Visual and haptic recognition of objects: effects of transfer and viewpoint

We investigated the nature of object recognition in two sensory systems, namely the visual and haptic systems. Specifically, we investigated view dependent recognition performance in these systems. Our stimuli were 3D, unfamiliar objects, which were constructed from six, red Lego parts stacked randomly. The objects were presented to the subjects in a fixed orientation and position. Subjects learned four target object either visually or haptically. Subjects then performed an old/new recognition task either within the same modality or in the other modality as learning. Furthermore, the target objects were presented in either the same orientation as during the learning session or rotated by 180 deg.. In Experiment 1 the objects were rotated about the vertical axis and in Experiment 2 we tested rotations about the horizontal or depth axis. In all experiments we found an effect of view within both the visual and haptic systems: Objects were easier to recognize when there was no change in orientation from learning to test. That is, recognition performance dropped from 75 to 65. However, when recognition was performed across modalities then recognition was better if the object was rotated but only when the rotation involved an exchange of the front and back of the object. Our results suggest that orientation specificity is a common property of both the visual and haptic systems. Furthermore, the view of the object facing the observer is more salient for the visual system, whereas the side of the object, which is best accessible to the fingers and thus easier to be explored, is more salient for the haptic system. With the stimuli we used this was the back side of the objects. This suggests that integration of information across the fingers is equivalent to seeing an object from a particular view