Aspectos estruturais da variabilidade no processo de aprendizagem de uma tarefa de controle de força isométrica

Abstract

The aim of this study was to investigate the changes in the dynamics of motor output due practice in a isometric force production task. An experiment was conducted with 10 subjects to provide a data bank with the time series representing the subjects intrinsic dynamics before, immediately after and 24 h after practice. Along the analysis, we looked for (i) insights on the relationship and concepts of Motor Behavior, Dynamics Systems theory and aspects of Complex Systems and (ii) a better understanding of the theoretical fundaments, limits and pitfalls of usual analytical techniques applied in the study of physiological signals. First, an exploratory data analysis was performed in the context of a case study with the data of the participant V5. The results suggested: (i) non-stationarity of the time series; (ii) the coexistence of two different dynamics operating in different time scales (macro and micro-temporal time scales, with an estimated crossover around 1 s). The results suggested that the structure of isometric force is dominated in the temporal macroscale by a dynamics with strong periodic component, insofar the temporal microscale has characteristics of a stochastic process. Therefore, one proposed that any analysis that captures aspects of only one of these kind of dynamics shall be incomplete in the characterization of motor behavior. As a consequence of that premise, it's pertinent to apply complementary analytical techniques simultaneously, in order to capture signatures of both deterministic and stochastic dynamics. In parallel, it was presented evidence that recurrence quantification analysis (RQA) and detrended fluctuation analysis (DFA) are analytical techniques that satisfy that complementary requisite. Finally, a group analysis was performed with that analytical and conceptual paradigm stated previously. The performance of the participants improved in a relative permanent way due to practice, being better in the pos-test and in the 24 h retention test than in the pre-test. The dynamics of the temporal microscale (stochastic) changed in a transient way due practice, as estimated by the Hurst exponent (DFA), and we highlighted that it was not possible to infer changes in the temporal correlation due the lack of stationarity of the time series increments. Regarding the temporal macroscale, the results suggested that occurs a succession of dynamical regimes along the same trial. The complexity of that regime, inferred from the entropy S (RQA) and in a comparative analysis with DET% and LM (RQA), was affected in a transient way due practice: it was higher immediately after practice, and after 24 h returned to a similar value showed before practice. It's concluded that the structure of the isometric force has a dual character, being generated by (i) a stochastic process that dominates the dynamics in time scales under 1 s and (ii) a deterministic periodic dynamic (superposition of oscillators) dominating the dynamics in time scales above 1s. Regarding the complexity of that dual dynamics, it is affected only in a transient way due practice, as it returns to a basal value 24 h after practice. Both aspects suggested by the analysis, succession of dynamical regimes along the same trial and the apparent return of the dynamic's complexity to a "basal value", suggest the relevance to search for concepts and techniques in order to allow the deepening in these questions.