Colour, contrast and the visual evoked potential Dorothy A. Thompson* and Neville Drasdof Department of Vision Sciences, Aston University, Astron Triangle, Birmingham B4 7ET. UK (Received 14 October 1991) Visual evoked potentials exhibit interesting morphological changes when they are elicited by checkerboards of difFerent spatial and chromatic contrast, counterphasing in the foveal and lower macula Iield. The characteristic, positive wave of the phase-reversal visual evoked potential, for example, is preceded by an increasingly prominent negative peak as luminance contrast progressively increases above 10% and, at isoluminance, the response lo red and green checkerboards becomes a predominantly monophasic negative wave. To study the nature of ihc morphological change we synthesized these waveforms with a computer simulation consisting of Gaussian componenls. The amplitudes of positive and negative components were altered until the synthesized response was closely similar to the recorded data. These Gaussian components have response characteristics which are idenlilied with those of magnocellular and parvocellular neurones.
A changing visual scene elicits ionic changes in the tissue of the visual cortex. The summation of these changes is detected at the scalp as the visual evoked potential, (VEP). The efTect of contrast on this signal has been the subject of much interest since the early observations of Campbell and Kulikowski'. Theoretical models of the origin of the VEP suggest that the cortical tissue forms a dipoie generator sheet. Those parts radially oriented in the gyral crests contribute most to the recorded potential, because of their proximity to the scalp^. If stimulation extends into the more peripheral visual field the complexity of the VEP morphology increases; as potentials from various orientations of retinotopically organized tissue on the calcarine fissure, are included in the summated response. This has contributed to the considerable interstudy differences of analyses of the VEP. Many studies, for example, have used large stimulus fields, mixing foveai and peripheral responses, and sampled from a small number of points neglecting the possibility of inter-individual cortical variation. It has now become possible to record simultaneously from numerous points on the scalp and to display the scalp potential distribution, at a given time, as a colour contour 'brain map'. In addition, there have been advances in our understanding of both the functional architecture of the visual cortex and the physiological characteristics of its underlying processing streams. These enable us to select visual stimuli that will preferentially enhance activity of specialized visual areas. We have incorporated these developments in our detailed study of the VEP. Our strategy is to confine stimulation so that the recipient striate cortex is limited to the foveal •MBCO. + FBCO.
(0 1992 Butterworth-Heinemann for British College of Optometrists O275-54OX/9:/O2O225-O4
representation at the occipital poles, and to examine the changes in morphology and scalp topography which accompany selective stimulation with patterns varying in chromatic, spatial and temporal contrast. This paper specifically addresses the morphological changes that occur to the pattern reversal VEP under these conditions. Methods and subjects Pattern reversing checkerboards with side subtense of 10 and 20 min arc were presented with a Venus Neuroscientific Stimulator (Farmingdale, NY, USA). This has independent control of red, green and blue guns and a refresh rate of 119 Hz. Black and white patterns, with a range of Michelson contrasts, were phase-reversed at 1 Hz, 2 reversals s " ' ; red and green checkerboard patterns were also used. An isoluminant point was determined for each individual using a heterochromic flicker technique. The square field subtended 3 degrees and subjects were asked to fixate binocularly a point 0.5 degrees above the centre. All subjects had normal colour vision determined by an Ishihara test and 6/9 or better acuity. Hemispheric arrays of 16 electrodes were designed to cover the visual areas with maximum density. Each hemisphere was recorded from in turn, the midline column of electrodes was common to each recording session. The VEPs were averaged on a Nicolet Pathfinder Brain Mapping system (Madison, WI, USA) to a frontal (Fz) reference, with a bandpass of 0.5-30 Hz; 40 sweeps were averaged for each stimulus condition. Topographically the full field pattern reversal response is maximal over the midline, for both achromatic and chromatic patterns. Individual signals selected as templates for modelling, were therefore recorded from an electrode position 4.5% above the inion on the midline.
Ophthal. Physiol. Opt., 1992, Vol. 12, April
225
Colour, contrast and the VEP: D. A. Thompson and N. Drasdo Results and analysis
where
For foveal retinal stimulation the maximal signal has a triphasic morphology on the midline. A dominant positive peak was observed, but as contrast was progressively increased above 10%, a negative peak, preceding the positive, became apparent {Figure I). At low contrast levels the pattern reversal VEP was a predominantly monophasic, positive peak with a latency of about 124 ms. This is 15-20 ms later than the positive peak latency at high contrast. Isoluminant red and green checkerboards elicited a predominantly monophasic, negative VEP with a latency of 165 ms. This negative wave at isoluminance has been previously observed with counterphasing gratings"', but not reported with checkerboards. Inspection of these morphological changes suggests that the reversal VEP comprises a combination of positive and negative peaks that varies with contrast. This is only one way of regarding the data. There are alternative methods of analysing in Fourier techniques, or for example with Principle Component Analysis (PCA), which can reproduce the underlying components with some accuracy*. The principal component of the pattern reversal response is a composite of positive and following negative waves^. However a limitation of PCA is the mislocation of component variance which can occur with the overlap of PCA components. The reversal data described in this study appear to demonstrate a variable overlap of components which depends upon contrast. We have modelled the morphology of the recorded data with a Sum of Gaussians using positive and negative components. It seemed a natural assumption that an additive process should occur in ionic summation, and propagation of ionic changes should approximate a temporal Gaussian function. Positive and negative Gaussian components were described by the general equation: X -
A is the amphtude of the component, W the width at half height and L the time to peak. The sum of positive and negative components was graphically displayed as an amplitude time plot. Component amplitudes and latencies were altered until the closest fit to the recorded data was obtained. The results are shown in Figure 2. The synthesized waveforms displayed in the middle column closely match the recorded data, shown in the first column. The variation of Gaussian component amplitude with contrast was noted and plotted to give the function shown in Figure 3. Discussion The synthesized waveforms from the Sum of Gaussian model closely approximated the recorded data. A tentative parallel between positive and negative component amplitude/contrast functions. Figure 3, and those of magno- and parvo- units can be drawn*". The positive component amplitude is greater than the negative at low contrast, a trait that mirrors the response of magnocellular neurones. The negative component amplitude mirrors parvocellular behaviour, being less at lower contrast, but at high contrast the negative component amplitude exceeded the positive component. This could reflect the additional stimulation of a greater number of parvoresponding elements. There have been previous suggestions that negative VEP waveforms relate to parvocellular activity, using onset of presentation of chromatic or fine patterns^''^'^. In this study a mixed stimulus, comprising a reversal mode corresponding to magnocellular activity** and high spatial frequency corresponding to parvocellular activity, was biased towards either subsystem by a variation in contrast. The Sum of Gaussian model appears to provide a simple and effective method of separating these
FVG
70% 47.5 ms/div Figure 1 Examples of visual evoked potentials recorded fmm an electrode 4.5% above the inion on the midline in two individuals. Positive is an upward deflection. The top trace is the negative wave recorded in response lo red and green checkerboard reversal at isoluminance. The other traces arc the responses to black and while checks at the Michelson contrasts shown. At the highest contrast levels a negative wave precedes the dominant positive wave
226
Ophthal. Physioi. Opt., 1992, Vol. 12, April
Colour, contrast and the VEP: D. A. Thompson and N. Drasdo
10X
\
Figure 2 Visual evoked potentials recorded to different levels of Michelson contrast in another individual are shown in the first column. The second column displays the synthesized responses from the computer simulation. The third column shows examples of the relative amplitude and latencies of the positive and negative Gaussian components used to model the recorded signals
contributions. Its principle advantage is a true impression of amplitude rather than the transient peak-to-peak measure or indeed baseiine-to-peak measure, which only occurs after interaction of positive and negative peaks. Only two Gaussian components were required to successfully model the achromatic reversal data. When red and green patterns are used, isoluminance can be regarded as the lower limit of Michelson luminance contrast. In the Gaussian model both component amplitudes are attenuated at low contrast, yet a delayed, relatively large amplitude negative reversal VEP is obtained at isoluminance. It becomes necessary to consider the inclusion of a second, independent negative component to account for the isoiuminant VEP. Evidence from intracortical experiments would support a separate origin of the negative and positive peaks.
1001
•o 3
E
40 contrast % Figure 3 The amplitudes of the positive and negative Gaussian components arc shown as a function of contrast. The negative component amplitude. • . is smaller than the positive component, Q, at all but the highest contrast level
Schroeder et al.^^ investigated the macaque VEP by recording simultaneously at incremental depths in area 17 using multi-contact electrodes. They interpolate to a human homologue and suggest that the preceding negative peak arises primarily from sinks in laminae 4c; corresponding to the initial activation of the thalmorecipient sub-divisions of 4c. The characteristic positive response arises from large current sources in supragranular laminae. They propose that the later negative activity is a more complex summation of local and more peripheral activity, i.e. from other visual areas. This could be speculatively linked to a parvocellular input to shape detection occurring at high luminance contrast, high spatial frequencies or isoluminance". There have been previous accounts of a negative peak preceding the characteristic positive peak of both flash and pattern VEPs. Pauius et al.^~, for example, suggested that the negativity (N87) occurred maximally when a yellow and red 4 degree field were interchanged, but noted a similar morphology for a red luminance decrement which indicates that the observation was not a chromatic specific characteristic. Bodis-Wollner et ai^^ recently studied the preceding negativity termed N70, with sine wave gratings and found it to be spatially tuned and optimized with foveal fields, but the Gaussian model reveals the hidden amplitudes and latency of the true components that generate the N- P - N sequence. The range of inter-individual amplitude variation in the preceding negative peak might be a reflection of hemispheric asymmetry. A topographic display of hemifield data showed a tendency for an ipsilateral (to the stimulated field) distribution of the preceding negativity ^^ ^^. The positivity occurred over the contralateral hemisphere, i.e. the correctly projected hemisphere. The peaks are temporally separated by about 20 ms. These results would suggest that spatially distinct regions of neural tissue are activated. Bodis-Wollner et a/.'^ suggested the ipsilateral distribution reflects an ipsilateral origin. This would require a bilateral cortical representation
Ophthal. Physiol. Opt., 1992. Vol. 12, April
227
Colour, contrast and the VEP: D. A. Thompson and N. Drasdo of the retinal midline, as proposed in macaque by Leventhal et al.^^. More simply it may be that the N70 results from VI which has a slightly medial or even oblique orientation in some individuals''', whilst the PIOO has a generator in V2 situated on the cortical surface ^ ^.
4.
5. 6.
Conclusions This study of the pattern reversal VEP has demonstrated a contrast related morphology change which can be modelled using a Sum of Gaussians. This method separates positive and negative components allowing inspection before their interaction or ionic summation. The component amplitude was studied as a function of contrast and a similarity between magno- and parvocellular behaviour and positive and negative components, respectively, was observed. This type of variation with contrast, to our knowledge, has not been reported previously.
7. 8. 9. 10.
11.
12.
Acknowledgements Dr D. A. Thompson is supported by a Wellcome Trust Vision Research Fellowship.
13.
14.
References t.
Campbell, F. W. and Kulikowski, J. J. The visual evoked potential as a function of contrast of a grating pattern. J. Physiot. 222, 345-356(1972). 2. Stok.C.J. The inverse problem in EEG and MEG with application to visual evoked responses. Dissertation, Rijks University, Leiden, Netherlands, 1986. 3. Berninger, T. A., Arden, G. B., Hogg. C. R. and Frumkes, T. Separable retinal and cortical potentials from each major visual pathway: preliminary results. Br. J. Opihalmol. 73,502-511 (1989).
228
Ophthal. Physiol. Opt., 1992, Vol. 12, April
15.
16. 17.
Skrandies, W, Data reduction of multichannel fields: global field power and principle component analysis. Brain Topography 2, 73-80(1989). Maier, J., Dagnelie. G., Spekreijse, H. and Van Dijk. B. W. Principal component analysis for source localisation of VEPs in man. Vision Res. 27, 165-177 (1987). Kaplan, E. and Shapley. R. The primate retina contains two ty[>es of ganglion cells, with high and low contrast sensitivity. Proc. Mat. Acad. Sci. 83, 2755-2757 (1986). Drasdo, N. Cortical potentials evoked by pattern presentation in the foveal region. In Evoked Potentials (ed C. Barber) MTP Press, Lancaster, UK, pp. 167-174(1980). Murray, I. J., Parry. N. R. A., Garden, D. and Kulikowski, J. J. Human visual evoked potentials to chromatic and achromatic gratings. Clin. Vis. Sa. 1, 231-244(1987), Kulikowski, J. J. Pattern and movement detection in man and rabbit: separation and comparison of occipital potentials. Vision Res. \S. 183-189(1978). Schroeder, C. E., Tenke, C. E,, Givre, S. J,, Arezzo, J. C. and Vaughan Jr, H. G. Striate cortical contribution to the surface recorded pattern reversal VEP in the alert monkey. Vision Res. 7/8, 1143-1157(1991). Schiller, P. H. The color opponent and broad-band channels of primate visual system. In Front Pigments lo Perception: Advances in Understanding Visual Proces.ses (eds A. Valberg and B. B. Lee) NATO ASI Series, volume 203. Plenum, NY, pp. 127 - 132 (1991). Pauius, W,, Homberg, V,, Cunningham, K. and Halliday, A. M. Colour and brightness coding in the central nervous system: theoretical aspects and visual evoked potentials to homogeneous red and green stimuli. Proc. R. Soc. Umd. B 111, 53-66 (1986). Bodis-Wollner, I., Mylin, L. and Frkovic, S. The topography of the N70 component of the visual evoked potential in humans. In Topographic Mapping of EEG and Evoked Potential {sd K. Maurer) Springer-Verlag, Berlin, pp. 396-406(1989). Pauius. W. M., Plendl, H. and Krafczyk, S. Spatial dissociation of early and late colour evoked potentials, Eleciroenceph. Clin. Neurophy.siol. 11. 81 88 (1988). Onofrj, M., Bazzano, S., Malatesta, G. and P'ulgente, T. Mapped distribution of pattern reversal VEPs to central field and lateral half field stimuli of different spatial frequencies. FJectroenceph. Clin. Neurophysiol. 80. 167-180 (1991). Leventhal. A. G.. Ault, S. J. and Vitek. D. J. The nasotemporal division of the primate retina: the neural bases for macular sparing and splitting. Science 240, 66-67 (1988). Polyak, S, The Vertebrate Visual System. University of Chicago Press. Chicago USA, pp. 391-405 and 461-485 (1957).
A changing visual scene elicits ionic changes in the tissue of the visual cortex. The summation of these changes is detected at the scalp as the visual evoked potential, (VEP). The efTect of contrast on this signal has been the subject of much interest since the early observations of Campbell and Kulikowski'. Theoretical models of the origin of the VEP suggest that the cortical tissue forms a dipoie generator sheet. Those parts radially oriented in the gyral crests contribute most to the recorded potential, because of their proximity to the scalp^. If stimulation extends into the more peripheral visual field the complexity of the VEP morphology increases; as potentials from various orientations of retinotopically organized tissue on the calcarine fissure, are included in the summated response. This has contributed to the considerable interstudy differences of analyses of the VEP. Many studies, for example, have used large stimulus fields, mixing foveai and peripheral responses, and sampled from a small number of points neglecting the possibility of inter-individual cortical variation. It has now become possible to record simultaneously from numerous points on the scalp and to display the scalp potential distribution, at a given time, as a colour contour 'brain map'. In addition, there have been advances in our understanding of both the functional architecture of the visual cortex and the physiological characteristics of its underlying processing streams. These enable us to select visual stimuli that will preferentially enhance activity of specialized visual areas. We have incorporated these developments in our detailed study of the VEP. Our strategy is to confine stimulation so that the recipient striate cortex is limited to the foveal •MBCO. + FBCO.
(0 1992 Butterworth-Heinemann for British College of Optometrists O275-54OX/9:/O2O225-O4
representation at the occipital poles, and to examine the changes in morphology and scalp topography which accompany selective stimulation with patterns varying in chromatic, spatial and temporal contrast. This paper specifically addresses the morphological changes that occur to the pattern reversal VEP under these conditions. Methods and subjects Pattern reversing checkerboards with side subtense of 10 and 20 min arc were presented with a Venus Neuroscientific Stimulator (Farmingdale, NY, USA). This has independent control of red, green and blue guns and a refresh rate of 119 Hz. Black and white patterns, with a range of Michelson contrasts, were phase-reversed at 1 Hz, 2 reversals s " ' ; red and green checkerboard patterns were also used. An isoluminant point was determined for each individual using a heterochromic flicker technique. The square field subtended 3 degrees and subjects were asked to fixate binocularly a point 0.5 degrees above the centre. All subjects had normal colour vision determined by an Ishihara test and 6/9 or better acuity. Hemispheric arrays of 16 electrodes were designed to cover the visual areas with maximum density. Each hemisphere was recorded from in turn, the midline column of electrodes was common to each recording session. The VEPs were averaged on a Nicolet Pathfinder Brain Mapping system (Madison, WI, USA) to a frontal (Fz) reference, with a bandpass of 0.5-30 Hz; 40 sweeps were averaged for each stimulus condition. Topographically the full field pattern reversal response is maximal over the midline, for both achromatic and chromatic patterns. Individual signals selected as templates for modelling, were therefore recorded from an electrode position 4.5% above the inion on the midline.
Ophthal. Physiol. Opt., 1992, Vol. 12, April
225
Colour, contrast and the VEP: D. A. Thompson and N. Drasdo Results and analysis
where
For foveal retinal stimulation the maximal signal has a triphasic morphology on the midline. A dominant positive peak was observed, but as contrast was progressively increased above 10%, a negative peak, preceding the positive, became apparent {Figure I). At low contrast levels the pattern reversal VEP was a predominantly monophasic, positive peak with a latency of about 124 ms. This is 15-20 ms later than the positive peak latency at high contrast. Isoluminant red and green checkerboards elicited a predominantly monophasic, negative VEP with a latency of 165 ms. This negative wave at isoluminance has been previously observed with counterphasing gratings"', but not reported with checkerboards. Inspection of these morphological changes suggests that the reversal VEP comprises a combination of positive and negative peaks that varies with contrast. This is only one way of regarding the data. There are alternative methods of analysing in Fourier techniques, or for example with Principle Component Analysis (PCA), which can reproduce the underlying components with some accuracy*. The principal component of the pattern reversal response is a composite of positive and following negative waves^. However a limitation of PCA is the mislocation of component variance which can occur with the overlap of PCA components. The reversal data described in this study appear to demonstrate a variable overlap of components which depends upon contrast. We have modelled the morphology of the recorded data with a Sum of Gaussians using positive and negative components. It seemed a natural assumption that an additive process should occur in ionic summation, and propagation of ionic changes should approximate a temporal Gaussian function. Positive and negative Gaussian components were described by the general equation: X -
A is the amphtude of the component, W the width at half height and L the time to peak. The sum of positive and negative components was graphically displayed as an amplitude time plot. Component amplitudes and latencies were altered until the closest fit to the recorded data was obtained. The results are shown in Figure 2. The synthesized waveforms displayed in the middle column closely match the recorded data, shown in the first column. The variation of Gaussian component amplitude with contrast was noted and plotted to give the function shown in Figure 3. Discussion The synthesized waveforms from the Sum of Gaussian model closely approximated the recorded data. A tentative parallel between positive and negative component amplitude/contrast functions. Figure 3, and those of magno- and parvo- units can be drawn*". The positive component amplitude is greater than the negative at low contrast, a trait that mirrors the response of magnocellular neurones. The negative component amplitude mirrors parvocellular behaviour, being less at lower contrast, but at high contrast the negative component amplitude exceeded the positive component. This could reflect the additional stimulation of a greater number of parvoresponding elements. There have been previous suggestions that negative VEP waveforms relate to parvocellular activity, using onset of presentation of chromatic or fine patterns^''^'^. In this study a mixed stimulus, comprising a reversal mode corresponding to magnocellular activity** and high spatial frequency corresponding to parvocellular activity, was biased towards either subsystem by a variation in contrast. The Sum of Gaussian model appears to provide a simple and effective method of separating these
FVG
70% 47.5 ms/div Figure 1 Examples of visual evoked potentials recorded fmm an electrode 4.5% above the inion on the midline in two individuals. Positive is an upward deflection. The top trace is the negative wave recorded in response lo red and green checkerboard reversal at isoluminance. The other traces arc the responses to black and while checks at the Michelson contrasts shown. At the highest contrast levels a negative wave precedes the dominant positive wave
226
Ophthal. Physioi. Opt., 1992, Vol. 12, April
Colour, contrast and the VEP: D. A. Thompson and N. Drasdo
10X
\
Figure 2 Visual evoked potentials recorded to different levels of Michelson contrast in another individual are shown in the first column. The second column displays the synthesized responses from the computer simulation. The third column shows examples of the relative amplitude and latencies of the positive and negative Gaussian components used to model the recorded signals
contributions. Its principle advantage is a true impression of amplitude rather than the transient peak-to-peak measure or indeed baseiine-to-peak measure, which only occurs after interaction of positive and negative peaks. Only two Gaussian components were required to successfully model the achromatic reversal data. When red and green patterns are used, isoluminance can be regarded as the lower limit of Michelson luminance contrast. In the Gaussian model both component amplitudes are attenuated at low contrast, yet a delayed, relatively large amplitude negative reversal VEP is obtained at isoluminance. It becomes necessary to consider the inclusion of a second, independent negative component to account for the isoiuminant VEP. Evidence from intracortical experiments would support a separate origin of the negative and positive peaks.
1001
•o 3
E
40 contrast % Figure 3 The amplitudes of the positive and negative Gaussian components arc shown as a function of contrast. The negative component amplitude. • . is smaller than the positive component, Q, at all but the highest contrast level
Schroeder et al.^^ investigated the macaque VEP by recording simultaneously at incremental depths in area 17 using multi-contact electrodes. They interpolate to a human homologue and suggest that the preceding negative peak arises primarily from sinks in laminae 4c; corresponding to the initial activation of the thalmorecipient sub-divisions of 4c. The characteristic positive response arises from large current sources in supragranular laminae. They propose that the later negative activity is a more complex summation of local and more peripheral activity, i.e. from other visual areas. This could be speculatively linked to a parvocellular input to shape detection occurring at high luminance contrast, high spatial frequencies or isoluminance". There have been previous accounts of a negative peak preceding the characteristic positive peak of both flash and pattern VEPs. Pauius et al.^~, for example, suggested that the negativity (N87) occurred maximally when a yellow and red 4 degree field were interchanged, but noted a similar morphology for a red luminance decrement which indicates that the observation was not a chromatic specific characteristic. Bodis-Wollner et ai^^ recently studied the preceding negativity termed N70, with sine wave gratings and found it to be spatially tuned and optimized with foveal fields, but the Gaussian model reveals the hidden amplitudes and latency of the true components that generate the N- P - N sequence. The range of inter-individual amplitude variation in the preceding negative peak might be a reflection of hemispheric asymmetry. A topographic display of hemifield data showed a tendency for an ipsilateral (to the stimulated field) distribution of the preceding negativity ^^ ^^. The positivity occurred over the contralateral hemisphere, i.e. the correctly projected hemisphere. The peaks are temporally separated by about 20 ms. These results would suggest that spatially distinct regions of neural tissue are activated. Bodis-Wollner et a/.'^ suggested the ipsilateral distribution reflects an ipsilateral origin. This would require a bilateral cortical representation
Ophthal. Physiol. Opt., 1992. Vol. 12, April
227
Colour, contrast and the VEP: D. A. Thompson and N. Drasdo of the retinal midline, as proposed in macaque by Leventhal et al.^^. More simply it may be that the N70 results from VI which has a slightly medial or even oblique orientation in some individuals''', whilst the PIOO has a generator in V2 situated on the cortical surface ^ ^.
4.
5. 6.
Conclusions This study of the pattern reversal VEP has demonstrated a contrast related morphology change which can be modelled using a Sum of Gaussians. This method separates positive and negative components allowing inspection before their interaction or ionic summation. The component amplitude was studied as a function of contrast and a similarity between magno- and parvocellular behaviour and positive and negative components, respectively, was observed. This type of variation with contrast, to our knowledge, has not been reported previously.
7. 8. 9. 10.
11.
12.
Acknowledgements Dr D. A. Thompson is supported by a Wellcome Trust Vision Research Fellowship.
13.
14.
References t.
Campbell, F. W. and Kulikowski, J. J. The visual evoked potential as a function of contrast of a grating pattern. J. Physiot. 222, 345-356(1972). 2. Stok.C.J. The inverse problem in EEG and MEG with application to visual evoked responses. Dissertation, Rijks University, Leiden, Netherlands, 1986. 3. Berninger, T. A., Arden, G. B., Hogg. C. R. and Frumkes, T. Separable retinal and cortical potentials from each major visual pathway: preliminary results. Br. J. Opihalmol. 73,502-511 (1989).
228
Ophthal. Physiol. Opt., 1992, Vol. 12, April
15.
16. 17.
Skrandies, W, Data reduction of multichannel fields: global field power and principle component analysis. Brain Topography 2, 73-80(1989). Maier, J., Dagnelie. G., Spekreijse, H. and Van Dijk. B. W. Principal component analysis for source localisation of VEPs in man. Vision Res. 27, 165-177 (1987). Kaplan, E. and Shapley. R. The primate retina contains two ty[>es of ganglion cells, with high and low contrast sensitivity. Proc. Mat. Acad. Sci. 83, 2755-2757 (1986). Drasdo, N. Cortical potentials evoked by pattern presentation in the foveal region. In Evoked Potentials (ed C. Barber) MTP Press, Lancaster, UK, pp. 167-174(1980). Murray, I. J., Parry. N. R. A., Garden, D. and Kulikowski, J. J. Human visual evoked potentials to chromatic and achromatic gratings. Clin. Vis. Sa. 1, 231-244(1987), Kulikowski, J. J. Pattern and movement detection in man and rabbit: separation and comparison of occipital potentials. Vision Res. \S. 183-189(1978). Schroeder, C. E., Tenke, C. E,, Givre, S. J,, Arezzo, J. C. and Vaughan Jr, H. G. Striate cortical contribution to the surface recorded pattern reversal VEP in the alert monkey. Vision Res. 7/8, 1143-1157(1991). Schiller, P. H. The color opponent and broad-band channels of primate visual system. In Front Pigments lo Perception: Advances in Understanding Visual Proces.ses (eds A. Valberg and B. B. Lee) NATO ASI Series, volume 203. Plenum, NY, pp. 127 - 132 (1991). Pauius, W,, Homberg, V,, Cunningham, K. and Halliday, A. M. Colour and brightness coding in the central nervous system: theoretical aspects and visual evoked potentials to homogeneous red and green stimuli. Proc. R. Soc. Umd. B 111, 53-66 (1986). Bodis-Wollner, I., Mylin, L. and Frkovic, S. The topography of the N70 component of the visual evoked potential in humans. In Topographic Mapping of EEG and Evoked Potential {sd K. Maurer) Springer-Verlag, Berlin, pp. 396-406(1989). Pauius. W. M., Plendl, H. and Krafczyk, S. Spatial dissociation of early and late colour evoked potentials, Eleciroenceph. Clin. Neurophy.siol. 11. 81 88 (1988). Onofrj, M., Bazzano, S., Malatesta, G. and P'ulgente, T. Mapped distribution of pattern reversal VEPs to central field and lateral half field stimuli of different spatial frequencies. FJectroenceph. Clin. Neurophysiol. 80. 167-180 (1991). Leventhal. A. G.. Ault, S. J. and Vitek. D. J. The nasotemporal division of the primate retina: the neural bases for macular sparing and splitting. Science 240, 66-67 (1988). Polyak, S, The Vertebrate Visual System. University of Chicago Press. Chicago USA, pp. 391-405 and 461-485 (1957).
