Limitation of spike responses in the model granular cell to the post-EOCD time window observed experimentally (Bell and Grant, 1992) required delivery of an IPSP with a fixed latency to arrival of the EOCD input. Although this result clearly suggests LMI involvement at the level of single primary afferent stimulation, the invariant timing of IPSP arrival described in the model cannot easily be accounted for by known anatomical connections between either primary afferent fibers, juxtalobar fibers, or the granular cell and LMI cells. While it is expected that granular cell excitation is able to directly stimulate GABA release from the large LMI terminals that surround the somata of granular cells (Meek, Grant, and Bell, 1999), it is unclear how the degree of this ephaptic excitation could be controlled independently of spike number in the granular cell. For instance, a certain number of spikes would increase extracellular K⁺ concentrations independently of the onset of the first spike relative to EOCD arrival at the granular cell unless the LMI cell has direct information regarding the delay between inputs at the granular cell. It is possible that control of LMI inputs to the granular cell depends on K⁺ accumulations due to the collective activation of a population of granular cells, although this was not addressed by the model.

Since LMI cells possess heavily myelinated dendrites and axons (Meek, Grant and Bell, 1999), unknown inputs from the juxtalobar or primary afferent fibers are likely to be located at the LMI soma. To test the influence of such a connection on granular cell responsiveness, ablation of the LMI cell could be done proximal to the soma with respect to the innervated granular cell, in concert with intracellular recordings of the granular cell following inputs from the juxtalobar and primary afferent fibers. Alternatively, the GABA_A antagonist bicuculine could be applied to a slice preparation under the same inputs. While elimination of the graded spike number response under these conditions would agree with the scheme described in model 1b, maintenance of the response would support the plausibility of primarily intrinsic burst offset control of the type described in model 2b.

Additional considerations involving LMI interactions include development of a quantatative method for modeling extracellular K⁺ dependent ephaptic stimulation of GABA release at the calyx onto the granular cell soma (including changes in release rate following repetitive stimulation), and establishing the role of inhibitory LMI syanpses onto granular cell dendrites. Although it seems likely that inclusion of the constant GABA_A dependent leak conductance in the model would capture the effects of LMI synapses onto granular cell dendrites, due to the low-pass filtering effects of dendrites that lack active, voltage-dependent conductances, it is possible that the dynamics of either K⁺ accumulation at the calyx or vesicular fusion and release at the presynaptic LMI terminal could partially explain the aforementioned control of variable burst offset in the granular cell.

PREVIOUS---NEXT---CONTENTS