HIPPOCAMPAL CELLS NON-SPATIAL CORRELATES?

Chapter 5 will consider the various theories and descriptions o f hippocampal function. Some o f these do not conceive hippocampal function in purely spatial terms, though all theories clearly have to try to account for the features o f hippocampal cells

described above. This section considers claims for non-spatial correlates. This is a less studied area, but one o f obvious theoretical imporance. It is clear that correlates may be interpreted in different ways. The following focuses on approach correlates, mismatch correlates, and sampling correlates.

1) Approach cells

One correlate that was found by Ranck (1973) concerns approach to a goal. The basic observation here is that cells may fire at increased rates when goals are approached. Thus W iener et al, (1989) found that some cells fired maximally as the rat approached an odor port or a water cup.

It is obvious that such correlates are not inconsistent with spatial views o f hippocampal function, since the ports and cups are in particular places. Increased firing towards a place can clearly be interpreted in terms o f a directional place field. Remember from the study by Markus et al (1995) that when trajectories become stereotyped as is very likely in the Wiener et al study, place fields become directional.

2) Mismatch cells

Again this phenomenon was one originally explored by Ranck (1973). In turn O ’Keefe (1976) reported “misplace” cells. It is probably fair to say that “misplace” cells have been underexplored by the field as a whole. Broadly two types were reported by O ’Keefe. The first may be described as sensitive to behaviour-or-object- in-place, eg. when cells increased firing when a cup expected to be in a place was removed. The second were more global in operation, eg. increasing firing when environmental conditions were changed. It is possible to reinterpret the second category o f cells as perhaps those associated with “remapping”. The first kind o f response is very interesting, and may be similar to the findings o f “non-match” correlates in non-match tasks (eg. Hampson et al, 1999). Again, however, the spatial framework in such responses is very obvious.

3) Cue-Sampling correlates

These have been found, among others, by Wiener et al (1989), and indeed O ’Keefe (1976). Here arguments focus on two issues: the consistent spatial framework, but also that sampling correlates may be associated with theta cells, rather than pyramidal cells.

In the study by Wiener et al (1989) no more than 20% o f cells had cue-sampling correlates. O f this fifth o f the total, only 13% fired differentially during sampling o f a particular pair o f odors, as opposed to another pair, in the same position. Thus only about 2.6% o f the total cells had clear responses to cue-sampling which cannot be explained in terms o f a consistent spatial framework.

O ’Keefe has argued that insufficient attention has been paid to the possibility that some o f these cue-sampling cells might be theta cells.

Wood et ah 1999

In order to rebut the criticisms that the interpretation o f non-spatial correlates is at least uncertain when the tasks involve behaviours directed at/in certain places, recent efforts have focused on tasks which consistently alter the position o f a particular cue. A study by Wood et al (1999) was designed precisely in the light o f these concerns. This study focused on pyramidal cells only, as far as could be judged by waveform peak-to-valley width. The task was a successive olfactory discrimination non-match task, in which cups containing odors were placed in different locations. Altogether there were 9 cup locations, and 9 odors.

Anova showed that about 8% o f cells had cue-sampling correlates (irrespective o f location), 11% had location-specific correlates, 13% had pure match/non-match correlates, 20% o f cells had “approach” correlates, and many more cells showed interactions between these variables. O ’Keefe has made comments on this study in a review on apparent non-spatial correlates o f hippocampal cells (O ’Keefe, 1999). The following comments are largely restricted to those not previously made. Chapter 5 contains further discussion o f this paper in specific relation to the theory o f hippocampal function espoused by Eichenbaum.

In my view, it is not good practice to look at firing only around chosen time-points, as was the basic analysis method in this study; this does not aid understanding.

especially when the task is only run once. Only 108 trials were recorded (much less than 2 (match/non-match) x 9 (locations) x 9 odors). This is additionally important for the reason that hippocampal pyramidal cells have variable firing rates. It may be suspected that these rates can contribute to spurious interpretations, in so much as a low proportion o f cells may vary by chance in tune with any variable and conjunction o f variables. This problem is exacerbated when the session is long, and criteria for inclusion in the analysis is only 100 spikes in total. In my experience, and that o f O ’Keefe (personal communication), a cell may may “shut o f f ’ after a while, or take a while to “start up”, for reasons that may not relate to the experimental variables in question, as far as can be understood. There is a concern that cells that had very few burst episodes would enter into the analysis.

The other problem with not analysing the data for all times is that a correlate such as “approach” may not be the most appropriate interpretation for a cell. These authors did not use the same statistical analysis as for the other cell characterizations, but simply compared “the firing rate during the 1-s period beginning 3 s before the behavioural response, when the animal initiates its approach to the cup, with that during the 1 s immediately before the behavioural response” . It is obvious that speed and possibly other correlates could explain such a difference, but speed was not even measured. This is an example o f the way in which the study is well designed but inadequately analysed.

One o f the simple things to do in such a situation is to do some subsequent testing. For instance, for a cell with a pure odor correlate, test the cell after the task. Does it still show the correlate?

At any rate, it is not surprising that hippocampal-lesioned animals are not impaired in this task (Burton, O ’Keefe and others, unpublished data). It will be interesting to see if the basic results o f the Wood et al study are replicated.