In external space, both somatosensory and visual systems perform the critical functions of object localization and recognition. In touch, the hand is a flexible exploring effector that combines motor capabilities with sensory capabilities to actively find objects in personal (on the body) and in peripersonal (within reaching distance) space (Lederman & Klatzky Reference Lederman and Klatzky1987; Previc Reference Previc1998). Once an object is located, the hand then executes a series of exploratory procedures to extract information about the material and geometric properties of the object (Klatzky et al. Reference Klatzky, Lederman and Reed1989; Lederman & Klatzky Reference Lederman and Klatzky1987). Thus, in haptics, action is inextricably part of perception. Most real, common objects can be recognized in less than 2 seconds (Klatzky et al. Reference Klatzky, Lederman and Metzger1985). Nonetheless, haptic perception is best suited to extract local, material properties from objects (Klatzky et al. Reference Klatzky, Lederman and Reed1987). Information regarding the complex contours and part relations of objects, especially for large objects that exceed the size of the hand, is more difficult to apprehend. Although touch physically interacts with portions of the outside world, it needs information about objects located in extrapersonal space, outside of reach.

In contrast, perception is separate from action in the visual system. Vision is well suited to obtain the information that touch lacks and vice versa. In vision, the spatial layout of objects in extrapersonal space is easily obtained, as are the spatial relations of one object to another and the spatial relations of object features. However, physical object properties need to be inferred. By itself, vision cannot perform actions in the world.

Although touch and vision can find objects in peripersonal and personal space and recognize them independently, they typically function together. The motor/somatosensory system uses spatial information from vision to help move the body through space in order to help bring objects in extrapersonal space into peripersonal space, in which the haptic system can be effective. This system also uses information regarding frames of reference and spatial relations to help construct representations of object-shape and object-part relations. In turn, the somatosensory system provides the visual system with the means to perform actions. The interaction of the systems expands the object localization and recognition capabilities of both. Thus, neuroanatomical divisions within the posterior parietal cortex (PPC) should aid this interaction by providing the lacking information to each.

First, consider known divisions of labors within the visual stream of processing. Creem and Proffitt (Reference Creem and Proffitt2001) proposed that visual information processing can be divided into “what,” “where,” and “how” pathways. The “what” pathway is involved in object identification and follows a ventral route from primary visual cortex to inferior temporal cortex. The “where” pathway is involved in the coordination of spatial maps and spatial frames of reference and follows a more dorsal route to the inferior PPC. Finally, the “how” pathway is involved in coordinating visual information with action on objects and follows the most dorsal route to the superior PPC.

Of interest is how these divisions of labor for vision within the PPC relate to divisions of labor within touch. For tactile object recognition, converging evidence from primate, patient, and neuroimaging literatures points to a somatosensory “what” processing stream: A modality-specific ventro-lateral stream moves from the sensorimotor cortex to the secondary somatosensory cortex (SII), the insula, and the inferior PPC before connecting with modality-general portions of the ventral object processing stream (Reed et al. Reference Reed, Shoham and Halgren2004). A patient with unilateral tactile agnosia (impaired tactile object recognition) had left hemisphere inferior parietal damage (Reed & Caselli Reference Reed and Caselli1994). Her object-recognition impairment did not influence other forms of spatial processing, but her specific errors indicated a problem in integrating the different parts of objects (Reed et al. Reference Reed, Caselli and Farah1996). For tactile object localization, data from fMRI studies suggest a somatosensory “where” processing stream that differentially involves bilateral superior PPC (Reed et al. Reference Reed, Klatzky and Halgren2005).

At first glance, the somatosensory “what” and “where” systems appear to be connected to the wrong parts of the visual system: The somatosensory “what” system appears most closely related neuroanatomically to the visual “where” system, and the somatosensory “where” system appears most closely related to the visual “how” system. However, when one considers what information each system needs most to perform specific functions, the divisions of labors may make sense. The somatosensory recognition system includes inferior PPC but requires information regarding spatial relations among objects and their parts. The “where” system for vision, also associated with the inferior PPC, contains this information. Thus, when somatosensory “what” meets visual “where,” more precise multimodal object recognition can occur. Similarly, the somatosensory “where” system, associated with superior PPC, contains information regarding sensorimotor information in relation to body space. The visual “how” system, also associated with the superior PPC, connects with visual properties for object interaction with sensorimotor information for more precise object-related actions. These processing streams in touch and vision allow for better transmission of information for multimodal functioning.

In summary, Dijkerman & de Haan's (D&dH's) model is underspecified with regard to the integration of the higher spatial information with the PPC and how the PPC might be functionally divided for the somatosensory system. These divisions have implications for how the somatosensory system interacts with the visual system for the important functions of object localization, recognition, and interaction. Nonetheless, the model has achieved what the authors intended – it has provided a model from which predictions can be made.