Abstract
This thesis explores possible roles of oscillations in the brain. Computer simulations demonstrate the robustness of dendritic voltage bistability, a mechanism for working mem- ory, when including GABAB-activated internal-rectifying potassium channels (Ch. 11). This mechanism is employed in a biophysically realistic neuronal network model that is capable of transforming a temporally extended sequence of inputs into a simultaneously represented spatial pattern of output (Ch. 22). Such a transformation is necessary as many external inputs or internal communications in the brain last longer than any individual neuron can process. This transformation relies on the discretizing potential of oscillations for segmenting multi-part messages, which is the first role discussed. The second role arises from recent work on the rodent hippocampus that has called past models into question. A theory is proposed here in which the two halves of theta-frequency oscillations have distinct processes associated with them: in the first half of each theta cycle, current experience is represented, and in the second half of each theta cycle, predictions of upcoming locations are made (Ch. 33). Anal- ysis of recordings from rat hippocampus during behavior demonstrates that two phenomena exhibited by hippocampal place cells, rate remapping and phase precession, dominate dur- ing di↵erent halves of the theta cycle (Ch. 4, in preparation). This work elaborates on the possible role of oscillations in temporally organizing multi-part computations.