The effect of task constraints on performance distribution

[[abstract]]Abstract Graduate student: Tsung Yu Hsieh Advisor: Yeou-Teh Liu Motor control literatures often used the standard deviation to describe the variability of human movement outcomes (e.g., schimdt et al., 1979). This method gradually became popular in motor behavior research, but using standard deviation alone does not seem to be enough to describe the variability of human behavior because it is implicitly assumed that human behavior outcomes are distributed normally. Hancock and Newell (1985) suggested that we still have to consider skewness and kurtosis in addition to mean and standard deviation in order to describe distribution completely. Although there has been direct or indirect evidence suggesting human movement performance may not be distributed normally (Lai et al., 2005), a systematic examination of the distribution under different task constraints are still lacking.. So the purpose of the study is to examine the distributions of movement performance under systematic manipulation of task constraints. There 12 adults ranging in age from 25.167?2.23. There are 300 trials in each condition and there are 5 conditions, every participant has to finish one condition per day and it takes approximately 5 days to complete the experiment. All participants were follow the methods and procedure to finish the experiment. The results show that used repeated measure ANOVA on movement time in different conditions, conditions were significant F(1.892, 20.810)=24.318, p<.05, for spatial error, conditions were significant F(1.308,14.385)=12.226 p<.05, for performance score, conditions were F(2.303,25.330)=38.647, p<.05. Feedback function could guide participant to the specific area by predetermined weights of time and error. According to performance score, extreme conditions were more difficult, especially on accuracy condition. Entropy derived from the normal distribution was larger than the data entropy in all conditions when SD and bin size were controlled. The peak skewness value of the movement trajectory was systematically influenced by the task constraints. The conclusion was the feedback function can be applied on tasks with two movement goals to guide the movement performance and Information entropy provides a good way to examine the deviation of performance distribution under different task constraints