Probabilistic noise and human electrophysiology.

This article discusses probabilistic noise and human electrophysiology. Denoising procedures are usually needed and fully applicable to reduce the noise from data recording or processing procedures. Yet, human electrophysiological techniques deal with a variety of brain signals that are often overlooked or classified alternatively as signal or noise depending on the study rationale, target, techniques/sensors/equipment used for data recording and processing. This occurs, for instance, with stimulus- or event-related transient responses or with frequency-selective signals (like the gamma band activities), whereby the spontaneous electroencephalographic (EEG) or other signals may be treated as noise and canceled out to emphasize the investigated events. In these cases, however, noise collectively mislabels a combination of physical/electronic noise and of random or unpredictable interferences that originate from sources ubiquitous in the nervous system and would qualify as real brain signals. Common sources are for instance the fluctuations in neurotransmitter release, number of activated postsynaptic receptors, ion concentrations, membrane conductance, effects of previous action potentials, and so forth. To date, the signal structure of brain noise remains undefined; its potential as a source of information about synaptic activities is to a significant extent limited to stochastic models or speculation. A working hypothesis appears nevertheless practicable in human research wh...
Source: Journal of Psychophysiology - Category: Psychiatry & Psychology Source Type: research