Abstract
We previously reported experiments detailing how human actors plan and execute movements in briefly presented scenes containing a green reward region and one or two partially overlapping red penalty regions (Trommershauser et al., 2003, JOSA, 20, 1419). Participants earned money by touching the reward region within 700 ms of presentation but lost money if they inadvertently touched a penalty region. The positions of the regions and their overlap were varied randomly from trial to trial. As we varied penalties and degree of overlap, we found that actors modified their movement strategies so as to maximize expected gain: they chose mean end points that correctly balanced the risk of hitting the penalty region(s) and the benefit of gaining the reward. We report new experiments using stimulus configurations with up to three overlapping regions that could have distinct monetary values: a reward region, a lesser penalty region, and a greater penalty region. Under these conditions optimal movement strategies corresponded to non-obvious shifts of mean movement end points from the center of the reward region, which in some cases were predicted to be inside the lesser penalty region. Four naive participants were paid to take part in the experiment. In each of the two experimental sessions, the number of penalty regions (from zero to two) and the spatial location of penalty region(s) were randomly varied across trials. Three out of four subjects shifted movement end points so as to maximize their expected gain and they did so in the more complex configurations as well as in the simpler. They chose mean end points near or within a penalty region when it was appropriate to do so. The results suggest that human movement planners are able to select optimal movement strategies as predicted by a complex two-dimensional maximization of expected gain.
NIH EY08266, HFSP RG0109/1999-B, DFG Emmy-Noether-Programm