Haptic perception of shapes and line drawings

Abstract

In this thesis various aspect of haptic perception were studied. The first part of the thesis is mainly concerned with haptic perception of two-dimensional shapes and line drawings. We first studied the angular acuity of two-dimensional shapes an found that the manner of exploration as well as the local and global stimulus properties influence angular acuity. Secondly we studied identification of line drawings by touch. We found that the size of the picture influences identifiability. We also found that observers seem to use a hypothesis driven strategy: on average 23% of the total exploration time was spend on confirming the final hypothesis. In the next chapter on line drawing identification we report a finding that helped to explain why identifying a line drawing by touch is such a difficult task. We found that if observers were not able to identify a picture and were given the opportunity to sketch what they had just felled, in 30% of the cases they could identify their own sketch. A line drawing is easily processed with vision, but if the input is made sequential instead of simultaneous, identification becomes very difficult. This is because the structure of the input has changed and cannot be used to match the internal representations. Similar to sequential vision, if a line drawing is explored by touch, then the structure of the percept is what could be called `one-dimensional'; that is, a sequential description. Observers experience difficulty in mentally switching between these two structures. What can be done is restructuring the representation from sequential to simultaneous by producing a sketch. This explains the recognition-after-sketching-effect. In the second part of the thesis we aspect of haptic perception of three-dimensional curvature. First we studied real, solid shapes and virtual shapes generated by a robotic interface. One of the purposes was to study the contribution of two isolated geometric cues. We found that the surface orientation is a much more dominant cue than the position. Besides measuring the discrimination thresholds, we also measured whether a virtual stimulus would feel as curved as a real stimulus. We did not find a systematic difference and even found that it did not matter whether or not zeroth order information was available when observers compared the curvedness of real and virtual stimuli. In the last chapter, we found that if a curved shape is touched with one finger, it will induce an illusory curvature on the other finger. Interestingly, this phenomenon was present for both raised lines and solid shapes. In a subsequent experiment we wanted to investigate whether the biomechanical constraints of the hand were responsible for the effect, but we did not find evidence for this hypothesis. Whereas in a previous study we found that the surface attitude is the important cue for curvature, we propose that the curvature contrast illusion is mainly caused by a shift of dominance from the attitude cue to the positional cue.