Extracellular electrical activity in the hippocampus of freely moving rats is dominated by a roughly 7-12 Hz pattern o f rhythmical slow activity called the theta
(0) rhythm. This component o f the hippocampal EEG, first described by Jung and Kornmiiller in 1938, is constructed by summed intracellular potentials (Green et al. 1961), and must be present for the hippocampus to function. The 0 rhythm is generated in the hippocampus (Green et al. 1960) from a pacemaker input located in the medial septum (Petsche et al. 1962). Lesions blocking the projection from the septum (Green and Arduini 1954; Becker et al. 1980) or reversible inactivation with local injections of procaine (Mizumori et al. 1989b) produce behavioural deficits akin to those resulting from full hippocampal lesions.
The striking transition to the sinusoidal 0 rhythm from the form of EEG called large irregular activity (LIA) (Routtenberg 1968) and the involvement o f the 0 rhythm in hippocampal function have led to numerous suggested behavioural classifications of the EEG (e.g. Bennett 1975; Gray and Ball 1970; Berry and Thompson 1978) and to efforts to correlate variations in the 0 rhythm frequency and amplitude to observable characteristics of the animal's behaviour (e.g. Whishaw and Vanderwolf 1971; McFarland et al. 1975; Fontani and Vegni 1990).
During periods of immobility there is an occasional large amplitude signal, called a sharp wave, that coincides with a synchronised burst of cell activity. There is disagreement in the literature about which hippocampal neurons are active during LIA. O'Keefe and Nadel (1978) observed that LIA amplitude varies with depth and undergoes an inversion just below the pyramidal layer. The anatomical correlation of sharp waves was verified and mapped extensively by Buzsaki (1986). At the same time as a sharp wave, there is a 100-200 Hz ripple pattern in the pyramidal layer (O'Keefe 1976).
The behaviour patterns that correspond to the hippocampal 0 rhythm vary between species, but for a few behaviours there is nearly general agreement (for a review, see Robinson 1980). In rats, Vanderwolf (1969) found that 0 is not normally present during immobility, but it can be induced by midbrain stimulation or with certain drugs such as ethyl urethane. Based on results from a series o f experiments, Vanderwolf and colleagues proposed that the 0 rhythm is correlated with voluntary
movement, but not automatic movements.
The correlation between the 0 rhythm and movement was a central part o f the evidence for the cognitive map model (O'Keefe and Nadel 1978). However, they proposed that the 0 rhythm coincides with displacements rather than all types of movements. An interesting example is the righting reflex, which is classified as an automatic movement, but is associated with the 0 rhythm. In the cognitive map theory the 0 rhythm is a clock that gates the activity of place cells.
The relationship of the 0 rhythm to normal function of the hippocampus can be studied by measuring the temporal relationship between phase o f the EEG and the activity patterns of hippocampal neurons. It was already pointed out, in Section 3.2, that the relationship between EEG and cellular activity is part of the classification of hippocampal cell types. Also, both theta and complex spike cells have been reported to fire in correlation with the phase o f the 0 rhythm (Sinclair et al. 1982; Buzsaki et al. 1983; Fox et al. 1986; Otto et al. 1991). O'Keefe found that theta cell firing is correlated with either the low-high or the high-low phase o f the EEG. Two independent studies of the phase correlation of complex spike cells (Buzsaki et al. 1983; Fox et al. 1986) found that the spikes occur with a higher probability at a particular phase. However, the value for the most likely phase for complex spike activity differed by over 100° in the two studies.
The synchronisation found between hippocampal 0 rhythm and complex spike cells should be evident in place cell firing. As described above, complex spike cells increase their firing rate with increased speed of locomotion (McNaughton et al. 1983b). If these two results are true, then the frequency of the rhythm should increase with the rat's speed of movement. In one of the early studies, Whishaw and Vanderwolf (1971) found a brief (100-200 msec) correlation between the running
speed of a rat on a motorised treadmill and the 0 frequency, but there was no long-lasting correlation. In the same study, the initial 0 rhythm frequency was shown to correlate with the height of a jump. The largest frequency occurred at the beginning of the jump, and it decreased during the jump. From these data they concluded that the frequency of the 0 rhythm is correlated with vigour. A more recent study found that, during jumping, the instantaneous 0 rhythm fequency correlates better with the initial speed than with impulse or acceleration (Morris et al.
1976; Morris and Hagan 1983).
In rats the hippocampal 0 rhythm can be recorded only in C A l and the dentate gyrus, and not in CA3. There is evidence, however, of 0 rhythm in the activity of single cells in CA3. The 0 rhythm is also found in the subiculum and the entorhinal cortex. Using a grid of electrodes, Bullock et al. (1990) measured the phase correlation along the pyramidal layer. They found long-range correlation that stopped abruptly at the subiculum.
The amplitude and phase of the 0 rhythm change with position of the recording electrode perpendicular to the cell layer. In the rabbit, the phase of the 0 rhythm undergoes an abrupt reversal below the pyramidal layer, whereas in the rat, the phase changes gradually.
Recording of the 0 rhythm poses several difficulties. Much of the recording is differential, in which one electrode is stationary in the dentate, while the other electrode is placed in the test location. This results in the problem that the waveforms will add when in phase and will cancel when 180° out of phase. If the recording is single ended, then more noise is recorded. Several other oscillations can be misinterpreted as the 0 rhythm, including thalamo-cortical spindles during quiet sitting and jaw muscle movement. The difficulties in recording the rhythm may account for some o f the reported differences.
Most of the studies of hippocampal 0 rhythm have been perform ed using acute preparations. From the response of the 0 rhythm properties to pharmacological manipulation, two different forms have been identified (Kramis et al. 1975; Vanderwolf 1975). One form of the 0 rhythm is present during immobility and can be
blocked by atropine, but the second form o f the 0 rhythm is resistant to atropine. These two forms of the 0 rhythm may have different roles in hippocampal function.
3.11 Summary
This review has examined the experimental results that provide the current view of place cell properties, in the context of methodological issues. The properties are divided into the static characteristics of cell firing, and the results from manipulations to the environment. Setting aside the more difficult problem of dynamic environments, there are substantial variations in the reported properties in "unchanging" environments. A significant part of this variation may be due to the difficulties of extracellular recording in freely moving animals. Also, measurement of several of the static properties, including tests of the holistic firing hypothesis, and the distribution of place cells within an environment depend on simultaneous, accurate recording from several place cells. These properties are very important for the development of a neuron level model of the cognitive map.
The stereotrode method provides both improved isolation of single neurons and simultaneous records from several cells. In the stereotrode paper (McNaughton et al. 1983a), McNaughton and co-workers clearly point out that four electrodes in a tetrahedral arrangement can improve on two electrodes.
In the present research a four electrode recording method is developed, called the tetrode (Chapter 4). A new recording apparatus has been constructed in which all o f the signal processing stages have been improved. This method is used to re-examine the properties of place cells in several different environments (Chapter 5). Particular attention is paid to the shape of the firing field of place cells, the homogeneity of place cells and their distribution within an environment.
In addition, as a test of the cognitive map model, the relationship between the rhythm and the speed of movement is re-examined (Chapter 6). As discussed above, the prior measurements were made in a treadmill, and without the benefit of computerised methods. Finally, the temporal relationship between the activity of place cells and the hippocampal rhythm is examined in more detail than was possible in prior published work (Chapter 7). The impact of these new data on the cognitive map model is discussed in Chapter 8.