A method for analysis of encoding of stimulus separation in ensembles of afferents

This paper describes a new method for the analysis of ensemble coding in populations of receptor afferents. The method is based on principal component analysis (PCA) combined with algorithms for calculation of stimulus separation (discrimination). It allows a quantification of the ability of ensembles of afferents to discriminate between stimuli of different intensity, and it can also be used to calculate to which extent each afferent in an ensemble contributes to the discrimination. Since the emphasis is put on discrimination of stimuli rather than on similarities between the representations of the stimulus in the ensemble response and the stimulus itself, it involves no implicit assumption about the unknown decoding mechanisms in the CNS. Examples of results obtained with analysis of and comparison of simultaneously recorded and pooled sequentially recorded populations of primary muscle spindle afferents are given. These results support some of the general arguments in favour of the ensemble coding theory, and indicate that the results of studies made on populations of simultaneously recorded receptor afferents may differ considerably from results obtained in studies on pooled sequentially recorded afferents. In contrast to when ensemble responses are based on parallel recordings, population responses constructed from sequential recordings will be distorted as a result of unavoidable temporal variations in physiological conditions.