Channels and circuits: biophysical and network models of neuronal function

Neuroscience research studies the brain at various different levels of detail. Experimental work explores everything from the molecular machinery within each neuron, to the behavioral output of an organism. Similarly, computational models are designed to operate at many scales, and can address different questions depending upon the complexity of the description. Detailed biophysical modelling incorporates findings about neuronal physiology and structure in order to test hypotheses and generate predictions. However, due to the complexity and number of modelling studies, it is difficult to compare models and to assess biological fidelity. In this work we present a curated database of nearly 3000 voltage-gated ion channel models used in published neuronal simulations, called ICGenealogy. Furthermore, we present a standardized formulation for ion channel dynamics which we fit to all models on this database. Through these two endeavors, we facilitate better experimentally-constrained modelling, while also providing insight into the diversity and complexity of ion channel dynamics seen in single neurons. At a higher level, neuronal network models integrate findings about neuronal activity and cell types in order to explain representations and computations. Here, we present a model of context-dependent associative memory, which incorporates known principles of memory function from psychology research and proposes concrete functional roles for different components of neural circuits. Through analytics and simulations, we show that a contextual memory system not only provides benefits for memory capacity and robustness, but also enables control of memory expression. This work provides new conceptual ideas for memory research, suggests functional roles for different inhibitory cell types, and may help us to understand the interaction of different memory systems in the brain. Together, the work presented here spans two distinct levels of detail and addresses current challenges both in biophysically detailed models, as well as computation in abstract network models.