In 1983, Jon Kaas published a paper titled, in part, “What, if anything, is S-I?” This paper was published at a time when there was a controversy raging about the functional organization of the primary somatosensory cortex (SI). SI had been mapped in a large number of species, as had a second somatosensory map that came to be known as the secondary somatosensory cortex (SII). The numerology denotes the order of the discovery of these fields, and was meant by the founding fathers to convey nothing with respect to whether the areas were organized hierarchically or in parallel. Traditional, and rather crude, recordings suggested that each of these representations contained complete and separate representations of the contralateral body surface, and SI and SII were thought to be homologous across species. Kaas asked the question in 1983 because electrophysiological mapping studies conducted in his laboratory suggested that the four cytoarchitectonic areas of the primate SI (i.e., areas 3a, 3b, 1, and 2) each contains a separate and complete map of the contralateral body. Kaas and colleagues showed that areas 3b and 1 contained complete and separate cutaneous representations of the contralateral body, and, further, that there were patterns of receptive field progression reversals at the border between these two areas that were comparable to the reversals that Allman and Kaas (Reference Allman and Kaas1971) had found at the borders of separate representations of the visual field in monkey extrastriate cortex. In addition, they suggested that areas 3a and 2 contained maps of deep (e.g., muscle spindle, joint) receptors. These findings and suggestions complicated the issue of homology of SI across species, and raised questions about information processing in anterior parietal cortex. Similarities in somatotopic organization, cytoarchitectural features, location relative to motor cortex, and connections with the ventroposterior nucleus of the thalamus led Kaas and colleagues to suggest that area 3b is the homologue of SI in most other mammals. If one accepts that proposal, the question remains as to how the remaining areas of anterior parietal cortex (3a, 1, and 2) are contributing to somesthetic sensibility.
Some years ago, Garraghty et al. (Reference Garraghty, Florence and Kaas1990) suggested that area 1 represents a “higher” stage of processing than area 3b because the receptive fields of neurons in area 1 are larger and have more complex response properties than those of area 3b. To test this idea, we recorded in the hand representation of area 1 before and after acute ablations of specific parts of the hand representations in areas 3a and 3b. We found that such ablations immediately deactivated the corresponding part of the hand representation in area 1, and we concluded that area 1 is located at a higher level in a hierarchically arranged processing sequence in anterior parietal cortex. We further noted that this hierarchical view of anterior parietal cortex was not novel, but rather was supported by a number of other observations. First, feed-forward corticocortical projections typically terminate predominantly in layer IV, and this is the pattern displayed by the projections from area 3b to area 1. Second, relay cells in the ventroposterior nucleus project in much larger numbers onto area 3b than onto area 1, and the plexus of thalamic input to area 1 is sparser than in area 3b and largely avoids layer IV. Third, stimuli are represented less isomorphically by neuronal firing patterns and have more complex receptive fields in area 1 than in area 3b, suggesting additional processing in area 1. Fourth, area 1 ablations are followed by less severe behavioral consequences than are area 3 ablations. Finally, in humans, the latencies of evoked potentials attributed to areas 3b and 1 differ by an amount consistent with sequential processing. Despite all of these observations, all too often it is the case that “anterior parietal cortex” and “SI” are used interchangeably as though nothing is transpiring in these four cytoarchitectonic regions (for references, see Garraghty et al. Reference Garraghty, Florence and Kaas1990). At least Dijkerman & de Haan (D&dH) devote two boxes to the anterior parietal cortex (APC) in their Figure 1.
The issue of whether sensory information is processed serially or in parallel has proven to be a thorny and complicated one. The complications arise primarily because there is no general consensus about what constitutes the appropriate level of analysis; and because neural interconnections between the putative parallel processing streams, and the temporal patterns of neural responsiveness of neurons at different levels of a putative serial processing stream, can introduce ambiguities. An additional complication is the inclination to view this issue rather myopically by concentrating only on data from “higher” nonhuman primates and humans. This largely eliminates evolutionary considerations – and, clearly, present-day sensory systems are evolved solutions to the problem of detecting and effectively responding to ecologically relevant stimuli. Adopting an evolutionary perspective casts the issue in a different light, and offers the opportunity to ponder the computational implications of somatosensory systems that differ fundamentally in their organization, and what the concepts of “serial” versus “parallel” processing actually mean.
In primates, SII is dependent upon inputs from the fields of anterior parietal cortex for its activation (Pons et al. Reference Pons, Garraghty, Friedman and Mishkin1987). If the hand representations in areas 3a, 3b, 1, and 2 are ablated, neurons in the corresponding hand representation in SII can no longer be activated. Such is not the case in all mammals. In cats, for example, ablation of SI does not deactivate the SII cortex. Indeed, in cats, no fewer than five somatosensory cortical fields (SI to SV) have been identified. Although each of these fields may well modulate processing in neighboring areas, no clear hierarchical arrangement exists for any of them. Interestingly, comparable differences exist between the visual cortices of cats and primates, with the lateral geniculate nucleus projecting almost exclusively to area 17 in primates, but to areas 17, 18, and 19 in cats. Thus, sensory processing in general in primates might rely more heavily on hierarchically arranged processing modules, whereas parallel processing strategies might be predominant in cats. While one might presume that some advantage has been conferred on primates by the emergence of serial processing capabilities, one can also note that cats, too, appear to “fit” quite well into their niches. Finally, while we can imagine that the somesthetic world of the higher nonhuman primate is much like our own, we cannot begin to imagine what the cat's solution to the problem would feel like. Perhaps, computational approaches attempting to model these two quite different strategies for processing somatosensory information might reveal how their outcomes differ, as surely these must.
