Fig. 4

Heat plots demonstrating interaction effects for proprioceptive errors made with the active arm at a various combinations of speeds and distances generated by the passive, robotically-moved arm. In each panel, the color and value of the heatmap indicates the group level average for the respective proprioceptive outcome measure. We observed a significant interaction effect for Path Length Ratio (PLR, C) for both control (F = 10.66, p = 0.003) and stroke participants (F = 18.85, p < 0.001), and near significance for Response Latency in control participants (F = 4.00, p = 0.054). These interaction effects, particularly for PLR, suggest that participants systematically overshoot the target at low speeds and short distance combinations and undershoot reference movements that have fast speeds and long distance combinations