Computational modeling of neuronal networks

TL;DR

This paper discusses the complexity of neuronal networks in the human brain, emphasizing the importance of computational models that range from detailed biological realism to simplified abstractions for understanding neural dynamics.

Contribution

It provides an overview of different computational modeling approaches for neuronal networks, highlighting their relevance for studying brain function across multiple scales.

Findings

Neuronal groups can perform complex tasks reliably despite individual noise.

Models vary from detailed Hodgkin-Huxley to simplified integrate-and-fire types.

Appropriate models depend on the specific research questions.

Abstract

Human brain contains about 10 billion neurons, each of which has about 10~10,000 nerve endings from which neurotransmitters are released in response to incoming spikes, and the released neurotransmitters then bind to receptors located in the postsynaptic neurons. However, individually, neurons are noisy and synaptic release is in general unreliable. But groups of neurons that are arranged in specialized modules can collectively perform complex information processing tasks robustly and reliably. How functionally groups of neurons perform behavioural related tasks crucial rely on a coherent organization of dynamics from membrane ionic kinetics to synaptic coupling of the network and dynamics of rhythmic oscillations that are tightly linked to behavioural state. To capture essential features of the biological system at multiple spatial-temporal scales, it is important to construct a suitable computational model that is closely or solely based on experimental data. Depending on what one wants to understand, these models can either be very functional and biologically realistic descriptions with thousands of coupled differential equations (Hodgkin-Huxley type) or greatly simplified caricatures (integrate-and-fire type) which are useful for studying large interconnected networks.