Is Movement Trajectory Generation the Key to the Paradox of Refractoriness Beyond Selection of a Response?

Yalchin Oytam & Peter D. Neilson

Neuroengineering Laboratory, School of Electrical Engineering & Telecommunications, University of New South Wales, Sydney, Australia


The pioneering works of Craik, Hick, Welford and more recently Pashler on double stimuli reaction time (RT) and tracking experiments indicate a central Response Selection bottleneck. RT experiments primarily deal with two of the three stages of motor production, Perceptual Processing and Response Selection, with Response Execution and therefore movement control receiving little attention. Generally, the concept of Response Selection consists only of working out the desired end point, and not the movement trajectory required to get there. However, it has always been noted, originally by Welford and later by Pashler that refractoriness may exist even when the second stimulus is presented immediately after the bottleneck.

Based on our tracking studies we propose a three-stage motor production model as part of the Adaptive Model Theory. This differs from the classical formulation in that the middle stage not only formulates the desired end point but also the movement trajectory for execution. We postulate that the duration of such a trajectory is finite, 100-150 msec. If, for some reason, the desired end point is not reached within this period, another trajectory is planned to move the response from its resulting location to the desired one, the end result being that the bottleneck is occupied for another cycle.

To test this theory an RT experiment was set up, where 10 subjects, using joysticks, made ungraded manual responses of duration less than 100 msec to two visual stimuli with random interstimulus intervals. This time duration is equivalent to a single push button response. During the experiment the gain (sensitivity) of the joysticks was changed from 1 to 2.5 resulting in graded movements of duration greater than 200 msec. Reaction times to S2 were measured and compared for 3 conditions before and after the gain change—when S2 arrives before response movement to S1 is initiated (i.e., during the bottleneck), during response movement to S1, and after it has ended. The reaction times after the gain change were 345, 342 and 290 msec respectively, whereas no increase in the second condition was recorded before the gain change. This result vindicates the proposed theory.