Dijkerman & de Haan (D&dH) synthesize a diverse literature in the development of their two-stream model of somatosensory neural processing. I have few concerns regarding the value or utility of their model, and I appreciate the synthesis. However, I raise questions that have long lurked in the literature in which investigations of perception by touch, blindness, and neuroscience overlap, and which may be pertinent in judging the comprehensiveness of D&dH's model.
D&dH state in their introduction that tactile exploration “informs us about the characteristics of external objects” (sect. 1, para. 1), which, while true, glosses over an important achievement: making an ontological transition from patterns of proximal stimulation to objects and environment, making a distal attribution. The basis by which perceivers come to attribute patterns of proximal stimulation to objects and events with which they are in contact, directly or indirectly, is fundamental to perception in all modalities. How it arises in the somatosensory system is important because (among other things) the sentient skin serves as the boundary between the self and the environment. Perhaps it could be argued that any contact at the skin surface, by definition, must be caused by a distal object. But, as I illustrate further on, this is not the case. Moreover, as D&dH explain, somatosensation has multiple roles: to inform of the body's internal states, as well as to inform of the external world; and to yield descriptions of the environmental objects, as well to contribute information for guided movements with them. If the somatosensory system reliably and accurately assigns causes to stimulation, an important issue is by virtue of what operations, in what neural complexes, these different functions take place.
What psychological or neural operations give rise to the perception that the proximal stimulation is caused by a distal object in the environment making contact with the skin? To cite an everyday example: A pencil held by the contact with five fingerpads of one hand produces five distinct patterns of stimulation but is perceived as an external, single object (Gibson Reference Gibson1962). Moreover, it can be perceived as an object of certain size, held at a certain location with respect to its length (e.g., see Carello & Turvey Reference Carello, Turvey and Heller2000). Does the model give clues as to how or where these important perceptual achievements occur? Might the model explain how unusual grips (such as crossed fingers) can give rise to the illusory perception of multiple objects when only one is present (e.g., Benedetti Reference Benedetti1985; Reference Benedetti1986)? It is not clear how these questions can be answered by the data marshalled by D&dH in developing their model.
It seems not to be the case that distal attributions arise because such stimulation can only have been caused by an external object literally contacting the skin. For example, the tactile-vision substitution system (TVSS) developed by Bach-y-Rita and his colleagues (e.g., Bach-y-Rita Reference Bach-y-Rita1972; White et al. Reference White, Saunders, Scadden, Bach-y-Rita and Collins1970) revealed that digital camera images converted into isomorphic vibrotactile patterns on the skin are perceived by congenitally blind perceivers qualitatively differently, depending on whether or not the perceiver has active control of the camera. Proximal, skin-based attributions (e.g., “sharp” and “ticklish”) predominate when the camera is still or passively moved, but distal attributions (objects perceived “out there”) are rapidly made once sensorimotor control of the camera is established. However, in these instances, the distal attributions may well have been based on the users' intellectual understanding of the system that they were operating: someone told them how it all worked. Would a naïve perceiver ever be able to make distal attributions without some independent knowledge of the cause of stimulation at the skin; that is, by being told how it works or by bringing other sensory modalities into play (vision or audition) (e.g., Epstein et al. Reference Epstein, Hughes, Schneider and Bach-y-Rita1986)? What conditions would be necessary for this to arise? Epstein et al. (Reference Epstein, Hughes, Schneider and Bach-y-Rita1986) suggested that although conscious awareness of the sensorimotor contingencies takes place and may be necessary for distal attribution, it does not appear be sufficient. Perceivers tend to understand that self-movements cause proximal changes in stimulation without necessarily attributing the proximal changes to distal objects. There appears to be some other factor(s) that are required for the transition to occur.
Systems such as the TVSS raise important questions about the nature and ontogenesis of perception in the somatosensory system of the blind. The philosophically inclined may ask whether such perception through the skin counts as “seeing” (e.g., Morgan Reference Morgan1977; O'Regan & Noë Reference O'Regan and Noë2001). Experimental psychologists seek the informational bases for the qualitative shift from proximal to distal attributions. In terms of the neuroscience of the somatosensory system, what neural structures would support any long-term changes in perception via systems such as the TVSS? It is conceivable that areas of cortex beyond those identified by D&dH as part of the somatosensory processing network are recruited for perceptual purposes in the blind. There is already considerable evidence that extensive areas of occipital cortex (e.g., Amedi et al. Reference Amedi, Raz, Pianka, Malach and Zohary2003; Sadato et al. Reference Sadato, Pascual-Leone, Grafman, Deiber, Ibañez and Hallett1998) and cerebellum (Gizewski et al. Reference Gizewski, Timman and Forsting2004) are active in Braille-related tasks by the blind. Hence, although D&dH show how useful neural lesion data are to our understanding, the blind (both congenital and adventitious) constitute a research population whose increased reliance on somatosensation indicates another crucial source of data for testing models of somatosensory processing.
