Hippocampal theta activity related to elicitation and inhibition of approach locomotion

Transcription

1 Behavioural Brain Research 160 (2005) Research report Hippocampal theta activity related to elicitation and inhibition of approach locomotion Harry M. Sinnamon Neuroscience and Behavior Program, Wesleyan University, Judd Hall, 207 High Street, Middletown, CT , USA Received 22 September 2004; received in revised form 4 December 2004; accepted 6 December 2004 Abstract This study determined if the hippocampal theta rhythm showed phase relationships or changes in amplitude and frequency with the onset of stimuli and locomotion in a task in which auditory cues initiated and suppressed approach locomotion. Rats with electrodes in the dorsal hippocampus lapped at a milk dipper and were presented a tone which predicted the delivery of a food pellet. In some trials the pellet cue tone was negated by 60-Hz clicks beginning 0.3 s after onset, and no pellet was delivered. A video capture system (20-ms sampling) synchronized to the hippocampal recording system (10-ms sampling) was used to determine the onset of locomotor approach to the pellet area. The findings failed to support proposals that phase-related mechanisms play a role in encoding and retrieval of movement-related information. Neither the pellet cue nor the negating cue reset the theta rhythm, and they did not produce differential evoked potentials. During milk lapping, theta amplitude increased in the 1/2 s prior to all pellet cues regardless of their locomotor effect. Frequency also rose but only when a nonnegated pellet elicited short-latency locomotion. During locomotor execution, theta peak amplitude peaked earlier than theta frequency by approximately one period. In general during performance of this task, increasing theta amplitude reflected a general preparation to process the cue and increasing theta frequency reflected the readiness to respond to the cue with locomotion Elsevier B.V. All rights reserved. Keywords: Theta; Hippocampus; Approach; Locomotion; Inhibition; Reward; Attention There is general agreement that behavioral states characterized by prominent hippocampal theta activity are associated with increased levels of information processing related to the control of behavior [4,14,38 40]. However, the mechanisms used by the processes, the types of information processed, and the circuitry implementing behavioral control all remain elusive. Particular interest has developed in the temporal and phase dynamics of the theta rhythm because of the correspondence between theta frequencies and optimal parameters for long term potentiation [19,20,24]. Proposals have been made for theta phase-related mechanisms both for the encoding [21] and for the retrieval of movement-related information [19]. If the theta rhythm does operate in this way, it would be expected that phase-related patterns in theta Tel.: ; fax: address: hsinnamon@wesleyan.edu. would be present for stimuli that are being encoded, or that initiate retrieval of conditioned movements. Consistent with this idea, resetting of the theta rhythm was found in some cases with the presentation of an auditory cue that elicited approach behavior in a classical conditioning situation [9]. Consistent resetting has been found for auditory cues that required processing in working memory and that controlled differential instrumental behavior [16]. The general relationship between theta activity and movement is well-established. Theta activity is prominent during the performance of locomotor and orienting behaviors and is minimal during immobility and performance of repetitive consummatory and instrumental behaviors [5,15,36,41] in the absence of postural adjustments [36]. Theta activity of a lower frequency also appears during fear-related immobility in the rat [30]. Synchronization of theta to movements could represent either dependency on either the motor act or the /$ see front matter 2004 Elsevier B.V. All rights reserved. doi: /j.bbr

2 H.M. Sinnamon / Behavioural Brain Research 160 (2005) implicit sensory events related to it. Such movement-related phasing has been reported for sniffing [14,22,23]; and instrumental bar pressing behavior [8,31]. Synchronization of theta rhythm to the onset of locomotion has not been reported, but less time-locked modulations of theta activity prior to locomotion have been found. Theta frequency increases prior to the rapid onset of locomotor behaviors in the behaving rat [25,37,41]. In the anesthetized rat, low frequency theta activity increases prior to the initiation of locomotion [32,34]. It also appears prior to the initiation of a locomotor aversive movement in competitive feeding situation [26]. The purpose of the present study was to specify the temporal dynamics of the relationship between theta activity and sensory and locomotor events in a task in which auditory cues initiate and inhibit approach. Specifically, the question was whether a theta response to a cue was time-locked and could be described as rhythm resetting or synchronization, or alternately had a more graded modulated pattern. An analogous question was asked of theta patterns related to the onset of approach locomotion. The focus of the study was the theta response to cues that inhibit approach locomotion. Although the hippocampus appears to be involved in the processing of simple appetitive cues eliciting approach locomotion [7], a recent study [33] found no phase-related responses to auditory cues that elicited spatially directed locomotor approach. One of the earliest proposed functions [12] for the hippocampus is the inhibition of behavior. Recent considerations of this idea have implicated the hippocampus in latent inhibition [18], pre-pulse inhibition [3], the inhibition of attention to interfering stimuli [10,29], and have reinterpreted the wellestablished spatial function of the hippocampus in terms of inhibition [11]. Therefore, a cue suppressing well-learned approach locomotion would be expected to engage the encoding and control mechanisms of the hippocampus, and affect the theta generation processes related to locomotion. Rats were pre-trained in a classical conditioning task to associate an auditory cue with the delivery of a food pellet until the cue reliably produced approach locomotion to the pellet location. Next, in certain trials a delayed negating cue was superimposed on the pellet cue, and the pellet was withheld. This presentation procedure is a challenging variant of the stopsignal paradigm [13] intended to maximize active encoding. Because of its difficulty, the inhibition procedure allowed the comparison of trials in which the same negating cue was either effective or not effective in suppressing approach locomotion. 1. Methods and materials 1.1. Subjects and surgery All surgical and testing procedures were approved by the Wesleyan Animal Welfare Committee. Five Male Sprague Dawley rats bred at Wesleyan University were housed on a 12:12 reversed light cycle in individual cages with ad libitum access to water. They were provided a daily food pellet ration to produce a weekly weight gain of g up to a weight of 400 g. After preliminary training in the test chamber, the rats were anesthetized by intraperitoneal injections of Nembutal (40 mg/kg). During surgery, anesthesia was supplemented as needed by 10 mg/kg intraperitoneal injections of Nembutal. Eight 1-mm holes were drilled into the skull to receive anchor screws and electrodes. The electrodes were twisted pairs of Teflon-insulated stainless steel wires (125- m diameter) with a vertical tip separations of 1.5 mm. They were placed bilaterally in the dorsal hippocampus at approximately 4.0 mm posterior to bregma, 3.0 mm lateral to the midline, and at various depths but with the superficial pole always dorsal to CA1 cell layer. An uninsulated copper wire was wrapped around the anchor screws to serve as ground connection. The electrode wire terminated in Amphenol pins that were inserted into a plastic strip secured to the skull with dental cement. The incision was infiltrated with Marcaine (0.5%), treated with topical antibiotics, and closed with a wound clip. The rat was returned to ad libitum feeding for 5 days before resuming training Apparatus Test chamber The test chamber, illustrated in Fig. 1, had a floor 61 cm 25 cm and sidewalls 35 cm high that slanted outward 15. The front wall (panel B) contained an acrylic window for the video camera (Hitachi Denshi KP-M2U, 6-mm lens). A stainless steel tube protruded from a lower corner of the window to deliver a 45-mg food pellet to a circular tray recessed in the floor. The rat initiated the trial by using its forepaw to depress a clear acrylic treadle (3.5 cm 24 cm) centered in the floor 24 cm in front of the front wall (panel A). In the up position, the treadle protruded 1.3 cm above the floor, and an embedded light emitting diode (LED) was lit. Depression of the treadle extinguished the LED, and raised a dipper through an access hole in the floor to present 0.1 cc of sweetened condensed milk and water (1:1 by volume) for 1.36 s. When the dipper retracted, a 4- khz tone at 65 db was produced for 0.7 s by a speaker located above the camera window (panel B). On pellet trials (panel C), the pellet dispenser was activated at the offset of the tone resulting in the presentation of the pellet approximately 0.7 s later. On negation trials, the speaker above the pellet cue speaker produced clicks (60 Hz) beginning 0.3 s after the onset and continuing for its duration; no pellet was delivered on these trials (panel D). The rat restarted the trial by turning away from the pellet area and entering the rear of stall formed by two acrylic walls that lead to the treadle (panel A) Video recording A frame grabber board (Data Translation 3152) captured single monochrome video fields of pixels at rate of 50 fields/s. The video acquisition and storage was controlled by a custom program (BEProbe) running on a standard PC (Dell Optiplex GT110). The program is available in binary executable and Visual Basic 6 source code form at Each 5.12-s trial was associated with 256 frames Analog recordings A cable equipped with two dual operational amplifiers (LMC 6482) configured as voltage followers was connected to the strip on the rat s head. The other end terminated at a 9-channel slip-ring mounted on a counterbalanced arm. The paired outputs of the operational amplifiers were led to Grass P15 differential amplifiers

3 238 H.M. Sinnamon / Behavioural Brain Research 160 (2005) Fig. 1. Schematic illustration of the test apparatus. In the self-paced trial, the rat entered the stall to depress an illuminated treadle and lapped at a milk dipper presented for 1.36 s at a hole in the floor. On pellet trials (panel C), the retraction of the dipper was simultaneous with the onset of a 0.7-s, 4-kHz tone from the lower speaker. At the offset of the tone, a dispenser was activated and a food pellet exited the delivery tube to rest in the recessed tray approximately 0.7 s later. The rat could initiate locomotor approach to the pellet either during or after the pellet cue. After obtaining the pellet, the rat entered the rear of the stall to start another trial. On approximately half of the trials, a negating cue (clicks from the upper speaker) was superimposed on the pellet cue starting at 0.3 (panel D) and no pellet was dispensed. Correct behavior on these trials was suppression of locomotor approach to the pellet area and a direct turn to the rear of the stall. The depression of the treadle triggered two synchronized computers to acquire the data. One stored the current 64 video frames in a circulating buffer and then stored the subsequent 192 frames; the other acquired the parallel analog measures at a sampling rate of 1 khz, and stored them at a sampling interval of 100 ms. and low and high half-amplitude filter settings at 1 and 30 Hz. A Microstar DAP 2400 A/D board mounted in a separate computer sampled these signals at 1 khz, and to reduce the data storage demands, the digitized values were averaged over 10 samples to yield an effective sampling rate of 100 Hz. An Analog Devices ADXL05 accelerometer was mounted near the rat end of the recording cable. The single axis was oriented so that forward and upward movements produced upward deflections. Two force transducers (WP Instruments) mounted under a movable section of the floor in the stall registered the rat s presence in the stall. These signals and the various event markers were sampled similarly to the hippocampal activity. The video and A/D recording acquisition systems used continuously updated buffers containing 1.28 s of data. When the rat depressed the treadle, the two acquisition systems stored the data from their buffers and then stored the next 3.84 s of their respective data. For each 20-ms video frame there were two values for each recording channel. Each was the mean of 101-kHz samples, one corresponded to the first 10-ms period and the other corresponded to the second 10-ms period.

4 H.M. Sinnamon / Behavioural Brain Research 160 (2005) Procedure After surgery, the rats were pre-trained in multiple (range 9 19), 20-min sessions, each approximately 40 trials, to enter the stall, depress the treadle, lap the milk, and approach the pellet area on pellet cue trials. To strengthen the association between the pellet cue and the pellet presentation, a portion of the trials were doubledipper trials in which the pellet cue was omitted, and instead of a pellet delivery, the dipper was presented again. The number and density of double dipper trials was customized for each rat. Pretraining was complete when the rats consistently approached the pellet area on trials with the pellet cues and consistently remained at the dipper during its absence on the double dipper trials. Subsequent recording sessions (range 12 19) of trials with both pellet cues and negated pellet cues provided the data for the study. The rats found it difficult to withhold approach to the pellet area on negated trials. To facilitate training, trials with negated pellet cues after the first session were presented consecutively until locomotion to the pellet area was suppressed. Negated cue trials were limited to approximately 1/3 of the total trials in a session to avoid generally extinguishing the approach behavior to the pellet cue. Recordings from approximately 24 trials were stored for each session; half were from negated cue trials and included trials in which the rat both incorrectly approached the pellet area and correctly suppressed locomotion Histology The rat was given a lethal dose of Nembutal and perfused through the heart with normal saline followed by 10% Formalin. After several weeks of additional fixation, the brain was sectioned transversely every 100 m with a vibratome. Unstained sections were viewed with a microscope at 40 magnification. Recording sites were localized with reference to the atlas of Paxinos and Watson [28] Analysis of hippocampal activity Peak-by-peak measures of amplitude and frequency were used to relate theta activity to the onset of locomotor and stimulus events. The hippocampal record was transformed into standard scores, band pass filtered without phase lag between 4 and 14 Hz, and twice smoothed with a period 3 running average. Filtering, without phase lag, was implemented on acquired signals by performing Fourier transforms on the records for each s trial, setting the appropriate coefficients to 0, and performing the inverse transform. Positive peaks (relative positivity at the deeper electrode) above a selectable threshold were detected and their amplitudes determined. Detection accuracy was checked visually and corrected manually. The peak amplitude and inter-peak interval values were aperiodic and to make these records compatible with 10-ms sampling period of the analog records, they were interpolated. For the interpolation, the amplitude and interval values of a peak were replicated for each 10-ms sampling period up to the next peak. The onsets and offsets of behavioral events were determined (time marked) primarily by inspecting replays of video frame sequences, and as needed by inspecting the trajectories of manually digitized points representing the nose, eyes and forepaws of the rat. The onset times provided indexes for excerpting the analog records and generating peri-event averages. Averages of hippocampal slow wave activity were considered to show phasing relationships or synchronization if the peaks of the average exceeded 2 standard errors of the mean (S.E.M.). Because averages of non-synchronous signals approach 0, plots of the average along with ±2S.E.M. provides a graphic approximation of consecutive 1-sided t-tests of the null hypothesis of no synchrony at a significance level <0.05. Averages of peak-by-peak measures were evaluated for trends around locomotor initiation by repeated measures analysis of variance comparing values at 400, 200 ms, and the values at +200 and +400 to the values at 0 ms. For trends around the onset of the cues where more resolution was needed, comparisons were made at intervals of 100 ms. Comparisons at specific time points were made by paired t tests. 2. Results 2.1. Overview Each of the five rats had bilateral electrodes which had a superficial pole dorsal to the CA1 cell layer of the dorsal hippocampus, and a deeper pole at variable depths, including the hippocampal fissure, CA4, CA3 dendritic area, and dendritic region of the dentate gyrus. The analysis here was based on one bipolar electrode selected for each rat as the most consistent theta through the trial. Representative hippocampal activity is shown in panel A of Fig. 2 which also shows the key behavioral and stimulus events in a pellet cue trial. Positive peaks (deeper electrode relatively positive) of hippocampal activity filtered in the theta band are indicated in panel B, and the interpolated peak amplitudes and interpeak intervals are shown in panels C and D. The 5.12-s trial included 256 frames each of 20-ms duration. In the first frame shown (50) the rat entered the rear of the stall (note depression in force plate record, panel J), and began the approach to the treadle to start a trial. At frame 62, the rat s right forelimb (Rt FL) contacted the treadle (panel K) and the rat oriented to begin lapping (panel E) at the milk dipper (panel N). The pellet cue occurred (panel M) at the retraction of the dipper. The rat initiated locomotor approach to the pellet with an upward head movement (frame 152) that was apparent in the single frame inspection and well reflected in the accelerometer record (panel F). The initiation was not well reflected in the horizontal movement (panels G and H). Within few frames, the accelerating rightward head movement appeared in the horizontal velocity (panel H) and later in the lower resolution horizontal displacement (panel G). Approximately 200 ms later, the rat lifted the left forelimb (arrow, frame 162) to begin the stepping phase of the approach to the pellet (frame 167). Fig. 3 shows a representative trial with a negated pellet cue in which the rat suppressed locomotion to the pellet area. In addition the rat showed withdrawal from the treadle, a pattern appearing only in negated cue trials. After contacting (frame 60) and depressing the treadle, the rat lapped (panel E) at the milk dipper (panel N). After the dipper retracted and the pellet cue occurred (panel M), the rat continued to lap at the access hole (panel E). The rat responded to the negating

5 240 H.M. Sinnamon / Behavioural Brain Research 160 (2005) Fig. 2. Representative pellet trial showing behavioral measures, hippocampal activity measures, and stimulus events. Five video frames selected from the 256 frames in the 5.12-s trial show behaviors leading to the approach to the pellet (Rt FL, right forelimb). (Panel A) Hippocampal activity high pass filtered at 4 Hz. (Panel B) positive peaks in smoothed band pass (4 14 Hz) activity. (Panels C and D) Peak-by-peak amplitude of positive peaks and inter-peak interval, values interpolated from left to right. (Panel E) Manually digitized record of tongue protrusions at the milk dipper. (Panel F) Accelerometer trace, upward and forward movements produce upward deflections. (Panels G and H) Manually digitized trajectory and velocity of the left eye on the horizontal plane. (Panel J) Force plate in stall, note the depression when the rat enters the stall prior to frame 50 and the elevation when the rat has left to approach the pellet around frame 167. (Panels K P) Stimulus event markers. cue (panel P) by making a rightward head movement (HM, frame 158), followed by a succession of steps (frames 187, 199, 220) back into the stall reflected in the depression in the force plate record (panel J). Suppressing locomotion to the pellet area on the negated pellet cue trials was difficult, and no rat showed complete mastery of the task. In pretraining the rats had been presented with pellet cues exclusively associated with the pellets, and in the negated cue recording sessions, the majority of trials involved non-negated pellet cues followed by pellet delivery. Despite the difficulty, all rats showed indications that they were attending to the negating cue and responding appropriately with suppression of approach locomotion on some trials. Fig. 4, panel A shows the proportion of trials with negated pellet cues in which the rats approached the pellet area. In the first session, the rats were generally unresponsive to the negating cue, persisting in approach, but by the last three sessions, all had increased the proportion of trials without approach (median = 0.51, range: ). When the pellet cue was not negated, all rats continued to approach the pellet area on virtually every trial. Panel B shows that four of the five rats showed an increase in backward locomotion from the treadle on negated cue trials. Withdrawal behavior never occurred on non-negated trials. It persisted

6 H.M. Sinnamon / Behavioural Brain Research 160 (2005) Fig. 3. Representative negated cue trial. The selected video frames show suppression of locomotor approach to the pellet and withdrawal from the treadle during the negating cue (Rt FL, right forelimb; HM, head movement, Lt FL, left forelimb). Other panels as in Fig. 2 except panel (P) which indicates a negating cue. despite the delay it caused in the starting of the next trial, as the rat could enable the treadle only by returning to the front of the stall, locomoting around the stall, and entering it from the rear. Panel C shows redirected locomotion, another behavior that only appeared on negated pellet cue trials. It was a short-latency approach that veered away from the pellet area during the negating cue. Redirected locomotion increased in frequency for all rats with continued experience with the negating cue. The absence of complete locomotor suppression with the negated cue was useful because it provided trials in which the negating cue was not effective for comparison to trials in which the cue suppressed locomotion. The qualitative indexes illustrated in Fig. 4 were more effective in showing the development of locomotor suppression by the negating cue than was the latency of locomotor initiation. It was anticipated that after pretraining with the pellet cue alone, the rats would initiate locomotion within 0.3 ms of the cue and then after experience with the negating cue, they would progressively delay locomotion until the negating period had passed. Contrary to expectation, all rats showed a variety of latencies throughout the training with the negating cue Hippocampal activity related to pellet and negating cues To relate hippocampal activity to locomotion and cues in this task, trials with similar locomotor behaviors were combined across sessions. Thus, pellet cues were classified in terms of the time that the locomotion to the pellet occurred.

7 242 H.M. Sinnamon / Behavioural Brain Research 160 (2005) Fig. 4. Increase in behaviors related to locomotor suppression in five subjects with experience with a negating cue superposed on a pellet cue. Comparison to behavior in the first session with the negating cue to the mean of the last three sessions. (A) Approach locomotion to the pellet area. (B) Backward locomotion from the treadle (withdrawal). (C) Redirected locomotion, shortlatency approach that veered away from the pellet area during the negating cue. Locomotion initiated within 0.3 s of the onset of the pellet cue, i.e., prior to the time that the negating cue could have occurred, represents maximal locomotor activation by the pellet cue. In Fig. 5, pellet cues with this short latency locomotion are represented by the thick lines. In panels E and F, the accelerometer (Accel) (thick lines) show the rise associated with the locomotor head movement first appearing during the period before negation period (arrow) and becoming more prominent during the negation period. The force plate traces rose later as the subsequent stepping of the hindlimbs moved the rat out of the stall. The short latency trials were compared to trials in which locomotion appeared at longer latencies, i.e., during the negation period (panels A and C), and after the negation period (panels B and D). Locomotion occurring during the negation period (in the absence of the negating cue) reflects weaker activation of locomotor initiation; it is represented on the left side of Fig. 5. Locomotion after the offset of the pellet cue reflects the weakest activation of locomotor activation; it is represented on the right side of Fig. 5. The values at 100-ms intervals for each of the slower locomotion conditions were compared to the fast locomotion condition by two-way analysis of variance, with repeated measures over time, followed by individual paired t tests. Theta amplitude progressively rose within the 0.5 s period prior to all pellet cues. During this time, the dipper was present and the rats continued to lap the milk. As shown in Fig. 5, panel A, when locomotion started prior to the negating period, the pre-cue amplitude rise continued steeply into the pellet cue period to reach a maximum associated with rapid phase of the locomotion. When the locomotion was initiated during the negation period, the pre-cue rise continued more slowly (panel A), and when the locomotion was initiated after the pellet cue, the rise in amplitude stopped and maintained a lower level throughout the pellet cue (panel B). Inter-peak intervals decreased prior to onset of the cues that elicited shorter latency locomotion either before or during the negation period (Fig. 5, panel C). The pre-onset decline was not present when the locomotion occurred after the cue offset (panel D). The inter-peak intervals prior to cues which elicited locomotion during the negation period (panel C) were generally lower but the differences were small and not significant for any of the individual comparisons. After the onset of the pellet cue, inter-peak interval decreased according to the elicited locomotor patterns. When locomotion started prior to the negating period (panel C), the decline that begun prior to cue onset continued. When locomotion started during the negating period, the decline reversed and then resumed (panel C). When locomotion started after the pellet cue offset, the decline was not apparent during the pellet cue. These patterns indicate that theta amplitude generally, and theta frequency more selectively, increased prior to an expected cue that elicited locomotion. After onset of the cue, the time course of theta amplitude and frequency tracked the execution of locomotion. For negated pellet cues, trials in which locomotion to the pellet area was absent reflects maximal suppression of locomotor initiation. They provide the reference and are represented by heavy lines in Fig. 6. Trials with locomotion starting after the negating cue reflects lesser locomotor suppression and are represented on the right side of Fig. 6. Trials with locomotion starting during the negating cue reflect least locomotor suppression and are represented on the left side of Fig. 6. Similar to the pattern found for non-negated pellet cues, amplitude rose during the 0.5-s period prior to the negated cues regardless of whether locomotion was suppressed, or whether it started during the negating cue (panel A) or after it (panel B). When locomotion was suppressed, the rise stopped prior to the negating cue and remained level throughout the negating cue. When locomotion was incorrectly initiated during the negating cue, the pattern was similar but there was small continuation of the rise into the negation period (panel A). When incorrect locomotion occurred after the offset of the negating cue, amplitude maintained a level throughout the negation cue that was lower than the suppressed locomotion case. Note that this pattern is a reverse of the general positive association between amplitude and locomotion. Inter-peak interval did not show pre-onset trends prior to the negated pellet cues, but it was generally lower when locomotion was suppressed (panels C and D). The magnitude of the difference was small, with only one of the individual comparisons significant (panel D). However, the pre-onset inter-peak intervals on suppressed trials were also low compared to the trials with non-negated cues eliciting short latency locomotion (F(1,76) = 17.70, P < 0.001; Fig. 6). This pattern is a reversal of the general positive association of lower inter-peak intervals with locomotion.

8 H.M. Sinnamon / Behavioural Brain Research 160 (2005) Fig. 5. Theta peak amplitude and inter-peak interval averaged around the onset of non-negated pellet cues that elicited locomotion at different latencies. Means of five rats. The reference condition, represented by thick lines in each panel, is the cue eliciting locomotion within 0.3 s of onset, i.e., prior to the negating period. Circles and triangles, 100-ms points of comparison. Filled, significant difference (paired t-test, P < 0.05) from the corresponding mean for the reference condition. The dashed vertical lines represent, respectively, the onset of the pellet cue, and the onset and offset of the negation period. (A) Thin line, mean theta amplitude around a pellet cue eliciting locomotion within the potential negation period at latencies ( s). Pre-onset increase over time for both cues was significant (F(9,76) = 4.21, P < 0.001) and similar in pattern (F(9,76) = 0.51, P = 0.87). Post-onset time course of two conditions differed (F(9,76) = 9.67, P < 0.001), (B) Thin line, mean theta amplitude for trials in which locomotion to the pellet area appeared after the offset of the pellet cue (after the potential negating cue). Pre-onset increase over time for both cues was significant (F(9,76) = 2.61, P = 0.01) and similar in pattern (F(9,76) = 1.06, P = 0.40). Post-onset time course of two conditions differed (F(9,76) = 4.35, P < 0.001). (C) Mean theta inter-peak intervals for cues eliciting locomotion during the negation period (thin line) compared to cues eliciting locomotion earlier (thick line). Pre-onset, the cue eliciting locomotion during the negation period was generally lower (F(1, 76) = 21.22, P < 0.001) but none of the individual differences were significant. Both cues showed a decrease leading up to onset (F(9,76) = 2.45, P = 0.01) that was similar (F(9,76) = 0.54, P = 0.83). Post-onset, the time course for the two cues differed (F(9,76) = 9.67, P < 0.001). (D) Mean inter-peak intervals for cues eliciting post-offset locomotion compared to earlier locomotion. Pre-onset, no differences between cues (F(1,76) = 0.68, P < 0.41), and no trend over time (F(9,76) = 1.60, P = 0.13). Post-onset, time course for the two cues differed (F(9,76) = 3.21, P = 0.002). (E and F) Mean accelerometer (Accel) and force plate measures for a representative rat for the three cue conditions. Arrows indicate start of rise of accelerometer trace associated with the onset of locomotion. Hippocampal slow wave activity was averaged to determine if cues which controlled locomotion produced differential evoked responses or synchronization of theta activity. Fig. 7 shows the averages for the five recording sites around the onset of pellet cues eliciting short latency locomotion (left panels) and pellet cues with a superimposed negating cue which suppressed locomotion (right panels). The vertical dashed lines indicate the pellet cue onset and the boundaries of the negating period. Consistent evoked responses to the onset of any cue were infrequent and none differentiated the cues with different locomotor responses. For example, one site (vt 52-2, panels C and D) in which the deep electrode was located in the dendritic region of the dentate gyrus showed a prominent evoked response to the onset of the pellet cue. It showed similar waveforms in response to non-negated pellet cues that elicited locomotion at various latencies and to effective and non-effective negated pellet cues. Averages for effective negating cues are shown in the right panels of Fig. 7. Synchronization of theta activity, indicated by peaks exceeding the 2S.E.M. limits, was found but relations to the properties of the cues showed no consistent pattern among the sites. Of particular interest, there were no

9 244 H.M. Sinnamon / Behavioural Brain Research 160 (2005) Fig. 6. Theta peak amplitude and inter-peak interval averaged around the onset of a pellet cue with a superimposed negating cue. Means of five rats, format similar to Fig. 5.The reference condition represented by the thick lines includes trials in which the negating cue effectively suppressed locomotion to the pellet area. (A) Averaged peak amplitude around cues in which incorrect locomotion occurred during the negating cue (thin line) compared to cues with correctly suppressed locomotion (thick line). Pre-onset, cues did not differ (F(1,76) = 0.62, P = 0.43), and showed increasing trends (F(9,76) = 2.47, P = 0.01) that were similar (F(9,76) = 1.48, P = 0.17). Post-onset, overall differences between cues not significant (F(1,76) = 0.07, P = 0.80), and the trend over time (F(9,76) = 2.47, P = 0.01) was similar (F(9,76) = 1.48, P = 0.17). (B) Averaged peak amplitude averaged around cues in which incorrect locomotion occurred after the offset of the negated cue (thin line) compared to cues with correctly suppressed locomotion (thick line). Pre-onset, cues did not differ (F(1,76) = 0.44, P = 0.51), and showed increasing trends (F(9,76) = 8.47, P < 0.001) that were similar (F(9,76) = 0.35, P = 0.96). Post-onset, amplitude generally lower when locomotion occurred after the cue (F(1,76) = 21.15, P < 0.001), and no significant trend over time (F(9,76) = 1.53, P = 0.15). (C) Averaged inter-peak interval for cues with incorrect locomotion during the negating cue (thin line). Pre-onset, inter-peak interval generally lower when locomotion was suppressed (F(1,76) = 12.95, P < 0.001) and neither cue showed significant trend (F(9,76) = 1.52, P = 0.16). Post-onset, the trend over time for two conditions differed (F(9,76) = 4.93, P < 0.001). (D) Averaged inter-peak interval for cues with incorrect locomotion after the negated cue (thin line). Pre-onset, inter-peak interval generally lower when locomotion was suppressed (F(1,76) = 5.38, P = 0.02) and neither cue showed significant trend (F(9,76) = 0.65, P = 0.75). Post-onset, inter-peak interval generally lower when locomotion was suppressed (F(1,76) = 7.04, P = 0.009), and both conditions showed a decline (F(9,76) = 2.35, P = 0.02) that was similar (F(9,76) = 0.42, P = 0.92). (E and F) Averaged accelerometer (Accel) and force plate traces for one rat. Arrows indicate start of rise of accelerometer traces associated with the onset of locomotion. indications that the onset of the effective negating cue produced a resetting or synchronization of the theta pattern. The right panels of Fig. 7 show these averages. One site (vt62-1, panel K) showed synchronization during the locomotor suppression by the negating cue. However, the synchronization started prior to the onset of the negating cue, and also appeared prior to correct and incorrect locomotion. Finally, averages around the negating cues producing withdrawal from the treadle, arguably the most extreme suppressive response, showed no indication of phase relations. These patterns indicate that the cues controlling the initiation and suppression of locomotion in this task did not produce differential evoked responses or synchronization of theta activity Hippocampal activity related to locomotor initiation Locomotor approaches to the pellet area were classified according the type of cue (non-negated and negated) and according to latency (prior to, during and following the negating period). The various locomotor bouts produced simi-

10 H.M. Sinnamon / Behavioural Brain Research 160 (2005) Fig. 7. Averaged hippocampal activity (high pass filtered 4 Hz) around the onset of pellet cues eliciting short latency locomotion (left panels) and pellet cues with a superimposed negating cue that suppressed locomotion (right panels). The points above and below the averages indicate 2 standard errors of the mean (S.E.M.) above and below 0. (A K) Each row represents averages for one recording site for the two conditions. (L and M) Accelerometer (Accel) and force plate averages for rat vt62. The vertical dashed lines indicate the onset of the pellet cue, and the negating period. lar accelerometer and force plate recordings (Fig. 8, panels L Q). The leftmost panels of Fig. 8 show theta amplitude and inter-peak intervals around the shortest latency locomotion which was initiated prior to the negating period. Theta amplitude increased (panel A) and inter-peak interval decreased (panel F) in the 0.5 s period prior to the onset of this rapid onset locomotion, and the trends continued during the execution. All types of locomotion showed the general pattern of an increase in amplitude (panels B E) and a decrease in inter-peak interval (panels G K) during execution. These patterns indicate that amplitude and frequency of theta activity did not differentiate between correct and incorrect locomotion in this task. The pre-locomotor decrease in inter-peak interval was found for only the short-latency locomotion elicited prior to the negation period. Note that the period prior to this short-latency locomotion overlapped the period prior to the onset of the pellet cue. As shown in Fig. 5, the amplitude rise and inter-peak interval decline prior to this type of approach began in the period prior to the pellet cue. In Fig. 8, the time courses of the changes in amplitude and inter-peak interval were similar for the various types of locomotion. Amplitude peaked closely in time with the maximal acceleration of the locomotor approach and interpeak interval reached a minimum at a later point. This lag of inter-peak interval relative to amplitude was tested by comparing the times of the peaks in the accelerometer record, theta amplitude, and theta inter-peak interval. For each rat, the three values were averaged for the five types of locomotor bouts. Fig. 9 shows that the maximum in the accelerometer record appeared at approximately 150 ms after the onset of the locomotor head movement. Theta amplitude reached its maximum less than 50 ms later which was not significantly longer (t(4) = 1.19, P = 0.30). Inter-peak interval reached a minimum approximately 300 ms after the start of locomotion which was later than the accelerometer maximum (t(4) = 6.97, P = 0.002) and the amplitude maximum (t(4) = 2.62, P = 0.06). The differences were consistent for all of the recording sites. Hippocampal slow wave activity averaged around the onsets of the various classes of locomotor approach were examined for indications of phase relations, synchronization or resetting of theta activity. Fig. 10 provides examples for two of the recording sites. Site vt52-2 which displayed the largest evoked response to the pellet cue onset is illustrated in panels A E. It showed a synchronized pattern following the onset of correct approach (panel D) and incorrect approach (panel E) to the pellet area but no sustained pattern in the other conditions. Site vt62-1 which had theta records illustrated in Figs. 1 and 2 showed no indication of synchronization during any of the locomotor types. Overall, initiation of locomotion in this task was not associated with phase-related changes in theta activity that were consistent across recording sites or consistently differentiated between the classes of locomotion. 3. Discussion This study determined how hippocampal theta activity related to cue events and locomotion in a task involving both

11 246 H.M. Sinnamon / Behavioural Brain Research 160 (2005) Fig. 8. Theta peak amplitude and inter-peak interval averaged around onset of the onset of the locomotor head movement initiating the approach to the pellet area. Mean of five recording sites. The left panels (A, F, and L) represent short-latency (<0.3 s) locomotion initiated prior to the potential negating cue. Panels B, G, and M: locomotion on pellet trials initiated correctly during the period when the negating cue could have occurred but did not. Panels C, H, and N: incorrect locomotion initiated during the negating cue. Panels D, J and P: locomotion correctly initiated after the offset of non-negated pellet cue. Panels E, K, and Q: incorrect locomotion initiated after a negated pellet cue. The average traces are bracketed by ±1 standard error traces. Filled circles indicate significant differences before or after the onset tested by analysis of variance at 200-ms intervals. (A) Pre-onset: F(2,8) = 8.50, P = 0.01; (B) pre-onset: F(2,8) = 5.84, P = 0.03; (C) pre-onset: F(2,8) = 8.46, P = 0.01; (D) pre-onset: F(2,8) = 11.82, P = 0.004; (E) pre-onset: F(2,8) = 6.21, P = 0.024; (F) pre-onset: F(2,8) = 4.38, P = 0.05; post-onset, F(2,8) = 20.95, P = 0.001; (G) post-onset: F(2,8) = , P < 0.001; (H) post-onset: F(2,8) = 24.28, P < 0.001; (J) post-onset: F(2,8) = 30.36, P < 0.001; (K) Post-onset: F(2,8) = 21.58, P = (Panels L Q) Accelerometer and force plate records averaged for one rat (vt62). Number of trials for the five rats in each condition: (A and F) 47, 73, 59, 108, 37; (B and G) 95, 45, 57, 21, 39; (C and H) 47, 56, 35, 19, 55; (D and J) 34, 25, 13, 11, 24; (E and K) 23, 40, 29, 14, 21. Fig. 9. Comparison of times following the onset of locomotion for the maximum in the accelerometer record, the maximum in theta peak amplitude and the minimum in the theta inter-peak interval. Each data point is the mean (±1S.E.M.) of five rats. The value for each rat was the mean collapsed over the five types of locomotor bouts. The number of trials for each case is given in the caption for Fig. 8. the initiation and suppression of well-learned approach locomotion. The absence of specific responses to the cue consistently associated with a food pellet was not surprising. However, the putative role of the hippocampus in behavioral inhibition suggested that theta activity would selectively respond if the pellet cue was negated by a delayed superimposed cue never associated with the pellet. The theta rhythm did not synchronize either to the pellet cue when it could be negated or to the negating cue itself. Theta activity was present throughout the trial although amplitude and frequency were relatively low during milk lapping. During milk lapping, starting at approximately 0.5 s prior to the onset of pellet cues, theta amplitude and frequency showed anticipatory changes. Average amplitude increased prior to all cue conditions but frequency increased only when a nonnegated pellet elicited locomotion at shorter latencies. Following the onset of the cues, both theta peak amplitude and frequency depended on whether and when locomotion was executed or suppressed. Both amplitude and frequency increased with the execution of locomotion, but increases in

12 H.M. Sinnamon / Behavioural Brain Research 160 (2005) Fig. 10. Averaged hippocampal activity (high pass filtered 4 Hz) around the onset of locomotion. Locomotion conditions same as Fig. 8. Recording sites and format same as Fig. 7. theta peak amplitude peaked earlier. These results indicate that changes in theta patterns in this inhibition task were not closely time-locked to the stimulus or behavioral events. Rather, they appeared to be graded modulations in amplitude and frequency related to the preparation for, and execution of, movement. Consistent with related work [33], theta activity was nearly continuous throughout the trial with modulations of amplitude and frequency during the various behavioral sequences in the trial, which included approach to the treadle, lapping of the milk, orientation and locomotor approach to the pellet area. Although the theta rhythm did not reset or consistently synchronize with any of the stimulus or locomotor events, amplitude and frequency changed prior to the cue onsets, and therefore theta activity was sensitive to the features of the task. A recent study [42] that found changes in theta power corresponding to transitions in instrumental behavior generally similar to the present patterns also did not report cue related phasic changes in theta activity. The factors required for a cue to reset or synchronize theta activity appear not to be present in this situation even though it incorporated several of the information processing functions proposed for the hippocampus. When optimally performing the task, the rat withheld approach at the onset of the pellet cue, waited for a possible negating cue, and selected either an approach or an alternative to it. Performance would seem to involve attention, working memory, inhibitory control, and response selection. Moreover, the absence of complete mastery of this difficult task makes it reasonable to infer that active information processing occurred throughout the recording sessions. It seems that these factors are not sufficient for an approach cue or a negating cue to reset or synchronize the theta rhythm. The specification of factors that differentiate tasks in which cues produce resetting ([16,17,24] and tasks in which cues do not [33,42] will further understanding the function of theta activity. The negating cue used here was behaviorally effective and the absence of resetting can not be due to its lack of salience. Other types of inhibitory cues, perhaps those requiring working memory [17], would produce a resetting of theta activity. A factor that could work against producing theta resetting tendency to develop response predispositions. The rats frequently appeared to enter a trial with a movement program pre-selected on the basis of the outcomes of recent trials. Tasks structured to minimize the opportunity to pre-program responses might minimize theta activity prior to the cue onset and accordingly increase the likelihood of resetting. Vinogradova [39] has proposed that sustained theta associated with focused attention is resistant to resetting and represents the filtering out of distracting information. Another factor possibly working against resetting is that a rise in theta amplitude (and in some trials frequency) occurred prior to the onset of the predictable pellet cue. If cue onset were unpredictable, the increase in theta amplitude or frequency might be sufficiently abrupt to be characterized as resetting. Until research supports these conjectures, the available evidence leads to the conclusion that theta activity during performance of a behavioral inhibition task is modulated by processes that have relatively low time resolution. At least in this task, theta activity did not seem to reflect the

13 248 H.M. Sinnamon / Behavioural Brain Research 160 (2005) processing of higher time resolution phasic events like cue onset. Theta amplitude rose prior to the onset of the pellet cues that elicited locomotion at short and long latencies and prior to negated pellet cues which suppressed locomotion. Therefore, the pre-onset amplitude trends did not predict the behavior evoked by the cue. The rise in amplitude appeared during continuous milk lapping, and appears to reflect expectancy of cue onset rather than specific movements. The half-second period in which it occurred corresponds to approximately two to three theta cycles at the frequency typical during lapping of milk. This finding is consistent with the idea that the theta rhythm is involved in the coding of sensory information that is relevant to preparation for movement. In cats tested in an omitted stimulus paradigm, rising expectancy was associated with an amplitude increase in the theta components of the response to a cue that predicted the onset of a significant period [2]. In rats, theta power increases appeared prior to a bar press which initiated discriminatory cues to which the rats were uncertain of the appropriate response but not when the cues were well learned [42]. In the present study, theta frequency also rose during the pre-onset period but only in trials in which a nonnegated pellet elicited locomotion at shorter latencies. Frequency did not rise prior to the cue onset in trials in which the cue elicited locomotion at longer latencies or when the negating cue effectively suppressed approach. It may therefore reflect the activation of initiation processes that would antagonize suppression. Overall, the patterns indicate that the rise in theta amplitude reflected a general preparation to process the cue [42], and the rise in theta frequency reflected the readiness to respond to the cue with locomotion [4]. The changes in theta amplitude and frequency that appeared prior to the onset of the pellet cue reflected both the structure and the ordering of the trials. The trials were structured so that a fixed period of milk lapping preceded the onset of the pellet cue. It was effective in producing relative immobility and a low baseline level of theta activity. With the dipper fixed in duration and always terminating with the onset a pellet cue, the rats could predict the onset of the cue. The detection of rises in theta amplitude and frequency that correlated with the anticipation of the cue onset was facilitated by the low baseline theta activity during milk lapping. In a related study [33] that also used predictable locomotor cues but did not find these pre-onset trends, overt orienting behavior was present and baseline theta activity was higher than the present study. Following the onset of the cues, the patterns of theta peak amplitude and frequency depended on whether locomotion was executed or suppressed. Both amplitude and frequency increased with the execution of locomotion. When the latency was short, the locomotor-related trends continued those begun prior to the onset of the pellet cue. When the locomotion was delayed, the pre-onset trends leveled until locomotor onset. One finding did not fit this straightforward pattern: amplitude was maintained at a higher level on trials with locomotor suppression compared to trials with incorrect locomotion after the offset of the cue. With the inhibition of conditioned approach, various search behaviors emerge [35], and it is likely that these alternatives to locomotion also would be associated with theta activity. Further studies of locomotor suppression, should differentiate the suppression of approach locomotion associated with immobility versus that associated with alternative active behaviors such as orienting and the redirected locomotion which restarts the trial. With the execution of locomotor approach, the time courses of increases in theta peak amplitude and forward acceleration were similar whereas the increase in theta frequency peaked later. The lag of frequency relative to amplitude corresponded to approximately one theta cycle. Different sensitivity and recovery time courses for theta amplitude and frequency after reversible lesions have been described [6,27]. Differences in the short-term time courses of the two theta parameters apparently has not been explicitly reported, but a similar pattern is apparent in the time course of theta activity with locomotor onset described in the cat [1]. Several features of the amplitude frequency difference need to be clarified. One is the degree to which the times of the peaks are determined by the kinematics of the locomotion. In this study, the locomotor episodes were of short duration, involving only a few steps to reach the pellet, and therefore the apparent earlier peak in amplitude could reflect a greater sensitivity of amplitude to deceleration. Another possibility is that amplitude reduction corresponds to the arrival at the goal of the approach [42]. A fundamental question for further work is how amplitude and frequency covary during the execution of a range of behavior patterns. In this study, amplitude increases appeared prior to the onset of cues regardless of their behavioral effects, whereas pre-onset increases in frequency were related to the subsequent behavior. Combined with this finding, the temporal priority of amplitude over frequency with the execution of locomotion suggests different functions for amplitude and frequency modulation. Consistent with sensory-oriented theories [29,39] and recent findings [42], the anticipatory changes in amplitude are consistent with a role in attention and in processing antecedent information such as feed-forward signals. Consistent with the sensorimotor theory of Bland and Oddie [4], the lagging changes in frequency are consistent with a role in processing movement-related information such as reafference, feedback, and consequences. In general, the patterns observed suggest that theta amplitude and frequency might differentiate along a cognitive-motor dimension. Acknowledgements Supported by a Wesleyan University Program grant. Thanks to Esther Schlegel, Leah Pransky, Seth Shipman, Bruce Strickland, David Boule, and Greg Pare for contributions to this work.