Shape perception by touch is not yet well understood. This is partly because it has been studied much less than vision. But there is also another reason. The information we receive about the features of tactual forms comes from a number of sources. That is not unique to touch. But in touch the relative influence of these sources differs with the size, depth and composition of the forms. That is important. The processes that underlie the recognition of characters in one tactual reading system may not apply to the recognition of the characters in a reading system that uses symbols with characteristics that provide a different balance of perceptual information. In reading by touch, therefore, the question how perceptual processes affect and are affected by the linguistic and cognitive aspects of reading is an integral part of the questions that need to be answered in order to understand language through touch.
I am proposing that tactual shape perception has to be understood as intersensory processing. As such, processing depends on the balance of complementary information from touch, posture and movement. These converge to produce the reference organization on which shape coding depends. The size, depth and composition of the shapes, as well as prior knowledge and task conditions, determine the balance of information from the complementary sources.
I shall first discuss the reasons for regarding shape perception by touch as a form of intersensory processing which differs for different types of shapes. It is argued that at least six categories of tactual configurations have to be distinguished in terms of the range and type of inputs they require for spatial coding. Some neuropsychological aspects of sensory acuity and movement processing are presented next. Another section deals briefly with three broad classes of spatial reference that can be involved in the perception of shape.
The remainder of the chapter focuses on the perception of single small raised dot patterns, and more particularly on braille characters. The hypothesis that the small size, lack of salient features, and the paucity of reference cues make shape coding difficult is tested against the traditional assumption that braille patterns are perceived initially and directly as