Abstract
When two estimators capable of measuring the same environmental property are put into constant conflict, recalibration occurs and the apparent conflict is reduced. The prevailing wisdom in the adaptation literature is that vision is the ‘gold standard’; i.e. the non-visual estimator changes to match vision. However, recent theoretical models suggest that both estimators should adapt by amounts and at rates determined by their relative reliability. Related findings have been made in the cue-combination literature- i.e. ‘visual capture’ has been rejected in favor of a Bayesian probabilistic approach (Ernst & Banks, 2002; see Girshick, Burge, and Banks, VSS07). We examined visuo-haptic adaptation by manipulating the reliability of visually and haptically specified stimuli. The stimulus was a 3D hinge specified visually by random-dot stereograms and haptically by PHANToM force-feedback devices. The task was to indicate whether the hinge angle was greater than or less than 90 degrees. In a pre-adaptation phase we first measured uni-modal (vision-alone and haptic-alone) percepts. Then, in the adaptation phase we exposed subjects to a visuo-haptic conflict for an extended period. Finally, in a post-adaptation phase we measured uni-modal percepts again. Adaptation was assessed by comparing uni-modal percepts in the pre and post-adaptation phases. We found a role for relative reliability in adaptation, a result suggesting that traditional thinking should be re-examined.
AOF William C. Ezell Fellowship