Abstract
Manual estimation (ME) is frequently used to assess human processing of visual size information. Since the required actions are very similar to grasping, ME and grasping are often compared. The main difference between ME and grasping is the goal of the action (indicating a size with two digits vs. picking up an object) and the feedback participants receive (no direct feedback vs. haptic feedback on whether a comfortable grip was achieved). Surprisingly, the influence of feedback on ME is not very well known. We investigated whether feedback affects accuracy and precision in ME at all. N=33 participants performed ME tasks with varying feedback about whether the goal of the action was achieved. In two within-subject conditions, participants either viewed reference objects and indicated their sizes using index finger and thumb (visual-input condition; object sizes: 20, 60, and 100 mm), or indicated a freely-chosen size (no-visual-input condition). In both conditions, participants then reproduced their original estimate (which we will call 'reference estimate'), without any further visual input. First, participants repeated their estimates without any feedback, followed by a block with automated verbal feedback reflecting the accuracy of the estimate, followed by another block without feedback. Our main dependent variables were the accuracy (i.e., mean difference to the reference estimate) and the precision (i.e., standard deviation) of the estimates. We found systematic biases both in the visual-input and in the no-visual-input conditions. Biases were correlated with the reference estimate, as was estimation precision. These biases were greatly reduced by the verbal feedback but reappeared when verbal feedback was removed again, while precision was the same in verbal feedback and no-verbal feedback blocks. We conclude that systematic biases in ME depend on the magnitude of the ME response. Direct feedback can alleviate these biases, but does not calibrate ME lastingly or improve precision
Meeting abstract presented at VSS 2016