Computational neuroscience is a contemporary, dynamic, and interdisciplinary field that aims to apply tools from mathematics, computer science, and other fields to the study of neuroscientific problems. The creation of computational models of a neuroscientific system of interest, be it a single cell, a microcircuit, or the entire brain, allows for the performance of in silico experiments (i.e. experiments performed on a computer, in a simulated environment). Often, these models allow for the performance of manipulations and perturbations to the in silico system that would be infeasible, intractable, or impossible in vivo or in vitro. Such experiments facilitate the creation of novel hypotheses that can be tested experimentally (that already contain some "a priori" support via the computational experiment), and are often necessary to begin to untangle the non-linear and complex cause and effect relationships present throughout the brain.
The Neuron To Brain Laboratory uses Computational Neuroscience at three different scales at which the brain can be observed and manipulated. Work at each scale not only adds to our general knowledge of how the brain functions, but also has direct applications to the understanding and treatment of epilepsy.
Micro-scale: Single Neurons
Using the lab's unique access to single-unit patch clamp recordings from human cortical cells, we have created a detailed, multi-compartment model of such a cell. By focusing on the differences in such models based on human versus rodent data, we can begin to discern what is distinct about the human brain, and ensure any conclusions drawn with relation to the treatment of disease are relevant in the human setting. Moreover, utilizing such models in tandem with electrophysiological experiments creates a mutually-beneficial "cycle" in which in silico experiments can inform in vitro experiments, which can then improve the model, and so on.
Microcircuits are thought to be a crucial component of the brain's capability to perform complex tasks, but the dynamics of such networks are inherently non-linear. Mathematical tools are uniquely suited to understanding these complicated dynamics and propose potential mechanisms of action explaining various observed behaviors in such circuits in vivo and in vitro. Furthermore, by modeling these microcircuits in detail, manipulations can be performed that might not be feasible experimentally, often allowing for a more direct investigation of very specific aspects of these circuits.
Macro-scale: Mean-field Models
The activity of large regions of the brain are often characterized by coarse measurements such as LFPs or EEGs. Mathematically, mean-field models can encapsulate the activity of a large number of neuron models into a more simple structure that is amenable to analysis and oftentimes approximate these experimental measurements. These tools allow for an in silico investigation of dynamics that occur on a much larger scale throughout the brain, as well as the potential to create tools that are better suited to diagnose and understand these measures often used clinically.