Abstract
In the early 1960's Gibson and Caviness performed various experiments on visual and haptic shape discrimination — experiments whose data and detailed results were, unfortunately, never published. Recently, we have acquired and duplicated the original Gibson 'feelie' stimuli using 3D scanning and printing technologies. In these experiments we examine the visual and haptic perception of the feelies along with the well-studied, naturalistic stimuli (bell peppers) of Norman et al.
Method: The stimuli consisted of the 10 Gibson feelies and the original set of Norman's 12 bell peppers. The task was a simple same/different shape discrimination using pairs of objects selected randomly on each trial. There were 52 subjects; each judged 50 vision trials and 50 haptic trials. Modality was varied within-subjects while object type was varied between-subjects. For both modalities stimuli were presented sequentially and exploration was limited to three seconds per stimulus. Visual objects were presented via OpenGL depicted with motion, shading, and specular highlights. Haptic objects were explored behind an occluding curtain. For all presentations the objects had a randomly-determined orientation.
Results: In terms of discriminability, performance for the bell peppers was higher than for feelies (d’ of 2.62 vs 2.03, respectively). Judging the shape of the feelies was more difficult than for the bell peppers (F1, 50 = 39.7, p <.0001). With regards to modality there was no effect: haptics were equivalent to vision (d’ = 2.35 haptic vs. d’ = 2.3 vision, F1, 50 = 0.34, p = .56). Finally, there was no interaction — overall effect of object type was similar for both modalities.
Discussion: For these classes of stimuli there is apparently no effect of modality on performance but the type of object matters. This is likely due to the relative complexities of the stimuli which would be consistent with Phillips et al. previous findings.
Meeting abstract presented at VSS 2012