Figure 1

APCoP (mm) plots from a 30 trial run for subject f50z031. Categorization of trials into CORRECT (a, b) and INCORRECT (e, f) for Interval 1 (a, e) and Interval 2 (b, f). Plots (c, d, g, h) give the point-by-point ensemble average. The scale of (c, d, g, h) also applies to the individual plots in the raster. The individual time series in each raster are sorted such that those with the highest acceleration value are at the top of each plot. The platform position (mm) signals from the highest and lowest acceleration values tested are shown in i and j. The horizontal dashed lines illustrate the experimentally determined detection threshold, while the vertical dashed lines are set at the end of the averaged position signal in Interval 1 and 2. Observe that the ensemble average of CORRECT versus INCORRECT runs shows very similar APCoP response to the stimulus, indicating that classifying based on correctness of response may be of limited value for data mining of the average signal.