Abstract

Theta oscillations in the hippocampus support cognitive processing. Theta-range rhythmicity has also been reported in frontal and posterior cortical areas—where it tends to show consistent phase-relations with hippocampal rhythmicity. Theta-range rhythmicity may, then, be important for cortico-cortical and/or cortico-hippocampal interactions. Here, we surveyed the rat frontal and posterior midline cortices for theta-related oscillations and examined their relationships with hippocampal activity in freely moving rats. Variation in electroencephalography across 4 general classes of spontaneous behavior demonstrated different profiles of theta-like activities through the rat midline cortices. Analysis of cortico-cortical and cortico-hippocampal coherences showed distinct, behavior-dependent, couplings of theta and delta oscillations. Increased theta coherence between structures was most obvious during nonautomatic behaviors and least during immobility or grooming. Extensive coupling of theta oscillations throughout the rat midline cortices and hippocampus occurred during rearing, and exploratory behavior. Such increases in coherence could reflect binding of cortico-hippocampal pathways into temporary functional units by behavioral demands. Extensive coupling of frontal delta, which lacked coherence with posterior areas (including the hippocampus), suggests that different frequencies of rhythmicity may act to bind groups of structures into different functional circuits on different occasions.

Introduction

Theta rhythm is a high-amplitude, sinusoidal, 4- to 12-Hz electroencephalographic (EEG) activity that has been extensively studied in the rodent hippocampus. Rodent hippocampal theta has highly consistent behavioral correlates (Vanderwolf 1969) and has been suggested to be involved in sensorimotor integration (Bland and Oddie 2001). But it can also occur in the absence of movement when the animal is aroused (Sainsbury 1998). Many theories have postulated a contribution of theta activity to cognitive processing by the hippocampus (O'Keefe and Nadel 1978; Fox and Ranck 1979; Gray 1982; Miller 1991; Worden 1992; Carpenter and Grossberg 1993; Cohen and Eichenbaum 1993; Gray and McNaughton 2000) and to emotion (Gray 1982; Gray and McNaughton 2000). Recently it has been shown that restoring theta-range rhythmicity restores hippocampal function (McNaughton et al. 2006).

Theta-range activity is not exclusive to the hippocampus. Many cortical (Feenstra and Holsheimer 1979; Mitchell and Ranck 1980; Alonso and Garcia-Austt 1987a, 1987b; Borst et al. 1987; Leung and Borst 1987; Bullock et al. 1990; Bilkey and Heinemann 1999; Collins et al. 1999; Talk et al. 2004; Hyman et al. 2005; Jones and Wilson 2005a, 2005b; Siapas et al. 2005) and subcortical (Faris and Sainsbury 1990; Kirk and McNaughton 1993; McNaughton et al. 1995; Kirk et al. 1996; Vertes et al. 2001; Seidenbecher et al. 2003; Woodnorth et al. 2003; Nerad and McNaughton 2006; Decoteau et al. 2007b) areas also exhibit theta-range activities. The occurrence of theta-range activity and its synchronization between the hippocampus and other parts of the brain appears to be essential for the correct expression of learned behaviors (Seidenbecher et al. 2003; Jones and Wilson 2005b; Decoteau et al. 2007a).

Cortico-hippocampal interactions have long been postulated to support memory formation, consolidation, and retrieval—and to involve theta-range oscillations (Miller, 1991). The hippocampus appears to be crucial in the early aspects of acquisition, whereas the medial prefrontal cortex (mPFC) is important during consolidation (Bontempi et al. 1999; Frankland et al. 2004; Maviel et al. 2004). Like the hippocampus, the posterior cingulate or retrosplenial (RS) cortex is more important during acquisition compared with the mPFC (Bussey et al. 1996; Maviel et al. 2004). Lesion and multiunit activity have shown that the learning-phase-specific interactions between the mPFC/rrACC (the most rostral part of the rostral anterior cingulate cortex) (rabbit area 24) and the RS (rabbit area 29) mirror those described between the mPFC/rrACC and the hippocampus (Gabriel and Sparenborg 1987; Gabriel et al. 1991): 1) lesions to area 29 abolish learning-related activities in area 24; 2) lesion of both areas severely attenuates learning but area 24 lesion alone only mildly attenuates learning but; 3) area 24 lesion also affects early developing activities in area 29 while sparing the late-developing activities. All these data suggest that the whole cingulate cortex and the hippocampus may coordinate their activities to form functional circuits that facilitate mnemonic processes.

At the individual cell level, activity in mPFC is modulated by the hippocampus (Hyman et al. 2005) and, like hippocampal place cells (O'Keefe and Recce 1993), mPFC cells show phase precession relative to hippocampal theta (Jones and Wilson 2005a). Field activity in mPFC becomes synchronous with the hippocampus at theta frequencies only when the animals make a correct arm choice in a T-maze task (Jones and Wilson 2005b). There is, therefore, a direct functional relationship of oscillatory activities, at both the cellular and field levels, between the hippocampus and the neocortex.

It remains to be seen to what extent the theta oscillations in the medial frontal cortex are homogeneous, and whether other theta oscillators exist, such as those found in the ACC in humans (Gevins et al. 1997; Asada et al. 1999; Ishii et al. 1999; Pizzagalli et al. 2003) and monkeys (Tsujimoto et al. 2003, 2006). Although theta oscillation has been reported in the RS, its origin has been hotly debated (Feenstra and Holsheimer 1979; Borst et al. 1987; Leung and Borst 1987; Colom et al. 1988; Talk et al. 2004) and functions unclear. To further understand how cortico-hippocampal circuits give rise to functional circuits that serve higher cognitive processes, the nature of the various local activities and their behavioral correlates need to be explored.

We, therefore, undertook preliminary concurrent recordings from the mPFC, cingulate cortices and hippocampal areas to: 1) identify possible multiple frontal midline theta generators in the rat; 2) observe behavioral correlates of theta-range activity in the midline (cingulate and mPFC) cortices; and 3) examine the relationship of field activity between the midline cortices and the hippocampus.

Materials and Methods

Electrode Construction

Two types of electrodes were used in this study. Simple bipolar electrodes were used to record EEG from the hippocampus. These were made by twisting together a pair of annealed, Teflon-coated, stainless steel wires (200 μm diameter). Amphenol gold pins were soldered to the wires with the aid of phosphoric acid flux and the wires were pulled along each other to create a 0.7 mm separation between the recording tips. This allowed unipolar recordings from both the CA1 and DG subfields of the hippocampus against a separate common reference. The electrode arrays used to sample rat midline cortical activity were made from 100 μm diameter Teflon-coated stainless steel wires. Four to 6 of these wires were placed next to each other to form an array. A small amount of water was applied to the array, so that surface tension held the wires together. Excess water was removed and an instant adhesive—Loctite 401 (Loctite Australia Pty. Ltd, Australia)—was carefully distributed along the array. This was allowed to set overnight before a 2nd layer of adhesive was applied to the body of the array. On the next day, the arrays were cut into the desired lengths for implantation. Additionally, the recording tips were cut with a pair of very sharp scissors at an angle (of approximately 9° for 6-electrode arrays and 45° for 4-electrode arrays) to create even, stepped, tip separations within an array. Like the bipolar electrodes, the wires of the arrays were soldered to gold pins with the aid of phosphoric acid flux. The gold pins were then inserted into a McIntyre connector and a digital image was acquired of the tips of each array to allow accurate assessment of the final separation between the individual tips. Finally, the arrays were checked for their tip resistance and insulation integrity before implantation.

Subjects and Surgical Procedures

Thirty-six Sprague–Dawley rats obtained from the University Of Otago Department of Laboratory Animal Sciences were kept in a holding room with fluorescent lights set to a 12-h light/dark cycle supplemented by illumination from an external window, which coincided with the light part of the cycle. Temperature was controlled at 21 ± 2 °C. The rats were maintained in the colony room for at least 2 weeks after their arrival before surgery commenced. Animals weighed 400–500 g at the time of surgery and were anesthetized subcutaneously (s.c.) with ketamine (0.75 mg/kg, 100 mg/mL, Parnell Laboratories New Zealand, Ltd, New Zealand) and medetomidine (0.5 mg/kg, 1 mg/mL, Novartis Animal Health Australia, Ltd, Australia). After the animals had become thoroughly anesthetized (tested by the foot-pinch withdrawal reflex), hair in the area of surgery was removed and Betadine (Faulding Pharmaceuticals, Australia) was applied to the skin. Tricin gel (Jurox Pty., Ltd, Australia) was applied around the rat's eyes to avoid dehydration and infection. The surgery was done in aseptic conditions where a protective, sterile, clear plastic barrier was placed over the rat and pressed into contact with the Betadine on the scalp and the single midline incision of the scalp was made through the barrier. All surgical instruments were kept bathed in 70% alcohol. Items that were implanted into the rat (electrodes, screws, and suture threads) were bathed in a mixture of 70% alcohol and chlorhexidine (Pharmacia, Australia). One 6-electrode array, one 4-electrode array and a bipolar electrode were implanted in each rat. The target of the 6-electrode array was always in the frontal midline cortex, and that of the 4-electrode array was always in the posterior midline cortex; whereas the bipolar electrode targeted the CA1 and DG subfields of the hippocampus. The 6-electrode array was used rostrally to accommodate the elongated shape of the mPFC in the dorsoventral axis whereas the shorter 4-electrode was used caudally to accommodate a smaller brain space sampled. The 2 electrode arrays were implanted at different coordinates, and at different A–P displacements between each other, across rats, to maximize the amount of medial cortex sampled and the capacity to assess frontal–posterior cortical relationships in all possible combinations. The hippocampal recording electrodes were implanted at −3.80 mm posterior to Bregma, 2.50 mm lateral from midline and with the lower tip 3.50 mm ventral from the surface of the skull. All electrode arrays were implanted ipsilaterally and hippocampal recording electrodes contralaterally. Additionally, a silver wire attached to the skull just anterior to lambda on the right side served as the earth electrode. The indifferent electrode was a screw anchored just anterior to lambda on the left side. All the electrodes were held by 2 McIntyre connectors and anchored by dental cement to 6 jeweller's screws fixed to the skull. Upon completion of the surgery, suturing of the wound was followed by the injection of atipemazole (s.c., 0.25 mg/kg, 5 mg/mL, Novartis Animal Health Australia, Ltd) to assist anesthetic recovery. The animals were monitored until they regained consciousness and basic motor abilities and then Temgesic (0.5 mg/kg, s.c.) was administered for analgesia. The administration of Temgesic was repeated approximately 12 h later on the following day at the same dose. The animals were given at least a 14-day recovery period before further manipulation. Ethics approval for these experiments was obtained from the Otago University Animal Ethics Committee (approval number 35/04).

Electrophysiology and Behavioral Testing

Each recording electrode was connected to an impedance transformer that was connected to a channel of an EEG-4214 (Nihon Kohden, Japan). The amplified and filtered (0.3- to 30-Hz band pass) signals were fed into a personal computer through National Instruments interfaces (SCB-68 and PCI-MIO-16E-4, National Instruments Corporation, Austin, TX). The signals were sampled by a custom written Labview program at 128 Hz.

The animals were tested in a 73 cm × 73 cm × 51 cm box with black interior walls. Each animal received 4, 6-min trials in the box. The animals were connected to the signal acquisition system described above immediately before each trial and disconnected at the end of each trial. Ten minutes after the 1st trial, a 0.9% saline injection (i.p.) was administered as a part of another study for which the animals reported here acted as a control group. Another 10 min after the injection, the animals received the 2nd trial. Grooming, immobility, movements (locomotion and head movements) and rearing were recorded together with the EEG records throughout all trials by 2 manually operated digital triggers using a binary code in real time by the experimenter. Trials 3 and 4 followed 40 and 70 min after the saline injection. The testing apparatus was wiped down with cleaning spray between each pair of trials to maintain hygiene and remove animal odors. At the end of the experiment, animals were returned to their home cages.

Data Reduction

The data files collected by Labview were imported into Spike2 (CED, UK) for further analysis. Coherence analysis (block size of 512) was used, creating a single coherence spectrum from multiple contiguous blocks of EEG recording for any pair of electrodes from all trials from a single animal (at least 360 contiguous blocks). Coherence spectra were calculated for all combinations of pairs of recording electrodes for each EEG recording period irrespective of behavioral state. The main coherence analyses were divided into 3 parts corresponding to 3 different frequency bands: 2–5 Hz, 5–10 Hz, and 10–15 Hz. The coherences within these 3 frequency bands were averaged across all frequencies within the band and sorted into coherence matrices.

Because coherences may not be normally distributed (instead following a power function) the mean coherences (for all the coherence matrices as a population) were tested for any systematic relationship with their standard deviations. There was no such trend (r2, i.e., proportion of total variance attributable to such a trend: 2–5 Hz = 0.015; 5–10 Hz = 0.0002; 10–15 Hz = 0.0451).

The coherence matrices were submitted to factor analysis with a custom internal script in SPSS 12.0.1 (SPSS, Chicago, IL). The factor analysis was performed using an oblique (promax, Kaiser normalized) rotation with a minimum Eigenvalue of 1 and a maximum of 25 iterations for convergence. In this analysis the matrix of coherences of electrode pairs (which has each electrode generating both a row and a column) is decomposed into a new matrix in which each electrode generates a row and in which a small number of factors generate the columns. The data in the factor matrix represent then the phase correlation of each individual electrode with the estimated representative activity that is common to all electrodes that share a loading on that factor.

The promax rotation generates factors each of which has the highest possible loadings for a set of electrodes on any particular factor (that is the factor maximizes the extent to which it captures a set of electrodes) while allowing the extracted factors to be nonorthogonal. A 2nd factor maximizes capture of a 2nd set of (potentially overlapping) electrodes. That is, each factor best represents common activity in a distinct set of electrodes and, where 2 factors are not orthogonal this represents the extent to which the 2 inferred sources of neural activity (each captured by a separate electrode set) are themselves partially coherent. The loading of any electrode on any 1 factor can then be broken down into 2 components a unique 1 (predicted only by that factor and not shared with any other factor) and a shared 1 (predicted by the common variance of 2 or more factors).

Because the factors are not orthogonal a number of factor loading matrices can be constructed, depending on the treatment of shared and unique variance. We used the pattern matrix, which contains only the component of the variance in activity at a particular electrode that is uniquely predicted by a single factor, excluding any shared components. In the current situation, this unique factor loading can be interpreted as the extent to which the activity at that electrode is driven by the presumed common source that is driving activity in all electrodes that share a loading on that factor after filtering out the contribution from the 2nd source. So, an electrode that receives unique contributions from 2 factors can be thought of as definitely receiving a signal from both sources—because the common variance of the 2 sources has been mathematically excluded.

We, therefore, used the values in the pattern matrix as an indication of how an EEG recording location was related to other recorded local activities. There were no differences in the factor loadings assigned by the factor analysis between different frequency bands (data not shown). Once all the individual EEG from each recording electrode was classified, the recording electrode with the highest factor loading in the averaged pattern matrix for a particular factor was designated as the “representative” electrode within that array. If the highest factor loading observed in the analysis was shared by 2 or more electrodes, the electrode with the highest spectral power was designated the “representative” electrode. The data from the “representative” electrode was then used for further analyses and the data from adjacent electrodes loading on the same factor were omitted. It should be noted that the precise choice of electrode is not important here—the critical point is that we did not include in our analysis more than 1 electrode from any electrically homogenous area within a rat and so did not inflate the significance of subsequent statistics.

Each “representative” electrode, therefore, acted as an independent, nominal “best,” estimator of electrical activity in a specific area relative to other spatially distinct representative electrodes in the same animal and relative to other representative electrodes in the same area across animals. In many arrays, there were clear transitions between sets of electrodes loading on 1 factor and sets loading on a 2nd factor (Fig. 1) and these arrays were then assigned 1 “representative” electrode for each factor. Overall, all “representative” electrodes within structures of interest from the 19 animals that survived the surgery and had intact head stages at the end point of the experiment were included in this report. The loss of animals was mainly related to intracerebral hemorrhage due to the relative large sizes of the electrode array implanted and closeness to the midline. Only 4 animals lost their implants before the end of the study.

Figure 1.

Schematic illustration of data reduction by factor analysis and power spectra. The graph on the left plots the unique factor loadings for each electrode in a 6-electrode array on 2 extracted factors. The unique loading includes only the variance in power for that electrode that can be attributed solely to the factor (which can be thought of as a mathematically estimated source of unknown location). The variance common to the 2 factors has been eliminated. In this example, the 6-electrode array delivers 2 separate profiles. Electrodes 1–3 are largely driven by 1 source—electrodes 5 and 6 by the other. Electrode 4 receives unique contributions from both sources. Brain drawings are displayed in the center, with the electrode placements depicted by diamonds through the rrACC and the mPFC (+2.2 mm from Bregma). The electrodes with the largest factor loading values of the set corresponding to a particular factor profile (largest symbols) are designated as the “representative” electrode, representing the field activity of the given area and are the only ones used for subsequent analysis of activity from the relevant histologically defined area (rrACC and mPFC in this case). Power spectra (on the right) from each electrode are also used to corroborate the designation of “representative” electrodes by spectral power with those showing the largest discursion in the theta range (thick lines) being chosen as the representative electrodes when factor loadings were similar among a set of electrodes.

Figure 1.

Schematic illustration of data reduction by factor analysis and power spectra. The graph on the left plots the unique factor loadings for each electrode in a 6-electrode array on 2 extracted factors. The unique loading includes only the variance in power for that electrode that can be attributed solely to the factor (which can be thought of as a mathematically estimated source of unknown location). The variance common to the 2 factors has been eliminated. In this example, the 6-electrode array delivers 2 separate profiles. Electrodes 1–3 are largely driven by 1 source—electrodes 5 and 6 by the other. Electrode 4 receives unique contributions from both sources. Brain drawings are displayed in the center, with the electrode placements depicted by diamonds through the rrACC and the mPFC (+2.2 mm from Bregma). The electrodes with the largest factor loading values of the set corresponding to a particular factor profile (largest symbols) are designated as the “representative” electrode, representing the field activity of the given area and are the only ones used for subsequent analysis of activity from the relevant histologically defined area (rrACC and mPFC in this case). Power spectra (on the right) from each electrode are also used to corroborate the designation of “representative” electrodes by spectral power with those showing the largest discursion in the theta range (thick lines) being chosen as the representative electrodes when factor loadings were similar among a set of electrodes.

Data Analysis

Each 6-min EEG record for each “representative” electrode was separated into 4 smaller sets of records by a custom-made Spike2 script, so that each set contained segments of EEG related to 4 separate behavioral states: grooming, immobility, movement and rearing. Overall, all behavioral epochs consisted of roughly 17 000 s of data, with the exception of grooming (2000 s). These behaviorally segregated EEG records were then submitted to fast Fourier transform (FFT) and coherence analyses.

The FFT was performed at a bin size of 512 and averaged over successive segments for each trial. The resultant spectrum was log10 transformed and averaged across 4 trials to yield a single log power spectrum for any 1 “representative” electrode. Additionally, the inherent exponential decay noise function embedded in the FFT was partially normalized by subtracting the exponential decay trend from each FFT. This allowed better visualization of the FFT at the frequencies of interest. Analysis of variance (ANOVA) was used to compute (Genstat, VSN, UK) the differences in mean power during the 4 behavioral epochs in both delta (1–4 Hz) and theta (6–10 Hz) bands. Pair-wise comparisons were assessed using Student's t-test (with Bonferroni correction of P values and a minimum corrected significance level of P < 0.05).

The coherence analysis was also computed by a custom-made Spike2 script. Coherences between 0 and 10 Hz using a Hanning window were obtained to examine the synchronization between areas of interest. The data from the 0–10 Hz frequency band were then averaged across 4 trials to yield coherence measures for each “representative” electrode in relation to other “representative” electrodes in the same animal. Electrodes were then grouped together as independent samples of the activity in any particular structure. Coherence was calculated only where there were 2 or more cases of electrodes sampling a particular pair of structures across the nineteen animals examined. The 95% confidence intervals of the overall coherence were calculated for each brain region pair from each animal, regardless of behavioral state, to establish statistical significance of the coherences. The coherence spectra within the delta band (1–4 Hz) and theta band (6–10 Hz) between brain regions during specific behavioral epochs were submitted to an ANOVA procedure with extraction of orthogonal polynomial components to assess differences in forms of polynomial functions of frequency among behavioral conditions. Where several polynomial components of behavior versus frequency interaction were significant, the 1 with the highest F-ratio only was further examined by post hoc comparisons among the 6 pair-wise combinations of the 4 behavioral classes using the Bonferroni correction (see above for details).

Although the coherence analysis probes the stationary coupling of any 2 recording sites, it does not divulge the phase relation between 2 coupled local field potentials (LFPs). To visualize the phase relationships between the hippocampus and the cortex, field-triggered field averages were calculated. A custom Spike2 script was written that detected negative peaks in the dentate gyrus (DG) LFP during theta activity and used this as a trigger for cortical waveform averaging. The mean and standard error were calculated for each field-triggered average. This was done for all available data collected in the respective areas. Additionally, phase histograms were also calculated by using the DG theta wave negative peak as the trigger for each theta cycle to estimate the phase of any theta wave negative peak co-occurring in the cortex. The data used to construct the phase histograms were chosen from the high-power theta band (5–10 Hz) activity in the DG spectrogram (Matlab, Natwick, MA). Circular statistics were performed to determine the circular mean, phase uniformity (Rayleigh's test) and phase concentration coefficient (c), which was taken as the reciprocal of the phase dispersion (circular variance) to represent the clustering of phase values around the mean.

Histology

At the end of the experiment the animals were deeply anaesthetized and transcardially perfused with 0.9% saline and then 10% formalin. The brains of the animals were removed and fixed in 10% formalin-distilled water overnight before being placed in 30% sucrose/10% formalin-saline solution.

Once the brains were adequately fixed, a freezing microtome was used to obtain 90 μm sections. The brain sections were stained with thionin mounted on gelatine slides, cover-slipped with DPX mountant and digitally photographed for archiving. The positions of electrode tips were reconstructed in reference to Paxinos and Watson (1998). The bipolar electrode tips were easily identified. Only tips in the areas of stratum oriens/pyramidale (CA1) and of hippocampal fissure/stratum granulosum of the buried blade (DG) were used in the analysis. For the identification of the array tip positions, the location of the deepest tip was 1st determined. The locations of the rest of the tips in the array were then reconstructed from the known dorsoventral and rostro-caudal distances between the tips calculated from the preimplantation images of the arrays. Finally, the electrodes were pooled into anatomical groups devised for this study (Fig. 2). The anatomical groupings are based on connectional, cytoarchitectonic, and biochemical grounds (Palomero-Gallagher and Zilles 2004; Vogt et al. 2004). The final groupings were decided on the balance of reducing the number of nominal structures examined (and so increasing the number of samples within a “structure”) and the true diversity of the anatomical differences through the medial cortex.

Figure 2.

Histological preparations used to determine electrode positions. (A) A 6-electrode array transecting the mPFC area. (B) A 4-electrode array transecting the cRS area. (C) A hippocampal bipolar recording electrode track. Despite the size of the arrays, no extensive injuries or necrosis were observed in the brain sections. The arrows point to the deepest recording tips from the arrays and the bipolar electrode tips in the hippocampus.

Figure 2.

Histological preparations used to determine electrode positions. (A) A 6-electrode array transecting the mPFC area. (B) A 4-electrode array transecting the cRS area. (C) A hippocampal bipolar recording electrode track. Despite the size of the arrays, no extensive injuries or necrosis were observed in the brain sections. The arrows point to the deepest recording tips from the arrays and the bipolar electrode tips in the hippocampus.

Results

Histology

Examples of histological preparations used to determine electrode placements are shown in Figure 3. As described above, the positions of each individual recording tip from an array were calculated post hoc from the determination of the deepest point of the array track (as shown by the arrows in Fig. 3A,B) or, for bipolar hippocampal recording electrode placements, from identifiable tip positions (see arrows in Fig. 3C). The collated electrode placements in the midline cortex are presented in Figure 4, minus the electrodes that were faulty (i.e., yielded unusable recordings, n = 2 in the cortex, n = 2 in the hippocampus). The filled diamonds correspond to the “representative” recording electrodes (n = 62) determined by the factor analysis and power spectrum data. The unfilled diamonds (n = 124) correspond to recording tips that did not have a high unique factor loading, or, that had a higher factor loading but had relatively lower spectral power compared with another electrode with a similar loading on the same extracted factor. As seen in Figure 4, there were many “representative” electrodes outside of the regions of interest and the arrays in the mPFC sampled only as far as the prelimbic (PrL) cortex. Note that only “representative” electrodes (see Methods) that are in the regions of interest in the cortex or the hippocampus (CA1/DG) were included for the results described below (see Table 1).

Table 1

Summary of electrode designations for all brain regions sampled

Brain region Number of representatives Number of nonrepresentatives 
mPFC 17 
rrACC 
rACC 23 
cACC 
rRS 19 
cRS 12 
CA1 17 
DG 16 
Others 22 39 
Total 95 129 
Brain region Number of representatives Number of nonrepresentatives 
mPFC 17 
rrACC 
rACC 23 
cACC 
rRS 19 
cRS 12 
CA1 17 
DG 16 
Others 22 39 
Total 95 129 
Figure 3.

Median section of the rat brain illustrating the grouping of structures into cortical areas investigated in this study. The basis of anatomical grouping was on the rostro-caudal differences in connectivity, cytoarchitecture and biochemistry of the midline cortices. The anterior half of the medial cortex contains the medial prefrontal region (mPFC), which included the infralimbic (IL) and PrL cortices. The ACC was divided into the most rostral region just dorsal to the mPFC (rrACC), the rostral anterior cingulate with dorsal and ventral aspects (rACC), and their caudal counterparts (cACC). More caudally, the RS cortex was divided into rRS and cRS, which contain both the granular and agranular divisions. Figure redrawn from Palomero-Gallagher and Zilles (2004).

Figure 3.

Median section of the rat brain illustrating the grouping of structures into cortical areas investigated in this study. The basis of anatomical grouping was on the rostro-caudal differences in connectivity, cytoarchitecture and biochemistry of the midline cortices. The anterior half of the medial cortex contains the medial prefrontal region (mPFC), which included the infralimbic (IL) and PrL cortices. The ACC was divided into the most rostral region just dorsal to the mPFC (rrACC), the rostral anterior cingulate with dorsal and ventral aspects (rACC), and their caudal counterparts (cACC). More caudally, the RS cortex was divided into rRS and cRS, which contain both the granular and agranular divisions. Figure redrawn from Palomero-Gallagher and Zilles (2004).

Figure 4.

Collated electrode placements for the study. The filled diamonds represent the “representative” electrodes and the unfilled diamonds represent other electrodes used to determine the representative electrodes but were not included in the analyses of the differences between anatomical areas. In addition, “representative” electrodes falling outside of the midline cortex were not included in the analysis. For a detailed account of the number of electrodes in structures of interest, see Table 1. The distance to Bregma increases progressively from top to bottom in each row and then from left to right across rows.

Figure 4.

Collated electrode placements for the study. The filled diamonds represent the “representative” electrodes and the unfilled diamonds represent other electrodes used to determine the representative electrodes but were not included in the analyses of the differences between anatomical areas. In addition, “representative” electrodes falling outside of the midline cortex were not included in the analysis. For a detailed account of the number of electrodes in structures of interest, see Table 1. The distance to Bregma increases progressively from top to bottom in each row and then from left to right across rows.

Theta Rhythm in the Midline Cortex

The basic properties of the observed theta oscillations are described in Figure 5. A single particularly clear example of typical raw EEG is shown for each cortical region together with concurrent accompanying hippocampal records to provide an impression of morphology and amplitude. The typical activity in the population of electrodes overall is represented later by averaged FFT power in relation to specific behaviors. Phase properties of the population as a whole are provided in Figure 5 as DG field-triggered averages and via a rose diagram with mean phase and phase concentration scores. All phase distributions were found to be nonuniform (Rayleigh's test, P < 0.0001).

Figure 5.

Examples of individual raw EEG recordings, together with group field-triggered-field averages and rose diagrams showing cortical phase relative to DG. They are arranged rostro-caudally: (A) mPFC, (B) rrACC, (C) rACC, (D) cACC, (E) rRS, and (F) cRS. Each section of raw EEG contains 10 s of the record and was digitally filtered (1 Hz high pass). The top traces for each raw record are from CA1, middle traces are from DG and bottom traces are from the relevant midline cortex. Clear theta oscillations can be seen in the raw examples for all cortical recording sites. The field-triggered field averages were calculated between the cortical and DG recordings using DG theta troughs as triggers. A predominant sinusoidal waveform in the theta frequency range suggests cortical theta is closely coupled with hippocampal (DG) theta. The rose diagrams contain the number of cortical theta troughs (each annulus = 100 incidences) falling on particular phases of a DG theta cycle. The mean phase is represented as the black bar radiating from the center of the plot, with the actual value just outside the rose plot. Phase concentration coefficients (c) are displayed on the bottom right corner of each rose plots. Rayleigh's test of uniformity revealed the phase distribution of all cortical theta negative peaks were nonuniform within a DG theta cycle (P < 0.0001).

Figure 5.

Examples of individual raw EEG recordings, together with group field-triggered-field averages and rose diagrams showing cortical phase relative to DG. They are arranged rostro-caudally: (A) mPFC, (B) rrACC, (C) rACC, (D) cACC, (E) rRS, and (F) cRS. Each section of raw EEG contains 10 s of the record and was digitally filtered (1 Hz high pass). The top traces for each raw record are from CA1, middle traces are from DG and bottom traces are from the relevant midline cortex. Clear theta oscillations can be seen in the raw examples for all cortical recording sites. The field-triggered field averages were calculated between the cortical and DG recordings using DG theta troughs as triggers. A predominant sinusoidal waveform in the theta frequency range suggests cortical theta is closely coupled with hippocampal (DG) theta. The rose diagrams contain the number of cortical theta troughs (each annulus = 100 incidences) falling on particular phases of a DG theta cycle. The mean phase is represented as the black bar radiating from the center of the plot, with the actual value just outside the rose plot. Phase concentration coefficients (c) are displayed on the bottom right corner of each rose plots. Rayleigh's test of uniformity revealed the phase distribution of all cortical theta negative peaks were nonuniform within a DG theta cycle (P < 0.0001).

Theta frequency rhythms were recorded throughout the midline cortex. In the anterior cortex, theta field potentials with lower amplitude (compared with hippocampal theta) were recorded in the PrL cortex and subdivisions of the ACC. In the most rostral regions, the theta field potentials have the smallest amplitude (Fig. 5A). Prevailing ∼3-Hz oscillations are also prominent in this part of the brain, usually during hippocampal large irregular activity (LIA) accompanying behavioral immobility. High-amplitude theta field potentials in the mPFC (mean = 0.27 mV) and rrACC (mean = 0.31 mV) appear more sporadically relative to hippocampal theta, where theta amplitude remains constantly high (>0.90 mV). Although frontal theta field potentials usually co-occur with the ongoing hippocampal theta, their absence can also be observed during robust hippocampal theta (e.g. Fig. 5AC).

The rose diagram insert in Figure 5A shows that the mPFC theta wave negative peak tends to occur on average around ∼261° of the DG theta negative peak. The field-triggered field average illustrates an apparent lag (24 ms) of mPFC theta compared with the ongoing DG theta. Both the lack of a strong phase preference (c = 0.26) and the smaller mPFC theta amplitude (mean = 0.27 mV) contribute to the low amplitude of the triggered average waveform.

The rrACC shows a relatively high amplitude (mean = 0.31 mV) and the most continuous theta of the frontal areas mapped (Fig. 5B). Overall, the presence of theta in rrACC is tightly coupled with the presence of hippocampal theta but, nonetheless was occasionally absent during robust hippocampal theta. Theta activities at this site have a similar phase relationship with the DG to that of mPFC with the DG. The phase histogram suggests that the rrACC theta, like mPFC, has a weak (c = 0.34) phase preference at around ∼256° with respect to DG. Interestingly, the rrACC theta appears to have a greater (31 ms) lag relative to DG theta than does mPFC and the averaged wave lacks a symmetrical sinusoidal component with a relatively small standard error.

Continuing caudally, the rACC has high amplitude (mean = 0.28 mV), sporadic theta (Fig. 5C). Theta-range frequencies tend to co-occur in the rACC and the hippocampus, but during hippocampal LIA, rACC tends to display ∼3 Hz activity. There is somewhat less phase concentration (c = 0.27) of where the rACC theta negative peak falls on the DG negative peak (214°) compared with rrACC. Also, the wave-triggered wave average revealed a small amplitude waveform with large standard errors.

In the caudal part of the ACC (cACC), longer trains of theta can be observed (Fig. 5D). These potentials are higher in amplitude (mean = 0.44 mV) compared with potentials observed in the more rostral regions. Once again, theta activities recorded in the cACC co-occur with ongoing hippocampal theta. During a long train of hippocampal theta, the cACC theta is often present, interleaved with less-sinusoidal theta waves. However, it is the only region in the frontal half of the medial cortices that exhibits continuous and robust theta field potentials. The field-triggered field average shows that cACC theta is closely coupled with DG theta in terms of phase and occurrence, as indicated by the symmetrical, sinusoidal waveform average with small standard errors. The cACC theta negative peak appears to be completely phase-locked to the DG theta negative peak (7 ms lead). The cACC theta negative peak, compared with other frontal sites, demonstrated a much stronger phase concentration (c = 0.41) to the DG theta negative peak at an average of 207°.

The rostral part of the RS (rRS) exhibits theta activity that is almost identical to CA1 hippocampal theta (Fig. 5E). The amplitude of the rRS theta is mostly lower than that observed in CA1. However, some cases showed higher theta amplitudes in the rRS than CA1 (mean = 0.81 mV). These differences may be dependent on the exact locations of recording tips in each structure. Most rRS records also show a “cleaner” theta record than the hippocampus, that is, the theta activity has superimposed on it less high frequency activity and the wave morphology appears more symmetrical and sinusoidal. Clear sinusoidal components and larger amplitudes can be seen in the DG negative peak triggered waveform average, suggesting theta amplitudes are higher in this part of the posterior midline cortex and more robust. The highest phase concentration coefficient (c = 0.66) out of all cortico-DG theta activities was observed around the mean phase of 167°, consistent with the observation that rRS theta is mostly in-phase with CA1 theta (3 ms lag to DG theta trough).

The most caudal of the medial cortices sampled was the caudal RS (cRS). In this area, theta activity is visibly different from hippocampal theta (Fig. 5F). The co-occurrence of theta activity in the cRS and the hippocampus is less consistent compared with that observed between the rRS and the hippocampus. Once again, theta with higher amplitudes than CA1 was occasionally observed in the cRS (mean = 0.77 mV). The DG theta triggered cRS field average shows clear and robust sinusoidal waves at the theta frequency and a lead of 7 ms to the DG trough. The predominant cRS theta negative peak falls around 176° from the DG theta trough with a slightly lower concentration coefficient (c = 0.52) than the rRS.

In Fig. 6A, the phase relationship between the DG and CA1 negative peaks is presented in a rose diagram. As to be expected, the CA1 theta trough falls roughly 180° from the corresponding DG trough (mean = 166°) with a relatively high concentration coefficient (c = 0.53). Figure 6B summarizes the findings of Figure 5. It can be seen that the preferred theta phase in respect to DG theta recesses from ∼270° to ∼180° with the concentration coefficient increasing in the rostro-caudal axis.

Figure 6.

Intrahippocampal phase relations (CA1–DG) are depicted in (A), where a near 180° (166° averaged) phase preference was found. A summary of circular statistics is presented in (B), where a shift of preferred phase from ∼270° to ∼180° in the rostro-caudal axis was observed between cortical and DG theta troughs. Note that the phase of posterior cortical areas was similar to that of area CA1, but anterior cortices differed from CA1 and, to some extent, from each other. Also, phase concentration increased progressively in the rosto-caudal direction.

Figure 6.

Intrahippocampal phase relations (CA1–DG) are depicted in (A), where a near 180° (166° averaged) phase preference was found. A summary of circular statistics is presented in (B), where a shift of preferred phase from ∼270° to ∼180° in the rostro-caudal axis was observed between cortical and DG theta troughs. Note that the phase of posterior cortical areas was similar to that of area CA1, but anterior cortices differed from CA1 and, to some extent, from each other. Also, phase concentration increased progressively in the rosto-caudal direction.

Behavioral Correlates of Midline Cortical Theta

Variation in FFT power with time for 1 animal is shown as an example of spectral power changes across a single trial (Fig. 7). The 4 representative electrodes (see Methods) from this single animal were from the rACC, the rRS, and the 2 hippocampal subfields (CA1 and DG). The time-FFT analysis is for a behavioral session where the animal is actively exploring initially and this is accompanied by higher theta frequency and power. Later in the trial, theta appears in a more sporadic fashion, with lower frequency and power, which parallels the acclimatization of the animal to the environment and an associated decrease in exploration. This temporal progression of more activity in the theta range earlier to less activity later in the session is most evident in the rACC in this example. Qualitative observations from these spectrograms suggest that the rACC theta is initially well coupled with the rest of the recording sites with most of the high-power spectral peaks in the time domain matching their hippocampal and posterior cortical counterparts. However, as the session progresses, there appears to be an increase in slower oscillations in rACC as well as decoupling from the rest of the recording sites. In the rRS, theta activity is very closely coupled with the hippocampus, as indicated by the consistent power spectrum peaks across time, which were also seen in the hippocampus.

Figure 7.

An example of changes in the power spectrum with time taken from a single animal within a single trial at several recording sites: (A) CA1, (B) DG, (C) rRS, and (D) rACC. Spectral power is indicated by color (with the scale to the right of each panel). Within the 1- to 15-Hz frequency band (the signal was high-pass filtered at 1 Hz), all sites exhibit the highest power between 5 and 8 Hz. The power of theta-ranged oscillations is higher in hippocampal subregions and rRS. In contrast, the rACC has a more diffuse band of oscillatory activities that is predominantly theta, but also include slower oscillations down to 2 Hz. Down pointing white arrows illustrate a point in time at which the frequency structure is similar across all electrodes. Up pointing white arrows illustrate relative inconsistencies in theta frequency power across the electrodes. The discrepancy between the 2 types of case argues for distinct cortical generators of theta frequency rhythms.

Figure 7.

An example of changes in the power spectrum with time taken from a single animal within a single trial at several recording sites: (A) CA1, (B) DG, (C) rRS, and (D) rACC. Spectral power is indicated by color (with the scale to the right of each panel). Within the 1- to 15-Hz frequency band (the signal was high-pass filtered at 1 Hz), all sites exhibit the highest power between 5 and 8 Hz. The power of theta-ranged oscillations is higher in hippocampal subregions and rRS. In contrast, the rACC has a more diffuse band of oscillatory activities that is predominantly theta, but also include slower oscillations down to 2 Hz. Down pointing white arrows illustrate a point in time at which the frequency structure is similar across all electrodes. Up pointing white arrows illustrate relative inconsistencies in theta frequency power across the electrodes. The discrepancy between the 2 types of case argues for distinct cortical generators of theta frequency rhythms.

A plot of spectral power, averaged across all electrodes within each anatomical area (Fig. 8), shows prominent low frequency, delta-range activity in the anterior half of the midline cortex. Very prominent ∼1- to 2-Hz oscillations are present in the cACC but not in other areas. Together with other frontal recording sites, mPFC to cACC also exhibit a ∼2- to 3-Hz delta oscillation. However, these delta activities are largely absent in the posterior midline cortex and the hippocampus. When the animal is immobile or engaged in behavior such as grooming, there is a large amount of delta activity and some low frequency theta activity.

Figure 8.

FFTs, averaged across animals and epochs, from different brain regions under various behavioral states: (A) grooming, (B) immobility, (C) movement (head turn and locomotion, and (D) rearing. Oscillatory processes in the delta and theta frequencies dominate the power spectra. During grooming and immobility modest theta is present in all areas, more so in the posterior cortex and the hippocampus than the anterior cortex. But, delta oscillations dominate anterior cortical activities. When animals are engaged in locomotion, head movements, and rearing, delta oscillations decrease in power, whereas theta oscillations increase in power. This modulation of oscillatory activities in different frequency bands is more prominent in the anterior cortex compared with the posterior cortex and the hippocampus.

Figure 8.

FFTs, averaged across animals and epochs, from different brain regions under various behavioral states: (A) grooming, (B) immobility, (C) movement (head turn and locomotion, and (D) rearing. Oscillatory processes in the delta and theta frequencies dominate the power spectra. During grooming and immobility modest theta is present in all areas, more so in the posterior cortex and the hippocampus than the anterior cortex. But, delta oscillations dominate anterior cortical activities. When animals are engaged in locomotion, head movements, and rearing, delta oscillations decrease in power, whereas theta oscillations increase in power. This modulation of oscillatory activities in different frequency bands is more prominent in the anterior cortex compared with the posterior cortex and the hippocampus.

Delta/low frequency theta (3–6 Hz) power is higher during immobility compared with grooming. During the same behavioral epochs, the RS and the hippocampus collectively display 5–6 Hz theta without the delta component. As expected, theta power is higher during grooming compared with immobility in the hippocampus and the RS, most likely due to the small postural changes associated with grooming. When the animals are engaged in overt motor behaviors the delta and lower frequency (5–6 Hz) theta power is suppressed, whereas theta at a higher frequency dominates. The suppression of lower frequency oscillations and an enhancement of theta activities at higher frequencies are observed similarly in the RS and the hippocampus. The switching from delta to theta is more evident, and the theta power is higher, when the animals are rearing compared with when they are showing general locomotion and head movements.

When the spectral power is averaged into either delta (1–4 Hz) or theta (6–10 Hz) bands (Fig. 9) for comparisons between structures sampled, significant structure × band power × behavior interactions can be detected by ANOVA (F3, 309 = 115.42, P < 0.001, F3, 405 = 111.09, P < 0.001 for delta and theta bands, respectively). Figure 9A summarizes the differences of spectral power between structures in the form of the t-statistic. After Bonferroni correction, delta band power was significantly different between the anterior and posterior (including the hippocampus) halves of the cortex. In the theta band, spectral power is roughly equivalent between the structures examined, with the exception of rACC, which demonstrated significantly lower power compared with the rRS and the hippocampus (see insert for actual log power in each frequency band). It is clear that theta-ranged oscillations have more power during locomotion and rearing (Fig. 9B) compared with grooming and immobility. Conversely, delta power is highest during grooming and immobility but lowest during locomotion and rearing. This reciprocal coupling of delta and theta oscillations with ongoing behavior was seen consistently throughout the regions examined to be statistically significant (see Table 2), with the exception of rACC, where statistically significant behavioral modulation of oscillation power was only seen in the delta band.

Table 2

Repeated-measures ANOVA of theta and delta band power for variation across the 4 behavioral conditions of Figure 8

Brain region Delta band power
 
Theta band power
 
mPFC F3, 36 = 11.24 P < 0.001 F3, 48 = 7.36 P < 0.001 
rrACC F3, 36 = 16.72 P < 0.001 F3, 48 = 31.57 P < 0.001 
rACC F3, 36 = 4.92 P < 0.01 F3, 48 = 1.73 P = 0.174 
cACC F3, 36 = 8.24 P < 0.001 F3, 48 = 10.6 P < 0.001 
rRS F3, 36 = 28.38 P < 0.001 F3, 48 = 22.98 P < 0.001 
cRS F3, 36 = 14.41 P < 0.001 F3, 48 = 15.88 P < 0.001 
CA1 F3, 36 = 55.96 P < 0.001 F3, 48 = 23.52 P < 0.001 
DG F3, 36 = 86.83 P < 0.001 F3, 48 = 28.37 P < 0.001 
Brain region Delta band power
 
Theta band power
 
mPFC F3, 36 = 11.24 P < 0.001 F3, 48 = 7.36 P < 0.001 
rrACC F3, 36 = 16.72 P < 0.001 F3, 48 = 31.57 P < 0.001 
rACC F3, 36 = 4.92 P < 0.01 F3, 48 = 1.73 P = 0.174 
cACC F3, 36 = 8.24 P < 0.001 F3, 48 = 10.6 P < 0.001 
rRS F3, 36 = 28.38 P < 0.001 F3, 48 = 22.98 P < 0.001 
cRS F3, 36 = 14.41 P < 0.001 F3, 48 = 15.88 P < 0.001 
CA1 F3, 36 = 55.96 P < 0.001 F3, 48 = 23.52 P < 0.001 
DG F3, 36 = 86.83 P < 0.001 F3, 48 = 28.37 P < 0.001 
Figure 9.

Average log power of delta (1–4 Hz) and theta (6–10 Hz) band oscillations throughout the neocortex and the hippocampus. (A) Spectral power across all structures sampled irrespective of behavior (vertical histograms). The triangular matrices plot the t-statistics for differences between structures: top right for the delta band and bottom left for the theta band. The cell intensities in the matrix show structure grouping among the anterior cortices, up to cACC and among the posterior cortices up to and including the cACC. White asterisks denote statistical significance (P < 0.05) after Bonferroni correction. (B) Spectral power during different behavioral epochs. The data clearly show that theta band power was most prominent during voluntary behaviors and delta band power was most prominent during grooming and behavioral immobility. ANOVA revealed, with the exception of rACC in the theta band, that delta and theta band power were modulated by behavior. Specifically, post hoc comparisons revealed that the differences in power only existed between 2 main classes of behavior, voluntary (bodily movements and rearing) and grooming/behavioral immobility. *Significant differences observed in delta band power only between the 2 behavioral classes; **significant differences observed in both theta and delta band power between the 2 behavioral classes.

Figure 9.

Average log power of delta (1–4 Hz) and theta (6–10 Hz) band oscillations throughout the neocortex and the hippocampus. (A) Spectral power across all structures sampled irrespective of behavior (vertical histograms). The triangular matrices plot the t-statistics for differences between structures: top right for the delta band and bottom left for the theta band. The cell intensities in the matrix show structure grouping among the anterior cortices, up to cACC and among the posterior cortices up to and including the cACC. White asterisks denote statistical significance (P < 0.05) after Bonferroni correction. (B) Spectral power during different behavioral epochs. The data clearly show that theta band power was most prominent during voluntary behaviors and delta band power was most prominent during grooming and behavioral immobility. ANOVA revealed, with the exception of rACC in the theta band, that delta and theta band power were modulated by behavior. Specifically, post hoc comparisons revealed that the differences in power only existed between 2 main classes of behavior, voluntary (bodily movements and rearing) and grooming/behavioral immobility. *Significant differences observed in delta band power only between the 2 behavioral classes; **significant differences observed in both theta and delta band power between the 2 behavioral classes.

Cortico-Cortical and Cortico-Hippocampal Interactions at Theta Frequencies

Cortico-cortical coherences are illustrated in Figure 10 and cortico-hippocampal coherences are illustrated in Figure 11, except where there were fewer than 3 available electrode pairs.

Figure 10.

Cortico-cortical coherences between 0 and 10 Hz. The coherence plots are organized in a triangular matrix. Frequency is on the x-axis and coherence is on the y-axis of each coherence plot. Missing plots are due to the lack of sufficient data to construct reliable coherence plots. Coherences under each behavioral state are plotted in differently styled lines, as seen in the legend at the top right. Dashed horizontal lines in each coherence spectrum represent the 95% confidence interval for coherence values across all behavioral epochs. The anterior half of the cortex examined has high coherences across the 0–10 Hz range and the posterior half has preferential frequency coupling in the theta range. Between the rostro-caudal parts, only voluntary behavior associated theta activities were found to be above the set confidence interval. Interestingly, cACC appears to be preferentially coupled in the delta range with the anterior half of the midline cortices and preferentially coupled at theta frequencies with the posterior cortices.

Figure 10.

Cortico-cortical coherences between 0 and 10 Hz. The coherence plots are organized in a triangular matrix. Frequency is on the x-axis and coherence is on the y-axis of each coherence plot. Missing plots are due to the lack of sufficient data to construct reliable coherence plots. Coherences under each behavioral state are plotted in differently styled lines, as seen in the legend at the top right. Dashed horizontal lines in each coherence spectrum represent the 95% confidence interval for coherence values across all behavioral epochs. The anterior half of the cortex examined has high coherences across the 0–10 Hz range and the posterior half has preferential frequency coupling in the theta range. Between the rostro-caudal parts, only voluntary behavior associated theta activities were found to be above the set confidence interval. Interestingly, cACC appears to be preferentially coupled in the delta range with the anterior half of the midline cortices and preferentially coupled at theta frequencies with the posterior cortices.

Figure 11.

Cortico-hippocampal coherences between 0 and 10 Hz. Coherences (y-axis) across 0–10 Hz (x-axis) are plotted for each hippocampal subfield (CA1 and DG) compared with the midline cortex. Coherences under different behavioral states are represented by different line styles shown in the legend on the top left. Also, CA1 versus DG coherences are shown below the median section of the cortex for comparison. Dotted horizontal lines in each coherence spectrum represent the 95% confidence interval for coherence values across all behavioral epochs. Coherences within the theta range (dark gray shade) have prominent peaks well over the set significance level. A shift of higher cortico-DG coherences in the anterior cortex to higher cortico-CA1 coherences in the posterior cortex can be observed.

Figure 11.

Cortico-hippocampal coherences between 0 and 10 Hz. Coherences (y-axis) across 0–10 Hz (x-axis) are plotted for each hippocampal subfield (CA1 and DG) compared with the midline cortex. Coherences under different behavioral states are represented by different line styles shown in the legend on the top left. Also, CA1 versus DG coherences are shown below the median section of the cortex for comparison. Dotted horizontal lines in each coherence spectrum represent the 95% confidence interval for coherence values across all behavioral epochs. Coherences within the theta range (dark gray shade) have prominent peaks well over the set significance level. A shift of higher cortico-DG coherences in the anterior cortex to higher cortico-CA1 coherences in the posterior cortex can be observed.

The most anterior regions of the medial cortex, comprising the mPFC and rrACC, appeared to demonstrate uniformly high coherence across 0–10 Hz (Fig. 10). However, because the anterior sample size is smaller, only coherences in the theta band during rearing showed significant coupling consistently between these areas above the 95% confidence interval. Interestingly, there appears to be a double dissociation in delta/theta coupling and automatic/voluntary behaviors between the mPFC and the rrACC. Delta was more prominent during immobility and grooming with coherence values above the 95% confidence interval but theta was more prominent during movement and rearing (delta: behavior × frequency, linear, F3, 120 = 65.68, P < 0.001; theta: behavior × frequency, quadratic, F3, 120 = 88.23, P < 0.05). This behavior-dependent frequency coupling was significant during grooming and rearing, whereas immobility and movement demonstrated similar coherence spectral profiles for grooming and rearing respectively but fell just short of the established 95% confidence interval. The cACC appears to be coupled with the anterior structures in the 3 Hz range during immobility and grooming. There was no preferential theta coupling of cACC to other anterior structures, as opposed to the weak coupling seen between the cACC and the RS.

Both the rRS and cRS showed clear coupling at theta frequencies with their immediately anterior counterparts. There is also delta coupling between rRS and cRS at 1.5–2 Hz. This slower component of delta coupling appears to be most prominent during periods of immobility, where movement-related artifact should be at a minimum. Additionally, there appears to be a broader delta coupling between the cRS and the mPFC around the 1.5–2 Hz peak, and a 3 Hz component coupling between the cRS and the rrACC. Finally, when the animal is moving, the coherences within RS are uniformly high, particularly at the theta range. There is no preferential frequency coupling during grooming or immobility.

Generally, the cortico-hippocampal coherences are maximal within the theta range (Fig. 11). The peak coherence is always found within the theta band and well above the 95% confidence interval. Coherence in the theta range is generally higher between the RS and the hippocampal subfields than between the hippocampal subfields and the anterior midline cortices. The amount of cortico-hippocampal theta synchronization appears to be dependent on the behavioral state, where rearing elicited the highest level of coherence, followed by (in order) bodily movements, grooming and then immobility. This trend is more consistent with cortico-DG coherences than cortico-CA1 coherences (see Table 3). Additionally, the frequency range of theta coherence appears to be modulated by behavior. During immobility or grooming, the increase of theta coherence can be seen at a lower frequency (∼7.5 Hz); whereas peak theta coherences during voluntary behaviors are at around 8.5 Hz. This is likely to reflect the shift in predominant theta frequency between the behaviors. Unfortunately, data comparing between rACC and the rostral areas (mPFC, rrACC) were not available due to the lack of sufficient data for the analysis. Effort was made to collect the relevant data, but the proximity of necessary implants to the Bregma and midline sutures, and hence to the major blood vessels, meant that insufficient animals with such implants survived long enough for testing.

Table 3

Repeated-measures ANOVA with orthogonal polynomials of changes in delta and theta coherences between cortical and hippocampal LFPs in different behavioral epochs

Structure pairs Delta coherences
 
Theta coherences
 
mPFC–rrACC (6) behv.lin F3, 846 = 65.68 P < 0.001 behv.quad F3, 799 = 88.23 P < 0.001 
mPFC–cACC (3) behv.cub F3, 144 = 58.39 P < 0.001 behv.lin F3, 136 = 26.11 P < 0.001 
mPFC–rRS (3) behv.lin F3, 144 = 12.26 P < 0.001 NS NS NS 
mPFC–cRS (3) behv.cub F3, 126 = 23.25 P < 0.001 behv.quad F3, 119 = 9.09 P < 0.001 
rrACC–cACC (2) behv.cub F3, 144 = 34.57 P < 0.001 behv.lin F3, 136 = 32.34 P < 0.001 
rrACC–rRS (3) behv.cub F3, 144 = 11.68 P < 0.001 NS NS NS 
rrACC–cRS (3) behv.lin F3, 126 = 33.07 P < 0.001 behv.quad F3, 119 = 9.09 P < 0.001 
rACC–rRS (3) behv.cub F3, 306 = 23.97 P < 0.001 behv.quad F3, 289 = 19.02 P < 0.001 
rACC–cRS (2) behv.lin F3, 144 = 15.60 P < 0.001 behv.quad F3, 136 = 29.11 P < 0.001 
cACC–rRS (2) NS NS NS behv.quad F3, 62 = 8.848 P < 0.001 
cACC–cRS (3) behv.cub F3, 252 = 12.32 P < 0.001 behv.quad F3, 238 = 56.26 P < 0.001 
rRS–cRS (2) NS NS NS NS NS NS 
mPFC–CA1 (5) behv.cub F3, 666 = 24.56 P < 0.001 behv.quad F3, 629 = 42.11 P < 0.001 
mPFC–DG (3) behv.cub F3, 306 = 21.29 P < 0.001 behv.quad F3, 289 = 101.92 P < 0.001 
rrACC–CA1 (5) behv.cub F3, 666 = 11.39 P < 0.001 behv.quad F3, 629 = 41.59 P < 0.001 
rrACC–DG (4) behv.cub F3, 522 = 26.24 P < 0.001 behv.quad F3, 493 = 95.08 P < 0.001 
rACC–CA1 (6) behv.cub F3, 1101 = 16.08 P < 0.001 behv.quad F3, 765 = 37.74 P < 0.001 
rACC–DG (6) behv.quad F3, 810 = 7.75 P < 0.001 behv.quad F3, 765 = 55.51 P < 0.001 
cACC–CA1 (3) behv.quad F3, 270 = 4.10 P < 0.01 behv.lin F3, 255 = 87.38 P < 0.001 
cACC–DG (3) behv.quad F3, 270 = 7.29 P < 0.001 behv.lin F3, 255 = 110.83 P < 0.001 
rRS–CA1 (4) behv.quart F3, 468 = 10.98 P < 0.001 behv.quad F3, 442 = 53.33 P < 0.001 
rRS–DG (5) behv.quad F3, 702 = 14.16 P < 0.001 behv.quad F3, 663 = 46.80 P < 0.001 
cRS–CA1 (8) behv.lin F3, 1170 = 32.05 P < 0.001 behv.quad F3, 1105 = 143.79 P < 0.001 
cRS–DG (6) behv.lin F3, 1082 = 29.76 P < 0.001 behv.quad F3, 714 = 231.41 P < 0.001 
CA1–DG (14) behv.lin F3, 1872 = 77.51 P < 0.001 behv.quad F3, 1768 = 249.40 P < 0.001 
Structure pairs Delta coherences
 
Theta coherences
 
mPFC–rrACC (6) behv.lin F3, 846 = 65.68 P < 0.001 behv.quad F3, 799 = 88.23 P < 0.001 
mPFC–cACC (3) behv.cub F3, 144 = 58.39 P < 0.001 behv.lin F3, 136 = 26.11 P < 0.001 
mPFC–rRS (3) behv.lin F3, 144 = 12.26 P < 0.001 NS NS NS 
mPFC–cRS (3) behv.cub F3, 126 = 23.25 P < 0.001 behv.quad F3, 119 = 9.09 P < 0.001 
rrACC–cACC (2) behv.cub F3, 144 = 34.57 P < 0.001 behv.lin F3, 136 = 32.34 P < 0.001 
rrACC–rRS (3) behv.cub F3, 144 = 11.68 P < 0.001 NS NS NS 
rrACC–cRS (3) behv.lin F3, 126 = 33.07 P < 0.001 behv.quad F3, 119 = 9.09 P < 0.001 
rACC–rRS (3) behv.cub F3, 306 = 23.97 P < 0.001 behv.quad F3, 289 = 19.02 P < 0.001 
rACC–cRS (2) behv.lin F3, 144 = 15.60 P < 0.001 behv.quad F3, 136 = 29.11 P < 0.001 
cACC–rRS (2) NS NS NS behv.quad F3, 62 = 8.848 P < 0.001 
cACC–cRS (3) behv.cub F3, 252 = 12.32 P < 0.001 behv.quad F3, 238 = 56.26 P < 0.001 
rRS–cRS (2) NS NS NS NS NS NS 
mPFC–CA1 (5) behv.cub F3, 666 = 24.56 P < 0.001 behv.quad F3, 629 = 42.11 P < 0.001 
mPFC–DG (3) behv.cub F3, 306 = 21.29 P < 0.001 behv.quad F3, 289 = 101.92 P < 0.001 
rrACC–CA1 (5) behv.cub F3, 666 = 11.39 P < 0.001 behv.quad F3, 629 = 41.59 P < 0.001 
rrACC–DG (4) behv.cub F3, 522 = 26.24 P < 0.001 behv.quad F3, 493 = 95.08 P < 0.001 
rACC–CA1 (6) behv.cub F3, 1101 = 16.08 P < 0.001 behv.quad F3, 765 = 37.74 P < 0.001 
rACC–DG (6) behv.quad F3, 810 = 7.75 P < 0.001 behv.quad F3, 765 = 55.51 P < 0.001 
cACC–CA1 (3) behv.quad F3, 270 = 4.10 P < 0.01 behv.lin F3, 255 = 87.38 P < 0.001 
cACC–DG (3) behv.quad F3, 270 = 7.29 P < 0.001 behv.lin F3, 255 = 110.83 P < 0.001 
rRS–CA1 (4) behv.quart F3, 468 = 10.98 P < 0.001 behv.quad F3, 442 = 53.33 P < 0.001 
rRS–DG (5) behv.quad F3, 702 = 14.16 P < 0.001 behv.quad F3, 663 = 46.80 P < 0.001 
cRS–CA1 (8) behv.lin F3, 1170 = 32.05 P < 0.001 behv.quad F3, 1105 = 143.79 P < 0.001 
cRS–DG (6) behv.lin F3, 1082 = 29.76 P < 0.001 behv.quad F3, 714 = 231.41 P < 0.001 
CA1–DG (14) behv.lin F3, 1872 = 77.51 P < 0.001 behv.quad F3, 1768 = 249.40 P < 0.001 

Note: Number of pairs in each analysis is provided in brackets. lin: linear; quad: quadratic; cub: cubic; quart: quartic; NS: nonsignificant.

Figure 12 is a summary of observed coherences in the delta (1–4 Hz) and theta (6–10 Hz) ranges under the 4 behavioral states studied. The most striking pattern was the tight clustering of the 3 most anterior areas of cortex examined. As described earlier, these 3 areas are tightly coupled in both frequency bands examined. In all cases, an increase in theta band coherence with a decrease in delta band coherence can be seen between automatic and voluntary behaviors. There is an overall coherence increase when the animals switched from grooming and immobility to movement and rearing—as evidenced by the overall shift from cooler to warmer colors in the figure. More specifically, theta-range activities clearly became more synchronized as the animals switched from immobility, to grooming, to movement, and then to rearing. Although this tendency for theta coherence to increase appears to be a global effect, post hoc analysis was only able to demonstrate statistical significance for cRS and anterior cortical interactions when the animals were switching to rearing from any of the other 3 behaviors.

Figure 12.

Mean coherences within the delta (1–4 Hz) and theta (6–10 Hz) bands in all structural-pair comparisons. The 4 matrices represent the coherences in the delta and theta bands during the 4 behavioral states. Coherences in the theta band are represented in the top-left triangular matrix and coherences in the delta band are in the bottom right. The coherences in each frequency band were scaled and displayed as colors. White cells in the matrix denote missing data. The level of coherence between any pair is roughly proportional to the physical distance between the 2 structures. A general increase in synchronization can be seen when the animals engaged in motor activities compared with immobility and grooming. A clear increase in theta-range activities can be seen between the switch from immobility and grooming to motor behaviors. Post hoc statistical differences (P < 0.05, Bonferroni corrected) within a particular frequency band across the behaviors are presented as: *coherences are significantly higher than grooming; †coherences are significantly higher than immobility; ‡coherences are higher than movement.

Figure 12.

Mean coherences within the delta (1–4 Hz) and theta (6–10 Hz) bands in all structural-pair comparisons. The 4 matrices represent the coherences in the delta and theta bands during the 4 behavioral states. Coherences in the theta band are represented in the top-left triangular matrix and coherences in the delta band are in the bottom right. The coherences in each frequency band were scaled and displayed as colors. White cells in the matrix denote missing data. The level of coherence between any pair is roughly proportional to the physical distance between the 2 structures. A general increase in synchronization can be seen when the animals engaged in motor activities compared with immobility and grooming. A clear increase in theta-range activities can be seen between the switch from immobility and grooming to motor behaviors. Post hoc statistical differences (P < 0.05, Bonferroni corrected) within a particular frequency band across the behaviors are presented as: *coherences are significantly higher than grooming; †coherences are significantly higher than immobility; ‡coherences are higher than movement.

Cortico-hippocampal coherences demonstrate a clear-cut relationship that divides the halves of the midline cortex. From mPFC to cACC, the general coherences were much lower compared with those seen between the hippocampus and the posterior cortices. The CA1-anterior cortical interactions exhibit voluntary-behavior-related increase in theta coherence. However, only DG-anterior cortical theta coherence increases significantly from immobility and grooming to rearing. Although the rACC has little theta oscillation, a general increase in synchronicity is nevertheless present across behavioral transitions (i.e., from immobility/grooming to movement/rearing). Cortico-hippocampal interactions involving the posterior cortex demonstrate statistically significant increases of theta coherence from immobility and grooming to movement and rearing as discussed above. Instead the predominant DG involvement with coherence differences in the anterior cortex, the same pattern is observed in both DG and CA1 with the posterior cortex.

Discussion

Field potentials recorded from the rat midline cortices confirmed the existence of hippocampal theta-like activities of modest amplitude throughout the rostro-caudal axis. Recordings from segregated regions supported the view that a number of different theta-generating loci exist in the rat, as in monkeys (Tsujimoto et al. 2003, 2006) and men (Asada et al. 1999; Pizzagalli et al. 2003); but we did not attempt discrete identification and isolation of independent sources.

The power and frequency of rhythmic activity varied with the class of ongoing spontaneous behavior. Theta field potentials from the anterior and posterior halves of the midline cortex showed increases in higher frequency theta power (∼8 Hz) during locomotion and rearing compared with grooming or immobility. This was also true of the hippocampus, as would be expected from previous data. The increase in theta power during locomotion and rearing was coupled with a suppression of slower oscillations in the anterior midline cortices. These were strong during grooming and immobility in frontal but not posterior regions.

The behavior-related increases in higher frequency (6–10 Hz) theta activity were accompanied by an increase in cortico-cortical and cortico-hippocampal coherence. This was high during nonautomatic movements compared with grooming or immobility; and, with rearing, there was extensive coherence across the entire midline cortex and hippocampus.

Field Potential Recordings and Volume Conduction

In this study, electrode arrays were used to collect field potential data from the rat midline cortices. With such field potentials a major issue for interpretation is volume conduction. Field recordings are only possible because changes in local electrical charge at each individual neuron are conducted through the extracellular volume and so are, effectively, averaged by the gross recording electrode. A problem here is that a specific amplitude of field potential may be the result of activity in a weaker closer source or of a stronger more distant one. Further, large changes in current at a distant site may “contaminate” the representation of local field activities.

Much attention has been focused on the reliability of theta field potential recordings outside of the hippocampus, with many believing theta oscillations recorded elsewhere in the brain are all a product of volume conduction from the hippocampus (see Introduction for an example of this regarding the RS theta oscillations). In the current experiments, therefore, we need to address the question of the extent to which the recorded rhythmicity is generated close to the electrode as opposed to being volume conducted, particularly from the hippocampus.

If a component of a signal recorded at 1 location is the result of volume conduction from a more distant source, 1 would expect it to be a carbon copy of the source signal with zero phase difference, identical wave morphology and a reduction in amplitude that is related to the distance from the source Below we will use qualitative data (in the forms of 1) fluctuating theta amplitude/power along the rostro-caudal axis; 2) dissimilarities of wave morphology between neocortical and hippocampal recorded field potentials; and 3) behavioral state-dependent changes in theta power and coherence), as well as the statistically verified quantitative data on cortico-hippocampal theta phase differences, to argue against the notion that all our recordings may be a product solely of volume conduction from the hippocampus.

Theta-like Activities Differ Throughout the Rat Anterior Midline Cortex

Theta field potentials recorded from the various locations within the anterior midline cortices were different in amplitude, phase, prevalence and background activities at different sites. In the mPFC, small amplitude theta was recorded as reported by Siapas et al. (2005). Larger amplitudes of theta field potentials and an increase in prevalence were observed in the rrACC and rACC. At the cACC, there was medium amplitude, but nonetheless continuous, theta activity, which became progressively larger toward RS. Most importantly, rACC and cACC show similar phase profiles in terms of where their theta troughs fell on the DG theta cycle. This is completely different from the phase relationships observed in the more rostral regions of the frontal cortex (mPFC and rrACC). These data indicate that, in frontal cortex, there is at least 1 rostral site that displays distinct theta from more caudal sites. The relatively slow change in amplitude across rostro-caudal locations compared with that in the dorsoventral axis also suggests that, even when there is phase invariance, theta at different rostro-caudal locations is not due to volume conduction from a single source. Thus, a variety of cortical areas each appears to contribute, at least in part, to the theta rhythm recorded by electrodes placed within them.

Power in the theta range is largely equivalent in mPFC, rrACC, rACC, and cACC. The structure that demonstrates the highest theta-range power is the cACC. This trend is not reflected in theta coherences, where a relatively high coherence with the hippocampus is detected in the mPFC, rrACC, and cACC but not in the rACC. These findings are supportive of separate sources of theta generation in the anterior half of the midline cortex, because the extent to which a cortical site and hippocampal theta are coupled is independent of the total spectral power the particular cortical site demonstrates at theta frequencies. Functionally, it also suggests that some putative cortical sources of theta may act as, at least partially, independent oscillators from the hippocampus.

The data presented in this study have shown that theta field potentials do exist in the rat anterior midline cortex. Critically, characteristics of the recorded anterior midline EEG in terms of state transitions, amplitude, phase differences and wave morphology are all quantitatively and qualitatively different to the hippocampus. Therefore, the current observations suggest that the recorded field potentials in the anterior midline regions are not volume conducted from the hippocampus. This conclusion is consistent with the observed single-cell spike coupling to ongoing hippocampal theta in the mPFC (Hyman et al. 2005; Jones and Wilson 2005b; Siapas et al. 2005). The sources and sinks associated with this phasic single unit activity could provide the cellular basis for local theta generation.

Evidence for Independent Theta Generation in the RS

The nature of the field potentials recorded from the RS region is highly controversial. One of the earliest studies investigating the depth profile of hippocampal theta also examined the field potentials of the overlying cortex. A phase-reversal through the rRS was found and used as evidence for a local theta generator (Feenstra and Holsheimer 1979). This phase reversal has not been replicated in awake or anesthetized preparations (Borst et al. 1987; Leung and Borst 1987; Colom et al. 1988).

In the present study, a cortical layer cross-sampling revealed that, unlike the frontal cortices, RS theta is always in-phase with CA1 theta, suggesting LFPs recorded form the RS may be the product of volume conduction. However, not all the theta field potentials recorded from the RS were completely congruent, in wave shape or phase, with hippocampal theta. Also, the large amplitude irregular activity observed in the hippocampus was not correlated with that in RS (data not shown). Logically, if theta activities in the RS are volume conducted from the hippocampus, larger amplitude irregular activities, identical to that in the hippocampus, should also be observed in the RS. This is clearly not the case. The occasional observation of RS theta in the absence of any hippocampal theta rhythmicity also suggests that the RS generates theta locally. The above observation corroborate the theta-like multiunit activities in the RS found in rabbits with hippocampal lesions (Talk et al. 2004), where RS theta may be locally generated and become independent from the hippocampus. Our data cannot tell whether, when both RS and CA1 show theta, RS generates its theta locally and this happens to be in-phase with the hippocampus, or whether the RS theta record is, at such times, volume conducted from the hippocampus. It may be that the only time RS generates theta locally is during times when the hippocampus displays irregular activity.

Midline Cortical Theta has Behavioral Correlates Similar to the Hippocampus

Vanderwolf and colleagues conducted a series of experiments where hippocampal EEG was correlated with spontaneous behavior (Vanderwolf 1969; Kramis et al. 1975). They concluded that during locomotion or large movements of the head, high frequency theta (6–10 Hz) is always present in the rat hippocampus. Conversely, during immobility or smaller postural changes, lower frequency theta (4–6 Hz) could be observed (Whishaw and Vanderwolf 1973). The data presented here support the original findings on hippocampal theta rhythm. Additionally, our data also suggest that the RS shows approximately the same behavioral correlates of theta with the ongoing behavior compared with the hippocampus. This is not entirely surprising because many hippocampal-dependent tasks have correlates in the RS and the integrity of the RS also appears to be important for hippocampal function (Aggleton et al. 2000; Aggleton and Pearce 2001; Harker and Whishaw 2004).

The anterior portion of the midline cortex appears to follow the same behavioral relationships as the hippocampus and the RS with respect to the higher frequency theta characteristic of the hippocampus. The FFTs suggest that a delta frequency component present during grooming and immobility disappears when the animals are engaged in large bodily movements associated with exploration. Instead, high theta frequency at about 7.7 Hz increases in power in the anterior midline cortex, which is consistent with the dominant frequency that the hippocampus and the RS exhibit. Curiously, qualitative observations from many trials on video and EEG recordings revealed that instead of exhibiting strong, continuous theta activity during voluntary behaviors as does the hippocampus or the RS, the amplitude and robustness of theta oscillations in the anterior midline cortex appear to wax and wane during voluntary behaviors (as exemplified by spectrograms in Fig. 7). This is also illustrated by raw traces in Figure 5 where not all the highest power at theta frequencies at other sites was mirrored in the anterior cortex. This waxing and waning is similar to that reported for human theta/alpha EEG recorded superficially from frontal sites (Mizuki et al. 1980).

As suggested by Figure 7, frontal theta activity is the most robust and prevalent during the 1st half of the trial, suggesting behavioral modulation when animals decrease their exploratory behaviors as the trial progresses. Qualitatively, the highest amplitude, most continuous, frontal theta is often seen during rearing, active whisking and extension of the trunk during horizontal exploration. These preliminary observations are consistent with the proposed functional involvement of the area (mPFC, ACC) in attention, and related higher cognitive functions such as decision making and working memory. As mentioned above, coherence between the mPFC and the hippocampus (CA1) is selectively increased in the theta range during correct trials of a working memory paradigm (Jones and Wilson 2005b). The mPFC theta and theta from other anterior midline structures, and their coherences in the theta range with the hippocampus may reflect a more fundamental, nontask specific process (such as attention) that could be modulated by experience to serve behavioral demands on the animal.

Hippocampal and Anterior Midline Functions May Be Coordinated by Theta-Range Oscillations

There is evidence for functional connectivity between the hippocampus and the midline cortices; but it is unclear how the activities in these structures are synchronized. Wilson and colleagues (Jones and Wilson 2005b; Siapas et al. 2005) have suggested that the monosynaptic influences from the ventral CA1/subiculum areas on the mPFC interneurons (Tierney et al. 2004) may modulate the synchronization between the 2 areas. The results from the present study suggest that theta activity in the anterior midline cortex may have multiple loci that generate theta field potentials with differing delay, amplitude and prevalence—which could be the result of separate hippocampo-cortical projections with different conduction times.

There are a range of routes through which there could be reciprocal hippocampal–cortical interaction. The hippocampal–mPFC projections are known to mostly innervate the deeper layers of the mPFC but are more diffuse dorsally (Jay and Witter 1991; Hoover and Vertes 2007). It is unlikely that the CA1/subiculum can modulate theta in all parts of the anterior midline cortex due to the diffuse nature of innervation immediately outside of mPFC and rrACC and the lack of direct projections to the rACC and cACC. Alternatively, anterior thalamus is a possible candidate for theta relay/modulation in the mPFC because it projects extensively and topographically throughout the whole prefrontal cortex (Groenewegen 1988; Berendse and Groenewegen 1991) and displays theta unit activities (Vertes et al. 2001; Talk et al. 2004). Another set of structures providing afferents into the entire midline cortex (Finch et al. 1984; Gaykema et al. 1990), and that have been observed to display theta field activity (Nerad and McNaughton 2006), are the septal nuclei. It has been shown that lesions in the medial septum (MS) that do not damage the diagonal band nuclei abolish hippocampal theta but have no effect, or even enhances RS theta (Borst et al. 1987). Perhaps the MS/diagonal band of Broca (DBB) topographical projections to the hippocampus and related structures (such as the midline cortex) modulate theta generation and synchrony at all respective sites. In addition, there is a topographical organization of reciprocal anterior midline cortical and RS connections (van Groen and Wyss 1990, 1992, 2003; Fisk and Wyss 1999; Shibata et al. 2004; Jones et al. 2005). The RS also receives projections from both the hippocampus and the MS/DBB (Gaykema et al. 1990; Wyss and van Groen 1992) that are reciprocated through its projections to the entorhinal cortex (Wyss and van Groen 1992; van Groen and Wyss 2003).

The areas examined in this study could, thus, form an extensive network that contains numerous possible functional loops—with specific structures bound into different subcircuits at different times by rhythmic activity. For example, it is known that initial learning of trace eye-blink conditioning in the rabbit depends on both the hippocampus (Akase et al. 1989; Moyer et al. 1990; Kim et al. 1995) and cACC (Weible et al. 2000, 2003); and initial learning of conditional discrimination and spatial tasks depends on both the hippocampus and RS (Bussey et al. 1996; Bontempi et al. 1999; Maviel et al. 2004). Consolidation of these various tasks involves the more rostral regions of the ACC (Bontempi et al. 1999; Frankland et al. 2004; Maviel et al. 2004). A similar shift, but in the opposite rostro-caudal direction, occurs in hippocampal-independent tasks such as discriminative avoidance in rabbits (Gabriel et al. 1987; Talk et al. 2004). Both lesion and neurophysiological data show that, whereas the ACC is involved in early phases of learning, the RS is involved in the consolidation of learned behavior (for a review, see Gabriel and Orona 1982; Gabriel et al. 1987; Gabriel et al. 1991; Gabriel 1993). Interestingly, even though hippocampal damage does not affect simple discriminated avoidance behavior, it can alter the topographic pattern of cellular firing in cingulate cortex in the task (Gabriel et al. 1987)—suggesting some degree of interaction among the various structures even when 1 or more is not functionally critical. This interaction is likely to represent the encoding of the experimental context by the hippocampus—which in simple discrimination is not required for correct control of behavior by the cingulate. When context is made a critical aspect of the task, lesions of hippocampal input that eliminate the context-related topographic pattern of cingulate firing also impair context-specific task acquisition (Smith et al. 2004). The occurrence of the various dissociations, and the evidence for functional linkage between the structures under specific behavioral conditions, suggests that the differential phase dispersion profile seen between the mPFC/rrACC, rACC/cACC, and rRS/cRS groups (cf. Fig. 6) in the current study may reflect variation in functional grouping of cortical and hippocampal structures with cognitive demands.

In summary, this study uncovered at least 2 potential theta-generating loci in the frontal midline cortex. The data collected also show that RS is capable of generating theta locally. Behavioral correlates and coherences in the theta range during voluntary behaviors suggest animals may recruit different theta-generating structures preferentially depending on environmental demands. These data suggest that it will be important to elucidate the mechanisms of theta generation in these areas, and how they are modulated to gain insight into not only the functional importance of the individual structures but also how oscillatory activities can bind multiple brain structures into different circuits to carry out different functions.

Funding

Otago Medical Research Foundation Summer Scholarship to C.K.Y.

Conflict of interest: None declared.

References

Aggleton
JP
Pearce
JM
Neural systems underlying episodic memory: insights from animal research
Philos Trans R Soc Lond B Biol Sci
 , 
2001
, vol. 
356
 (pg. 
1467
-
1482
)
Aggleton
JP
Vann
SD
Oswald
CJ
Good
M
Identifying cortical inputs to the rat hippocampus that subserve allocentric spatial processes: a simple problem with a complex answer
Hippocampus
 , 
2000
, vol. 
10
 (pg. 
466
-
474
)
Akase
E
Alkon
DL
Disterhoft
JF
Hippocampal lesions impair memory of short-delay conditioned eye blink in rabbits
Behav Neurosci
 , 
1989
, vol. 
103
 (pg. 
935
-
943
)
Alonso
A
Garcia-Austt
E
Neuronal sources of theta rhythm in the entorhinal cortex of the rat I: Laminar distribution of theta field potentials
Exp Brain Res
 , 
1987
, vol. 
67
 (pg. 
493
-
501
)
Alonso
A
Garcia-Austt
E
Neuronal sources of theta rhythm in the entorhinal cortex of the rat II: Phase relations between unit discharges and theta field potentials
Exp Brain Res
 , 
1987
, vol. 
67
 (pg. 
502
-
509
)
Asada
H
Fukuda
Y
Tsunoda
S
Yamaguchi
M
Tonoike
M
Frontal midline theta rhythms reflect alternative activation of prefrontal cortex and anterior cingulate cortex in humans
Neurosci Lett
 , 
1999
, vol. 
274
 (pg. 
29
-
32
)
Berendse
HW
Groenewegen
HJ
Restricted cortical termination fields of the midline and intralaminar thalamic nuclei in the rat
Neuroscience
 , 
1991
, vol. 
42
 (pg. 
73
-
102
)
Bilkey
DK
Heinemann
U
Intrinsic theta-frequency membrane potential oscillations in layer III/V perirhinal cortex neurons of the rat
Hippocampus
 , 
1999
, vol. 
9
 (pg. 
510
-
518
)
Bland
BH
Oddie
SD
Theta band oscillation and synchrony in the hippocampal formation and associated structures: the case for its role in sensorimotor integration
Behav Brain Res
 , 
2001
, vol. 
127
 (pg. 
119
-
136
)
Bontempi
B
Laurent-Demir
C
Destrade
C
Jaffard
R
Time-dependent reorganization of brain circuitry underlying long-term memory storage
Nature
 , 
1999
, vol. 
400
 (pg. 
671
-
675
)
Borst
JG
Leung
LW
MacFabe
DF
Electrical activity of the cingulate cortex II: Cholinergic modulation
Brain Res
 , 
1987
, vol. 
407
 (pg. 
81
-
93
)
Bullock
TH
Buzsaki
G
McClune
MC
Coherence of compound field potentials reveals discontinuities in the CA1-subiculum of the hippocampus in freely-moving rats
Neuroscience
 , 
1990
, vol. 
38
 (pg. 
609
-
619
)
Bussey
TJ
Muir
JL
Everitt
BJ
Robbins
TW
Dissociable effects of anterior and posterior cingulate cortex lesions on the acquisition of a conditional visual discrimination: facilitation of early learning vs. impairment of late learning
Behav Brain Res
 , 
1996
, vol. 
82
 (pg. 
45
-
56
)
Carpenter
GA
Grossberg
S
Normal and amnesic learning, recognition and memory by a neural model of cortico-hippocampal interactions
Trends Neurosci
 , 
1993
, vol. 
16
 (pg. 
131
-
137
)
Cohen
NJ
Eichenbaum
H
Memory, amnesia, and the hippocampal memory system
 , 
1993
Cambridge (MA)
The Massachusetts Institute of Technology Press
Collins
DR
Lang
EJ
Pare
D
Spontaneous activity of the perirhinal cortex in behaving cats
Neuroscience
 , 
1999
, vol. 
89
 (pg. 
1025
-
1039
)
Colom
LV
Christie
BR
Bland
BH
Cingulate cell discharge patterns related to hippocampal EEG and their modulation by muscarinic and nicotinic agents
Brain Res
 , 
1988
, vol. 
460
 (pg. 
329
-
338
)
Decoteau
WE
Thorn
C
Gibson
DJ
Courtemanche
R
Mitra
P
Kubota
Y
Graybiel
AM
Learning-related coordination of striatal and hippocampal theta rhythms during acquisition of a procedural maze task
Proc Natl Acad Sci USA
 , 
2007
, vol. 
97
 (pg. 
3800
-
3805
)
Decoteau
WE
Thorn
CA
Gibson
DJ
Courtemanche
R
Mitra
P
Kubota
Y
Graybiel
AM
Oscillations of local field potentials in the rat dorsal striatum during spontaneous and instructed behaviors
J Neurophysiol
 , 
2007
, vol. 
104
 (pg. 
5644
-
5649
)
Faris
PD
Sainsbury
RS
The role of the pontis oralis in the generation of RSA activity in the hippocampus of the guinea pig
Physiol Behav
 , 
1990
, vol. 
47
 (pg. 
1193
-
1199
)
Feenstra
BW
Holsheimer
J
Dipole-like neuronal sources of theta rhythm in dorsal hippocampus, dentate gyrus and cingulate cortex of the urethane-anesthetized rat
Electroencephalogr Clin Neurophysiol
 , 
1979
, vol. 
47
 (pg. 
532
-
538
)
Finch
DM
Derian
EL
Babb
TL
Afferent fibers to rat cingulate cortex
Exp Neurol
 , 
1984
, vol. 
83
 (pg. 
468
-
485
)
Fisk
GD
Wyss
JM
Associational projections of the anterior midline cortex in the rat: intracingulate and retrosplenial connections
Brain Res
 , 
1999
, vol. 
825
 (pg. 
1
-
13
)
Fox
SE
Ranck
JB
Jr
Hippocampal field potentials evoked by stimulation of multiple limbic structures in freely moving rats
Neuroscience
 , 
1979
, vol. 
4
 (pg. 
1467
-
1478
)
Frankland
PW
Bontempi
B
Talton
LE
Kaczmarek
L
Silva
AJ
The involvement of the anterior cingulate cortex in remote contextual fear memory
Science
 , 
2004
, vol. 
304
 (pg. 
881
-
883
)
Gabriel
M
Vogt
BA
Gabriel
M
Discriminative avoidance learning: a model system
Neurobiology of cingulate cortex and limbic thalamus
 , 
1993
Boston (MA)
Birkhäuser
(pg. 
478
-
527
)
Gabriel
M
Kubota
Y
Sparenborg
S
Straube
K
Vogt
BA
Effects of cingulate cortical lesions on avoidance learning and training-induced unit activity in rabbits
Exp Brain Res
 , 
1991
, vol. 
86
 (pg. 
585
-
600
)
Gabriel
M
Orona
E
Parallel and serial processes of the prefrontal and cingulate cortical systems during behavioral learning
Brain Res Bull
 , 
1982
, vol. 
8
 (pg. 
781
-
785
)
Gabriel
M
Sparenborg
S
Posterior cingulate cortical lesions eliminate learning-related unit activity in the anterior cingulate cortex
Brain Res
 , 
1987
, vol. 
409
 (pg. 
151
-
157
)
Gabriel
M
Sparenborg
SP
Stolar
N
Hippocampal control of cingulate cortical and anterior thalamic information processing during learning in rabbits
Exp Brain Res
 , 
1987
, vol. 
67
 (pg. 
131
-
152
)
Gaykema
RP
Luiten
PG
Nyakas
C
Traber
J
Cortical projection patterns of the medial septum-diagonal band complex
J Comp Neurol
 , 
1990
, vol. 
293
 (pg. 
103
-
124
)
Gevins
A
Smith
ME
McEvoy
L
Yu
D
High-resolution EEG mapping of cortical activation related to working memory: effects of task difficulty, type of processing, and practice
Cereb Cortex
 , 
1997
, vol. 
7
 (pg. 
374
-
385
)
Gray
JA
The neuropsychology of anxiety: an enquiry into the functions of the septo-hippocampal system
 , 
1982
New York
Oxford University Press
Gray
JA
McNaughton
N
The neuropsychology of anxiety: an enquiry into the functions of the septo-hippocampal system
 , 
2000
New York
Oxford University Press
Groenewegen
HJ
Organization of the afferent connections of the mediodorsal thalamic nucleus in the rat, related to the mediodorsal-prefrontal topography
Neuroscience
 , 
1988
, vol. 
24
 (pg. 
379
-
431
)
Harker
KT
Whishaw
IQ
A reaffirmation of the retrosplenial contribution to rodent navigation: reviewing the influences of lesion, strain, and task
Neurosci Biobehav Rev
 , 
2004
, vol. 
28
 (pg. 
485
-
496
)
Hoover
WB
Vertes
RP
Anatomical analysis of afferent projections to the medial prefrontal cortex in the rat
Brain Struct Funct
 , 
2007
, vol. 
212
 (pg. 
149
-
179
)
Hyman
JM
Zilli
EA
Paley
AM
Hasselmo
ME
Medial prefrontal cortex cells show dynamic modulation with the hippocampal theta rhythm dependent on behavior
Hippocampus
 , 
2005
, vol. 
15
 (pg. 
739
-
749
)
Ishii
R
Shinosaki
K
Ukai
S
Inouye
T
Ishihara
T
Yoshimine
T
Hirabuki
N
Asada
H
Kihara
T
Robinson
SE
, et al.  . 
Medial prefrontal cortex generates frontal midline theta rhythm
Neuroreport
 , 
1999
, vol. 
10
 (pg. 
675
-
679
)
Jay
TM
Witter
MP
Distribution of hippocampal CA1 and subicular efferents in the prefrontal cortex of the rat studied by means of anterograde transport of Phaseolus vulgaris-leucoagglutinin
J Comp Neurol
 , 
1991
, vol. 
313
 (pg. 
574
-
586
)
Jones
BF
Groenewegen
HJ
Witter
MP
Intrinsic connections of the cingulate cortex in the rat suggest the existence of multiple functionally segregated networks
Neuroscience
 , 
2005
, vol. 
133
 (pg. 
193
-
207
)
Jones
MW
Wilson
MA
Phase precession of medial prefrontal cortical activity relative to the hippocampal theta rhythm
Hippocampus
 , 
2005
, vol. 
15
 (pg. 
867
-
873
)
Jones
MW
Wilson
MA
Theta rhythms coordinate hippocampal-prefrontal interactions in a spatial memory task
PLoS Biol
 , 
2005
, vol. 
3
 pg. 
e402
 
Kim
JJ
Clark
RE
Thompson
RF
Hippocampectomy impairs the memory of recently, but not remotely, acquired trace eyeblink conditioned responses
Behav Neurosci
 , 
1995
, vol. 
109
 (pg. 
195
-
203
)
Kirk
IJ
McNaughton
N
Mapping the differential effects of procaine on frequency and amplitude of reticularly elicited hippocampal rhythmical slow activity
Hippocampus
 , 
1993
, vol. 
3
 (pg. 
517
-
525
)
Kirk
IJ
Oddie
SD
Konopacki
J
Bland
BH
Evidence for differential control of posterior hypothalamic, supramammillary, and medial mammillary theta-related cellular discharge by ascending and descending pathways
J Neurosci
 , 
1996
, vol. 
16
 (pg. 
5547
-
5554
)
Kramis
R
Vanderwolf
CH
Bland
BH
Two types of hippocampal rhythmical slow activity in both the rabbit and the rat: relations to behavior and effects of atropine, diethyl ether, urethane, and pentobarbital
Exp Neurol
 , 
1975
, vol. 
49
 (pg. 
58
-
85
)
Leung
LW
Borst
JG
Electrical activity of the cingulate cortex I: Generating mechanisms and relations to behavior
Brain Res
 , 
1987
, vol. 
407
 (pg. 
68
-
80
)
Maviel
T
Durkin
TP
Menzaghi
F
Bontempi
B
Sites of neocortical reorganization critical for remote spatial memory
Science
 , 
2004
, vol. 
305
 (pg. 
96
-
99
)
McNaughton
N
Logan
B
Panickar
KS
Kirk
IJ
Pan
WX
Brown
NT
Heenan
A
Contribution of synapses in the medial supramammillary nucleus to the frequency of hippocampal theta rhythm in freely moving rats
Hippocampus
 , 
1995
, vol. 
5
 (pg. 
534
-
545
)
McNaughton
N
Ruan
M
Woodnorth
MA
Restoring theta-like rhythmicity in rats restores initial learning in the Morris water maze
Hippocampus
 , 
2006
, vol. 
16
 (pg. 
1102
-
1110
)
Miller
R
Cortico-hippocampal interplay and the representation of contexts in the brain
 , 
1991
Berlin
Springer-Verlag
Mitchell
SJ
Ranck
JB
Jr
Generation of theta rhythm in medial entorhinal cortex of freely moving rats
Brain Res
 , 
1980
, vol. 
189
 (pg. 
49
-
66
)
Mizuki
Y
Tanaka
M
Isozaki
H
Nishijima
H
Inanaga
K
Periodic appearance of theta rhythm in the frontal midline area during performance of a mental task
Electroencephalogr Clin Neurophysiol
 , 
1980
, vol. 
49
 (pg. 
345
-
351
)
Moyer
JR
Jr
Deyo
RA
Disterhoft
JF
Hippocampectomy disrupts trace eye-blink conditioning in rabbits
Behav Neurosci
 , 
1990
, vol. 
104
 (pg. 
243
-
252
)
Nerad
L
McNaughton
N
The septal EEG suggests a distributed organization of the pacemaker of hippocampal theta in the rat
Eur J Neurosci
 , 
2006
, vol. 
24
 (pg. 
155
-
166
)
O'Keefe
J
Nadel
L
The hippocampus as a cognitive map
 , 
1978
Oxford (United Kingdom)
Clarendon Press
O'Keefe
J
Recce
ML
Phase relationship between hippocampal place units and the EEG theta rhythm
Hippocampus
 , 
1993
, vol. 
3
 (pg. 
317
-
330
)
Palomero-Gallagher
N
Zilles
K
Paxinos
G
The rat isocortex
The rat nervous system
 , 
2004
San Diego
Academic Press
(pg. 
729
-
757
)
Paxinos
G
Watson
C
The Rat Brain in Stereotaxic Coordinates
 , 
1998
Sydney (Australia)
Academic Press
Pizzagalli
DA
Oakes
TR
Davidson
RJ
Coupling of theta activity and glucose metabolism in the human rostral anterior cingulate cortex: an EEG/PET study of normal and depressed subjects
Psychophysiology
 , 
2003
, vol. 
40
 (pg. 
939
-
949
)
Sainsbury
RS
Hippocampal theta: a sensory-inhibition theory of function
Neurosci Biobehav Rev
 , 
1998
, vol. 
22
 (pg. 
237
-
241
)
Seidenbecher
T
Laxmi
TR
Stork
O
Pape
HC
Amygdalar and hippocampal theta rhythm synchronization during fear memory retrieval
Science
 , 
2003
, vol. 
301
 (pg. 
846
-
850
)
Shibata
H
Kondo
S
Naito
J
Organization of retrosplenial cortical projections to the anterior cingulate, motor, and prefrontal cortices in the rat
Neurosci Res
 , 
2004
, vol. 
49
 (pg. 
1
-
11
)
Siapas
AG
Lubenov
EV
Wilson
MA
Prefrontal phase locking to hippocampal theta oscillations
Neuron
 , 
2005
, vol. 
46
 (pg. 
141
-
151
)
Smith
DM
Wakeman
D
Patel
J
Gabriel
M
Fornix lesions impair context-related cingulothalamic neuronal patterns and concurrent discrimination learning in rabbits (Oryctolagus cuniculus)
Behav Neurosci
 , 
2004
, vol. 
118
 (pg. 
1225
-
1239
)
Talk
A
Kang
E
Gabriel
M
Independent generation of theta rhythm in the hippocampus and posterior cingulate cortex
Brain Res
 , 
2004
, vol. 
1015
 (pg. 
15
-
24
)
Tierney
PL
Degenetais
E
Thierry
AM
Glowinski
J
Gioanni
Y
Influence of the hippocampus on interneurons of the rat prefrontal cortex
Eur J Neurosci
 , 
2004
, vol. 
20
 (pg. 
514
-
524
)
Tsujimoto
T
Shimazu
H
Isomura
Y
Direct recording of theta oscillations in primate prefrontal and anterior cingulate cortices
J Neurophysiol
 , 
2006
, vol. 
95
 (pg. 
2987
-
3000
)
Tsujimoto
T
Shimazu
H
Isomura
Y
Sasaki
K
Prefrontal theta oscillations associated with hand movements triggered by warning and imperative stimuli in the monkey
Neurosci Lett
 , 
2003
, vol. 
351
 (pg. 
103
-
106
)
van Groen
T
Wyss
JM
Connections of the retrosplenial granular a cortex in the rat
J Comp Neurol
 , 
1990
, vol. 
300
 (pg. 
593
-
606
)
van Groen
T
Wyss
JM
Connections of the retrosplenial dysgranular cortex in the rat
J Comp Neurol
 , 
1992
, vol. 
315
 (pg. 
200
-
216
)
van Groen
T
Wyss
JM
Connections of the retrosplenial granular b cortex in the rat
J Comp Neurol
 , 
2003
, vol. 
463
 (pg. 
249
-
263
)
Vanderwolf
CH
Hippocampal electrical activity and voluntary movement in the rat
Electroencephalogr Clin Neurophysiol
 , 
1969
, vol. 
26
 (pg. 
407
-
418
)
Vertes
RP
Albo
Z
Viana Di Prisco
G
Theta-rhythmically firing neurons in the anterior thalamus: implications for mnemonic functions of Papez's circuit
Neuroscience
 , 
2001
, vol. 
104
 (pg. 
619
-
625
)
Vogt
BA
Vogt
L
Farber
N
Paxinos
G
Cingulate cortex and disease models
The rat nervous system
 , 
2004
San Diego
Academic Press
(pg. 
705
-
727
)
Weible
AP
McEchron
MD
Disterhoft
JF
Cortical involvement in acquisition and extinction of trace eyeblink conditioning
Behav Neurosci
 , 
2000
, vol. 
114
 (pg. 
1058
-
1067
)
Weible
AP
Weiss
C
Disterhoft
JF
Activity profiles of single neurons in caudal anterior cingulate cortex during trace eyeblink conditioning in the rabbit
J Neurophysiol
 , 
2003
, vol. 
90
 (pg. 
599
-
612
)
Whishaw
IQ
Vanderwolf
CH
Hippocampal EEG and behavior: changes in amplitude and frequency of RSA (theta rhythm) associated with spontaneous and learned movement patterns in rats and cats
Behav Biol
 , 
1973
, vol. 
8
 (pg. 
461
-
484
)
Woodnorth
MA
Kyd
RJ
Logan
BJ
Long
MA
McNaughton
N
Multiple hypothalamic sites control the frequency of hippocampal theta rhythm
Hippocampus
 , 
2003
, vol. 
13
 (pg. 
361
-
374
)
Worden
R
Navigation by fragment fitting: a theory of hippocampal function
Hippocampus
 , 
1992
, vol. 
2
 (pg. 
165
-
188
)
Wyss
JM
van Groen
T
Connections between the retrosplenial cortex and the hippocampal formation in the rat: a review
Hippocampus
 , 
1992
, vol. 
2
 (pg. 
1
-
11
)