Model-based reconstruction of neural networks

Nykamp

Higher-level functions, such as visual object recognition or

memory, emerge from interactions within the brain's neural

networks. The ability to determine the patterns of connections

among neurons would facilitate discovering how such computations

are performed. Because many connectivity patterns could

theoretically produce neural output with only subtle differences,

characterizing even basic circuitry elements within the brain

remains a formidable challenge. The goal of this project is to

develop and analyze mathematical models of neuronal networks in

order to design tools that can distinguish among network

configurations. Preliminary studies have demonstrated a new

approach that shows promise of leading to tools to infer

connectivity patterns in spiking neuronal networks. The key idea

motivating this approach is to exploit models of neural response

to an experimentally controlled stimulus in order to analyze

neuron output. The explicit structure imposed by mathematical

models helps overcome a major difficulty in estimating network

connectivity from experiments: the fact that only a small subset

of neurons can be measured simultaneously. Hence, the central

challenge addressed by this project is accounting for the effects

of unmeasured neurons. Mathematical models lead to predictions of

the effects of connections from unmeasured neurons. The aim of

the work is to use these predictions to factor out the effects of

unmeasured neurons and better estimate the connectivity among the

measured neurons. The resulting tools help neuroscientists

understand local circuits that underlie neural processing and

function within the brain.

The goal of this work is to develop mathematical tools that

neuroscientists can use to determine how neurons in the brain are

connected to each other. Understanding how neurons are connected

is an important first step toward discovering how neurons

communicate with each other to process information. Such

knowledge could help scientists better understand the effects of

the degradation of such connections, such as in neurodegenerative

diseases, and devise strategies to mitigate or reverse these

effects.