Date of Award
Spring 2010
Document Type
Thesis
Degree Name
Master of Science (MS)
Department
Biomedical Engineering
First Advisor
Scheidt, Robert A.
Second Advisor
Schmit, Brian D.
Third Advisor
Winters,Jack M.
Abstract
Linear, limited-memory models capture many important features of adaptive motor performance during reaching, stepping and pointing. A recent study in our lab found that a model fitted to data obtained from subjects reaching against elastic loads which varied from trial-to-trial later failed to fit the steady-state response behavior of subjects exposed to deterministic, step changes in load. Does motor adaptation depend on statistical properties of the environment (eg. mean load strength and variability)? Neurologically intact volunteers (n=14) made 6 blocks of 100 planar, ballistic, 10cm, out-and-back reaching movements against spring-like loads having equilibrium positions at the hand's starting point. View of the limb was not allowed. Load stiffness varied trial-by-trial, and each block of movements differed in mean and/or variance such that three, 3-block contrasts were evaluated: increasing standard deviation (VAR), increasing mean (MEAN), and proportionally increasing standard deviation and mean (WEBER). In the VAR and MEAN contrasts, either the mean or the standard deviation of the load stiffness sequence was held constant while the other parameter varied systematically. In WEBER contrast, mean and standard deviation scaled proportionally over the contrast. The zero location of the transfer function moved toward the origin as variability increased. This trend in the zero location was the result of an unbalance in the decrease in the influence of previous load and the decrease of effective limb compliance with increasing variability. Specifically, the decrease in the influence of prior load was greater than the decrease in effective limb compliance. Effective limb compliance decreased to a larger extent in the MEAN and WEBER contrasts, which both presented an increase in mean load. In the MEAN contrast, the decrease in effective limb compliance with increasing mean load was balanced by an equivalent decrease in the influence of prior load, resulting in no significant change in the transfer function zero location. No changes in the influence of prior errors were observed in any of the contrasts. Thus, motor adaptation adjusts in two ways: the influence of prior load on subsequent movements decreases both when the environment is more variable and when effective limb compliance decreases with the mean load.