was significant either during correct retrieval or during incorrect retrieval, or during both if the responses were of opposite sign. For the neuron to be considered ‘performance selective’ during encoding (i.e. ‘performance predictive’) we applied these same criteria for the activity of the neuron during stimulus encoding based on the subsequent performance of recognizing those stimuli during retrieval.

The proportion of units in a region showing relevant responses were compared across regions, stimuli or tasks, using chi-square methods with exact permutational P values (Statxact, Cytel Inc., Cambridge, MA, USA). McNemar's test was used to compute P values for paired comparisons. For partially paired comparisons, a logistic model allowing correlations was used (SAS procedure GENMOD, SAS Inc., Cary, NC, USA).

It should be pointed out that generalization about normal neuronal function from recordings in patients with epilepsy is a potential limitation; however, most of the units recorded in this study (91.5%) were contralateral to or outside the focus of seizure onset. For responsive neurons (to faces), spontaneous firing rates were compared between the hemisphere with the epileptogenic region and the contralateral hemisphere, and there were no significant differences (2.9 and 4.2 Hz for the epileptogenic and normal hemispheres respectively; P = 0.201). There were also no significant hemispheric differences in the firing rates of these neurons during T1 or during T2 with respect to baseline between the two hemispheres (P = 0.42 and 0.89 for T1 and T2, respectively, using independent t-test comparison).

Results

We recorded the activity of 225 units during the encoding session (46 in the amygdala, 97 in the hippocampus and 82 in the entorhinal cortex) and 177 units during the recognition session (35 in the amygdala, 73 in the hippocampus and 69 in the entorhinal cortex).

Encoding of Faces

During encoding of faces 25.3% of the neurons responded significantly to faces in the second of stimulus presentation (T1). There was no significant difference in the percentage of responding units in the amygdala (28.3%), hippocampus (24.7%) and entorhinal cortex (24.4%; χ2 = 0.26, d.f. = 2, P = 0.88). However, there were significant differences in the proportions of units responding with increased firing rate (‘excitatory responses’) and decreased firing rate (‘inhibitory responses’) (Fig. 3, Table 1). In the amygdala faces evoked only excitatory responses (Fig. 4b), whereas in the hippocampus and entorhinal cortex, 62.5% and 50% of the responding neurons, respectively, were inhibited by face stimuli (χ2 = 13.85; d.f. = 2; P = 0.001; examples in Fig. 4a). Hippocampal neurons typically responded with a decrease in activity, or even complete ‘silence’, that often outlasted stimulus presentation by several hundred milli-seconds (46.7% of inhibited neurons; Fig. 6c). As for the entorhinal cortex, only in 10% of neurons inhibited by faces did suppression of activity outlast stimulus presentation. The predominance of excitatory responses in the amygdala during encoding (13/46 excitatory and 0 inhibitory) was present only during the first second of stimulus presentation. Later (during T2) there were less neurons responding to faces, and both inhibitory and excitatory responses were observed (Table 1).

Of note is that there was no difference between the neurons with excitatory and inhibitory responses in the mean firing rate throughout the entire duration of the experiment (P > 0.5; Mann–Whitney test for comparing the distributions of firing rates). There was also no difference between these two neuronal groups in discharge frequency during the prestimulus baseline alone (T0). During encoding of faces the mean prestimulus baseline firing rate was 3.21 spike/s for the cells with excitatory responses and 4.17 spike/s for the cells with inhibitory responses (P > 0.6).

Recognition of Faces

During recognition, 32.2% of MTL neurons responded to faces. The proportion of neurons responding to faces during recognition was greater in the hippocampus (41.1%) and entorhinal cortex (33.3%) than in the amygdala (11.4%; χ2 = 9.8; d.f. = 2; P = 0.0075; Fig. 3). The proportion of responsive hippocampal neurons was greater during recognition than during encoding (41.1% vs 24.7%; χ2 = 5.14, P = 0.023), an increase that can be accounted for by the rise in the number of neurons with inhibitory responses. The percentage of hippocampal neurons with inhibitory responses doubled during recognition (31.5% of all hippocampal neurons) compared to encoding (15.5%, Fig. 3). Considering only the responding hippocampal units, 76.7% had inhibitory responses. A similar pattern, although less robust, was apparent in the entorhinal cortex (65.2%).

Encoding of Objects

During encoding, 25.3% of the cells responded significantly to object stimuli (during the second of stimulus presentation), the same percentage of cells as during encoding of faces (Fig. 3). Only 8% of the units responded significantly to both faces and objects. Of MTL cells responding to objects, a substantial proportion (33.3%) had decreased firing rate below prestimulus baseline. The difference between encoding of faces and objects was not in the total number of responding cells, but in the distribution of cells with excitatory and inhibitory responses. In the hippocampus and entorhinal cortex the relative prevalence of neurons with inhibitory responses was higher for faces (e.g. in the hippocampus, 62.5% and 27.3% for faces and objects, respectively; χ2 = 6.20, d.f. = 1, P = 0.013; Fig. 3).

Recognition of Objects

During recognition, 35% of the units responded to objects, a proportion not significantly different from that responding to faces (32.2%). As noted for faces, the increase in proportion of units responding during recognition compared to encoding could be accounted for by a two-fold increase in number of neurons with inhibitory responses (8.4% during encoding and 16.4% during retrieval) (Fig. 3). As in encoding, the percentage of cells with inhibitory responses was higher for faces than objects in the hippocampus and entorhinal cortex (e.g. in the hippocampus 76.7% and 43.5% for faces and objects, respectively; χ2 = 8.45, d.f. = 1, P = 0.0036), whereas in the amygdala fewer cells responded to faces than objects (11.4% vs 40%; χ2 = 6.73, d.f. = 2, P = 0.0094).

We found no significant difference between the left and right side of the brain in the proportion of cells with excitatory, inhibitory and null responses in any of the three regions studied.

Selective Responses

We considered a neuron ‘selective’ for a particular stimulus category (i.e. faces or objects) if (i) the activity during the first second of stimulus presentation (T1) or the ensuing second (T2) was significantly different from the baseline activity, (ii) the neuron's activity in response to the other category was not significantly different from baseline (in the same direction), and (iii) the response to the preferred category was significantly different from the response to the other stimulus category. During encoding there were altogether 32 MTL neurons (14.2%) that fulfilled these stringent criteria: 12 selective to faces, 18 to objects and two responding to both but with opposite responses (inhibitory to faces, excitatory to objects). All the selective responses to faces in the hippocampus and entorhinal cortex were inhibitory (nine neurons) and in the amygdala were excitatory (five neurons; example Fig. 4b). The responses to objects in the hippocampus and entorhinal cortex were mostly excitatory (14 of 16 neurons), and in the amygdala all inhibitory (although only three neurons responded).

During recognition there were 32 selective units (18%), seven selective for faces, 15 for objects, and 10 responding with opposite signs to these two categories [eight of these had inhibitory responses to faces and excitatory to objects (Fig. 5a), and two responded with the opposite pattern (Fig. 5b)]. In the hippocampus and entorhinal cortex, 12 neurons responded to faces with inhibition of firing rate and three with excitation, whereas 22 responded to objects with excitation and only one with inhibition. Only two neurons in the amygdala responded selectively during recognition.

Reponses to Stimulus Features

We next examined the relationship of the neuronal responses, excitatory and inhibitory, during presentation of faces to the specific attributes of the faces, i.e. gender and emotional expression. We performed analysis of variance on the discharge rates of all the neurons that responded significantly to faces during T1 or T2 (see Materials and Methods). There were 73 such neurons, 38 with excitatory responses and 35 with inhibitory responses (not all were selective to faces only). Of these 73 neurons, 19 discriminated between facial expressions (Fig. 2) or the gender of faces (Fig. 6b) or between conjunctions of gender and expression. These neurons were considered ‘discriminative’ neurons. Of the 38 neurons with excitatory responses 15 (39.5%) were discriminative cells, while of the 35 neurons with inhibitory responses only four (11.4%) were discriminative (Fig. 6a; see Fig. 6c for an example of a nondiscriminative cell). Thus the proportion of discriminative cells was higher for neurons with excitatory responses compared to those with inhibitory responses (χ2 = 7.44; P = 0.008). Similar proportions were observed for the hippocampal neurons alone (46.7% and 15.8% for the cells with excitatory and inhibitory responses, respectively).

Recognition Performance

Is the neuronal activity during recognition related to the subject's performance? We considered the neuron activity ‘performance selective’ if (i) the response (during T1 and/or T2) of the neuron during correct performance was different from the response during incorrect performance and (ii) the response of the neuron was significant either during correct retrieval or during incorrect retrieval (or during both if the responses were of opposite sign). During recognition of faces there were 26 (14.7%) ‘performance selective’ units, most of them (19) with inhibitory responses. The majority of these units responded during correct performance and not during incorrect performance (12 of 19 units). Seven units had the opposite pattern. Only seven performance-selective neurons had excitatory responses, and five of these responded during incorrect performance. No units exhibited opposite sign responses for correct and incorrect performance, i.e. excitatory response for correct responses and inhibitory for incorrect responses or vice versa. Object recognition showed a similar but less striking pattern of performance-related responses. Of 30 ‘performance-selective’ units, 17 had inhibitory responses (10 during correct performance), and 13 had excitatory responses (only six during correct performance).

Were the unit responses during encoding related to the subsequent performance on the recognition task? For the neuron to be considered ‘performance selective’ (i.e. ‘performance predictive’) during encoding of stimuli, we applied the same criteria mentioned above (see also Materials and Methods section) to the activity of the neuron during encoding based on the subsequent performance on recognition of these stimuli. There were 21 such units for encoding of faces and 21 for encoding of objects. The majority of these units had excitatory responses (for faces, 12 excitatory and 9 inhibitory; and for objects, 13 excitatory, six inhibitory and two inhibitory for correct and excitatory for incorrect performance). There were no significant differences between the number of units responding to items that were subsequently recognized and of those responding to items that were not recognized.

Neuronal Responses in Dominant versus Nondominant MTL

Of the 225 neurons recorded during encoding, 88 were in the dominant hemisphere, 117 in the nondominant hemisphere and 20 in a patient with bilateral hemispheric dominance. Of the 88 dominant MTL neurons 35 (40%) were responsive to faces during either T1 or T2 (24% excitatory and 16% inhibitory), whereas of the 117 nondominant neurons 34 (29%) were responsive to faces (13% excitatory and 16% inhibitory). There was no significant difference between the proportions of neurons responding to faces in the dominant versus non-dominant hemisphere (χ2 = 2.58; P = 0.14). In the nondominant hemisphere a relatively uniform sampling of the regions was obtained (38 amygdala units, 42 hippocampal units and 37 units from entorhinal cortex), thus enabling a comparison of the distribution of excitatory and inhibitory responses among the regions.

Altogether, 32 cells in the nondominant temporal lobe responded to faces (not all were selective to faces only). In these neurons a pattern of predominantly excitatory responses to faces in the amygdala and inhibitory responses in the hippocampus and entorhinal cortex (recorded during T1 and/or T2) was evident. In the amygdala 12 of the 14 responding cells had excitatory responses and two had inhibitory responses. In the hippocampus only two out of the 12 responding neurons showed excitatory responses and 10 cells showed inhibitory responses. In entorhinal cortex only one of eight responding cells showed excitatory responses and seven had inhibitory responses [overall χ2 = 16.74, d.f. = 2, P = 0.0002; for pairwise comparison there are significant differences in the distribution of excitatory and inhibitory responses between the amygdala and hippocampus (χ2 = 12.4, P = 0.001) and between the amygdala and entorhinal cortex (χ2 = 11.3; P = 0.0015) but not between the hippocampus and entorhinal cortex)].

Discussion

Inhibitory Responses in the Hippocampus

The main finding in this study is the striking prevalence of stimulus-related inhibitory responses in the hippocampus and to a lesser degree in the entorhinal cortex during the recognition memory paradigm. These responses were more prevalent for faces compared to objects and during retrieval compared to encoding. The hippocampus is known to have an extensive inhibitory network, and the pyramidal cells are influenced by inhibitory local interneurons as well as by extrahippocampal inhibitory input [see Buhl et al. (Buhl et al., 1994), and for a review, Freund and Buzaki (Freund and Buzaki, 1996)]. Yet to our knowledge the abundance of inhibitory responses found in our recordings in the human hippocampus has not been observed in cognitive paradigms in the monkey hippocampus. These inhibitory responses reflect either a reduction in excitation or an increase in inhibition of the responding cells. A reduction in excitatory responses has been described in nonhuman primate studies upon repeated stimulus presentation (‘match suppression’), and is thought to represent stimulus recognition (Desimone, 1996; Suzuki, 1999). However, these responses were diminished compared to the first response, whereas the inhibitory responses observed in the present study involved reduction of firing rate below prestimulus baseline. Often the neurons were virtually ‘shut down’ for several hundred milliseconds following stimulus presentation (Fig. 6c). This observation is of particular interest since inhibitory responses have received less attention than excitatory responses in cognitive studies of non-human primates, although such responses have been described in human temporal neocortex during language measures [for a review, see Ojemann et al. (Ojemann et al., 1998)]. While the abundance of inhibitory responses during the memory task could be unique to the human hippocampus and entorhinal cortex, another plausible explanation for the discrepancy is methodological, namely a sampling bias in many animal extracellular recordings, where the electrode is commonly moved to find cells with increased firing rate using audio monitoring (Miller et al., 1993). In contrast, for clinical reasons, the microelectrodes in the present study were fixed and could not be moved in search of an optimal excitatory response. It is possible, however, that the prevalence of inhibitory responses is a function of the epileptic brain. Although we have not seen differences in prevalence of inhibitory responses between the seizure focus and other areas or even between neurons in the hemisphere which included the seizure focus compared with the other hemisphere, it is still possible that these patients with intractable seizures have a disturbance of normal inhibitory function (Engel, 1996) even outside the area of the seizure focus.

Selectivity of Responses

There is evidence that neurons in the human MTL respond selectively to particular stimuli (Heit et al., 1988, 1990), to stimulus categories (Fried et al., 1997; Kreiman et al., 2000a,b) and to specific stimulus features such as the gender and emotional expression of faces (Fried et al., 1997). The present study confirms these specific responses with respect to faces and objects. Single unit recordings in the monkey MTL have shown selective responses to stimulus categories and stimulus features. Neurophysiological studies of nonhuman primates have shown selective responses of amygdalar neurons to faces (Rolls, 1981, 1984; Leonard et al., 1985; Nakamura et al., 1992) and to objects of biological significance [see the review by Ono et al. (Ono et al., 1993)]. Several animal studies reported coding of stimulus and task variables and their conjunctions by single hippocampal neurons (Cahusac et al., 1989; Rolls et al., 1989; Ono et al., 1991; Eifuku et al., 1995). Suzuki et al. report that 11% of the neurons in monkey entorhinal cortex show selective responses (Suzuki et al., 1999). What is only rarely addressed in these studies is the issue of selectivity of inhibitory and excitatory responses.

We show that the inhibitory responses observed in the hippocampus and entorhinal cortex show relative specificity to faces. The overwhelming majority of the selective responses to faces during the memory task are inhibitory. Interestingly, the majority of specific responses to objects in this study were excitatory. At the same time, the more selective cells, i.e. those with selective responses to specific features of faces, were predominantly excitatory. These cells could be pyramidal neurons in the hippocampus, which in animal studies have been shown to have highly selective responses [e.g. ‘place cells’ in rodent hippocampus (O'Keefe et al., 1998)]. Conversely, the cells with inhibitory responses may be hippocampal interneurons (Buhl et al., 1994; Freund and Buzaki, 1996), or pyramidal cells that are strongly influenced by inhibitory neurons. Support for selective responses of interneurons, albeit with broader tuning than pyramidal cells, has been reported in studies of primate prefrontal cortex (Rao et al., 1999, 2000). It is important to point out though, that suggestions about the identity of the cells recorded in our study must be viewed with caution, since distinguishing between pyramidal cells and interneurons, based on extracellular recordings, is difficult. Several criteria, including discharge frequency, spike duration and the autocorrelation function, have been used in attempts to separate pyramidal cells from interneurons in the rodent hippocampus [for a review, see Csicsvari et al. (Csicsvari et al., 1999)]. In the present study, we have not seen any consistent difference between the cells with inhibitory and excitatory responses in terms of discharge frequency, spike duration or propensity for bursting.

Encoding versus Retrieval

The differential contribution of various MTL structures in encoding versus retrieval has been a major focus of neuroscience research of declarative memory. Based on neuroimaging studies, various suggestions have been made that anterior or posterior MTL regions (Stern et al., 1996; Lepage et al., 1998; Schacter and Wagner, 1999) or different MTL structures (Gabrieli et al., 1997; Aggleton and Brown, 1999; Brown and Aggleton, 2001) are functionally specialized for the memory processes of encoding and retrieval. There have also been suggestions of a differential role of the hippocampus in episodic memory and of para-hippocampal and perirhinal cortex in recognition memory (Vargha-Khadem et al., 1997; Aggleton and Brown, 1999; Eldridge et al., 2000) [for a review, see Brown and Aggleton (Brown and Aggleton 2001)]. Our study examined recognition memory and did not address episodic memory. The results support the importance of the MTL in both encoding and recognition, while also suggesting differences in the neuronal mechanisms involved in these processes. Similar to our previous findings with word encoding (Cameron et al., 2001), and in line with reports from fMRI studies (Brewer et al., 1998; Wagner et al., 1998), we found that the activity of some MTL neurons during encoding of faces and objects predicted subsequent performance of the subject during retrieval. The majority of these neurons had excitatory responses. During recognition, more neurons with inhibitory response were recruited, especially in the hippocampus and entorhinal cortex, suggesting a more prominent role for inhibition during recognition than during encoding. Overall, we found that the activity of ~15% of MTL neurons during recognition of faces reflected the subject's performance. Most of these neurons had inhibitory responses, more commonly during correct rather than incorrect performance. It is also noteworthy that in a previous study of human MTL neurons which employed a perceptual discrimination rather than a memory task the prevalence of inhibitory responses in the hippocampus was less prominent (Kreiman et al., 2000a). Yet, in a recent paired associate memory task, we found that the majority (60%) of significant responses of hippocampal neurons to word stimuli were inhibitory (Cameron et al., 2001).

While the present study corroborates fMRI findings demonstrating MTL activation during encoding and retrieval [see, among others (Gabrieli et al., 1997; Dolan and Fletcher, 1999; Schacter and Wagner, 1999)], it also raises the question of how changing patterns of excitatory and inhibitory responses may be translated into hemodynamic responses. In that light, it is perhaps not surprising that early attempts at measuring activation of hippocampus by functional neuroimaging have met with difficulty (Ungerleider, 1995; Schacter and Wagner, 1999), and that some studies have suggested the functional importance of decreases in activation (Andreasen et al., 1995; Buckner and Tulving, 1995; Nyberg et al., 1996). Recent studies have also pointed out the complexity of establishing a ‘baseline’ of MTL activation, raising the possibility that stimulus-related changes during the task may represent less excitation than during the baseline (Eldridge et al., 2000; Stark and Squire, 2000, 2001).

Regional and Hemispheric Differences

In contrast to the sparse excitatory representation of faces in the hippocampus (and to a lesser extent the entorhinal cortex), the amygdala had a robust excitatory code with no inhibitory responses during the first second of face presentation. Moreover, when the category-selective responses are considered for both T1 and T2 all the selective responses to faces were excitatory (and all responses to objects inhibitory). The opposite pattern was observed in the hippocampus (all responses to faces inhibitory and all responses to objects excitatory). However, these results should be considered preliminary, since the number of selective neurons is small (~12% of all 225 recorded neurons during encoding), and this is particularly true for the amygdala where the number of sampled units was relatively small compared to the hippocampus and entorhinal cortex. Comparison of the responding neurons in the nondominant hemisphere where sampling across the three MTL regions was more uniform, confirms the pattern of excitatory responses to faces in amygdala and inhibitory responses in the hippocampus during encoding.

Excitatory coding for faces in the amygdala during encoding may reflect responses to the affective content of the facial expressions, confirming the wealth of evidence that the amygdala functions as a fast conduit for reaction to emotionally charged stimuli (LeDoux, 1998) and a modulator of declarative memory for emotionally arousing events (Cahill and McGaugh, 1990, 1998; McGaugh et al., 1996). The poor response of amygdala units on repeat presentation of faces during recognition lends further support to the observation of rapid habituation to face stimuli in this region (Breiter et al., 1996).

Despite the presumed specialization of the nondominant hemisphere to processing of faces, we did not observe significant differences in neuronal responses to faces in the dominant versus nondominant temporal lobe. This is not unexpected since responses of single units do not always reflect hemispheric differences. For instance, Ojemannn et al. (Ojemannn et al., 1992) found responses of single units to verbal stimuli in the nondominant hemisphere [reviewed in Ojemann et al. (Ojemann et al., 1998)].

Patterns of Excitation and Inhibition

Our results suggest that it is the distributed pattern of excitation and inhibition that reflects involvement of a region in a task. Such patterns have been observed in the olfactory cortex, where each odor has been shown to elicit a specific and widely distributed pattern of excitatory and inhibitory responses (Tanabe et al., 1975; Haberly and Bower, 1989). In visual and somatosensory cortex, inhibition may serve to highlight feature-specific excitatory responses and sharpen the distinction between neighboring stimulus patterns. Several studies suggest that GABAA-mediated inhibition plays an important role in spatial selectivity in primary visual and somatosensory cortices and in prefrontal cortex. Application of GABAA receptor antagonists results in (i) reduction of directional selectivity (Murphy and Humphrey, 1999) and broadening of orientation tuning (Sillito, 1984; Sato et al., 1996) in primary visual cortex, (ii) increase in size of receptive fields in somatosensory cortex (Alloway et al., 1989; Alloway and Burton, 1991), (iii) loss of spatial tuning in both pyramidal cells and interneurons in the monkey prefrontal cortex [although a minority of units, show establishment of previously absent spatial tuning (Rao et al., 2000)], and (iv) of particular relevance to our study, changes in the specificity of neuronal responses to specific visual stimuli in monkey inferotemporal cortex (Wang et al., 2000). Several studies also suggest that inhibition plays an important role in working memory. Disruption of spatial tuning of neurons during mnemonic tasks (Rao et al., 2000) as well as impairment on actual performance of a delayed response task (Sawaguchi et al., 1988, 1989) have been demonstrated in the monkey with application of GABAA receptor antagonist into the dorsal prefrontal cortex. Models of hippocampal function emphasize sparse coding (Marr, 1972; McNaughton and Morris, 1987), where a small proportion of the neurons are excited and information is transmitted with relatively few spikes. Such coding serves to minimize interference between stored representations of the environment (Trevis and Rolls, 1994; Eichenbaum, 1997), and also has been described for processing of faces in inferotemporal cortex (Young and Yamane, 1992). Our finding of a relatively small pool of neurons with excitatory responses coding for stimulus features suggests sparse coding of faces in the human hippocampus as well. Moreover, the prevalence of inhibitory responses may make such sparse coding more effective, highlighting the specific excitatory responses and thus reducing the probability of misfiring with resultant spurious encoding or recognition (e.g. mistaking one face for another).

Table 1

Distribution of neurons with excitatory and inhibitory responses to faces and objects during encoding and recognition

 Time after stimulus onset 
 T1 T2 
 Excitatory Inhibitory Excitatory Inhibitory 
T1 is the second of stimulus presentation, and T2 is the following second. The significance of response during T1 or T2 was calculated by a t-test comparison to the second preceding stimulus presentation. For each region, the number of recorded cells during a particular task is given in parentheses. 
Faces     
Encoding     
Amygdala (n = 46) 13 
Hippocampus (n = 97) 15 10 11 
Entorhinal cortex (n = 82) 10 10 
Total (225) 32 25 17 18 
     
Recognition     
Amygdala (n = 35) 
Hippocampus (n = 73) 23 24 
Entorhinal cortex (n = 69) 15 15 
Total (177) 17 40 18 43 
     
Objects     
Encoding     
Amygdala (n = 46) 
Hippocampus (n = 97) 16 
Entorhinal cortex (n = 82) 14 16 
Total (225) 38 19 27 20 
     
Recognition     
Amygdala (n = 35) 
Hippocampus (n = 73) 13 10 16 
Entorhinal cortex (n = 69) 12 13 14 
Total (177) 33 29 19 33 
 Time after stimulus onset 
 T1 T2 
 Excitatory Inhibitory Excitatory Inhibitory 
T1 is the second of stimulus presentation, and T2 is the following second. The significance of response during T1 or T2 was calculated by a t-test comparison to the second preceding stimulus presentation. For each region, the number of recorded cells during a particular task is given in parentheses. 
Faces     
Encoding     
Amygdala (n = 46) 13 
Hippocampus (n = 97) 15 10 11 
Entorhinal cortex (n = 82) 10 10 
Total (225) 32 25 17 18 
     
Recognition     
Amygdala (n = 35) 
Hippocampus (n = 73) 23 24 
Entorhinal cortex (n = 69) 15 15 
Total (177) 17 40 18 43 
     
Objects     
Encoding     
Amygdala (n = 46) 
Hippocampus (n = 97) 16 
Entorhinal cortex (n = 82) 14 16 
Total (225) 38 19 27 20 
     
Recognition     
Amygdala (n = 35) 
Hippocampus (n = 73) 13 10 16 
Entorhinal cortex (n = 69) 12 13 14 
Total (177) 33 29 19 33 
Figure 1.

Coronal MRI section depicting electrodes placed in the left hippocampus and right entorhinal cortex (left side of the brain is on the right and visa versa). Each electrode has platinum contacts for EEG recording along the electrode, and microwires at the tip. MRI-related artifact exaggerates the true diameter of the electrodes (1.3 mm) and the screw guides at the entry sites.

Figure 1.

Coronal MRI section depicting electrodes placed in the left hippocampus and right entorhinal cortex (left side of the brain is on the right and visa versa). Each electrode has platinum contacts for EEG recording along the electrode, and microwires at the tip. MRI-related artifact exaggerates the true diameter of the electrodes (1.3 mm) and the screw guides at the entry sites.

Figure 2.

Schematic of single unit recording during a visual memory task. The patient with depth probe targeted to the amygdala bilaterally as seen on coronal MRI, views a face with a distinct expression of surprise. Below are rasters and frequency histograms (200 ms bins) during encoding of a neuron in the left amygdala responding selectively to faces with expressions of surprise but not to other expressions. Vertical line (at time = 0 s) denotes the onset of the stimulus (duration 1 s).

Figure 2.

Schematic of single unit recording during a visual memory task. The patient with depth probe targeted to the amygdala bilaterally as seen on coronal MRI, views a face with a distinct expression of surprise. Below are rasters and frequency histograms (200 ms bins) during encoding of a neuron in the left amygdala responding selectively to faces with expressions of surprise but not to other expressions. Vertical line (at time = 0 s) denotes the onset of the stimulus (duration 1 s).

Figure 3.

Inhibitory and excitatory responses of neurons. Percentage of cells in the amygdala, hippocampus and entorhinal cortex responding during (a), initial presentation of faces (encoding), (b) recognition of faces, (c) initial presentation of objects, and (d) recognition of objects. Cells responding during T1 with increased firing rate (‘excitatory’) are depicted by gray bars, and those with decreased firing rate (‘inhibitory’) are depicted by black bars.

Figure 3.

Inhibitory and excitatory responses of neurons. Percentage of cells in the amygdala, hippocampus and entorhinal cortex responding during (a), initial presentation of faces (encoding), (b) recognition of faces, (c) initial presentation of objects, and (d) recognition of objects. Cells responding during T1 with increased firing rate (‘excitatory’) are depicted by gray bars, and those with decreased firing rate (‘inhibitory’) are depicted by black bars.

Figure 4.

Inhibitory and excitatory responses during encoding. Trial-by-trial rasters and average frequency histograms (200ms bins) of (a) two cells, in the left anterior hippocampus (left) of subject no. 312 and the right anterior hippocampus (right) in subject no. 306, during encoding of faces; (b) a cell in the right amygdala of the same patient (no. 306) during encoding of faces (left) and objects (right). Vertical line (at time = 0s) denotes the onset of stimulus (duration 1s). Note the inhibitory response in the hippocampus and the excitatory response in the amygdala during encoding of faces. The amygdala cell was considered selective as it responded to faces and not to objects, and the response to faces was significantly different from the response to objects.

Figure 4.

Inhibitory and excitatory responses during encoding. Trial-by-trial rasters and average frequency histograms (200ms bins) of (a) two cells, in the left anterior hippocampus (left) of subject no. 312 and the right anterior hippocampus (right) in subject no. 306, during encoding of faces; (b) a cell in the right amygdala of the same patient (no. 306) during encoding of faces (left) and objects (right). Vertical line (at time = 0s) denotes the onset of stimulus (duration 1s). Note the inhibitory response in the hippocampus and the excitatory response in the amygdala during encoding of faces. The amygdala cell was considered selective as it responded to faces and not to objects, and the response to faces was significantly different from the response to objects.

Figure 5.

Selective inhibitory and excitatory responses to faces and objects during recognition. Rasters and frequency histograms of (a) right anterior hippocampus neuron (subject no. 291) with inhibitory responses to faces and excitatory responses to objects. (b) Right entorhinal cortex cell (subject no. 308) with excitatory responses to faces and inhibitory responses to objects.

Figure 5.

Selective inhibitory and excitatory responses to faces and objects during recognition. Rasters and frequency histograms of (a) right anterior hippocampus neuron (subject no. 291) with inhibitory responses to faces and excitatory responses to objects. (b) Right entorhinal cortex cell (subject no. 308) with excitatory responses to faces and inhibitory responses to objects.

Figure 6.

(a) Percentage of cells discriminating stimulus attributes (gender and expression) among those neurons with inhibitory and excitatory responses to faces during encoding. (b) Trial-by-trial rasters and histograms of responses of a single neuron in the entorhinal cortex (subject no. 308) to female (left) and male (right) faces during recognition. Note selective excitatory responses to female faces. (c) Rasters and histograms for a cell in the right middle hippocampus (subject no. 309) with inhibitory responses to faces (female left and male right) during recognition. Note the lack of discriminative responses to gender of the faces.

Figure 6.

(a) Percentage of cells discriminating stimulus attributes (gender and expression) among those neurons with inhibitory and excitatory responses to faces during encoding. (b) Trial-by-trial rasters and histograms of responses of a single neuron in the entorhinal cortex (subject no. 308) to female (left) and male (right) faces during recognition. Note selective excitatory responses to female faces. (c) Rasters and histograms for a cell in the right middle hippocampus (subject no. 309) with inhibitory responses to faces (female left and male right) during recognition. Note the lack of discriminative responses to gender of the faces.

We thank Charles Wilson, Eric Behnke and Tony Fields for technical assistance and Irene Wainwright for editorial assistance. Supported by grants from the National Institutes of Health, National Institute of Neurological Disorders and Stroke (NS33221 and NS02808).

References

Aggelton JP, Brown MW (
1999
) Episodic memory, amnesia, and the hippocampal–anterior thalamic axis.
Behav Brain Sci
 
22
:
425
–489.
Alloway KD, Burton H (
1991
) Differential effects of GABA and bicuculline on rapidly- and slowly-adapting neurons in primary somatosensory cortex of primates.
Exp Brain Res
 
85
:
598
–610.
Alloway KD, Rosenthal P, Burton H (
1989
) Quantitative measurements of receptive field changes during antagonism of GABAergic transmission in primary somatosensory cortex of cats.
Exp Brain Res
 
78
:
514
–532.
Andreasen NC, O'Leary D, Arndt, S, Cidzadlo T, Hurtig R, Rezai K, Watkins L, Boles Ponto LL, Hichwa RD (
1995
) Short-term and long-term verbal memory: a positron emission tomography study.
Proc Natl Acad Sci USA
 
92
:
5111
–5115.
Breiter HC, Etcoff NL, Whalen PJ, Kennedy WA, Rauch SL, Buckner RL, Strauss MM, Hyman SE, Rosen BR (
1996
) Response and habituation of the human amygdala during visual processing of facial expression.
Neuron
 
17
:
875
–887.
Brewer JB, Zhao Z, Desmond JE, Glover GH, Gabrieli JDE (
1998
) Making memories: brain activity that predicts how well visual experience will be remembered.
Science
 
281
:
1185
–1187.
Brown MW, Aggleton JP (
2001
) Recognition memory: what are the roles of the perirhinal cortex and hippocampus?
Nature Rev
 
2
:
51
–61.
Buckner RL, Tulving E (1995) Neuroimaging studies of memory: theory and recent PET results. In: Handbook of neuropsychology (Boller F, Grafman J eds), pp. 439–466. New York: Elsevier.
Buhl EH, Halasy K, Somogyi P (
1994
) Diverse sources of hippocampal unitary inhibitory postsynaptic potentials and the number of synaptic release sites.
Nature
 
368
:
823
–828.
Cahill L, McGaugh JL (
1990
) Amygdaloid complex lesions differentially affect retention of tasks using appetitive and aversive reinforcement.
Behav Neurosci
 
104
:
532
–543.
Cahill L, McGaugh JL (
1998
) Mechanisms of emotional arousal and lasting declarative memory.
Trends Neurosci
 
21
:
294
–299.
Cahusac PMB, Miyashita Y, Rolls ET (
1989
) Responses of hippocampal formation neurons in the monkey related to delayed spatial response and object–place memory tasks.
Behav Brain Res
 
33
:
229
–240.
Cameron KA, Yashar S, Wilson CL, Fried I (
2001
) Human hippocampal neurons predict how well word pairs will be remembered.
Neuron
 
30
:
289
–298.
Csicsvari J, Hirase H, Czurkó A, Mamiya A, Buzsáki G (
1999
) Oscillatory coupling of hippocampal pyramidal cells and interneurons in the behaving rat.
J Neurosci
 
19
:
274
–287.
Desimone R (
1996
) Neural mechanisms for visual memory and their role in attention.
Proc Natl Acad Sci USA
 
93
:
13494
–13499.
Dolan RJ, Fletcher PF (
1999
) Encoding and retrieval in human medial temporal lobes: an empirical investigation using functional magnetic resonance imaging (fMRI).
Hippocampus
 
9
:
25
–34.
Eckmann P, Friesen WV (1976) Pictures of facial affect. Palo Alto, CA: Consulting Psychologists.
Eichenbaum H (
1997
) How does the brain organize memories?
Science
 
277
:
330
–331.
Eichenbaum H, Otto T (
1992
) The hippocampus – what does it do?
Behav Neurol Biol
 
57
:
2
–36.
Eifuku S, Nishijo H, Kita T, Ono T (
1995
) Neuronal activity in the primate hippocampal formation during a conditional association task based on the subject's location.
J Neurosci
 
15
:
4952
–4969.
Eldridge LL, Knowlton BJ, Furmanski CS, Bookheimer SY, Engel SA (
2000
) Remembering episodes: a selective role for the hippocampus during retrieval.
Nat Neurosci
 
3
:
1149
–1152.
Engel J Jr (
1996
) Excitation and inhibition in epilepsy.
Can J Neurol Sci
 
23
:
167
–174.
Felleman D, Van Essen DC (
1991
) Distributed hierarchical processing in primate cerebral cortex.
Cereb Cortex
 
1
:
1
–47.
Fernández G, Brewer JB, Zhao Z, Glover GH, Gabrieli JDE (
1999
) Level of sustained entorhinal activity at study correlates with subsequent cued-recall performance: a functional magnetic resonance imaging study with high acquisition rate.
Hippocampus
 
9
:
45
–53.
Fried I, MacDonald KA, Wilson CL (
1997
) Single neuron activity in human hippocampus and amygdala during recognition of faces and objects.
Neuron
 
18
:
753
–765.
Fried I, Wilson CL, Maidment NT, Engel J Jr, Behnke EJ, Fields TA, MacDonald KA, Morrow JW (
1999
) Cerebral microdialysis combined with single neuron and EEG recording in neurosurgical patients.
J Neurosurg
 
91
:
697
–705.
Freund TF, Buzsaki G (
1996
) Interneurons of the hippocampus.
Hippocampus
 
6
:
347
–470.
Gabrieli JDE, Brewer JB, Desmond JE, Glover GH (
1997
) Separate neural bases of two fundamental memory processes in the human medial temporal lobe.
Science
 
276
:
264
–266.
Haberly LB, Bower JM (
1989
) Olfactory cortex: model circuit for study of associative memory?
Trends Neurosci
 
12
:
258
–264.
Heit G, Smith ME, Halgren E (
1988
) Neural encoding of individual words and faces by the human hippocampus and amygdala.
Nature
 
333
:
773
–775.
Heit G, Smith ME, Halgren E (
1990
) Neuronal activity in the human medial temporal lobe during recognition memory.
Brain
 
113
:
1093
–1112.
Kreiman G, Koch C, Fried I (
2000
) Category-specific visual responses of single neurons in the human medial temporal lobe.
Nat Neurosci
 
3
:
946
–953.
Kreiman G, Koch C, Fried I (
2000
) Imagery neurons in the human brain.
Nature
 
408
:
357
–361.
LeDoux J (
1998
) Fear and the brain: where have we been, and where are we going?
Biol Psychiatry
 
44
:
1229
–1238.
Leonard CM, Rolls ET, Wilson FA, Baylis GC (
1985
) Neurons in the amygdala of the monkey with responses selective for faces.
Behav Brain Res
 
15
:
159
–176.
Lepage M, Habib R, Tulving E (
1998
) Hippocampal PET activations of memory encoding and retrieval: the HIPER model.
Hippocampus
 
8
:
313
–322.
Marr D (
1972
) Simple memory: a theory for archicortex.
Philos Trans R Soc Lond
 
262
:
23
–81.
McGaugh JL, Cahill L, Roozendaal B (
1996
) Involvement of the amygdala in memory storage interaction with other brain systems.
Proc Natl Acad Sci USA
 
93
:
13508
–13514.
McNaughton BL, Morris RGM (
1987
) Hippocampal synaptic enhancement and information storage within a distributed memory system.
Trends Neurosci
 
10
:
408
–415.
Miller EK, Li L, Desimone R (1993) Activity of neurons in anterior inferior temporal cortex during a short-term memory task.
J Neurosci
 
13
:
1460
–1478 (p. 1462).
Milner B, Squire LR, Kandel ER (
1998
) Cognitive neuroscience and the study of memory.
Neuron
 
20
:
445
–468.
Murphy A, Humphrey AL (
1999
) Inhibitory contributions to spatio-temporal receptive-field structure and direction selectivity in simple cells of cat area 17.
J Neurophysiol
 
81
:
1212
–1224.
Nakamura K, Mikani A, Kubota K (
1992
) Activity of single neurons in the monkey amygdala during performance of a visual discrimination task.
J Neurophysiol
 
67
:
1447
–1463.
Nyberg L, McIntosh AR, Cabeza R, Nilsson L-G, Houle S, Habib R, Tulving E. (
1996
) Network analysis of positron emission tomography regional cerebral blood flow data: ensemble inhibition during episodic memory retrieval.
J Neurosci
 
16
:
3753
–3759.
Ojemann JG, Ojemann GA, Letich E (
1992
) Neuronal activity related to faces and matching in human right nondominant temporal cortex.
Brain
 
115
:
1
–13.
Ojemann GA, Ojemann SG, Fried I (
1998
) Lessons from the human brain: neuronal activity related to cognition.
Neuroscientist
 
4
:
285
–300.
O'Keefe J, Burgess N, Donnett JG, Jeffery KJ, Maguire EA (
1998
) Place cells, navigational accuracy and the human hippocampus.
Philos Trans R Soc Lond B Biol Sci
 
353
:
1333
–1340.
Ono T, Nakamura K, Fukuda M, Tamura R (
1991
) Place recognition responses of neurons in monkey hippocampus.
Neurosci Lett
 
121
:
194
–198.
Ono T, Tamura R, Nishijo H, Nakamura K (1993) Neural mechanisms of recognition and memory in the limbic system. In: Brain mechanisms of perception and memory (Ono T, Squires LR, Raichle E, Perrett DI, Fukuda M, eds), pp. 330–355. Oxford: Oxford University Press.
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.
Rao SG, Williams GV, Goldman-Rakic PS (
2000
) Destruction and creation of spatial tuning by disinhibition: GABAA blockade of prefrontal cortical neurons engaged by working memory.
J Neurosci
 
20
:
485
–494.
Rolls ET (1981) Responses of amygdaloid neurons in the primate. In: The amygdaloid complex (Ben-Ari Y, ed.), pp. 383–393. Amsterdam: Elsevier.
Rolls ET (
1984
) Neurons in the cortex of the temporal lobe and in the amygdala of the monkey with responses selective for faces.
Hum Neurobiol
 
3
:
209
–222.
Rolls ET, Miyashita Y, Cahusac P, Kesner RP, Niki HD, Feigenbaum JD, Bach L (
1989
) Hippocampal neurons in the monkey with activity related to the place where a stimulus is shown.
J Neurosci
 
9
:
1835
–1845.
Sato H, Katsuyama N, Tamura H, Hata Y, Tsumoto T (
1996
) Mechanisms underlying orientation selectivity of neurons in the primary visual cortex of the macaque.
J Physiol Lond
 
494
:
757
–771.
Sawaguchi T, Matsumura M, Kubota K (
1988
) Delayed response deficit in monkeys by locally disturbed prefrontal neuronal activity by bicuculline.
Behav Brain Res
 
31
:
193
–198.
Sawaguchi T, Matsumura M, Kubota K (
1989
) Delayed response deficits in monkeys by local injection of bicuculline into the dorsolateral prefrontal cortex in Japanese macaque monkeys.
Exp Brain Res
 
75
:
457
–469.
Schacter DL, Wagner AD (
1999
) Medial temporal lobe activations in fMRI and PET studies of episodic encoding and retrieval.
Hippocampus
 
9
:
7
–24.
Scoville WB, Milner B (
1957
) Loss of recent memory after bilateral hippocampal lesions.
J Neurol Neurosurg Psychiatry
 
20
:
11
–21.
Sillito AM (1984) Functional considerations of the operation of GABAergic inhibitory processes in the visual cortex. In: Cerebral cortex, 2nd edn (Jones EG, Peters A, eds), pp. 91–117. New York: Plenum.
Squire LR (
1992
) Memory and the hippocampus: a synthesis from findings with rats, monkeys, and humans.
Psychol Rev
 
99
:
195
–231.
Stark CE, Squire LR (
2000
) Functional magnetic resonance imaging (fMRI) activity in the hippocampal region during recognition memory.
J Neurosci
 
20
:
7776
–77781.
Stark CE, Squire LR (
2001
) When zero is not zero: the problem of ambiguous baseline conditions in fMRI.
Proc Natl Acad Sci USA
 
98
:
12760
–12766.
Stern CE, Hasselmo ME (
1999
) Bridging the gap: integrating cellular and functional magnetic resonance imaging studies of the hippocampus.
Hippocampus
 
9
:
45
–53.
Stern CE, Corkin S, Gonzalez RG, Guimaraes AR, Baker JR, Jennings PJ, Carr CA, Sugiura RM, Vendantham V, Rosen BR (
1996
) The hippocampal formation participates in novel picture encoding: evidence from functional magnetic resonance imaging.
Proc Natl Acad Sci USA
 
93
:
8660
–8665.
Suzuki W, Miller E, Desimone R (
1997
) Object and place memory in the macaque entorhinal cortex.
J Neurophysiol
 
78
:
1062
–1081.
Tanabe T, Iino M, Takagi SF (
1975
) Discrimination of odors in olfactory bulb, pyriform–amygdaloid areas and orbitofrontal cortex of monkey.
J Neurophysiol
 
38
:
1284
–1296.
Trevis A, Rolls ET (
1994
) Computational analysis of the role of the hippocampus in memory.
Hippocampus
 
4
:
374
–391.
Ungerleider LG (
1995
) Functional brain imaging studies of cortical mechanisms for memory.
Science
 
51
:
1197
–1207.
Vargha-Khadem F, Gadian DG, Watkins KE, Conelly A, Van Paesshen W, Mishkin M (
1997
) Differential effects of early hippocampal pathology on episodic and semantic memory.
Science
 
277
:
376
–380.
Wada J, Rasmussen T (
1960
) Intracarotid injection of sodium amytal for the lateralization of cerebral speech dominance: experimental and clinical observations.
J Neurosurg
 
17
:
266
–282.
Wagner AD, Schacter DL, Rotte M, Koutstaal W, Maril A, Dale AM, Rosen BR, Buckner RL (
1998
) Building memories: remembering and forgetting of verbal experiences as predicted by brain activity.
Science
 
281
:
1188
–1191.
Wang Y, Fujita I, Murayama Y (
2000
) Neuronal mechanisms of selectivity for object features revealed by blocking inhibition in inferotemporal cortex.
Nat Neurosci
 
3
:
807
–813.
Young MP, Yamane S (
1992
) Sparse population coding of faces in inferotemporal cortex.
Science
 
256
:
1327
–1331.
Zola-Morgan S, Squire LR (
1990
) The neuropsychology of memory.
Ann N Y Acad Sci
 
608
:
434
–456.