Figure 3. No-feedback trials before audiomotor testing show that subjects on average do not differentiate between targets, but nevertheless they show some consistency in their reaching pattern. This consistency is strengthened upon presentation of the first anchor trial and then remains constant.

A subset of individuals performed probe-only trials (without feedback) before and after audiomotor testing. a, Movement endpoints for the various probes. The horizontal axis indicates what the probe position would be given linear interpolation and the grey line indicates the identity line where subjects would reach if they linearly interpolated the probes. Prior to learning on average subjects move to the same location for all probes but after audiomotor testing they distinguish the probes in their movements. b, Cycle-by-cycle distance between pairs of movements. Each panel represents the comparison of movements on two cycles. The horizontal axis represents the probes of a given cycle n and the vertical axis represent the probes of cycle n+1. The colour in each cell codes the absolute distance between the pair of reaching movements to those probes. Diagonal elements indicate movements to the same probe, off diagonal elements movements to different probes. It is evident that even before audiomotor testing (pre), there is some structure to the participant’s reaching reflecting an internal consistency in the way sounds are allocated in space but this is strengthened after testing (post). c, Tracking over time the distances between pairs of movements to same or different probes shows that immediately upon exposure to the first anchors, the distance between pairs of movements to different probes increases, and then remains constant for the rest of the experiment in spite of repeated exposure to the anchors.