Changes in the structure of children’s isometric force variability with practice

Abstract

This study examined the effect of age and practice on the structure of children’s force variability to test the information processing hypothesis that a reduction of sensorimotor system noise accounts in large part for age-related reductions in perceptual-motor performance variability. In the study, 6-year-olds, 10-year-olds, and young adults practiced on 5 consecutive days (15 trials/day), maintaining for 15-s trials a constant level of force (5 or 25% of maximum voluntary contraction) against an object using a pinch grip (thumb and index finger). With increasing age, the amount of force error and variability decreased, but the sequential structure of variability increased in irregularity. With practice, children reduced the amount of variability by changing the structure of the force output so as to be more similar to that of their older counterparts. The findings provide further evidence that practice-driven changes in the structure of force output, rather than a decline in the amount of white noise, largely account for age-related reductions in the amount of force variability.

Introduction

Information processing perspectives on development have postulated that reductions in the variability of perceptual-motor performance that occur with increasing age throughout childhood are due, in large part, to reductions in the amount of noise in the sensorimotor system (Kail, 1997; Plude, Enns, & Brodeur, 1994; Prechtl, 1970; Salmoni, 1983; Yan, Thomas, Stelmach, & Thomas, 2000). This perspective is based on assumptions about noise and signal integrity that can be traced to information theory and the transmission of signals through communication systems (Shannon & Weaver, 1949).

This view of signal transmission considers the output from a system as consisting of the original intended signal and noise. The noise is assumed to have the structure of white Gaussian noise (Pierce, 1963) and to represent interference that alters the original signal during information transmission. Information processing perspectives of perceptual-motor performance assume that the mean performance reflects the intended motor output, or neural control signal, the amount of variability represents the amount of noise in the sensorimotor system, and the signal-to-noise (mean/standard deviation) ratio reflects the strength of the signal relative to the amount of noise in the system (Attneave, 1959; Fitts, 1951, Fitts, 1954; Siebert, 1978; Tustin, 1947). However, the standard distributional outcome measures of mean and standard deviation do not provide sufficient information for making inferences about the amount of noise in the sensorimotor system because white Gaussian noise by definition has a specific time and frequency structure (independence between successive values and equal representation of all frequencies) that these measures cannot identify. Therefore, the structure of the variability produced during the performance of a perceptual-motor skill must also be examined to assess the role of noise in human movement variability (Newell & Slifkin, 1998).

The structure of variability in behavioral output can be revealed by techniques in either the frequency domain or the time domain. In the frequency domain, Fourier spectral analysis allows a decomposition of a signal to determine the relative contribution of different frequencies to the structure of the aggregate signal (cf. Gottman, 1981). The degree of structure in the time domain can be assessed through determination of the regularity of a signal (Pincus, 1991). These techniques, as well as others, have been used extensively to reveal the structure of variability in biological systems (Bassingthwaighte, Liebovitch, & West, 1994), including in human motor behavior (Newell & Slifkin, 1998; Riley & Turvey, 2002; Slifkin & Newell, 1999) and cognitive science (Gilden, Thornton, & Mallon, 1995; Ward, 2002).

There have been some demonstrations that the familiar age-related changes in mean level of performance are accompanied by changes in the structure of children’s motor output variability. For example, during postural standing, the center of pressure profile has been shown to exhibit an increase in irregularity in the sequential structure, with age-related reductions in the amount of postural sway (Newell, 1998; Newell, Slobounov, Slobounova, & Molenaar, 1997). Also, the stride-to-stride variability during walking is lower, and its structure includes higher frequency components, in older children than in younger children (Hausdorff, Zemany, Peng, & Goldberger, 1999). However, Smits-Engelsman and van Galen (1997) found that the structure of children’s handwriting variability did not change over the course of 1 year but instead depended on the quality of their handwriting (which also did not change), with the variability of the poor-quality handwriting signal exhibiting a broader frequency profile than that evident in good-quality handwriting. This contrast in findings between the latter study and the former two studies suggests that the structure of variability might depend less on age than on the performance level and task.

In a variety of motor tasks, younger children are typically more variable and/or less accurate than adults in controlling their force (Blank et al., 1999, Blank et al., 2000; Forssberg, Eliasson, Kinoshita, Johansson, & Westling, 1991; Harbst, Lazarus, & Whitall, 2000; Konczak, Jansen-Osmann, & Kalveram, 2003; Lazarus, Whitall, & Franks, 1995). However, the variability of younger children’s force output holds considerably more deterministic structure that is dominated by slower frequency components (Deutsch and Newell, 2001, Deutsch and Newell, 2002, Deutsch and Newell, 2003; Smits-Engelsman, Westenberg, & Duysens, 2003). This finding contrasts with a white noise information processing account of age-related reductions in performance variability and instead indicates that, with increasing age through adulthood, the force control strategy becomes more adaptive to the requirements of the task (Newell, Broderick, Deutsch, & Slifkin, 2003). That is, broadening the range of contributing frequency components to the force output by including a greater proportion of higher frequencies tends to dampen the amplitudes of the force oscillations, leading to a reduction of error and variability to a constant level force target.

When visual feedback information of the force output is unavailable, age differences (from 6 years through adulthood) in the signal-to-noise ratio and structure of the force signal are minimal (Deutsch and Newell, 2001, Deutsch and Newell, 2002, Deutsch and Newell, 2003). This finding is important because it challenges the notion that sensorimotor system noise may be evaluated through measures of performance outcome (e.g., the signal-to-noise ratio) alone as well as the idea that reductions in error and variability in motor skill performance throughout childhood are due to a decline of motor system noise. That is, if these types of performance improvements throughout childhood were due to reductions in motor system noise, age differences in the structure of the force output signal should not be influenced by the availability of visual feedback.

When they are given appropriate instructions, information feedback, and/or sufficient practice, children of all ages show reductions in variability and error in their performance of motor skills, with the performance of older children sometimes approximating that of adults (cf. Keogh & Sugden, 1985). In addition, children increase the smoothness of their force trajectories (Ferrel-Chapus, Hay, Olivier, Bard, & Fleury, 2002; Lazarus et al., 1995; Thomas, Yan, & Stelmach, 2000), adapt to externally applied force fields and damping during arm movements (Konczak et al., 2003), and learn to modify their grip force in anticipation of an increase in load force during pulling (Serrien, Kaluzny, Wicki, & Wiesendanger, 1999). These performance improvements occur with very little practice and suggest that children learn to adapt the organization of the motor output to that required by the task (Engelhorn, 1988; Simard, 1969), leading to the postulation that practice and experience in organizing the motor output, rather than a reduction of sensorimotor system noise, largely accounts for children’s age-related performance improvements.

The current study examined the viability of this postulation by examining the effect of practice and age on the structure of children’s force variability. A change in the age-related structure of the variability as a function of practice would provide further evidence that variability is not a direct consequence of the stochastic random-like processes of white Gaussian noise, as assumed by the information processing accounts of motor variability. It was hypothesized that the structure of the force signal produced by the children would change significantly with practice to become more like that of older children or even like that of adults. It was also expected that practice would lead to a greater contribution from higher frequency components and enhanced irregularity in the sequential properties of the time domain. These changes would reflect the introduction of additional and more rapid control processes to the regulation of force (Newell et al., 2003).

In the experiment, 6-year-olds, 10-year-olds, and young adults of college age practiced an isometric force task with the thumb and index finger in a pinch grip at a force level of either 5 or 25% of their maximal voluntary contraction (MVC) on 5 consecutive days, with visual feedback and knowledge of results of performance available. In an isometric task, the participant exerts force against an immovable surface so that there is little or no movement of the limb where force is measured. In this situation, the length changes of the muscle are very small and there are no apparent changes to the external load on the system (Latash, 1993). In a pinch grip, the pads of the thumb and index finger are opposed in the production of force on an object (MacKenzie & Iberall, 1994). The 5 and 25% MVC force levels were chosen because the signal-to-noise ratio of the force output of children tended to be lowest and highest, respectively, at these levels in previous experiments based on limited practice (Deutsch and Newell, 2001, Deutsch and Newell, 2002).