along multiple dimensions, including color, motion, spatial configuration, and perhaps cognitive components.
The natural images were digitized from magazines, including landscapes and housing interiors, and presented in 8-bit color. The images were presented as stationary scenes, for 2 sec/ presentation, in epochs 16 sec long, separated by epochs of uniform gray. Thus the stimulus timing and design was identical to that used for the moving gratings. Subjects were instructed to fixate the central point (as with the moving grating stimuli), but were given no additional instructions with regard to the naturalistic stimuli.
Fig. 5A shows the results of this type of ipsilateral stimulation, in comparison to ipsilateral stimulation with moving gratings in the same hemisphere (Fig. 5B). The activation produced by the naturalistic stimuli was located in similar regions of cortex (anterior to the areas showing contralateral retinotopy), compared with those activated by the moving gratings—thus our major conclusions about the location of ipsilateral activation were adequately supported.
However, finer details of the two activity patterns differ. In general, the activation produced by the naturalistic stimuli did not show the characteristic two-branched pattern produced by the moving gratings. Furthermore, the naturalistic activation extended more inferiorly in human cortex, further into the fusiform gyrus and other regions of the temporal lobe. Similar differences were obtained consistently in all subjects tested with these two stimuli. These results support our hypothesis that the correspondence between callosal and activity-driven maps actually may be slightly greater than was revealed in most of our tests, by using a single type of stimulus. Unfortunately, it is logically impossible to test the correspondence of the callosal maps to fMRI maps produced by all possible ipsilateral stimuli, and the available human callosal maps are likewise technically incomplete. Thus the degree of correspondence between ipsilateral visual activity and callosal maps mediating this activity remains unresolved.
The relative expansion of the ipsilateral activity into the temporal lobe (when produced by the naturalistic stimuli) is consistent with the idea (from macaques) that color and form are processed more in a temporal “stream,” whereas motion and spatial relations are processed more in a parietal “stream” (41–45). In human cortex, this idea has been supported in well-controlled experiments comparing attention to form vs. attention to spatial relations (46). The present experiments suggest that more direct tests of bilateral stimulus specificity for form/color vs. motion/spatial relations might also successfully differentiate temporal vs. parietal “streams.”
The results presented here, and previous results using other techniques, suggest the following generalities. Human visual information is processed first in the contralateral visual field, then gradually it crosses the vertical meridian as receptive fields become larger and extend into the ipsilateral visual field. Visual information is represented even more bilaterally in correspondingly more anterior areas, with much larger receptive fields and without demonstrable retinotopy. Converging fMRI evidence suggests that human area MT and the lateral occipital region have such bilaterally responsive, large, poorly retinotopic receptive fields.
The extent of ipsilateral influence can change abruptly at the border between cortical areas, as between human areas V3A/V4v vs. V3/VP (see Fig. 4). Thus these maps of the ipsilateral retinotopy may help to differentiate human cortical areas invisible by other means.
These results also indicate that psychophysical comparisons of stimuli in the two visual field must avoid the vertical meridian by significantly more than 40° (polar angle) for maximum independence. Complete interhemispheric independence may be impossible to achieve throughout visual cortex.
The ipsilateral visual representation is thus a highly organized system, topographically well integrated with other aspects of the human visual cortical organization. The communication across the interhemispheric “seam” in higher visual areas presumably is related to the construction of a unitary visual percept, uniting the two hemifield maps present in lower-tier areas. Though we focus here on this interhemispheric seam in visual cortex, a similar approach (using different stimuli) should make it possible to map the interhemispheric seam in additional cortical systems.
We thank Mary Foley, Terrance Campbell, William Kennedy, Bruce Rosen, and Thomas Brady for invaluable assistance during the course of this project. We are grateful to Martin Sereno for significant comments on a previous version of this manuscript. This work was supported by grants from the Human Frontiers Science Program and National Eye Institute to R.B.H.T., the Swiss Fonds National de la Recherche Scientifique to N.K.H., and the McDonnell-Pew Foundation to J.D.M.
1. Dow, B.M., Snyder, A.Z., Vautin, R.G. & Bauer, R. (1981) Exp. Brain Res. 44, 213–228.
2. Tootell, R.B.H., Siverman, M.S., Hamilton, S.L. & Switkes, E. (1988) J. Neurosci. 8, 1531–1568.
3. Zeki, S. (1970) Brain Res. 19, 63–75.
4. Van Essen, D.C. & Zeki, S.M. (1978) J. Physiol. (London) 277, 193–226.
5. Newsome, W.T. & Allman, J.M. (1980) J. Comp. Neurol. 194, 209–233.
6. Van Essen, D.C., Newsome, W.T. & Bixby, J.L. (1983) J. Neurosci. 2, 265–283.
7. Cusick, C.G., Gould, H.J. & Kaas, J.H. (1984) J. Comp. Neurol. 230, 311–336.
8. Beck, P.D. & Kaas, J.H. (1994) Brain Res. 651, 57–75.