The role of oscillations in the Hippocampal formation
John O'Keefe
University College London, London, UK

The electrical activity of the hippocampal formation is characterized by several oscillatory potentials of different frequencies: theta (6 - 11 Hz), gamma (30 - 60 Hz), and high-frequency ripples (100- 200 Hz). In this talk I will discuss the functions of theta activity, one of the best-known of these oscillations. I will argue that it has at least three functions: 1) to bind together the activity of cells in different parts of the hippocampal formation and more widely distributed brain regions; 2) to set up the conditions for LTP or LTP depending on the relationship of synaptic inputs to a given cell and the phase of theta; and finally, 3) to provide one of the two oscillators of different frequencies which interact to produce interference patterns within the principal cells of the hippocampal formation and which determine the temporal firing patterns in these cells and their relationship to that of other cells. I will concentrate primarily on the latter function. The first evidence for interference patterns came from the phase precession effect originally observed in CA1 pyramidal cells by O'Keefe and Recce in 1993, and now known to exist in CA3 pyramids, layer 2 stellate cells of the entorhinal cortex and, to a lesser extent, dentate granule cells. We found that hippocampal pyramidal cells fired in bursts at approximately the theta frequency as the rat ran through the place field but that the preferred phase of firing did not remain constant but instead moved to earlier phases with each successive burst. Furthermore the phase of firing correlated had a higher correlation with the animal's location within the field then with any other variable, for example, time since entry into the field. We speculated that this phase precession effect might be due to the interaction of two oscillators of slightly different frequencies summing within the pyramidal cell. This model would explain many of the observed phenomena and in particular the fact that the phase precession often continued in the second half of the field despite the fact that the firing rate decreased at same time. An alternative was that the precession effect was generated in the input structures to the hippocampus and propagated into the hippocampus. Evidence in favor of this latter view came from an experiment by Zugaro and Buzsaki who showed that temporary silencing of pyramidal cells as the animal ran through the field did not alter the relationship between phase and location. When the cell resumed firing, it did so at the appropriate phase for the animal's location in the field. The recent discovery of grid cells in the medial entorhinal cortex by Hafting and Fyhn in the Moser's lab opens up the possibility that they may be the source of the phase precession. Grid cells fire with multiple place fields located at the vertices of a triangular grid. Cells located more ventrally in the entorhinal cortex have larger spacing between firing fields than those in the dorsal entorhinal cortex. We now know that the grid cells located in layer 2 also show the phase precession effect. In a recent computational model, Neil Burgess, Caswell Barry and I have proposed a modification of the dual oscillator interference model which accounts for the firing patterns observed in grid cells. One feature of the model has already been tested and supported. This is that the resonant frequency of the intrinsic oscillator within the EC layer 2 stellate cells should vary systematically from dorsal to ventral entorhinal cortex in order to account for the change in scale of the grid size in that dimension. This has now been demonstrated by Hasselmo's group who have also shown that the time constants of the H current vary systematically from dorsal to ventral in a way which would explain the changes in natural frequency.