| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Physiology, Faculty of Medicine, Toyama Medical and Pharmaceutical University, Toyama 930-01, Japan; and 2 Institute of Physiology, Academy of Sciences, 142 20 Prague 4-Krc, Czech Republic
 |
ABSTRACT |
|---|
 Abstract
 Introduction
Methods
Results
Discussion
References
|
|---|
Kobayashi, Tsuneyuki, Hisao Nishijo, Masaji Fukuda, Jan Bures, and Taketoshi Ono. Task-dependent representations in rat hippocampal place neurons. J. Neurophysiol. 78: 597-613, 1997. It is suggested that the hippocampal formation is essential to spatial representations by flexible encoding of diverse information during navigation, which includes not only externally generated sensory information such as visual and auditory sensation but also ideothetic information concerning locomotion (i.e., internally generated information such as proprioceptive and vestibular sensation) as well as information concerning reward. In the present study, we investigated how various types of information are represented in the hippocampal formation, by recording hippocampal complex-spike cells from rats that performed three types of place learning tasks in a circular open field with the use of intracranial self-stimulation as reward. The intracranial self-stimulation reward was delivered in the following three contexts: if the rat 1) entered an experimenter-determined reward place within the open field, and this place was randomly varied in sequential trials; 2) entered two specific places, one within and one outside the place field (an area identified by change in activity of a place neuron); or 3) entered an experimenter-specified place outside the place field. Because the behavioral trails during navigation were more constant in the second task than in the first task, ideothetic information concerning locomotion was more relevant to acquiring reward in the second task than in the first task. Of 43 complex-spike cells recorded, 37 displayed place fields under the first task. Of these 37 place neurons, 34 also had significant reward correlates only inside the place field. Although reward and place correlates of the place neuron activity did not change between the first and second tasks, neuronal correlates to behavioral variables for locomotion such as movement speed, direction, and turning angle significantly increased in the second task. Furthermore, 6 of 31 place neurons tested with the third task, in which the reward place was located outside the original place field, shifted place fields. The results indicated that neuronal correlates of most place neurons flexibly increased their sensitivity to relevant information in a given context and environment, and some place neurons changed the place field per se with place reward association. These results suggest two strategies for how hippocampal neurons incorporate an incredible variety of perceptions into a unified representation of the environment: through flexible use of information and the creation of new representations.
Data demonstrating that the activity of complex-spike cells in the hippocampal formation increases when a rat or a monkey is in a specific location in the environment (Eichenbaum et al. 1987 The methods used for the behavioral aspects of the present experiments using rewarding intracranial self-stimulation were substantially the same as those described in our methodological paper (Fukuda et al. 1992 Subjects and surgery
Seventeen male albino Wistar rats were used. The rats weighed 270-320 g at the time of surgery, and were individually housed with food and water available ad libitum. All rats were treated in strict compliance with the US Public Health Service Policy on Human Care and Use of Laboratory Animals and the National Institutes of Health Guide for the Care and Use of Laboratory Animals.
Apparatus
Spatial behavior was investigated in a 150-cm-diam circular open field with a 45-cm-high wall, painted black on the inside (Fig. 1Aa). The experimenter could manually move or rotate the field on casters attached to the bottom. The open field was enclosed by a black curtain (180 cm diam and 200 cm high). The ceiling of the enclosure contained four small speakers mounted near the circumference, spaced 90° apart, four light bulbs individually mounted near the inner edge of each speaker, and a video camera (Behavioral Tracking Analyzer, BTA-2A, Muromachi Kikai, Tokyo) at the center (Fig. 1Ab). Usually a light bulb was turned on at the 3 o'clock position, and a speaker continuously emitted white noise at the 9 o'clock position. A small light bulb was mounted on the head of the rat. The video camera converted a real video image signal to a binary signal, and tracked the two-dimensional (horizontal) motion of the small bulb. A laboratory microcomputer (PC-9801, NEC, Tokyo) received the X and Y coordinates of the position of the head through an RS-232C serial port at 20 frames per second. A program delimited circular areas (reward places) in the open field, and triggered the delivery of current for intracranial self-stimulation when the rat entered the reward place. The experimenter monitored the locations of the reward place and the rat on a display screen, but no distinctive local cues marked the reward place for the rat.
Behavioral training
SELF-STIMULATION SCREENING.
The rats were screened to self-stimulate in an operant chamber (30 × 30 × 33 cm) equipped with a lever on one wall. Each lever press triggered the delivery of an 0.5-s train of 0.3-ms negative square-wave pulses at 100 Hz.
PLACE TASK TRAINING.
After the implantation of the recording electrodes, the rats were trained to perform a spatial behavior in the open field. In the first condition, current for intracranial self-stimulation was delivered when the cumulative distance traveled by the rat reached a given distance. The initial distance was 80 cm, and this was increased progressively to 120 or 160 cm. The rats were usually trained for ~1 h/day for 2-4 days until they learned to travel the open field continuously. They were then trained in the following three kinds of place tasks.
Recording procedures and data acquisition
After a few days of recovery from the implantation, the recording electrode assembly was driven at ~20-80 µm per day in 20-µm steps in the open field. If a complex-spike cell was found, the stability of the cell recording was tested by the running place tasks for 30-60 min. Neuronal activity was passed through a high-input impedance preamplifier made of a dual-channel field-effect transistor (2SK389, Toshiba Electric), amplified by a main amplifier, and converted to standardized pulses by means of a window discriminator. During intracranial self-stimulation, neuronal activity was isolated from the stimulus artifact by the window discriminator. The microcomputer summed the spikes (standardized pulses) over 50-ms intervals and combined this data with the X and Y coordinates of the head of the rat to construct a distribution map of the mean firing rate as a function of the position of the rat.
Data analysis
NEURONAL CORRELATES TO PLACE AND INTRACRANIAL SELF-STIMULATION.
Place fields were delineated by a process used previously (Breese et al. 1989 NEURONAL CORRELATES TO BEHAVIORAL VARIABLES.
We examined the behavioral correlates of the place neuron activity within the place field according to the procedure described by Wiener et al. (1989) Histology
After the recording electrodes were advanced below the hippocampal formation, the locations of the recording and stimulating electrodes were identified histologically. The rats were deeply anesthetized with pentobarbital sodium (50 mg/kg ip), and the stimulating site was marked by an iron deposit, created by passing a 20-µA positive current through the stimulating electrode for 30 s. The recording site was marked by an electrolytic lesion, created by passing a 20-µA negative current through the recording electrodes for 30 s. Each rat was then perfused through the heart with 50 ml of 0.9% saline containing heparin followed by 200 ml of 10% buffered Formalin-containing 2% potassium ferricyanide. The brain was removed and fixed in Formalin for Identification of place fields and behavioral correlates of place neurons by the random search task
The activity of 43 complex-spike cells was recorded from the CA1 (n = 22) and the CA3 (n = 21) subfields of the hippocampal formation of rats as they performed the random search task. Thirty-seven (86.0%) had place fields (CA1, 19; CA3, 18). The remaining six (CA1, 3; CA3, 3) had no place field that met the criterion described in METHODS. Figure 2 shows an example of a CA3 place neuron. Figure 2A shows an example of the trajectory and sites of application of intracranial self-stimulation (each dot) for a recording session, and the corresponding place field of the place neuron in the random search task. The rat traveled 20165.7 cm and obtained 50 rewards in 533.5 s. The trail and intracranial self-stimulation almost uniformly covered the open field (Fig. 2A, left). The firing rate map shows a place field as a rectangle in the center of the open field (Fig. 2A, right). The mean firing rate of this neuron in the pixels inside the place field was 3.2 ± 0.2 (SE) spikes/s, and the mean firing rate outside the place field was 1.4 ± 0.1 spikes/s. The mean firing rates of >37 place neurons inside the place fields were 0.9-11.6 (3.8 ± 0.4) spikes/s, and the mean firing rates outside the place fields were 0.1-3.1 (0.6 ± 0.1) spikes/s. Figure 2B displays a curve of the average movement speed (top) and a histogram of the average firing rate (bottom) inside (Ba) and outside (Bb) the place field shown in A; these were time locked to the start of the intracranial self-stimulation. The firing rates in preintracranial self-stimulation, intracranial self-stimulation, and postintracranial self-stimulation phases were significantly different from each other inside [ANOVA, F(2,12) = 5.76, P < 0.05] but not outside [ANOVA, F(2,34) = 1.77, P > 0.05] the place field. Of 37 place neurons tested in the random search task, 34 had significant reward correlates only inside the place field as shown in Fig. 2; the remaining 3 neurons had no significant reward correlates inside or outside the place field. Because a trial was terminated if 50 intracranial self-stimulation rewards had been delivered in one session, the insignificant reward correlates in these three neurons were ascribed to low numbers of total intracranial self-stimulation rewards inside the place fields. The low numbers of total intracranial self-stimulation rewards resulted in a low degree of freedom in the statistical analysis, which in turn resulted in an insignificant reported difference. However, it should be emphasized that no place neurons consistently had reward correlates both inside and outside the place field. Furthermore, changes in movement speed before and after delivery of intracranial self-stimulation were very similar inside and outside the place field, as shown in Fig. 2B, top. These findings indicated that the effects of intracranial self-stimulation on place neuron activity were due neither to the direct effects of intracranial self-stimulation nor to the direct effect of movement speed.
Spatial and reward correlates of place neuron activity during the random search task and place learning task 1
To compare the spatial and reward modulation of place neuron activity in different behavioral tasks, a place field was initially identified in the random search task, and the activity of the same neuron was then tested in place learning task 1, in which one of the two reward places was located in the place field. Figure 3 shows examples of the spatial and reward correlates of a CA3 place neuron activity in the random search task and place learning task 1 with 0- and 1.0-s delay. The neuron had one large and two small place fields located at 1-3 o'clock during the random search task (Fig. 3Aa). The mean firing rate of the neuron in the pixels inside the three place fields was 5.6 ± 0.3 spikes/s, and the mean firing rate outside the place field was 0.5 ± 0.1 spikes/s. The firing rates during the preintracranial self-stimulation, intracranial self-stimulation, and postintracranial self-stimulation phases were significantly different inside the large place field [ANOVA, F(2,11) = 19.01, P < 0.01], and decreased during reward delivery (Fig. 3Ab). The post hoc test indicated that the firing rate during the preintracranial self-stimulation phase was significantly higher than the firing rate during the intracranial self-stimulation phase (Newman-Keuls test, P < 0.05). However, there was no significant difference in neuronal activity among the three phases outside the place field [ANOVA, F(2,28) = 1.23, P > 0.05; Fig. 3Ac]. During place learning task 1, the activity of the neuron corresponded spatially with the activity observed in the random search task (Fig. 3, Ba and Ca). In place learning task 1 with 1.0-s delay, the same effects of intracranial self-stimulation on neuronal activity as shown in the random search task were observed. In place learning task 1 with delay, the activity of the neuron was significantly influenced by intracranial self-stimulation delivered in the reward place inside the place field [ANOVA, F(3,23) = 32.04, P < 0.05; Fig. 3Cb] but was not influenced by intracranial self-stimulation delivered outside the place field [ANOVA, F(3,23) = 0.76, P > 0.05; Fig. 3Cc]. The post hoc test indicated that the firing rate during the preintracranial self-stimulation phase was significantly higher than the firing rate during the intracranial self-stimulation phase inside the place field (Newman-Keuls test, P < 0.01). In place learning task 1 without delay, the neuronal activity was not significantly influenced by the intracranial self-stimulation inside [ANOVA, F(2,20) = 0.70, P > 0.05; Fig. 3Bb] or outside [ANOVA, F(2,20) = 2.11, P > 0.05; Fig. 3Bc] the place field. Although the effects of intracranial self-stimulation reward were not statistically significant in place learning task 1 without delay, the activity changes were very similar to those in the random search task and place learning task 1 with delay. Thus the activity of the neuron inside the place field for all three tasks (random search task, and place learning task 1 with or without delay) was higher before intracranial self-stimulation than during intracranial self-stimulation.
Behavioral correlates of place neuron activity during the random search task and place learning task 1
The neuronal correlates of the hippocampal place neurons to behavioral variables (speed, direction, and turning angle of movement) were analyzed only inside the place field identified during the random search task, and were compared within three tasks (random search task, and place learning task 1 without and with delay). Figure 7 shows the neuronal correlates of the same place neuron shown in Figs. 3 and 4 to the behavioral variables. In the random search task, the neuronal activity was significantly modulated by speed(
Effects of task difference on breadth of responsiveness
We analyzed the breadth of responsiveness of 21 place neurons (CA1, 9; CA3, 12) that were tested in three tasks in the same way. Because there was no significant difference between the mean breadth of responsiveness of the CA1 and CA3 place neurons for each behavioral variable (Student's t-test, P > 0.05), the data from both CA1 and CA3 place neurons were analyzed together. First, we compared the entropy measurements of breadth of responsiveness, calculated from actual responses to each parameter of the behavioral variables, with those calculated from the predicted responses to each parameter according to the formula by Muller et al. (1994)
Plastic changes of the spatial correlates of the place neuron activity in place learning task 2
Initially, the place fields were identified by the random search task in two or three sessions. Place learning task 1 was tested with and without delay in four or five sessions, and place learning task 2, in which the reward place corresponding to the place fields was changed to a nonreward place, was tested in four or five sessions. The changes in place field were analyzed in relation to reward place. We defined the place field changes as follows: 1) if a new place field during and/or after place learning task 2 moved away from the circular area where an initial place field was located during place learning task 1, which was the reward place in place learning task 1 and nonreward place in place learning task 2; and 2) only if the new place field did not overlap with that circular area. The place fields of the place neurons were very stable during both the random search task and place learning task 1. No place neurons shifted place fields during the random search task and place learning task 1, before place learning task 2 was introduced. During place learning task 2 and subsequent place learning task 1, 6 of 31 (19.4%) place neurons identified by the random search task and place learning task 1 developed new place fields associated with a reward place located outside the place fields that were initially identified by the random search task and place learning task 1. The other 25 (80.6%) did not change place fields. It must be emphasized that there were no relations between the place field changes and total numbers of sessions for all three tasks, and that all six place neurons shifted their place fields during or after place learning task 2.
Recording sites
The approximate anatomic locations at which the place neurons were recorded are shown in Fig. 13. These sites were determined histologically from small lesions made in the hippocampal formation after recording. The place neurons recorded were located in the CA1 and CA3 subfields of the hippocampal formation.
Effects of reward on place neuron activity
In the present study, intracranial self-stimulation reward influenced the neuronal activity of most place neurons only inside the place fields, and the effects of intracranial self-stimulation reward were essentially similar in three different tasks: random search task, place learning task 1 without delay, and place learning task 1 with delay. Speakman and O'Keefe (1990) Effects of behavioral tasks on spatial and behavioral correlates of place neuron activity
The random search task and place learning task 1 had a common association structure, in which intracranial self-stimulation reward was associated both with the place field and with multiple locations in the random search task and with a zone diametrically opposite the place field in place learning task 1. The same spatial and reward correlates were consistently observed in both the random search task and place learning task 1. These results suggest that spatial correlates do not change if the tasks share a common place reward association. In the random search task, intracranial self-stimulation was delivered if the rat entered areas that were randomly located on the open field, but only after a certain distance was traversed. Because the rat could not locate a fixed locus of intracranial self-stimulation, it seemed to explore the open field uniformly with no particular orienting directions. In place learning task 1, the rats' constant trails over several sessions suggest that behavioral factors such as a specific speed, direction, and turning in a specific place were important to accomplish the task effectively. Each behavioral variable during locomotion was less relevant in the random search task than in place learning task 1; place neurons, with relatively broad movement tuning in the random search task, became more selective in place learning task 1. The difference in behavioral requirements between the random search task and place learning task 1 resulted in an increased correlation of the place neuron to specific behavioral variables.
Plastic change of place neuron activity
The present study shows that, during several sessions of place learning task 2, ~20% of the place neurons shifted place fields to locations newly associated with intracranial self-stimulation rewards; these neurons displayed no increase in activity in the random search task and place learning task 1. The intracranial self-stimulation was associated with both place field and non-place field regions in the random search task and place learning task 1, but was only associated with a region outside the original place field in place learning task 2. However, the change of reward place alone cannot account for the shifting of the place field, because reward availability was not the sole determinant of hippocampal neuron activity (see above). Although most of the place fields identified during the random search task retained their locations in place learning task 2, some place fields shifted to the reward place in place learning task 2, and the locations of place fields changed gradually. The neurons did not demonstrate the original place field when the task was reversed from place learning task 2 to place learning task 1. These findings indicate that complete remapping can occur in the same behavioral task and environment, and suggest that former experience (insertion of place learning task 2) influenced the place field because it was stable during place learning task 1 if place learning task 2 was not imposed during recording.
Hippocampal function in path integration
It has been suggested that place cells are primarily involved in path integration (Knierim et al. 1995 Conclusions
The present study elucidates important characteristics of the hippocampal place neurons: they flexibly encode diverse and relevant information such as space, reward, behavioral variables, and former experience to create a new representation. The place field might work as a window to focus attention on relevant information to be processed in the hippocampal formation. The results also suggest that spatial (place), as well as other task-relevant information, is relationally (Cohen and Eichenbaum 1993
INTRODUCTION
Abstract
Introduction
Methods
Results
Discussion
References
; McNaughton et al. 1983; Muller et al. 1987; Nishijo et al. 1997; O'Keefe 1976; O'Keefe and Burgess 1996; O'Keefe and Dostrovsky 1971; Olton et al. 1978; Ono et al. 1991, 1993; Quirk et al. 1990; Wible et al. 1986; Wiener et al. 1989; Wilson and McNaughton 1993, 1994), combined with memory deficit after lesion of the hippocampal formation (e.g., Bunsey and Eichenbaum 1996; Mishkin 1978; Olton et al. 1979; Zola-Morgan and Squire 1990), have contributed to the understanding of functions of the hippocampal formation (Eichenbaum et al. 1992; McNaughton and Nadel 1989; O'Keefe and Nadel 1978). These complex-spike cells are called "place cells," and areas of the environment associated with firing increment are called "place fields." Spatial correlates of place cell activity can be influenced by controlling environmental cues (Breese et al. 1989; Gothard et al. 1996; Muller and Kubie 1987; Muller et al. 1987; O'Keefe and Conway 1978; O'Keefe and Speakman 1987) that affect the place cell activity.
; McNaughton et al. 1983; Wiener et al. 1989). In such studies rats approached reward locations through rather restricted trajectories in an eight-arm radial maze (McNaughton et al. 1983; Muller et al. 1994) or a square chamber (Breese et al. 1989; Wiener et al. 1989). On the other hand, Muller et al. (1987, 1994) showed relatively low directional selectivity in a behavioral paradigm in which rats searched for small food pellets scattered at random in a cylindrical apparatus. This discrepancy might be attributed to differences in the testing environments, because it has been suggested that the hippocampal neurons encode only salient cues (Muller and Kubie 1987; O'Keefe and Speakman 1987), local views (Leonard and McNaughton 1990; McNaughton et al. 1991), or structure of the maze (Muller et al. 1994). Furthermore, not only these sensory and/or motor factors but also cognitive factors such as behavioral context might affect place cell activity (Knierim et al. 1995; Kubie and Ranck 1983; Markus et al. 1995; Nishijo et al. 1997). Previously, we reported that monkey hippocampal neurons responded differently to the same percept (presentation of the same spatial cues and the same object) with the same behavioral response (go/no-go responses) when the object was presented in different contexts or situations (Eifuku et al. 1995; Nishijo et al. 1993; Ono et al. 1995). However, effects of context on neuronal correlates of rat place cells to movement variables (speed, direction, and turning angle) have not yet been tested in the same environment, except in a recent study that reported changes in neuronal correlates to direction between two different behavioral contexts (Markus et al. 1995). Reward availability might be also an important determinant of place cell activity. Two studies focused on whether or not place field shifted when the location of the reward was changed, and came to different conclusions: Breese et al. (1989) recorded place fields in rats as the rats explored a platform to obtain rewards available at five locations. When reward delivery was restricted, most of these place fields shifted to a single location. In contrast, Speakman and O'Keefe (1989) demonstrated that, after relocation of the goal, only 2 of 19 place fields changed, but the others retained their locations relative to cues in a cue controlled environment.
; Eichenbaum et al. 1992; Knierim et al. 1995; McNaughton et al. 1989; Young et al. 1994). The question is how the hippocampal formation encodes the ever-changing and surprising variety of perceptions to represent the environment. There are at least two strategies. The first strategy is that representation in the hippocampal formation remains unaltered when an animal is confronted with modest changes in sensory information within an environmental or a behavioral context. The lesion studies reported that rats with lesions in the hippocampal formation or fornix were impaired when they were confronted with a new situation in the Morris water maze or a stimulus-stimulus association task, and suggest that the hippocampal formation is important for processing information effectively and flexibly (Bunsey and Eichenbaum 1996; Cohen and Eichenbaum 1993; Eichenbaum et al. 1990). Unit recording studies indicate that some hippocampal neurons could maintain their place field after darkening the experimental room (O'Keefe and Speakman 1987; Quirk et al. 1990). The rats probably switch information to be processed in the hippocampal formation from visual in light to proprioceptive and/or vestibular sensation in darkness (Knierim et al. 1995; McNaughton et al. 1991). This first strategy for the hippocampal formation to process information, i.e., flexibility in selection of information, suggests that the hippocampal neuronal correlates would be susceptible to adapting to new relevant cues if a given context changes, even in the same environment. This change in sensitivity to movement variables should be tested in the same environment, because some changes in the environment can influence place cell activity (Muller and Kubie 1987; Quirk et al. 1990). The second strategy is that representation in the hippocampal formation completely changes when an animal is confronted with a relatively large change in an environmental or behavioral context. This second strategy corresponds to creation of a new representation, or "complete remapping" (creation of a new place field) (Muller and Kubie 1987; Quirk et al. 1990) in the hippocampal formation. Theoretical studies assert that sparse and conjunctive coding enables the hippocampal formation to represent a variety of episodes with relatively little overlap (McClelland et al. 1995; McNaughton and Morris 1987; O'Reilly and McClelland 1994). Previous studies propose that the extent of remapping reflects the magnitude of the difference between two environments (Muller and Kubie 1987; Quirk et al. 1990), although a recent study reported significant effects of behavioral contexts on the place fields in the same environment (Markus et al. 1995).
). The use of intracranial self-stimulation as a reward (Shizgal 1989) offers the following advantages over food or water rewards: 1) rapid learning of a task, 2) lack of satiation, and 3) absence of visual and olfactory information regarding the reward. This protocol enables us to perform many trials under different conditions within a short time and in a well-controlled environment. In the present study, using this paradigm, we identified the place fields of hippocampal place neurons in rats as they explored a circular open field to obtain an intracranial self-stimulation reward. To investigate the two strategies in the hippocampal formation noted above, we analyzed spatial, behavioral, and reward correlates of hippocampal place neurons of rats in the different behavioral and reward tasks but in the same environment. It should be noted, however, that changes in behaviors might falsely result in behavioral correlates of place neurons due to a sampling bias (Muller et al. 1994). For example, if the rat tends to exhibit a certain behavior (type X) in the place field of a place neuron, where neuronal activity is high, and another behavior (type Y) outside the place field, where neuronal activity is low, then correlation between the neuron activity and the type X behavior would be falsely high ("distributive hypothesis" of Muller et al. 1994). Therefore we further compared actual behavioral correlates of place neurons with predicted behavioral correlates (i.e., false behavioral correlates) based on the distributive hypothesis, in which neuronal correlates to behaviors are ascribed to the difference in time spent by the rat in different portions of the firing field of the place neurons in different behaviors. Finally, we assessed the plasticity of place neuron activity under conditions in which intracranial self-stimulation was delivered only outside the place field.
, 1996).
METHODS
Abstract
Introduction
Methods
Results
Discussion
References
). The general methods are described below.
. Electrodes were attached to the skull with dental acrylic and stainless steel screws, one of which also served as the indifferent electrode for stimulation.
View larger version (33K):
[in a new window]
FIG. 1.
Scheme of experimental setup. Aa: open field (150 cm diam) containing rat was viewed from top center by video camera that signaled rat's position in Cartesian coordinates. Open field was enclosed by black curtain. Under usual conditions, electric bulb at 3 o'clock position and speaker at 9 o'clock position on ceiling of enclosure were turned on. Video signal was sent to conventional TV monitor, and digital interface connected to video camera sent X and Y coordinates of miniature light bulb attached to head of rat through RS-232C serial port to microcomputer. Microcomputer plotted trail of rat, compared rat's behavior with preset criteria, and gated delivery of intracranial self-stimulation as reward to medial forebrain bundle at level of lateral hypothalamus (LHA) from stimulator when criteria were met. CCD, charge-coupled device. Ab: view of ceiling of enclosure. Four small speakers (S) and 4 electric bulbs (L) near inner edge of each speaker were mounted at 90° intervals. Video camera (C) was positioned at center. N, north. Ba: random search task [random reward place search task (RRPST)]. Computer program delimited circular reward place (thick circle) at some randomly selected coordinate. Rat was rewarded with intracranial self-stimulation when it entered reward place, which was then made inactive (changed to thin circle). After 5-s interval, reward place was moved to different location and reactivated. Bb: place learning task 1 (PLT1). Rat received rewards in 2 target areas (thick circles) when it returned to one reward place after visit to other reward place. Delay time before delivery of rewards was introduced from 0.5 to 2.0 s in 0.5-s increments. Bc: place learning task 2 (PLT2). Rat received rewards in target area (thick circle) only when return to target area was preceded by visit to place field, which was now nonreward place (thin circle). S: location of rat at start of session. Dots: locations of reward delivery. Convention of thick circle around reward place and thin circle around nonreward place is used in all figures. C: schematic representation of instantaneous direction ( 
n) and turning angle (
n) of movement. Arrows on curve: movement direction. Points indicated by (n 
1) × 100 ms and(n + 1) × 100 ms represent sequential locations 100 ms before and after passing through observed point (n × 100 ms). Instantaneous direction of movement at each location was calculated along vector between sequential locations 100 ms before and after passing through point observed (n). Instantaneous turning angle at each location within place field was estimated as arc subtended by 2 vectors connecting that location point and points 100 ms before and after passing through point observed (n). N, north (0°).
; Muller et al. 1987). Clusters of 4.5 × 4.5-cm pixels with firing rates exceeding twice the mean firing rate were identified. All pixels that did not satisfy this criterion were eliminated. A place field could be continued through any edge shared by two pixels meeting the criterion, but not through corners. If one or more neighboring pixels satisfied the criterion, the field was expanded to include the pixel(s). Each added pixel was then tested for the presence of a neighboring pixel that met the criterion. When no neighboring pixel satisfied the criterion, the limit of the field was identified. The minimum size for a place field was set at nine pixels. Boundaries of a place field were established by constructing a rectangle that had one diagonal connecting the minimum X and Y coordinates with the maximum X and Y coordinates.
. The instantaneous direction of movement at each location within the place field was calculated along the vector between sequential locations 100 ms before and after passing through the point observed (Fig. 1C, n). The direction of movement was determined only for those locations through which the rat traveled at a rate >5 cm/s. We examined the relationship between movement direction and neuronal activity by averaging firing rate samples within 45° angular bins. The directions whose firing rates were above the mean firing rate for all directions were defined as preferred directions. The instantaneous turning angle at each location within the place field was estimated as the arc subtended by two vectors connecting a location point and the points passed through 100 ms before and after the point observed (Fig. 1C, n). The turning angle was determined only for those locations through which the rat traveled at a rate >5 cm/s. We examined the relationship between the turning angle and neuronal activity by averaging firing rate samples within 45° angular bins. The turning angles whose firing rates were above the mean firing rate for all turning angles were defined as preferred turning angles. The speed of movement at each location in the place field was calculated from the distance traveled in the same 200-ms interval used for the direction determination. The firing rate at each location was calculated as the number of spikes in each 50 ms. Preferred speeds were defined as those whose firing rates were above the mean firing rate for all speeds. The relation between movement speed and firing rate was plotted in 20-cm/s bins. The statistical reliability of each movement variable was evaluated by 
2 analysis of the firing rate at various speeds, directions, and turning angles on the basis of at least five observations.
, which was derived from the equation for entropy based on information theory (Shannon and Weaver 1949). For each neuron, the breadth of responsiveness to each behavioral variable (i.e., speed, direction, and turning angle) was separately calculated from the following formula (Smith and Travers 1979)
where H is breadth of responsiveness, K is a scaling constant, and pi is the relative response to each parameter of speed, direction, or turning angle. The value of K was adjusted so that H = 1.0 when pi was 1/n. On the basis of this definition, H = 0 means that the activity of a given neuron is highly tuned to one specific parameter, whereas H = 1 means that activity of a neuron is broadly tuned to all of the parameters in each variable. Entropy measures of breadth of responsiveness between the random search task and place learning task 1 were compared by paired t-test (P < 0.05).
proposed a distributive hypothesis, in which neuronal correlates to direction are ascribed to the difference in time spent by the rat in different portions of the firing field of the place cells in different direction sectors. Because the behavioral trails during navigation were more constant in place learning task 1 than in the random search task, the changes in the entropy measurements of breadth of responsiveness from the random search task to place learning task 1 might be ascribed to "distributive errors." To analyze this issue, we calculated the predicted firing in each parameter of speed, direction, and turning angle, with the use of the data within place fields and on the basis of the formula by Muller et al. (1994) in which place neurons are supposed to have no correlation to speed, direction, and turning angle of movements. The predicted firing rate at a parameter of 
[R()] is
where Pi is the firing rate in a pixel, Ti(
![<IT>R</IT>(ϕ) = <LIM><OP>∑</OP><LL><SUB><IT>i</IT>=1</SUB></LL><UL><SUP><IT>i=n</IT></SUP></UL></LIM>[<IT>Pi × Ti</IT>(ϕ)]/<LIM><OP>∑</OP><LL><SUB><IT>i</IT>=1</SUB></LL><UL><SUP><IT>i=n</IT></SUP></UL></LIM>[<IT>Ti</IT>(ϕ)]](/content/vol78/issue2/fulltext/597/img002.gif)
) is the time spent at the parameter of in the pixel, and n is the number of pixels within the place field.
48 h. Sections (75 µm) were cut on a freezing microtome and were then stained with cresyl violet.
RESULTS
Abstract
Introduction
Methods
Results
Discussion
References
View larger version (47K):
[in a new window]
FIG. 2.
Example of spatial, reward, and behavioral correlates of place neuron activity during random search task. A: trail of exploration (left) and map of firing rate (right). This neuron increased its activity near center of open field. Large rectangle: place field of neuron (see detail in METHODS). Dots in trail map: locations of reward delivery. B: curve of average movement speed (top) and histogram of average firing rates (bottom), which are time locked to onset of intracranial self-stimulation within (Ba) and outside (Bb) place field. Horizontal bar below abscissa: duration of reward. C: relation between neuronal activity and speed (Ca), direction (Cb), and turning angle of movement (Cc) within place field. In Cb, 0° corresponds to direction of vector extending from center to 12 o'clock. In Cc, turning angles greater than and <0° correspond to right and left turns, respectively. H, numerical measure of breadth of hippocampal neuronal responses to each behavioral variable. A-C are from same neuron.
2 = 86.3, df = 5, P < 0.01; preferred speed = 50-90 cm/s), direction (2 = 64.3, df = 7, P < 0.01; preferred direction = 0-135°), and turning angle (2 = 53.0, df = 7, P < 0.01; preferred turning angle = 45-45°) over a wide range. Of the 37 place neurons, 28 had significant direction-dependent responses (CA1, 13; CA3, 15), 24 had significant speed-dependent responses (CA1, 11 neurons; CA3, 13), and 21 had significant turning-angle-dependent responses (CA1, 9; CA3, 12). To quantitatively assess the breadth of responsiveness to each behavioral variable, a numerical measure of the breadth of responsiveness was introduced. Although the neuron in Fig. 2 demonstrated significant behavioral variable-dependent responses, numerical measures of breadth of neuronal response to each behavioral variable were rather high (H 0.96).
View larger version (52K):
[in a new window]
FIG. 3.
Example of spatial and reward correlates of place neuron activity during random search task and place learning task 1. A-C: trail of rat, reward locations, and spatial distribution of neuronal activities in random search task (A), place learning task 1 without delay for intracranial self-stimulation (B), and place learning task 1 with 1.0-s delay (C). During random search task, place field was located in quadrant located between 1 and 3 o'clock (Aa), and activity increased before reward delivery within place field (Ab) but not outside place field (Ac). During place learning task 1 without (B) and with (C) delay for reward delivery, neuron had spatial correlates consistent with those observed in random search task (Ba and Ca). Activity of this neuron increased before reward delivery in reward place located within place field (Bb and Cb) but not outside place field (Bc and Cc). Inside: inside place field. Outside: outside place field. Speed: speed of movement. Thin lines below abscissas in Cb: delay periods in reward place. A-C are from same neuron. Other conventions as for Fig. 2.
View larger version (18K):
[in a new window]
FIG. 4.
Spatial correlates of place neuron activity shown in Fig. 3 in different phases of place learning task 1 with 1.0-s delay. A: firing rate histogram is same as that shown in Fig. 3Cb. B: trail of rat is shown as small dots, sampled at 50-ms intervals, and firing rate map is superimposed. Neuron fired at lower rate after reward delivery, although rat stayed in place field (Bc and Bd). Dashed rectangles: place field of neuron identified by random search task in Fig. 3. Other conventions as for Figs. 2 and 3.
View larger version (54K):
[in a new window]
FIG. 5.
Another example of spatial and reward correlates of place neuron activity during random search task and place learning task 1. A-C: trail of rat, reward locations, and spatial distribution of neuronal activities in random search task (A), place learning task 1 without delay for intracranial self-stimulation (B), and place learning task 1 with 1.0-s delay (C). During random search task, place neuron had place fields at 6 o'clock and at 8-10 o'clock (Aa). During place learning task 1 without (B) and with (C) delay time, neuron increased its activity during and after reward delivery in reward place within place field (Bb and Cb) but not outside place field (Bc and Cc). A-C are from same neuron. Other conventions as for Figs. 2 and 3.
View larger version (15K):
[in a new window]
FIG. 6.
Spatial correlates of place neuron activity shown in Fig. 5 in different phases of place learning task 1 with 1.0-s delay. A: same firing rate histogram as shown in Fig. 5Cb. B: trail of rat is shown as small dots sampled at 50-ms intervals, and firing rate map is superimposed. Neuron fired at lower rate before reward delivery, although rat stayed in place field (Ba and Bb). Dashed rectangles: place field of neuron identified by random search task in Fig. 5. Other conventions as for Figs. 2 and 3.
2 = 38.3, df = 4, P < 0.01; preferred speed = 30-90 cm/s; Fig. 7Aa), direction (2 = 85.6, df = 7, P < 0.01; preferred direction = 0° and 90-180°; Fig. 7Ab), and turning angle (2 = 53.79, df = 7, P < 0.01; preferred turning angle = 45-0-180°; Fig. 7Ac) over a wide range. However, neuronal correlates to some behavioral variables became more evident in place learning task 1. In place learning task 1 without delay, the neuronal activity was significantly modulated by speed (2 = 37.8, df = 4, P < 0.01; preferred speed = 10-30 cm/s; Fig. 7Ba), direction (2 = 20.7,df = 7, P < 0.01; preferred direction = 135-180° and 315°; Fig. 7Bb), and turning angle (2 = 12.47, df = 7, P < 0.01; preferred turning angle = 45°, 135-180°, and 45°; Fig. 7Bc). A comparison of the numerical measures of the breadth of responsiveness to behavioral variables between the random search task and place learning task 1 without delay indicated that the breadth of responsiveness in speed (H = 0.80) in place learning task 1 without delay became smaller than the breadth of responsiveness in speed in the random search task (H = 0.99), whereas there were no changes in the breadth of responsiveness in direction (0.98 vs. 0.98) and only a slight increase in the breadth of responsiveness in turning angle (0.98 vs. 0.99).
View larger version (33K):
[in a new window]
FIG. 7.
Neuronal correlates of behavioral variables of place neuron activity shown in Figs. 3 and 4 during random search task and 2 conditions of place learning task 1 within place field. A-C: neuronal correlates of speed (a), direction (b), and turning angle (c) of movement in random search task (A), place learning task 1 without delay (B), and place learning task 1 with 1.0-s delay (C). In random search task, preferred speed, direction, and turning angle were 30-90 cm/s (Aa), 0° and 90-180° (Ab), and 45-0-180° (Ac), respectively. In place learning task 1 without delay (B), neuronal correlates of firing shifted to ~10-30 cm/s for speed (Ba), to 135-180° and 315° for direction (Bb), and to 45°, 45°, and 135-180° for turning angle (Bc). In place learning task 1 with 1.0-s delay (C), neuronal correlates of firing shifted to ~10-50 cm/s for speed (Ca), to 45-0-90° and 180° for direction (Cb), and to 45-90° and 45° for turning angle (Cc). H, numerical measure of breadth of responsiveness in speed (a), direction (b), and turning angle (c) of movement. Note that numerical measures of breadth of responsiveness in speed in place learning task 1 without delay (H = 0.80; Ba) and with 1.0-s delay (H = 0.89; Ca) became smaller than in random search task (H = 0.99; Aa), and that numerical measures of breadth of responsivenessin direction in place learning task 1 with 1.0-s delay(H = 0.95; Cb) became smaller than in random search task (H = 0.98; Ab). Other conventions as for Fig. 2. A-C are from same neuron.
2 = 73.0, df = 5, P < 0.01; preferred speed = 10-70 cm/s; Fig. 7Ca), direction (2 = 119.5, df = 7, P < 0.01; preferred direction = 315-0-180°; Fig. 7Cb), and turning angle (2 = 11.62, df = 7, P < 0.01; preferred turning angle = 45-90° and 45°; Fig. 7Cc). The numerical measures of the breadth of responsiveness to speed (H = 0.89) and direction (H = 0.95) in place learning task 1 with 1.0-s delay became smaller than those in the random search task (Fig. 7, A and C). However, it should be emphasized that these changes in breadth of responsiveness were observed only inside the place field, but not outside the place field (not shown).
2 = 4.8,df = 5, P > 0.05; Fig. 8Aa), but the directional (2 = 46.2, df = 7, P < 0.01; preferred direction = 45-180°; Fig. 8Ab) and turning angle (2 = 10.97, df = 7, P < 0.01; preferred turning angle = 90-135° and 90°; Fig. 8Ac) modulation were significant over a wide range. In place learning task 1 without delay, the neuronal activity developed significant speed (2 = 35.3, df = 5, P < 0.01; preferred speed = 10-70 cm/s; Fig. 8Ba), directional (2 = 69.5, df = 6, P < 0.01; preferred direction = 45-135° and 225°; Fig. 8Bb), and turning angle (2 = 17.86, df = 6, P < 0.01; preferred direction = 45°, 135-180°, and 45°; Fig. 8Bc) modulations that narrowed to more limited ranges. In place learning task 1 with 1.0-s delay, the neuronal activity was significantly modulated by speed (2 = 20.0, df = 5, P < 0.01; preferred speed = 30-90 cm/s; Fig. 8Ca), direction (2 = 46.7,df = 7, P < 0.01; preferred direction = 45-180°; Fig. 8Cb), and turning angle (2 = 20.86, df = 7, P < 0.01; preferred direction = 90°, 180°, and 135°; Fig. 8Cc). Although the neuronal correlates to the behavioral variables, except for the speed in the random search task, were evident in all three conditions, a comparison of the breadth of responsiveness indicated that all three numerical measures of the breadth of responsiveness for speed, direction, and turning angle became smaller in place learning task 1 than in the random search task (Fig. 8). In accordance with the place neuron shown in Figs. 3, 4, and 7, the breadth of the neuronal responses changed only inside the place field.
View larger version (31K):
[in a new window]
FIG. 8.
Neuronal correlates of behavioral variables of place neuron activity shown in Figs. 5 and 6 during random search task and place learning task 1 within place field. A-C: neuronal correlates of speed (a), direction (b), and turning angle (c) of movement in random search task (A), place learning task 1 without delay (B), and place learning task 1 with 1.0-s delay (C). In random search task, neuron fired at all speeds (Aa), at all directions except for 0° (Ab), and at all turning angles, especially 90° (Ac). In place learning task 1 without delay (B), neuronal correlates of firing shifted to ~10-70 cm/s for speed (Ba), to 45-135° and 225° for direction (Bb), and to 45°, 135-180°, and 45° for turning angle (Bc). In place learning task 1 with 1.0-s delay (C), neuronal correlates of firing shifted to ~30-90 cm/s for speed (Ca), to 45-180° for direction (Cb), and to 90°, 180°, and 135° for turning angle. Note that numerical measures of breadth of responsiveness in speed in place learning task 1 without delay (H = 0.90; Ba) and with 1.0-s delay (H = 0.80; Ca) became smaller than in random search task (H = 0.99; Aa), that numerical measures of breadth of responsiveness in direction in place learning task 1 without delay (H = 0.83; Bb) and with 1.0-s delay (H = 0.79; Cb) became smaller than in random search task (H = 0.95, Ab), and that numerical measures of breadth of responsiveness in turning angle in place learning task 1 without delay (H = 0.87; Cb) and with 1.0-s delay (H = 0.85; Cc) became smaller than in random search task (H = 0.97, Ac). Other conventions as for Fig. 2. A-C are from same neuron.
(Fig. 9). In all three tasks (the random search task, place learning task 1 without delay, and place learning task 1 with delay), the entropy measures of breadth of responsiveness calculated from actual responses (filled columns) were significantly smaller than those calculated from predicted responses (open columns) to each of three parameters (paired t-test, P < 0.01 or 0.05). These results indicate that actual neuronal tuning to behavioral variables in all three tasks was significantly more selective than the neuronal tuning that was predicted according to the assumption that place neurons have no correlation to speed, direction, and turning angle of movements.
View larger version (32K):
[in a new window]
FIG. 9.
Comparison of numerical measure of breadth of responsiveness of hippocampal place neurons to each behavioral variable among random search task, place learning task 1 without delay, and place learning task 1 with delay. A-C: mean numerical measures of breadth of responsiveness (left; filled and open columns) to speed (A), direction (B), and turning angle (C) calculated with both predicted and actual responses, and mean differences between predicted and actual breadth of responsiveness (right; hatched columns). Open columns: entropy measures of breadth of responsiveness calculated with predicted responses based on formula by Muller et al. (1994) . Filled columns: entropy measures of breadth of responsiveness calculated with actual responses. Note that in all 3 tasks (random search task, place learning task 1 without delay, and place learning task 1 with delay), entropy measures of breadth of responsiveness calculated with actual responses (filled columns) were significantly smaller than those calculated with predicted responses (open columns) to each of 3 parameters (paired t-test, P < 0.01 or 0.05), and that breadth of responsiveness of hippocampal place neurons based on actual responses to speed, direction, and turning angle in place learning task 1 without and with delay was significantly smaller than that in random search task (paired t-test, P < 0.01 or 0.05). Single asterisk: significant at level of P < 0.05. Double asterisk: significant at level of P < 0.01. Hatched columns, mean differences in breadth of responsiveness between predicted and actual responses. Post hoc test after 2-way analysis of variance (ANOVA) indicated that mean difference between predicted and actual breadth of responsiveness in place learning task 1 with delay was significantly larger than that in random search task (Newman-Keuls test, P < 0.05) and larger than that in place learning task 1 without delay (Newman-Keuls test, P < 0.05). Data were sampled from 21 hippocampal place neurons tested at least with random search task and place learning task 1 without or with delay.
View larger version (38K):
[in a new window]
FIG. 10.
Example of spatial correlates of place neuron activity that did not change in place learning task 2. A: place field was located between 9 and 11 o'clock and subfield was located at 1 o'clock in random search task. B: spatial correlates were retained in place learning task 1, in which 1 reward place was located in place field at 9-11 o'clock, and was 1st observed in random search task. C and D: spatial correlates were not changed in 1st (C) and 4th sessions (D) of place learning task 2, where reward was delivered only in reward place at 4 o'clock. A-D are from same neuron. Other conventions as for Fig. 2.
View larger version (45K):
[in a new window]
FIG. 11.
Example of place neuron that developed new place field in place learning task 2 while it retained place field in random search task. A: place field was located at 5-6 o'clock in random search task. B: spatial correlates were retained in place learning task 1, where 1 reward place was located in place field observed in random search task. C: spatial correlates at 5 o'clock were retained, but new spatial correlates appeared at 11 o'clock in 1st session of place learning task 2, where reward was delivered only in reward place at 11 o'clock. D: spatial correlates at 5 o'clock were retained, but new spatial correlates at 11 o'clock persisted in 4th session of place learning task 2. A-D are from same neuron. Other conventions as for Fig. 2.
View larger version (58K):
[in a new window]
FIG. 12.
Example of place neuron that completely shifted its place field during sessions of 2 place learning tasks. A: place field was located at 6 o'clock in random search task. B: in place learning task 1, reward was presented in place field observed in random search task. C: 1st session of place learning task 2, in which reward was delivered only in reward place at 0 o'clock. D: 4th session of place learning task 2. E and F: 2 successive sessions of place learning task 1 after place learning task 2. G: gradual changes of center of place field. Random search task (1), place learning task 1 (2), 1st-4th sessions of place learning task 2 (3-6), and 1st-3rd sessions of place learning task 1 after place learning task 2 (7-9). A-G are from same neuron. Other conventions as for Fig. 2.
View larger version (26K):
[in a new window]
FIG. 13.
Recording sites of hippocampal place neurons. All recorded neurons were located in CA1 and CA3 subfields of hippocampal formation. Dots: place neurons. Number above each section: distance (mm) posterior to bregma.
DISCUSSION
Abstract
Introduction
Methods
Results
Discussion
References
suggested that the reward would work as a sensory cue or a landmark. This indicated that the place neurons in the present study might develop a spatial reference frame relative to the reward place. This does not appear likely from the results of the present study. First, the difference between random and fixed reward deliveries in the random search task and place learning task 1 did not affect the spatial correlates of the place neurons in both tasks. Second, the change in neuronal activity before or after delivery of intracranial self-stimulation was similar in both the random search task and place learning task 1 although locations of delivery of intracranial self-stimulation were different between the random search task and place learning task 1. In the random search task the animals could not use reward delivery as a landmark because it was delivered in random locations, but in place learning task 1 the animals could use it as a landmark because it was delivered in fixed locations (i.e., reward places). Therefore the similarity of changes in neuronal activity aligned with reward delivery in both tasks strongly suggests that the neuronal activity reflected the motivational aspects rather than the spatial significance of the intracranial self-stimulation reward. The effects of intracranial self-stimulation were also not ascribed to other coincidental factors such as behavioral changes, because the results from the three different behavioral tasks were consistent. Therefore no spatial, behavioral, or reward factors could account for all of the activity of a place neuron, because intracranial self-stimulation only influenced the activity inside the place field. These results suggest that the activity of a place neuron is determined by a constellation of all of the factors.
; Otto and Eichenbaum 1992; Sakurai 1990; Wible et al. 1986), odor discrimination tasks (Eichenbaum et al. 1987; Wiener et al. 1989), and a nonspatial radial maze task (Young et al. 1994), as well as with reward delivery (Hampson et al. 1993). Little attention was focused on the neuronal correlates to the reward itself in those studies. The present experiments show two types of neuronal responses synchronous with reward delivery in the place field during the random search task and place learning task 1. Neurons fired most 1) from the time of arrival in the place fields until the onset of reward (type 1) and 2) from the onset of reward until the start of approach to the opposite reward place (type 2). The type 1 neurons might correspond to the approach-consummate cells (Ranck 1973) and goal-approach cells (Eichenbaum et al. 1987) that fired when the rat approached the food/water or goal. Because these type 1 neurons continued to fire during the delay period before delivery of intracranial self-stimulation, they might also correspond to the approach-consummate mismatch cells (Ranck 1973), or to the misplace cells (O'Keefe 1976) that fired when a food reward was not found at the goal. The type 2 neurons might correspond to the hippocampal neurons, such as motion punctuate cells (Ranck 1973), and cup-approach cells (Wiener et al. 1989), that fired at the end of approach movements, or to consummatory cells that fired during consummatory behaviors (Ranck 1973).
in which they reported that the firing of hippocampal place neurons was about the same when the rat moved in opposite directions inside the place field. The present results, in which the place neuron activity was nondirectional, or had broad directional correlation in the random search task, are similar to the results of Muller and Kubie (1987) and Muller et al. (1994), and suggest that the place neuron activity is poorly correlated with the direction of rats' movement when the rats randomly move in all directions.
, Breese et al. (1989), Wiener et al. (1989), and Muller et al. (1994). McNaughton et al. (1983) tested place neurons in an eight-arm radial maze, and found place neurons that increased their firing rates only when the subject was passing through an arm of the maze in one direction. Breese et al. (1989) and Wiener et al. (1989) used square chambers in which reward was available in each corner, and reported place neuron activity correlated with the movement speed, direction, and turning angle of the animal. These results suggested that place neuron activity is correlated with the directional parameter when rats move linearly (Muller et al. 1994). Recently, Markus et al. (1995) also reported an increase in neuronal correlates to direction in a directed search task in comparison with a random search task in the hippocampal formation. However, there are some differences between our results and those of Markus et al. The increase in neuronal correlates to direction was largely ascribed to creation of new directional place fields rather than changes in neuronal correlates in the same place fields (Markus et al. 1995), whereas neuronal correlates to direction increased in the same place fields in the present study. The present results indicate that, even in the same environment, neuronal correlates not only to direction, but also to speed and turning angle, flexibly increase according to behavioral requirement in the tasks.
). Passive displacement of the restrained animals greatly reduced the place correlates of the hippocampal neurons (Foster et al. 1989). This reduction in place correlates may be attributed to disturbed or decreased attention to environmental cues. A monkey's active intention to move the cab that it drove was important to the formation of place-related activity by hippocampal neurons; passive movement of the monkey cab greatly reduced place-related activity (Nishijo et al. 1997). Furthermore, the amplitudes of sensory evoked potentials responding to auditory, visual, and somatosensory cues in the human hippocampal formation were markedly reduced when the subject's attention was directed elsewhere (McCarthy et al. 1989). These observations from other studies also suggest the existence of active information selection in the hippocampal formation.
and Speakman and O'Keefe (1990) focused their attention on the plasticity of spatial correlates of place neuron activity in relation to reward. Speakman and O'Keefe (1990) found that shifting the goal in a four-arm elevated plus maze by 180° relative to controlled cues elicited a goal-related change of place field in only 2 of 19 place neurons. Breese et al. (1989) showed that place fields were established during a place task in which a water reward was available at five loci on a rectangular platform. In 40 of 47 neurons, the old place fields disappeared and new place fields emerged at a single locus where reward was available. Wiener et al. (1989) compared the spatial and behavioral correlates of hippocampal complex-spike cell activity in spatial and nonspatial paradigms, and showed that of 52 complex-spike cells 41 demonstrated place fields at a different location during a place task than from an odor task performed on the same platform. Consistently, Markus et al. (1995) also reported that place fields changed under different behavioral tasks. These studies indicate that hippocampal place cells behave differently as animals perform different tasks.
, place fields were located in various parts of the apparatus; according to Breese et al. (1989), most place fields were located near the cups from which a water reward was delivered before testing, and new place fields emerged at a single locus where reward was available. These differences might be due to differences in the behavioral requirements of the two tasks. The behavioral task by Speakman and O'Keefe (1990) was a spatial reference memory task, in which the position of the rat, in a frame of reference to extramaze spatial cues, was important to the accomplishment of the task. In the behavioral tasks by Breese et al. (1989), the rats might determine their positions by using a local maze-based frame; here, the relation of intramaze cues to a specific cup associated with reward delivery was probably more relevant than extramaze cues.
; Bichot et al. 1996; Kirkwood et al. 1996). The present results, along with an earlier study (Markus et al. 1995), add evidence of learning-dependent changes in the place fields of the place neurons in the hippocampal formation. It has been suggested that ensemble, or population neuron, but not single-neuron, coding in the hippocampal formation effectively represents space (Wilson and McNaughton 1993), or various task-relevant features (Deadwyler et al. 1996). Therefore, if the population neuron coding of space is considered, an increase in the number of place neurons with reference to a reward area after place learning task 2 may contribute to more precise representation of the new reward area or the path integration to the new reward area.
; McNaughton et al. 1991). In the present study, the place neurons did not alter the locations of their place fields when the paths to the reward loci changed from random (i.e., random search task) to fixed places (i.e., place learning task 1). Although these results seem to contradict the path integration hypothesis, they do not refute it. For one thing, the mere existence of place cells may contribute to path integration. A computational study suggested that the entire synaptic connection along a chain of place cells, the initial and last two place cells of which represented start and goal locations, was comparable with path integration, and that this synaptic connection would be refined by learning or experience so that there would be a minimum number of place cells in the chain (Muller et al. 1996). The stability of the place fields between the random search task and place learning task 1, as shown in the present study, is a prerequisite for this computational hypothesis. Second, the increase in the breadth of tuning of place neuron responses in place learning task 1 indicated that the hippocampal place neurons became more sensitive to behavioral variables in place learning task 1 than in the random search task. Behavioral variables were more relevant and path integration was more important in place learning task 1 than in the random search task. These results strongly suggest that place neurons more effectively encode information regarding behavioral variables for path integration in place learning task 1, because ideothetic information regarding behavioral variables (i.e., internally generated sensation such as proprioceptive and vestibular sensation in contrast to externally generated sensation such as visual and auditory sensation) is the primary information to be processed for path integration (Knierim et al. 1996; McNaughton et al. 1991).
; Eichenbaum 1993; Eichenbaum et al. 1990, 1992; Young et al. 1994) or conjunctively (Knierim et al. 1995; McNaughton et al. 1989) encoded in the hippocampal formation, because hippocampal neuronal correlates of relevant information were associated with a specific area (i.e., place field) in the present study. These findings correspond to the neurophysiological bases of human episodic memory suggested by theoretical and computation studies (McClelland et al. 1995; McNaughton and Morris 1987; McNaughton and Nadel 1989).
| |
ACKNOWLEDGEMENTS |
|---|
We thank Drs. R.G.M. Morris and S. I. Wiener for valuable comments.
This work was supported in part by Japanese Ministry of Education, Science and Culture Grants-in-Aid for Scientific Research 08408036, 08279105, 08279215, 08234209, 07244103, and 08680884, and by Funds for Comprehensive Research on Aging and Health.
Present addresses: T. Kobayashi, Supermolecular Science Division, Electrotechnical Laboratory, Tsukuba, Ibaraki 305, Japan; M. Fukuda, Dept. of Behavioral Sciences, Faculty of Medicine, Toyama Medical and Pharmaceutical University, Sugitani 2630, Toyama 930-01, Japan.
| |
FOOTNOTES |
|---|
Address for reprint requests: T. Ono, Dept. of Physiology, Faculty of Medicine, Toyama Medical and Pharmaceutical University, Sugitani 2630, Toyama 930-01, Japan.
Received 8 November 1996; accepted in final form 22 April 1997.
| |
REFERENCES |
|---|
|
Abstract
Introduction
Methods
Results
Discussion
References
|
|---|
what does it do?
Behav. Neural. Biol.
57: 2-36, 1992.[Medline]New Vistas,
edited by
V. Chan-Palay,
and C. Kohler
. New York: Liss, 1989, p. 425-441This article has been cited by other articles:
|
|
 |
|
  M. Takahashi, J. Lauwereyns, Y. Sakurai, and M. Tsukada A Code for Spatial Alternation During Fixation in Rat Hippocampal CA1 Neurons J Neurophysiol, July 1, 2009; 102(1): 556 - 567. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. J. Siegel, J. P. Neunuebel, and J. J. Knierim Dominance of the Proximal Coordinate Frame in Determining the Locations of Hippocampal Place Cell Activity During Navigation J Neurophysiol, January 1, 2008; 99(1): 60 - 76. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. L. Griffin, H. Eichenbaum, and M. E. Hasselmo Spatial Representations of Hippocampal CA1 Neurons Are Modulated by Behavioral Context in a Hippocampus-Dependent Memory Task J. Neurosci., February 28, 2007; 27(9): 2416 - 2423. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. R. Bower, D. R. Euston, and B. L. McNaughton Sequential-Context-Dependent Hippocampal Activity Is Not Necessary to Learn Sequences with Repeated Elements J. Neurosci., February 9, 2005; 25(6): 1313 - 1323. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. H. Tran, R. Tamura, T. Uwano, T. Kobayashi, M. Katsuki, and T. Ono Dopamine D1 receptors involved in locomotor activity and accumbens neural responses to prediction of reward associated with place PNAS, February 8, 2005; 102(6): 2117 - 2122. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. A. P. Moita, S. Rosis, Y. Zhou, J. E. LeDoux, and H. T. Blair Putting Fear in Its Place: Remapping of Hippocampal Place Cells during Fear Conditioning J. Neurosci., August 4, 2004; 24(31): 7015 - 7023. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. P. Battaglia, G. R. Sutherland, and B. L. McNaughton Local Sensory Cues and Place Cell Directionality: Additional Evidence of Prospective Coding in the Hippocampus J. Neurosci., May 12, 2004; 24(19): 4541 - 4550. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. H. Tran, R. Tamura, T. Uwano, T. Kobayashi, M. Katsuki, G. Matsumoto, and T. Ono Altered accumbens neural response to prediction of reward associated with place in dopamine D2 receptor knockout mice PNAS, June 25, 2002; 99(13): 8986 - 8991. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. A. Gusev and D. L. Alkon Intracellular Correlates of Spatial Memory Acquisition in Hippocampal Slices: Long-Term Disinhibition of CA1 Pyramidal Cells J Neurophysiol, August 1, 2001; 86(2): 881 - 899. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. A. Hollup, S. Molden, J. G. Donnett, M.-B. Moser, and E. I. Moser Accumulation of Hippocampal Place Fields at the Goal Location in an Annular Watermaze Task J. Neurosci., March 1, 2001; 21(5): 1635 - 1644. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Matsumura, H. Nishijo, R. Tamura, S. Eifuku, S. Endo, and T. Ono Spatial- and Task-Dependent Neuronal Responses during Real and Virtual Translocation in the Monkey Hippocampal Formation J. Neurosci., March 15, 1999; 19(6): 2381 - 2393. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
