Visual evoked response as a function of grating spatial frequency

Transcription

1 Visual evoked response as a function of grating spatial frequency Ronald Jones and Max J. Keck Transient visual evoked responses (VER's) to the appearance-disappearance of sinusoidal gratings have been investigated for a range of spatial frequencies. Contrary, to the results of previous studies, the results show that the transient VER consists of a relatively simple waveform that is most easily characterized by the initial negative peak (Nj) whose latency and amplitude vary with the contrast and spatial frequency of the grating. At spatial frequencies less than 3 cycles/degree (eld) an additional short latency component appears in the response. This component is maximum at 1 to 2 eld, saturates at low contrast, and is insensitive to the precise position of the grating on the retina. The results are related to the properties of transient and sustained channels assumed to exist in the human visual system. Key words: visual evoked response, contrast sensitivity, sinusoidal grating, spatiotemporal interactions, transient-sustained channels, pattern perception use of sinusoidal gratings as visual stimuli has provided new insight into the spatial-temporal properties of the human visual system. Although a large and diverse psychophysical literature exists, transient visual evoked response (VER) studies using sinusoidal gratings as stimuli have been few in number. 1 " 4 Such studies would be desirable for making comparisons with psychophysical results. A transient VER is the response to a physiologically distinct stimulus. Response averaging is necessary to separate the transient response from background electrocerebral activity; the stimulus repetition rate must be low and preferably aperiodic. On the other From the College of Optometry, The Ohio State University, Columbus, Ohio, and the Department of Physics, John Carroll University, Cleveland, Ohio. This study was supported by a grant from the Ohio Lions Eye Research Foundation. Submitted for publication Sept. 27, Reprint requests: Dr. Ronald Jones, College of Optometry, The Ohio State University, 338 W. 10th Ave., Columbus, Ohio hand, steady-state VER's are obtained by presenting repetitive stimuli at a rate sufficient to cause overlap of the individual evoked responses. The amplitude and phase of the resulting oscillating potential constitutes the steady-state VER. Transient VER's are higher in information content than steady-state VER's 5 and thus may be more suitable for making general comparisons with psychophysical results. Kulikowski 1 and Kulikowski and Leisman 2 have investigated the transient VER obtained to an abrupt change in the contrast of a sinusoidal grating while the space average luminance is held constant. They reported the effect of changes in grating contrast for a limited range of spatial frequencies near the peak of the contrast sensitivity function. These transient responses are characterized as having an early and a late negative-positive complex whose amplitude and latency are contrast dependent. The psychophysical threshold contrast was successfully predicted by extrapolation to find the contrast giving zero amplitude /78/ S00.80/ Assoc. for Res. in Vis. and Ophthal., Inc.

2 Volume 17 Number 7 VER and grating spatial frequency b a CL < 6 10% 25% 40% TIME (ms) Fig. 1. Response of Subject S. L. to 50 msec contrast-impulses of a 4 c/d grating at five contrast values. Time t =0 corresponds to the onset of the contrast-impulse. More recently, Parker and Salzen 3 ' 4 studied the effect of spatial frequency on the transient VER for high-contrast sinusoidal gratings. They reported that the amplitude of the early complex exhibits a roll-off only at high frequency whereas the late complex shown a maximum tuned to the peak of the psychophysical contrast sensitivity function. However, their data are confounded by the fact that the contrasts used were probably above the saturation level for the response. 6 We have undertaken to investigate the effect of spatial frequency on the transient VER with a more physiologically meaningful range of contrasts. Method Visual stimulus. The visual stimuli were vertical sinusoidal gratings generated on a CRT monitor (Model 604 with P31 phosphor; Tektronix, Inc., Beaverton, Ore.) by a method similar to that of Campbell and Green. 7 The major modification to the method was to drive the z-axis of the CRT (phosphor intensity control) by a feedback circuit that linearized the control of CRT spot intensity. The system overcomes a major problem with CRT generation of visual patterns, i.e., the nonlinear relation of CRT spot intensity to z-axis voltage. The feedback control consists of a CRT spotintensity monitoring circuit achieved by optically imaging the CRT screen face onto a linear, fast photodetector (10DP; United Detector Technology, Inc., Santa Monica, Calif). The amplified photodetector output was continuously compared to the desired z-axis modulation voltage; any difference automatically effected a compensatory adjustment of the CRT z-axis control. The response time of the electronic circuitry was sufficient at the utilized frame rate of the display (100 Hz) to maintain desired display characteristics for contrasts up to 60% (contrast = amplitude/mean value) and for spatial frequencies up to 15 cycles/degree (c/d). The system permitted changes in contrast of the sinusoidal gratings to be made without introducing any measurable change in the space average luminance of the display. The area of the display was rectangular and subtended 10 by 6 degrees visual angle at the viewing distance of 50 cm. A fixation point was provided at the screen center which the subject was instructed to fixate during the experimental sessions. Binocular viewing with natural pupils and accommodation was employed. The type of stimulus presentation employed in

3 654 Jones and Keck Invest. Ophthalmol. Visual Sci. July GRATING CONTRAST 0.5 Fig. 2. The amplitude (o) and latency (a) of the N, peaks (of Fig. 1) as a function of grating contrast. IOc/d O.I 0.2 CONTRAST Fig. 3. Latency of N, (solid curves) as a function of grating contrast at the spatial freguencies indicated for Subject S. L. The dashed curves connect the latency values of N o, which was present only at low spatial frequencies. these experiments was an abrupt (between successive video frames) change in the contrast of the grating from zero to some suprathreshold level, while maintaining a constant over-all screen luminance of 10 cd/m 2. We will refer to this type of stimulus exposure as a "contrast-impulse. The duration of the contrast-impulse as well as the time interval between successive exposures was under on-line computer control. Recording. Bipolar recording electrodes (E5S; Grass Instrument Co., Quincy, Mass.) were employed in obtaining evoked cortical responses; the positive input was at O z (10% of the inion-nasion distance above the inion); the negative input was this distance (about 3.7 cm) to the left of O z ; the right earlobe was grounded. The electrode contact resistance was tested to assure that it was below 10kO. Signals were amplified with a bandwidth of 0.1 to 100 Hz. Transient VER's to the appearance of the gratings were averaged with a FORTRAN IV program on a digital computer (Super 8; Digital Equipment Corp., Marlboro, Mass.) which simultaneously controlled the CRT display parameters. The gratings were exposed at the average rate of 2 Hz. The actual interval between successive exposures was random between the limits of 350 to 650 msec as determined by a random number generator in the computer program. Each record was the average of 300 sweeps a sweep being the 300 msec response period following each stimulus presentation. Subjects. Complete data were taken on two subjects. The main findings were replicated on three others to confirm the general nature of the results. One subject, S. L., was a 23-year-old college

4 Volume 17 Number 7 VER and grating spatial frequency 655 M.K, UJ Q a. 4.0 c/d 4% 0.5 c/d 25% 0.5 c/d -25% TIME (ms) Fig. 4. Response of Subject M. K. to 50 msec contrast-impulses at the indicated spatial frequencies and contrasts. student who had considerable experience as an observer in evoked response studies. The other subject was one of the authors, M. K. All subjects had normal vision with refractive corrections. We found the VER waveform to be remarkably consistent across the subjects investigated; the substantial individual differences that usually plague VER interpretation for other types of stimuli were not observed. Results VER as a function of grating contrast. Preliminary experiments indicated that an exposure of the contrast-flash of 50 msec duration was sufficient to produce a VER having a maximum amplitude; therefore this duration was employed in the experiments reported. The responses elicited by a 4 c/d grating at five different contrast values are given in Fig. 1. Each VER consists of an early negative-positive complex, Nx,Pi, and a later complex, N 2, P2, which is less prominent under our recording conditions. These components have been identified in previous studies 1 " 4 which used monopolar recording leads. It can be seen that the amplitude of component Nj increases with contrast up to 25%, where saturation begins. The latency of Nj shows a similar course: it decreases with contrast up to 25% and then saturates. The effect of contrast changes on the other major components is not regular. In Fig. 2 the amplitude and latency of the Nj peak have been plotted as a function of grating contrast to show these relationships more clearly. The same general behavior of Nj as a function of contrast was found for all spatial frequencies within the 0.5 to 10 c/d range that was investigated. The effect of contrast change was most pronounced in the amplitude and latency of the first major component N v Systematic changes in other components are not apparent. VER as a function of spatial frequency. The contrast sensitivity function of the eye determined psychophysically shows a bandpass characteristic having a maximum at about 4 c/d. 8 We have shown that the amplitude of N x at a given spatial frequency is strongly influenced by grating contrast. If this component reflects information about contrast sensitivity, independent of spatial frequency, its amplitude is expected to be

5 656 Jones and Keck Invest. Ophthalmol. Visual Sci. July 1978 S.L. UJ O < ISO TIME(ms) Fig. 5. Response of Subject S. L. to 50 msec contrast-impulses at the indicated spatial frequencies and contrasts. maximum at 4 c/d and to diminish for higher and lower spatial frequencies. Similarly, the latency of Nj should exhibit a U-shaped function of spatial frequency with a minimum at 4 c/d. VER's were measured for a range of spatial frequencies, and the effect of spatial frequency on the latency of Nj is illustrated in Fig. 3. At any given contrast it may be seen that the latency of Nj is minimum at 4 c/d. The amplitude of N, also exhibited the expected trend. These results notwithstanding, it is pertinent to ask if the observed effects of spatial frequency on the VER are due only to variation of contrast sensitivity across the spatial frequency domain. If this is the case, VER's obtained at different spatial frequencies should be identical provided that the contrasts are suitably adjusted to compensate for the differences in contrast sensitivity. We manipulated the contrast of the gratings so as to produce VER's of the same amplitude at each of a number of different spatial frequencies. Responses from such an experiment are given in Fig. 4. Focusing only on the upper four curves of Fig. 4, it is apparent that as spatial frequency is reduced from 10 to 4 c/d there is a marked reduction in the latency of the VER. Since it was not possible to produce identical VER's for different spatial frequency gratings by any adjustment of contrast, it must be concluded that the VER encodes information about the spatial frequency of the grating as well as that about contrast sensitivity. Spatial frequency had another effect on the VER, which is apparent in the lower two curves of Fig. 4 and is shown more fully for a second subject in Fig. 5. The waveform of the VER for gratings having spatial frequencies less than 3 c/d contained an additional negative-positive component complex (labeled N 0,P 0 ). The latency of N o was shorter than N t for all contrasts (see Fig. 3). Included in Fig. 5 are VER's for three levels of contrast obtained from a 2 c/d grating. At the lowest contrast of 5%, both N 0,P 0 and NJ.PJ are clearly identifiable. However, as the contrast is increased, the early complex shows no growth in amplitude, whereas

6 Volume 17 Number 7 VER and grating spatial frequency 657 Nj increases in amplitude and moves to shorter latency. This causes N 2 to overlap P o, giving the waveform of the VER a much different appearance. The early and late complexes of the 1.0 c/d and the 0.5 c/d gratings did not overlap, so that the effects of contrast on the individual complexes could be observed. Changes in grating contrast were primarily reflected in the amplitude and latency of N 1; whereas the early complex was largely independent of contrast, i.e., saturating below 5%. These results indicate that N o and N t are independent. The proximity of the components for spatial frequencies below 3 c/d makes it difficult to accurately determine the amplitude of Nj because its baseline is altered by the presence of P o. An experimental means of isolating the components will be required before it will be possible to make valid amplitude measurements below 3 c/d. To a lesser extent the latency of the components is also distorted by the proximity of the components; however, the latter effect is not believed to be sufficient to alter the general nature of our results. Because N 0,P 0 was present only for relatively coarse gratings, there exists the possibility that it is the result of stimulation of a localized area of the retina (e.g., the fovea) by one of the cycles of the grating. If this is the explanation, the VER should be highly sensitive to the position of fixation relative to the grating. A control experiment was undertaken in which the grating was shifted laterally by l A cycle with respect to its previous position. The VER obtained in this position was found to be indistinguishable from that obtained previously. This result is illustrated for a grating of 0.5 c/d in Fig. 4. The curve labeled 25% was obtained from the "shifted" grating and is not significantly different from the previous response (labeled 25%). Because of this lack of position sensitivity we conclude that N 0,P 0 is not an artifact due to local stimulation. Discussion Our results show that the transient VER response consists of a relatively simple waveform pattern that is most easily characterized by the initial negative peak N u whose latency and amplitude varies with the contrast and spatial frequency of the grating stimulus. Further, at spatial frequencies less than 3 c/d an additional short-latency complex (N 0,P 0 ) appears in the response. This observation was not reported in previous investigations. Some differences in our experimental method may account for this. (1) Stimulation was aperiodic, the interstimulus interval being random between the limits of 350 to 650 msec. This would be expected to reduce VER data contamination by after-discharges and certain forms of noise as compared to the use of periodic stimulation. 9 (2) Bipolar recordings were used in these experiments, whereas midline monopolar leads were used in previous investigations of transient VER's. '~ 4 (3) We employed less restrictive bandpass filtering (0.1 to 100 Hz) to reduce loss of component detail from temporal integration of successive components. (4) Only relatively brief exposures were employed so as to avoid off-effects. The significance of the early complex N 0,P 0 is open to conjecture. We would like to suggest that it represents the response of a transient system. This is indicated by its early latency, its appearance at low spatial frequencies only, and its saturation at low contrast values. Psychophysical measurements 10 " 12 of the spatial frequency sensitivity function show that the sensitivity for moving gratings peaks near 1 to 2 c/d whereas the sensitivity for stationary gratings peaks near 4 to 5 c/d. To explain these results, Tolhurst 10 proposed the existence of two distinct classes of channels in human vision: form-analyzers and movement-analyzers, analogous to the X (sustained) and Y (transient) neurons, respectively, in the cat. Both classes are thought to operate in parallel, with the movement-analyzers dominating the detection at low spatial frequencies and the form-analyzers dominating the detection of high spatial frequencies. The brief contrast-impulses used as stimuli in the present experiments may well stimulate both transient and sustained neurons whose combined response is recorded as the transient VER. Thus we suggest that the

7 658 Jones and Keck Invest. Ophthalmol. Visual Sci. July I 2 5 SPATIAL FREQUENCY (c/deg) Fig. 6. Latency of N, (o) and N o (D) as a function of spatial frequency for 25% contrast gratings for Subject S. L. short latency peak N o can be attributed to transient responses and the N! peak to sustained responses. This interpretation conflicts with that drawn from previous studies, 1 " 4 where the N u F t complex was interpreted as a transient response and the later N 2,P 2 complex as the sustained response. The attenuation of N o relative to the amplitude of the contrast-sensitive component Nj is probably attributable to the fact that the VER is primarily a macular response. 13 Animal studies 14 suggest that the macular region is dominated by neurons of the form-analysis (sustained) type; thus, if we are correct in associating N 0,P 0 with the transient mechanism of movement analysis, it would be reasonable to see less emphasis of it in the VER response. Relationship of the present results to reaction time studies. The effect of spatial frequency on the latency of the transient VER is relevant to psychophysical investigations of reaction time (RT) to sinewave gratings. It has been shown psychophysically 15 " 17 that the RT to exposure of a grating of a fixed contrast is a monotonic increasing function of spatial frequency. The RT data obtained at a fixed contrast can be compared directly to the VER data for N, and N o latencies (Fig. 3). Selecting a criterion contrast of 25%, the latencies of Nj and N o have been replotted as a function of spatial frequency in Fig. 6. This figure indicates that the latency of Nj for a given grating contrast is not monotonically related to spatial frequency but rather that it 10 is a tuned function peaked at about 4 c/d. As such, it is in qualitative agreement with the known contrast sensitivity function, not the RT data. However, it is apparent from Fig. 6 that the latency of the "earliest" component (either N o or N x ) gives a monotonic relationship. Since we have attributed N o to movement detection, it would be of interest to know the response criterion of the subjects in the RT experiments. 15 " 17 Our results suggest that subjects used a flicker criterion at low spatial frequencies and shifted to formdetection criterion above 3 c/d. Relationship of the present results to steady state VER studies. Steady-state VER's are obtained by presenting repetitive stimuli at a rate sufficient to cause overlap of the individual evoked responses. Such studies 18 ' 19 have successfully demonstrated a correlation between psychophysically determined threshold and that extrapolated from the amplitude of the steady-state VER. However, Regan and Richards 20 have raised some questions about the validity of the steadystate VER amplitude as a measure of brightness-contrast for suprathreshold stimuli. Campbell and Maffei 18 reported a distinction between steady-state VER's for high and low spatial frequency gratings. They found that the relation of the VER amplitude to log contrast could be fit by a straight regression line at spatial frequencies above about 3 c/d but that for coarser gratings two straight lines were needed to fit the data (provided the parafoveal area was included in the stimulus field). Our finding of an additional component at low spatial frequencies is consistent with their result. At the high temporal stimulation rate (8 Hz) that was employed by Campbell and Maffei, the component nature of the response cannot be distinguished; rather, the VER appears like a sinewave having an amplitude which is the integrated sum of the contributing components. The amplitude of this steady-state VER would be expected to grow at different rates in the high and low contrast ranges because N o saturates at a lower contrast level than Nj does. Thus the presence of two independent inputs to the VER that saturate at different contrasts is

8 Volume 17 Number 7 VER and grating spatial frequency 659 consistent with the two-slope relationship found by Campbell and Maffei in the low spatial frequency range. The latency of the evoked response can also be deduced from a steady-state VER by examining the phase of the response relative to the stimulus. An interesting study by Williamson et al. 21 reported measurements of steady-state visually evoked magnetic fields (VEF's), using sinusoidal gratings of various spatial frequencies. Williamson et al. were able to deduce latency from the phase of the steady-state VEF and found that the latency increased with spatial frequency, being highly correlated with the RT results of Breitmeyer 15 discussed in the previous section. It must be kept in mind that the steady-state VEF is an integrative measure of response. We note that a latency increase in the steady-state response does not distinguish a latency shift of the entire transient response from an attenuation of its earlier components relative to its later ones. Such questions are better answered by an examination of the transient VER. REFERENCES 1. Kulikowski, J. J.: Relation between psychophysics and electrophysiology, Trace, Paris 6:63, Kulikowski, J. J., and Leisman, G.: The effect of nitrous oxide on the relation between the evoked potential and contrast threshold, Vision Res. 13: 2079, Parker, D. M., and Salzen, E. A.: The spatial frequency selectivity of the early and late waves within the human visual evoked response, Perception 6:85, Parker, D. M., and Salzen, E. A.: Latency changes in the human visual evoked response to sinusoidal gratings, Vision Res. 17:1201, Regan, D.: Evoked Potentials in Sensory Physiology, Psychology and Clinical Medicine, London, 1972, Chapman and Hall. 6. Spekreijse, H., Van der Tweel, L. H., and Zuidema, T. H.: Contrast evoked responses in man, Vision Res. 13:1577, Campbell, F. W., and Green, D. G.: Optical and retinal factors affecting visual resolution, J. Physiol. (Lond.) 181:576, Blakemore, C., and Campbell, F. W.: On the existence of neurons in the human visual system selectively sensitive to the orientation and size of retinal images, J. Physiol. (Lond.) 203:237, Ruchkin, D. S.: An analysis of averaged response computations based upon aperiodic stimuli, Trans. IEEE, BME 12:87, Tolhurst, D. F.: Separate channels for the analysis of the shape and the movement of a moving grating, J. Physiol. (Lond.) 231:385, van Nes, F. L., Koenderink, J. J., Nas, H., and Bouman, M. A.: Spatiotemporal modulation transfer in the human eye, J. Opt. Soc. Am. 57:1082, Kulikowski, J. J., and Tolhurst, D. J.: Psychophysical evidence for sustained and transient detectors in human vision, J. Physiol. (Lond.) 232:149, Reitveld, W. J., Tordir, W. E. M., and Duyff, J. VV.: Contribution of fbvea and parafovea to the visual evoked response, Acta Physiol. Pharmacol. Neerl. 13:330, Ikeda, M., and Wright, M. J.: Receptive field organization of sustained and transient retinal ganglion cells which subserve different functional roles, J. Physiol. (Lond.) 227:769, Breitmeyer, B. G.: Simple reaction time as a measure of the temporal response of transient and sustained channels, Vision Res. 15:1411, Lupp, U., Hauske, G., and Wolf, W.: Perceptual latencies to sinusoidal gratings, Vision Res. 16:969, Vassilev, A., and Mitov, D.: Perception time and spatial frequency, Vision Res. 16:89, Campbell, F. W., and Maffei, L.: Electrophysiological evidence for the existence of orientation and size detectors in the human visual system, J. Physiol. (Lond.) 207:635, Campbell, F. W., and Kulikowski, J. J.: The visual evoked potential as a function of contrast of a grating pattern, J. Physiol. (Lond.) 222:345, Regan, D., and Richards, W.: Brightness contrast and evoked potentials, J. Opt. Soc. Am. 63:606, Williamson, S. J., Kaufman, L., and Brenner, D.: Biomagnetism. In Schwartz, B. B., and Foner, S., editors: Superconductor Applications: Squids and Machines, New York, 1977, Plenum Publishing Corp.