sign in sign up
<<
Home , Cerebral cortex, Cerebrum, Cuneus, Superior temporal sulcus, Visual cortex

Coinciding early activation of the primary visual and anteromedial

Simo Vanni*, Topi Tanskanen†, Mika Seppa¨ , Kimmo Uutela, and Riitta Hari

Brain Research Unit, Low Temperature Laboratory, Helsinki University of Technology, P.O. Box 2200, FIN-02015 HUT, Espoo, Finland

Communicated by Olli V. Lounasmaa, Helsinki University of Technology, Espoo, Finland, December 18, 2000 (received for review October 17, 2000) Proper understanding of processes underlying visual stimuli were presented to the four quadrants (4–12° requires information on the activation order of distinct areas. from fixation, the peripheral stimuli covered a 45° sector, and We measured dynamics of cortical signals with magnetoencepha- were symmetrically off the horizontal and vertical meridians; the lography while human subjects viewed stimuli at four visual check size of the pattern stimuli increased from 0.8 to 1.9° with quadrants. The signals were analyzed with minimum current eccentricity). The luminance comprised of increase of estimates at the individual and group level. Activation emerged luminance from the gray background (30 cd͞m2) to the maxi- 55–70 ms after stimulus onset both in the primary posterior visual mum luminance (61 cd͞m2) for duration of 470 ms. The three areas and in the anteromedial part of the cuneus. Other cortical stimulus types covered the same visual field locations. The areas were active after this initial dual activation. Comparison of individual equiluminant values for the RG stimulus were deter- data between species suggests that the anteromedial cuneus mined with flicker photometry; the red was invariable but the either comprises a homologue of the monkey area V6 or is an area green was varied. The upper and lower visual field stimuli were unique to . Our results show that visual stimuli activate delivered in separate sessions. The left and right quadrants were two cortical areas right from the beginning of the cortical response. stimulated in random order within the same session, with The anteromedial cuneus has the temporal position needed to random 1–1.5 s interstimulus intervals. During the pattern interact with the primary V1 and thereby to modify reversal sessions, the pattern was continuously on, displaying the information transferred via V1 to extrastriate cortices. two upper or the two lower quadrant patterns, and the stimu- lation comprised of reversals of the contrast in one quadrant at natomy of connections between visual cortices a . Asuggests a hierarchical organization of signal processing (1). However, the order of processes at different functional areas Signal Recording and Analysis. Signals were recorded with a cannot be directly deduced from the anatomical hierarchy 306-channel Vectorview neuromagnetometer (Neuromag Ltd., without relevant timing information. In monkeys, several areas Helsinki, Finland), filtered (passband 0.1–200 Hz), sampled at become active immediately after the primary visual cortex (V1), 600 Hz, and averaged time locked to the pattern reversal or to while another set of areas is active at clearly longer latencies (2, the onset of the luminance stimulus. Vertical and horizontal 3). The areas showing early responses, such as V5͞middle electro-oculograms were recorded and epochs contaminated temporal area (MT), medial superior temporal area, and frontal with movements or blinks were rejected online. eye field, are specialized in analyzing dynamical visual informa- Environmental noise was attenuated in the measured signals tion and in visuomotor transformation (for reviews, see refs. 4 by extracting signal-space vectors with strongest noise fields. The and 5). The areas with longer latencies, such as V4, are sensitive signals were then preprocessed by omitting noisy channels, to object form and and participate in object recognition (6, applying a 200-ms prestimulus baseline and a 40-Hz low-pass 7). The dissimilarities in activation latencies and functional filter. The analysis applied MCE (9), an implementation of the properties of these areas suggest diversity in the type and amount minimum L1-norm estimate (17); MCE explains the measured of necessary input and preprocessing before activation. Given signals with a current distribution that has the smallest sum of the differences between monkey and human visual cortices (for current amplitudes. Estimates of individual signals were calcu- a review, see ref. 8), we aimed to explore the dynamics and lated separately for each time point. For the group average, the distribution of early cortical activation in humans. Neuromag- individual estimates were aligned on a standard brain (18) and netic signals were recorded while the subjects viewed pattern then averaged. The alignment phase applies first a 12-parameter reversal or luminance stimuli at the four visual quadrants. The affine transformation (11), followed by a refinement with an signals were analyzed with minimum current estimates (MCEs), elastic nonlinear transformation (10). The match is based on the which require minimal human intervention for determining the comparison of gray-scale values of the individual and the stan- location and orientation of the cortical currents (9). The indi- dard brain MR images. As a result, major sulci and other vidual three-dimensional estimates were aligned with a nonlin- important brain structures are aligned. ear transformation according to individual brain shapes (10, 11), For more detailed analysis, regions of interest (ROIs) were and both the individual and group average MCEs were com- selected to cover steadily active areas. The center and extent of pared with the existing maps of human visual cortices (12–16). each ROI was automatically adjusted to fit the estimated activity. The activity within a ROI was calculated as a weighted average Materials and Methods Subjects and Stimuli. We studied five female and five male subjects (mean age 27 years, range 20–42 years). The stimuli were Abbreviations: MT, middle temporal area; MCE, minimum current estimate; ROI, region of generated with a Macintosh computer and presented with a interest; BW, black and white; RG, red and green; V1, primary visual cortex; V6, visual dataprojector (VistaPro, Electrohome Ltd., Ontario, Canada) cortical area V6. *Present address: Centre de Recherche Cerveau & Cognition, UMR5549, Centre National de on a back projection screen, with viewing distance of 84 cm. The la Recherche Scientifique–Universite Paul Sabatier, Faculte de Medecine de Rangueil-133, ͞ 2 subjects fixated a small cross on a gray (30 cd m ) background. route de Narbonne, 31062 Toulouse Cedex, France. E-mail: [email protected]. ͞ 2 ͞ 2 Black (4 cd m ) and white (BW; 61 cd m ) or red and green †To whom the reprint requests should be addressed. (RG; red luminance 29.5, x ϭ 0.539, y ϭ 0.408 in CIE space, ϭ ϭ The publication costs of this article were defrayed in part by page charge payment. This green mean values luminance 31.4, x 0.331 and y 0.559) article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. checkerboard patterns with reversing contrasts or luminance §1734 solely to indicate this fact.

2776–2780 ͉ PNAS ͉ February 27, 2001 ͉ vol. 98 ͉ no. 5 www.pnas.org͞cgi͞doi͞10.1073͞pnas.041600898 Downloaded by guest on September 29, 2021 Fig. 1. Data of Subject 1 after right upper quadrant BW pattern reversal. Fig. 2. The mean location, its SD, and current orientation (blue dot, circle, Magnetic field pattern and the corresponding MCE results, viewed from the and line, respectively) of the cuneus and inferior occipital sources are dis- back of the head, are integrated from 50 to 70 ms. The red contours in the played in a standard brain (18). The red dot displays the mean location of left magnetic field display indicate the magnetic flux coming out from the head and right V1 from Hasnain et al. (13), who reported, however, the average and the blue contours the magnetic flux entering the head. The blue dots and location of the upper and lower visual field representations. lines on the individual MRIs represent the center and orientation of the estimated current. The red and black curves show the strengths of the an- teromedial cuneus and left occipital (Occ.) sources as a function of time. parieto-occipital , and the other in the more inferior left occipital . The rising edge of the response is 3–4 ms (about two samples) earlier at the left occipital cortex than at the of the estimate: the maximum weight was in the center of the cuneus, suggesting that these sources reflect two separate brain ROI and the weight extended to the neighboring locations with regions. the form of a three-dimensional generalized normal distribution. Fig. 2 presents the mean locations across subjects for the two The reported radius of ROI is the distance at which the weighting sources activated by BW pattern reversals. The source at or close factor is 60% of that at the center. A ROI was accepted for the anteromedial cuneus was found in 9 and 5 individuals after ⅐ further study if (i) the peak source activity exceeded 2 nA m and the left and right upper quadrant stimuli, respectively, but only was at least twice the peak activity during prestimulus baseline; in 5 and 3 subjects after the left and right lower quadrant stimuli. (ii) the activation duration exceeded 10 ms; and (iii) the location The inferior occipital source was found in 8 and 7 subjects after was stable for over 10 ms. An active brain area could remain the left and right upper quadrant stimuli and in 8 and 10 subjects undetected in our recordings and analysis because of a too short after the left and right lower quadrant stimuli, respectively. At distance between two active areas, because of cancellation of the best (after left upper quadrant stimuli), the simultaneous two magnetic signal attributable to source current configuration, or sources were found in 7 subjects. The absence of the cuneus because of masking of a deeper brain activity by simultaneous source in 5–7 subjects after lower visual field stimuli could reflect strong activation close to the sensors. Only BW pattern reversal the shorter intersource distance after lower than upper field data were used for source localization in individual subjects. stimuli. In subjects who showed two sources, the intersource distances were on average only 25–26 mm for lower field stimuli Determination of Cortical Areas. The sources were first located but 32–40 mm for upper field stimuli. If the distance had been from group average data to find common features of activation, smaller in other source pairs, the estimation method might have and then the mean coordinates across subjects were compared faced difficulties in differentiating the two sources. with earlier imaging and histology studies (12–16) for determi- The location of the inferior occipital source was more stable nation of the functional areas. Because our coordinate space across individuals after lower than upper visual field stimuli (Fig. corresponded to the standard brain coordinates of the European 2). After lower visual field stimuli, the mean location of the ‡ Computerized Database, the Talairach coordi- inferior source corresponded well to the human V1 in the medial nates (20) in the earlier studies had to be transformed to the (13), with the current directed upward. Assuming same coordinate system for comparison and display purposes. In that this activation occurred in the upper lip of the calcarine NEUROBIOLOGY addition, our source coordinates were transformed into Ta- , the current direction would correspond to cortical cur- lairach space; the rightward shift of the midline at the posterior rent flow from the surface to the depth of the cortex. After upper parts of (14) was taken into account by aligning the visual field stimuli, the inferior source was lateral to the midline. midsagittal plane, in addition to anterior and posterior commi- Although the inferior occipital response could receive additional sures, with the hemispheric midline at the . contribution from other visual areas, similar latencies for the Results and Discussion upper and lower visual field responses and the group mean location close to calcarine fissure suggest considerable V1 Fig. 1 presents the estimated brain activity in Subject 1 after right contribution to the signal. upper visual field BW pattern reversal stimulation. The magnetic Location of the anteromedial cuneal source in Talairach space field pattern between 50 and 70 ms cannot be explained with only (ref. 20; x, y, and z coordinates 2, Ϫ79, and 23 mm after the left one active area. Accordingly, the MCE display shows two and Ϫ2, Ϫ77, and 26 mm after the right visual field stimulation; maxima, one located in the anteromedial cuneus, abutting the average for the upper and lower quadrant epochs) falls outside many positron emission tomography (13) and functional MRI

‡Roland, P. E., Fredriksson, J., Svensson, P., Amunts, K., Cavada, C., Hari, R., Cowey, A., (12, 15, 16) studies mapping the human occipital areas. These Crivello, F., Geyer, S., Kostopoulos, G., et al., 5th International Conference on Functional maps do not report areas in the anteromedial cuneus. However, Mapping of the Human Brain, June 22–26, 1999, Du¨sseldorf, Neuroimage 9, S128 (abstr.). Corbetta et al. (21) reported deactivation in functional MRI in

Vanni et al. PNAS ͉ February 27, 2001 ͉ vol. 98 ͉ no. 5 ͉ 2777 Downloaded by guest on September 29, 2021 similar regions (their figure 2 and table 1; 21 and 8 mm mean distance from our left and right cuneus sources) when covert shifts of and saccadic eye movements were compared. Thus, although our subjects had no task during stimulation, our activation in the medial cuneus could reflect automatic activa- tion of a part of the attention- or visuomotor-related networks. Although our cuneus source showed both lower and upper visual field representations similarly as the human area V3a does (15), the source location was 1–2 cm too medial for V3a, regardless of the quadrant stimulated (refs. 13 and 15; Fig. 2). Neither can the cuneus source be at V3, because the human V3 does not have upper visual field representation above the (15), and because our source was 2–3 cm too medial to the known location of the left and right human V3 (13). In the monkey V6 (22), the representations of the upper and lower visual fields are clearly distinct, and in part even in different walls of the parieto-occipital sulcus, with a trend of more anterior locations for upper than lower visual field repre- sentations. Interestingly, the mean current directions at the cuneus source were opposite during the upper and lower visual field responses (Fig. 2), suggesting both that occurs at the source area and that the responses could arise from opposite walls of a sulcus or a . In the stereotaxic isocontours displayed on the Visible Man atlas (8), the coordinates of the anteromedial cuneus source are located at the posterior bank of the parieto-occipital sulcus, which corresponds to the macaque area V6 (also known as PO) in the comparison map between monkey and man (23). In addition, based on histological evidence, this source area could § Fig. 3. (a) Group average MCEs after left upper quadrant pattern reversals, be a homologue of monkey area V6. The backward shift of integrated over successive 10-ms intervals after stimulus. (b) Spherical smooth- histologically defined area 19 during cortical phylogenesis (for edged regions of interests with 8-mm radius are placed close the higher and reviews, see refs. 14 and 25) should have pushed the functional lower current clusters and between. The ROIs are presented on the standard area V6 from the medial fundus of the parieto-occipital sulcus brain MRI (18). The right column shows the signal strength of these ROIs as a in monkeys (22) to the upper posterior wall of the parieto- function of time. occipital sulcus and thus to the anteromedial cuneus in humans. The of V6 have large receptive fields (22), and lack enhanced foveal magnification both in monkeys (22) and hu- 100–200 ms the mean activation at the posterior brain was about mans (26), and the area has been suggested to participate in four stronger in the right than the left hemisphere (vol. 310 3 visuospatial analysis of environment for arm reaching and eye cm ). Whereas between 110 and 150 ms multiple structures movements (27, 28). The human anteromedial cuneus could also contributed to the signal, later only few areas were prominently contain an area or multiple areas with no equivalent in the active at the right hemisphere. A movie of this activation is macaque brain. The larger amount of neurons and longer published on the PNAS web site as supplementary material distances between brain areas in humans than in monkeys (www.pnas.org). Fig. 4a depicts the strongest activations be- tween 50 and 400 ms after left upper and lower quadrant BW implicate different transmission latencies between areas and Ј thus different requirements for cortical organization in the two stimuli. The posterior areas (a, a , and b) were active clearly species when rapid processing is of essence. earlier than the temporo-occipital (c) and superior temporal (d) Fig. 3 presents the dynamics of processing in the two areas. regions. The temporo-occipital source region was about 1 cm anterior and inferior to human MT͞V5 (x, y, and z Talairach The group average MCE to left upper quadrant stimuli showed Ϫ Ϫ the two maxima during the earliest response from 60 ms onward coordinates for the ROI c are 42, 67, and 4 mm, respectively; (Fig. 3a). The lower source fades out before 80 ms, whereas the data have been compared with a summary of V5 coordinates in ref. 32). However, because of its size, the ROI c should include upper source continues to be active up to about 100 ms. ͞ Attenuation of the MCE strength at sites located between the activation of the human V5 MT. This area has been shown to be two maxima (middle trace in Fig. 3b) further supports discrete active in monkeys almost immediately after V1 (2) and suggested clusters during the initial activation; these clusters also display to be active even before V1 in humans (33). However, the 170- clearly different temporal behaviors. After upper quadrant BW to 220-ms latencies in our study are in line with earlier magne- stimuli, the occipital source became active on average at 56 toencephalographic data showing V5 activation at 160 ms for motion stimuli (32, 34). It is possible that our stimuli did not (mean half-amplitude latency from individual data), and the ͞ cuneus source at 58 ms; the latency difference between the two activate V5 MT, but the signals with longer latencies at the ROI sites was not statistically significant. This timing information c come from other functional areas. After left upper BW stimuli, leaves the pathway of retinal information to anteromedial cuneus the signals of the right superior temporal region increased unclear. Signals could either go through V1 and be rapidly fed slightly simultaneously with the signals of the posterior regions, forward (2, 29), or bypass V1 (30, 31). but only in a few subjects. Of course, our data alone cannot After 90 ms, the activation spread to other areas, predomi- exclude the possibility of other early responses, which might have nantly in the contralateral hemisphere; in the time window of been detected if more area-specific stimuli would have been applied or a much larger number of responses would have been averaged to improve the signal-to-noise ratio. §Zeki, S. Proceedings of The Physiological Society, July 25–26, 1986, Cambridge, J. Physiol. The initial dual activation of the occipital cortex and of the (London) 381, 62P (abstr.). anteromedial cuneus was best distinguished after reversals of

2778 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.041600898 Vanni et al. Downloaded by guest on September 29, 2021 nocortical signal at equiluminance (36). At longer latencies, the cortical dynamics was different for each stimulus. The novel finding of early activation in the human anterome- dial cuneus is interesting in the light of earlier monkey data. Activation latencies in monkey V6 have not been reported. Whereas in monkeys the V5͞MT and respond to stationary stimuli shortly after V1 (2, 3), we found prominent activation considerably later in the human V5͞MT and superior temporal sulcus areas, suggesting differences in the processing dynamics between humans and monkeys. In monkeys, the first responses at V1 occur around 31–34 ms (2, 29) to luminance stimuli and the first responses to pattern reversal stimuli peak in lamina 4C of V1 around 50 ms after the stimulus, corresponding most likely to the 70-ms in human scalp recordings (37). However, the cortical activation latencies depend on stimulus characteristics, such as contrast and luminance, hindering direct comparison across studies. Neuro- magnetic signals at the human parieto-occipital sulcus region have been detected earlier after luminance stimuli (26, 38), eyeblinks (39), and (40). The responses in those studies reached their maxima between 130 and 285 ms after the stimuli, and thus showed considerably longer latencies than what we found at the anteromedial occipital lobe. Apparently, our first 56-ms response can be considered early, and it could even correspond to disynaptic activation immediately after the first afferent signals coming through the Fig. 4. (a) Group average MCE after left upper and left lower quadrant magnocellular pathway (29). Furthermore, a relatively early pattern reversals, integrated from 50 to 400 ms after stimulus. Note activated higher-order area could interact with V1 and thereby that averaging the signal from 50 to 400 ms makes the earliest transient modulate the processing of the later arriving parvocellular activation poorly visible. The four distinct ROIs are in the posterior inferior (main) volley of afferent information (41), thus affecting the occipital cortex (a and aЈ), in the anteromedial cuneus (b), in the tempo- quality or quantity of visual information reaching later pro- rooccipital junction (c), and in the superior temporal sulcus (d). After left cessing stages. lower quadrant stimuli, the group average MCE separate poorly the cuneus According to our findings, an area at the human anteromedial and occipital sources because the two areas are close together and there is cuneus has the temporal position needed to interact with the some interindividual variance in locations—the two distinct sources were better separable in individual data. (b) Group average amplitude as a lowest levels of cortical hierarchy before processing proceeds function of time for BW pattern reversal, luminance onset, and equilumi- more widely to other visually responsive areas. In monkeys, the nant RG pattern reversal stimuli at the left upper quadrant. The dashed first responses in V1 specific to figure boundary arise with about vertical lines indicate latencies at 67 and 97 ms. The amplitude curves 60–70 ms latency, about 30 ms after the first response (19). In reflect group average activity at the mean locations of the individual our data, the initial dual activation lasted less than 30 ms. Thus sources, determined from BW data. these two areas would have time to interact before the edges of a figure are detected in V1. The methods and results of this study open one window for exploring the hypothesis of stimulus-driven BW checkerboard patterns, but it was also present after the early top-down of V1 processing (41, 24) in human luminance onsets and reversals of equiluminant RG patterns. subjects. Fig. 4b displays activation of the two sources for the three stimuli; the source locations and current orientations were determined We thank Jean Bullier for valuable comments on the manuscript and from the individual BW pattern reversal data. These two areas Jaakko Ja¨rvinen for help in the data analysis. The MRIs were obtained were always simultaneously active, but responses of both areas at the Department of Radiology, Helsinki University Central Hospital. were about 30 ms delayed for the equiluminant stimuli. This The source coordinate transformations are based on work conducted within European Computerized Brain Database project, EC 4th Frame- delay may correspond to slower conduction of signals through work program͞Biotech 2, No EN D 100413. This study has been the wavelength-sensitive parvocellular than the broad-band supported by Academy of Finland, Ministry of Education, and Sigrid magnocellular retinocortical pathway (35) or to a weaker reti- Juse´lius Foundation. NEUROBIOLOGY

1. Felleman, D. J. & Van Essen, D. C. (1991) Cereb. Cortex 1, 1–47. 12. Sereno, M. I., Dale, A. M., Reppas, J. B., Kwong, K. K., Belliveau, J. W., Brady, 2. Schmolesky, M. T., Wang, Y., Hanes, D., Thompson, K. G., Leutgeb, S., Schall, T. J., Rosen, B. R. & Tootell, R. B. H. (1995) Science 268, 889–893. J. D. & Leventhal, A. G. (1998) J. Neurophysiol. 79, 3272–3278. 13. Hasnain, M. K., Fox, P. T. & Woldorff, M. G. (1998) Hum. Brain Mapp. 6, 3. Schroeder, C. E., Mehta, A. D. & Givre, S. J. (1998) Cereb. Cortex 8, 575–592. 301–315. 4. Orban., A. G. (1997) in , eds. Rockland, K. S., Kaas, J. H. & 14. Amunts, K., Malikovic, A., Mohlberg, H., Schormann, T. & Zilles, K. (2000) Peters, A. (Plenum, London), Vol. 12, pp. 359–434. NeuroImage 11, 66–84. 5. Schall, J. D. (1997) in Cerebral Cortex, eds. Rockland, K. S., Kaas, J. H. & 15. Tootell, R. B., Mendola, J. D., Hadjikhani, N. K., Ledden, P. J., Liu, A. K., Peters, A. (Plenum, London), Vol. 12, pp. 527–638. Reppas, J. B., Sereno, M. I. & Dale, A. M. (1997) J. Neurosci. 17, 7060–7078. 6. Zeki, S. (1980) (London) 284, 412–419. 16. DeYoe, E. A., Carman, G. J., Bandettini, P., Glickman, S., Wieser, J., Cox, R., 7. Desimone, R. & Schein, S. J. (1987) J. Neurophysiol. 57, 835–868. Miller, D. & Neitz, J. (1996) Proc. Natl. Acad. Sci. USA 93, 2382–2386. 8. Van. Essen, D. C. & Drury, H. A. (1997) J. Neurosci. 17, 7079–7102. 17. Matsuura, K. & Okabe, U. (1995) IEEE Trans. Biomed. Eng. 42, 608–615. 9. Uutela, K., Ha¨ma¨la¨inen, M. & Somersalo, E. (1999) NeuroImage 10, 173–180. 18. Roland, P. E., Graufelds, C. J., Wåhlin, J., Ingelman, L., Andersson, M., 10. Schormann, T., Henn, S. & Zilles, K. (1996) Lect. Notes Comput. Sci. 1131, Ledberg, A., Pedersen, J., Åkerman, S., Dabringhaus, A. & Zilles, K. (1994) 337–342. Hum. Brain Mapp. 1, 173–184. 11. Woods, R. P., Grafton, S. T., Watson, J. D., Sicotte, N. L. & Mazziotta, J. C. 19. Lamme, V. A., Rodriguez-Rodriguez, V. & Spekreijse, H. (1999) Cereb. Cortex (1998) J. Comput. Assist. Tomogr. 22, 153–165. 9, 406–413.

Vanni et al. PNAS ͉ February 27, 2001 ͉ vol. 98 ͉ no. 5 ͉ 2779 Downloaded by guest on September 29, 2021 20. Talairach, J. & Tournoux, P. (1988) Co-Planar Stereotaxic Atlas of the Human 31. Barbur, J. L., Watson, J. D. G., Frackowiak, R. S. J. & Zeki, S. (1993) Brain Brain (Thieme, New York). 116, 1293–1302. 21. Corbetta, M., Akbudak, E., Conturo, T. E., Snyder, A. Z., Ollinger, J. M., 32. Vanni., S., Uusitalo, M. A., Kiesila¨, P. & Hari, R. (1997) NeuroReport 8, Drury, H. A., Linenweber, M. R., Petersen, S. E., Raichle, M. E., VanEssen, 1939–1942. D. C. & Shulman, G. L. (1998) 21, 761–773. 33. Beckers, G. & Zeki., S. (1995) Brain 118, 49–60. 22. Galletti, C., Fattori, P., Gamberini, M. & Kutz., D. F. (1999) Eur. J. Neurosci. 34. Uusitalo, M. A., Jousma¨ki, V. & Hari, R. (1997) Neurosci. Lett. 224, 45–48. 11, 3922–3936. 35. Schiller, P. H. & Malpeli, J. G. (1978) J. Neurophysiol. 41, 788–797. Proc. Natl. Acad. 23. Van Essen, D. C., Drury, H. A., Joshi, S. & Miller, M. I. (1998) 36. Logothetis, N. K., Schiller, P. H., Charles, E. R. & Hurlbert, A. C. (1990) Sci. USA 95, 788–795. Science 247, 214–217. 24. Ullman, S. (1995) Cereb. Cortex 5, 1–11. 37. Schroeder, C. E., Tenke, C. E., Givre, S. J., Arezzo, J. C. & Vaughan, H. G., 25. Mountcastle, V. B. (1998) Perceptual : The Cerebral Cortex (Har- Jr. (1991) Vision. Res. 31, 1143–1157. vard Univ. Press, Cambridge, MA). 26. Portin, K. & Hari, R. (1999) Proc. R. Soc. London. B 266, 981–985. 38. Portin, K., Salenius, S., Salmelin, R. & Hari, R. (1998) Cereb. Cortex 8, 253–260. 27. Galletti, C., Battaglini, P. P. & Fattori, P. (1995) Eur. J. Neurosci. 7, 2486–2501. 39. Hari, R., Salmelin, R., Tissari, S. O., Kajola, M. & Virsu, V. (1994) Nature 28. Shipp, S., Blanton, M. & Zeki, S. (1998) Eur. J. Neurosci. 10, 3171–3193. (London) 367, 121–122. 29. Nowak, L. G., Munk, M. H., Girard, P. & Bullier, J. (1995) Vis. Neurosci. 12, 371–384. 40. Jousma¨ki, V., Ha¨ma¨la¨inen, M. & Hari., R. (1996) NeuroReport 7, 2961–2964. 30. Bullier, J., Girard, P. & Salin, P.-A. (1994) in Cerebral Cortex, eds. Peters, A. 41. Nowak, L. G. & Bullier, J. (1997) in Cerebral Cortex, eds. Rockland, K. S., Kaas, & Rockland, K.S. (Plenum, New York), Vol. 10, pp. 301–330. J. H. & Peters, A. (Plenum, London), Vol. 12, pp. 205–241.

2780 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.041600898 Vanni et al. Downloaded by guest on September 29, 2021