Tracing a parallel to the two visual streams model, Dijkerman & de Haan (D&dH) propose a similar distinction in the somatosensory processing system (SP), whose anatomical and physiological organization would ultimately reflect the action-related use of somatosensory inputs. Their model is built upon multiple evidence showing that different body representations are used for perceptual judgments about the structural relationships among body parts (body image) and for the guidance of action (body schema).

Somatosensation is highly complex, including different sub-modalities besides discriminative touch, such as proprioception, the information about the relative position of our different body parts in space. It is thus important to describe the possible dissociations, as well as the multiple interconnections, between the different somatosensory streams, as the authors stress. However, one may question the extent to which, nowadays, a dissociation-based approach would help our understanding of SP. For vision, this approach proved to be successful initially, until the need to go beyond dissociations and transcend a dichotomic model for a more interactive one was acknowledged (Rossetti & Revonsuo Reference Rossetti and Revonsuo2000). The main argument for the dichotomic view was the so-called double dissociation between visual agnosia (patient D.F.'s deficit in visual recognition), and optic ataxia (e.g., patient A.T.'s deficit in visuo-manual guidance), considered as consecutive to ventral and dorsal damage, respectively. However, the empirical foundation of such a dissociation has been largely reconsidered on the basis of both old evidence (Pisella et al. Reference Pisella, Binkofski, Lasek, Toni and Rossetti2006; Rossetti et al. Reference Rossetti, Pisella and Vighetto2003) and recent evidence from the very same patient (Schenk Reference Schenk2006), making the ventral-dorsal dissociation of functions less obvious than previously thought (see also Coello et al. Reference Coello, Danckert, Blangero and Rossetti2007). By showing that not only visuo-motor processing, but also perceptual processing is impaired in optic ataxia, Rossetti et al. (Reference Rossetti, Revol, McIntosh, Pisella, Rode, Danckert, Tilikete, Dijkerman, Boisson, Vighetto, Michel and Milner2005) have questioned the perception-action dissociation classically described in optic ataxia. By reporting that D.F.'s performance was impaired on both perceptual and motor tasks in conditions tapping allocentric, but not egocentric, coordinates, Schenk showed that D.F.'s behavior may actually reflect dissociation between different modes of visuospatial processing.

Related evidence against purely encapsulated functional modules in vision comes from studies on normal subjects' grasping at illusory displays (Gentilucci et al. Reference Gentilucci, Chieffi, Deprati, Saetti and Toni1996), such as in the Titchener circles illusion. In this size-contrast illusion, two identically sized disks are perceived as being different in size, depending upon the size of the surrounding inducers. It was recently demonstrated that the supposed dorsal stream immunity to visual illusion cannot be advocated to explain the different effects found for perception and action in this and other illusions (Coello et al. Reference Coello, Danckert, Blangero and Rossetti2007). In addition, Pavani and colleagues showed that both perceptual estimation and hand shaping while grasping are similarly influenced by the illusion (Franz Reference Franz2001; Pavani et al. Reference Pavani, Boscagli, Benvenuti, Rabuffetti and Farnè1999), challenging the conventional visual perception/action model in the normal brain. So the question arises as to whether action would resist somatosensory illusion (regarding vision, see, Bruno Reference Bruno2001).

Take, as an example, the “fixed-set” method (Uznadze Reference Uznadze1966) whereby, for 30 consecutive trials, subjects have to grasp two spheres of different size (one in each hand) simultaneously and report which one is bigger. In the subsequent trials, when two identical spheres are presented to the subjects, an aftereffect emerges: Typically, the sphere held by the hand previously holding the bigger sphere is perceived as being smaller than the sphere in the opposite hand. How would subjects grasp the illusory smaller sphere? Would the proposed two-route model for touch predict that their kinematics reflect the veridical or the illusory somatosensory percept? Besides its empirical interest, the answer to this question might have important theoretical implications for D&dH's model, particularly when considering neuropsychological evidence showing that this method can disclose unconscious residual somatosensory processing of objects' size. A right-brain–damaged patient with left tactile extinction (patient G.R.; see Maravita Reference Maravita1997) was unable to perceive the left sphere, but was deceived by the perceptual aftereffect and reported the size of the right, ipsilesional, sphere incorrectly. As D&dH say, the notion of separation of functional modules has to be taken with caution, and we think this would be the case particularly for SP, which is composed of not only separable, but also functionally and anatomically interconnected submodalities, and integrated at multiple levels with other sensory modalities (Calvert et al. Reference Calvert, Spence and Stein2004).

Orthogonal to the main dissociation, the proposed model further distinguishes between somatosensory processing occurring in the internal (bodily) space and external (extrapersonal) space. In this respect, an aspect that the model seems to not address fully is the following question: What is the relationship between a touch on the body and the body position in space? In other words, the “tactile where” is not only “where-on-the-body” but also “where-on-the-body-in-the-space.” This issue is tightly linked to the multisensory view of somatosensory perception (Brozzoli et al. Reference Brozzoli, Dematte, Pavani, Frassinetti and Farnè2006), because body position information may also be provided by visual input (Rossetti et al. Reference Rossetti, Desmurget and Prablanc1995a). Previous work on patients with cross-modal visuo-tactile extinction (di Pellegrino et al. Reference di Pellegrino, Ladavas and Farnè1997) provided support for multisensory representations of a third spatial medium, the peri-personal space, linking the body to external space. In this respect, Farnè et al. (Reference Farnè, Pavani, Meneghello and Ladavas2000) found that extinction of tactile stimuli delivered on the contralesional hand is strongest when visual stimuli are presented closest to the ipsilesional hand. A comparable effect was obtained when the visual stimulation was actually presented far from the patients' ipsilesional hand (concealed behind their back), but near a rubber hand that was visible and aligned with the patients' ipsilesional shoulder. Not only visual, but also proprioceptive, information about the distance between hands (body schema) can alter static touch perception, the subjects being faster and more accurate when they perform tactile tasks while keeping their hands far apart (Brozzoli et al. Reference Brozzoli, Dematte, Pavani, Frassinetti and Farnè2006; Driver & Grossenbacher Reference Driver, Grossenbacher, Innui and McClelland1996).

In sum, we suggest the model might benefit from a more integrated approach to SP, and incorporating the multisensory peri-personal space representation, to account for the aforementioned phenomena – which are, in addition, tightly related to “acting” in somatosensory terms. Sensorimotor non-conscious maps of the body are plastic, as suggested by the behavioral effects of tool use in human and nonhuman primates (Berti & Frassinetti Reference Berti and Frassinetti2000; Farnè et al. Reference Farnè, Iriki and Ladavas2005; Iriki et al. Reference Iriki, Tanaka and Iwamura1996), and their adaptive functional role might be better captured within such a wider framework.