Virtual blindness - a choice blindness experiment with a virtual experimenter

Abstract

How are people facing a virtual agent affected by the vividness and graphical fidelity of the agent and its environment? A choice blindness (CB) experiment – measuring detection rate of hidden manipulations – was conducted presenting a high versus low immersion virtual environment. The hypothesis was that the lower quality virtual environment (low immersion) would increase the detection rate for the CB manipulations. 38 subjects participated in the experiment and were randomized into two groups (high and low immersion). Both conditions presented a virtual agent conducting the CB experiment. During the experiment, 16 pairs of portraits were shown two at a time for the participants who were then asked to choose which portrait they found most attractive. For eight of the pairs, participants were asked to justify their choice while in four cases their choice had been secretly switched to the portrait they had not chosen. If a participant stated that the chosen portrait had been switched, it was annotated as a concurrent detection. The results revealed an increase in detection and earlier detection rate for the low immersion implementation compared to the high immersion implementation. Future research may involve experiments with higher degree of both immersion and presence, using for example head mounted display systems.