The Language of Neurons: Theory and Applications of a Quantitative Analysis of the Neural Code

摘要

Neurons are excitable cells that communicate with each other producing trains of action potentials, so-called spikes. These spike trains transmit information about the individual's external and internal environment. Furthermore, the processing of spike trains by neurons forms the fundamental basis of all neural computations, leading the individual to take actions and make decisions, and constituting the basis of the interaction with the environment. It is agreed in neuroscience that some form of spikes' organization must exist, serving the encoding and decoding of information. This has been called the neural code, and the specific form that it takes is a matter of current study. Understanding the neural code is highly relevant for achieving a thorough comprehension of the nervous system. Different coding schemes have been proposed in different systems, serving different purposes, and recently the co-existence of various codes in the same system has been recognized as well. We focus our discussion on currently available methods for the quantitative analysis of the neural code. In the field, a classic controversy is whether neurons code information using a time averaged measure (such as a rate code) or complex time patterns. We analyze different formal approaches to this problem, such as the use of a multifractal formalism, the temporal structure function, entropy analysis and information content. We show that an analysis of specific characteristics of the neural code is possible (e.g., rate vs. time coding properties), even in the absence of an experimental correlate with behavioral measures. To this purpose, a specific algorithm based on the calculation of the temporal structure function is presented. We also discuss the role of sparse coding, specifically population coding, and the need to consider both the space and the time domain. Finally, deep brain stimulation is discussed as a possible application. Nowadays, medical technology makes it possible to stimulate neurons lying in both deep and superficial brain structures. This kind of procedure allows directly influencing the neuronal output, and hence the neural code. To count with quantitative tools of analysis that help us better understand the coding properties of neurons and neuronal ensembles will be key in the future for the further development of stimulation therapies.