Cytochrome oxidase and functional coding in primate striate cortex: a hypothesis.

In 1978, Margaret Wong-Riley stained sections of squirrel monkey striate cortex for the activity of the mitochondrial enzyme, cytochrome oxidase, and noticed a periodic distribution of "puffs" of increased enzyme activity in layers 2 and 3 (letter to D. Hubel cited in Livingstone and Hubel [1984]). Her discovery anticipated a whole series of anatomical and physiological findings from many laboratories that correlated with the distribution of this enzyme in the striate cortex of primates, yet there has never been a satisfactory explanation as to why the distribution of this enzyme, crucial for aerobic energy metabolism, would be related to the functional organization of visual cortex (Martin 1988). The puffs have also been called blobs, spots, dots, and patches, with the term blob used most frequently. When the striate cortex is viewed from above, the blobs form a periodic array intercalated within a lattice of lower cytochrome oxidase concentration. We propose that the distinction between the blobs and the lattice is related to two different modes for representing stimulus variables. We submit that scalar variables related to the intensity of the stimulus are represented in the blobs. Intensity information is encoded explicitly over a very broad dynamic range, in which activity is proportional to the intensity variable (e.g., contrast). This encoding strategy requires that neurons have the energetic capacity to sustain a broad range of activity levels, which in turn is related to the high concentration of cytochrome oxidase. The situation is analogous to red muscle, which also is rich in cytochrome oxidase and which is able to maintain a sustained level of contraction over time (Needham 1971). We propose that in the surrounding lattice of lower cytochrome oxidase concentration, geometric variables are carried by neurons with orientation preferences and lead to different representational requirements. Each stimulus orientation is possible at every retinotopic location, and each is represented explicitly within an orientation hypercolumn. Activity varies with how well each individual orientation matches image structure at that location. However, there is rarely more than one orientation at any retinotopic location, so, on average, most oriented cells in each hypercolumn are quiet. The preference of these neurons for higher spatial frequencies (and possibly binocular disparities) further reduces the statistical probability that they will be active at any particular instant in time. The average level of neural activity over time is thus much less in the lattice than in the blobs, which is consistent with the lower levels of cytochrome oxidase in the lattice. There is an analogy with white muscle, which contains less cytochrome oxidase and typically has short bouts of rapid contraction interspersed with longer resting periods. We review the main empirical findings that have led us to this hypothesis. Tootell et al. (1988b) showed that the blobs were preferentially responsive to low spatial frequency gratings, whereas the lattice preferred higher spatial frequencies in experiments in which the functional activity of macaque monkey striate cortex was mapped with 2-deoxyglucose autoradiography. Silverman et al. (1989) found a strong negative correlation between cytochrome oxidase concentration and spatial frequency preference in electrophysiological recording experiments. Probably related to the spatial frequency organization of the blobs and the intervening lattice is the observation of Livingstone and Hubel (1984) that blob neurons tend to be much less sensitive to the orientation of elongated stimuli than are neurons in the lattice. Livingstone and Hubel (1984) also noted that many neurons in the blobs were preferentially activated by stimulus color, and they found "double-opponent" color cells in the blobs. Tootell et al. (1988a) found that the blobs were more activated by colored stimuli than by gray stimuli of equal luminance. However, there are comparative data which suggest that the blobs have important functions that transcend color vision. Galagos, lorises, and owl monkeys, which are nocturnal primates I and therefore live in dim lighting conditions in which color vision is virtually impossible, nevertheless have well-developed cytochrome oxidase