Showing papers by "Herbert Ugrinowitsch published in 2015"

Different Faces of Variability in the Adaptive Process of Motor Skill Learning

Abstract: This study investigated the variability by considering an action programme as hierarchically organized, which reconciles invariant and variant features of motor skills at the macro- and microstructural level of analysis. It was assumed that invariant aspects of skilled actions express the macrostructure and therefore measures of sequencing, relative size, relative timing, relative force and relative pause time. The microstructure was related to the variant aspects so that total size, total movement time, total force, and total pause time were selected as its measures. These propositions were tested in an experimental design comprised by three learning phases: a stabilisation phase that entailed a given number of trials to achieve the functional stabilization on a graphic task, followed by transfer and retention phases. In the transfer phase, the graphic task was modified to yield different demands upon skill reorganization. Two such modifications demanded parametric changes (i.e. microstructure changes), in which graphic size and drawing speed were altered. Another modification demanded structural alterations (i.e. macrostructure change), in which drawing was changed. Overall, results supported the main predictions by showing that parametric changes in the task affected the microstructure, but did not affect the macrostructure consistently. Furthermore, a structural change affected both macro- and microstructure.

A new training algorithm using artificial neural networks to classify gender-specific dynamic gait patterns.

Abstract: The aim of this study was to present a new training algorithm using artificial neural networks called multi-objective least absolute shrinkage and selection operator (MOBJ-LASSO) applied to the classification of dynamic gait patterns. The movement pattern is identified by 20 characteristics from the three components of the ground reaction force which are used as input information for the neural networks in gender-specific gait classification. The classification performance between MOBJ-LASSO (97.4%) and multi-objective algorithm (MOBJ) (97.1%) is similar, but the MOBJ-LASSO algorithm achieved more improved results than the MOBJ because it is able to eliminate the inputs and automatically select the parameters of the neural network. Thus, it is an effective tool for data mining using neural networks. From 20 inputs used for training, MOBJ-LASSO selected the first and second peaks of the vertical force and the force peak in the antero-posterior direction as the variables that classify the gait patterns of the different genders.

Thin Bandwidth Knowledge of Results (KR) Improves Performance Consistency on Motor Skill Acquisition

Abstract: The bandwidth knowledge of results in one way of supplying feedback during practice, which provides control over the amount and type of information given to the subject. The present study investigated the best distribution of this information throughout the practice. A task requiring to perform a sequence of movements on the keypad, in order to achieve a specific the target time. The experiment consisted of three stages: a) acquisition (50 attempts); b) transfer, ten minutes after the end of the acquisition (10 attempts); c) retention, twenty-four hours after the end of the acquisition (10 attempts). Participants were distributed into four groups: wide bandwidth, with a range of 20%; narrow bandwidth, with a range of 5%; increasing bandwidth, with a range of 5% on the first half of the acquisition and 20% on the second half; and decreasing bandwidth, with a range of 20% on the first half and 5% on the second one. The results show thin bandwidth improves performance consistency.