- Email: [email protected]
Spatial-frequency-tuned attenuation and enhancement of the steady-state VEP by grating adaptation
Vi.sionRes. Vol. 31, No. 7/8, pp. 1167-1175, 1991 Printed in Great Britain. All rights reserved
0042-6989/91 S3.00 + 0.00 Copyright 0 1991 Pcrgamon Prm pk
SPATIAL-FREQUENCY-TUNED ATTENUATION AND ENHANCEMENT OF THE STEADY-STATE VEP BY GRATING ADAPTATION S-I-EVE SUTER,’COLINA. ARMSTR~NG,~ PENELOPE S. SUTER’ and JUS~INA C. Powsas3 ‘Department of Psychology, California State University, Bakersfield, CA 93311, ‘San Diego State University/University of California San Diego Joint Doctoral Program in Clinical Psychology, San Diego, CA 92182 and Department of Psychology, University of California, Riverside, CA 92521, U.S.A. (Received 14 September 1989; in revised form
7
August 1990)
Abstract-Steady-state visual evoked potentials (VEPs) were recorded from adults using lO%C fast spatial frequency (SF) sweeps of horizontal gratings under two conditions: (a) after exposure to a 4O%C grating of 6 or 4 c/deg, and (b) after exposure to a blank screen equalling the adapting gratings in space-averaged luminance. SF adaptation attenuated VEP amplitude near the adapting SF, but maximum attenuation was displaced from the adapting SF for 6 c/deg adaptation. Small displacements in maximum attenuation would be expected if underlying neural subunits are tuned to a small number of different center SFs. In addition, SF adaptation caused amplitude enhancement 1.O-2.0 octaves below the adapting SF, providing electrophysiological evidence in humans for coinhibitory relationships among neural mechanisms that have been postulated on the basis of analogous psychophysical findings. The results are consistent with coinhibition between SF-tuned subunits and between transient and sustained mechanisms. VEP
Adaptation
Spatial frequency
INTRODUCTION
Several lines of evidence suggest that the visual world of humans and other mammals may be constructed from spatial frequency (SF) information extracted from the retinal image by a set of SF-tuned channels (Campbell & Robson, 1968). These neural mechanisms are thought to function as bandpass filters differing in center SF. The overall contrast sensitivity function is the envelope of the underlying channel sensitivities. This view is primarily supported by psychophysical studies using SF adaptation, masking and subthreshold summation paradigms (reviewed by De Valois & De Valois, 1988; Regan, 1982). Effects that are limited to the region of the SF spectrum near the adapting, masking or summation stimuli are suggestive of SF-tuned subunits. Psychophysical studies indicate that: (a) SF adaptation diminishes contrast sensitivity (e.g. Blakemore & Campbell, 1969) and the apparent contrast of suprathreshold stimuli (Blakemore, Muncey & Ridley, 1971) over a region of about 1.0 octave (half-amplitude bandwidth) near the adapting SF stimulus; (b) masking effects are confined to a similarly narrow range of SF differences between mask-
ing and test stimuli (e.g. Legge & Foley, 1980); and (c) subthreshold summation effects are usually seen only when the subthreshold stimuli are within a narrow SF region (e.g. Sachs, Nachmias & Robson, 1971). Based on such evidence, several quantitative models have been proposed that incorporate sets of SF-tuned neural subunits with particular performance characteristics (e.g. Klein & Levi, 1985; Legge & Foley, 1980; Watson, 1987; Watt & Morgan, 1985; Wilson & Gelb, 1984). Electrophysiological evidence bearing on the hypothesized neural mechanisms is quite limited, but critical since the psychophysical data themself cannot provide direct evidence about neural mechanisms (Regan, 1983). However, SF-tuned effects of SF adaptation have been noted for visual evoked potential (VEP) amplitudes in humans (e.g. Mecacci & Spinelli, 1976) and single cortical cells of monkeys (e.g. Sclar, Lennie & DePriest, 1989) and cats (e.g. MafIei, Fiorentini & Bisti, 1973). Masking paradigms yield SF-tuned VEP attenuation (e.g. Fiorentini, Pirchio & Spinelli, 1983). In summary, psychophysical and neurophysiological findings are consistent in a general sense as they relate to the multiple SF-tuned channel hypothesis.
1167
SlwE SumR et al.
1168
In the present study a VEP fast-SF sweep technique was used to explore for adaptation effects at many SFs in the absence of state changes. This allowed detailed SF-resolution of adaptation effects near the adapting SF while also making it possible to search for SF adaptation effects several octaves removed from the adapting SF. METHOD
Participants
Data were collected from 8 adult observers from the university community, ages 19-45 yr, with no overt strabismus, and normal corrected acuity. Observers wore their best refractive corrections for the viewing distance used. Design
Steady state VEPs were obtained under two binocular viewing conditions-after preexposure to: (a) a stationary adapting sinusoidal grating of 6 c/deg (n = 8 participants) or 4 c/deg (n = 3 participants)--the adapted condition; and (b) a uniform field matching the adapting grating in space-averaged luminance-the nonadapted condition. There were 10 trials per condition. Stationary gratings were used as adapting stimuli in order to avoid adaptation of movement processing mechanisms which may be associated with adaptation to contrast reversing gratings (Kulikowski, 1977). However, Bowker and Tulunay-Keesey (1983) have shown that psychophysical SF adaptation effects are insensitive to the temporal characteristics of the adapting grating for SFs of 3.0 c/deg and higher. Stimulus presentation
Gratings were vertically-generated on a Sony Model PVM-122 video monitor with 1000 lines of horizontal resolution, rotated 90 deg so that all gratings were horizontal. Horizontal gratings were used because persons with horizontal fixation disparity do not show VEP amplitude attenuation with vertical adapting and test gratings (Suter, Suter, Armstrong, Powers & Bass, 1990a). Each grating subtended 11 by 11 deg at the viewing distance of 95 cm with space averaged luminance of 38 cd/m2. Adapting sinusoidal gratings were presented at 4O%C, test gratings at lO%C. Pilot studies indicated that VEP adaptation effects are not obtained consistently at test grating contrasts above
lO%C, which is consistent with other published research (e.g. Mecacci & Spinelli, 1976). Depending upon the condition, observers viewed either the blank screen or grating for 5 min initially, and then for 30 set between the lo-set trials. Accurate accommodation and fixation were assisted by a small fixation target in the center of the screen. Participants were instructed and reminded to keep fixation moving one stripe-width above and below the fixation target during these periods in order to avoid formation of grating afterimages. During each trial, test SFs were swept from 2-3 octaves below, to approx. 1 octave above the adapting SF, increasing in equal log SF steps every 500 msec. The purpose of using the swept-SF technique was to obtain VEP data for many SFs under constant state conditions. In addition, it was reasoned that by presenting each SF only briefly during test trials, possible SF adaptation effects during VEP testing would be minimized. Test gratings were counterphase reversed at 7.5 Hz. This reversal rate yields a robust second harmonic EEG response in adults (Allen, Norcia & Tyler, 1986; Suter, Suter & Deegan, 1990b) in a region of the EEG temporal frequency spectrum that is uncontaminated by EEG alpha. Since there are spatio-temporal interactions in the steady-state VEP (Regan, 1983; Tyler, Apkarian & Nakayama, 1978) it should be noted that the present results may not hold for extremely slow or fast stimulation rates. VEP recording and pre -analysis
EEG was recorded from a bipolar montage of gold cup electrodes placed 3 cm above the inion and 6 cm laterally to the right, grounded over the left ear. This placement optimizes recording of pattern-evoked steady-state VEPs to gratings (Tyler & Apkarian, 1985). The raw EEG was monitored on an oscilloscope during all trials in order to detect and repeat trials containing gross deviations in EEG amplitude or temporal frequency which were considered artifacts. Altogether, 7% of test trials were repeated because of EEG artifacts, equipment failure, or EEG amplitude beyond the digitization limit. After the EEG had been pre-processed by a Coulbourn S75-01 amplifier, evoked responses were extracted from noise by a specialized microprocessor-based steady-state VEP system (Norcia & Tyler, 1985). Following 8-bit digitization at 200 HZ, a discrete Fourier transform (DFT) algorithm computed the EEG amplitude
VEP attenuation and enhancement
m
II“3;;.
20
3 % I. ,8
5
1169
L
,.(‘...__..___... .
.._....
~_..__._....__..____~_.._.
. .. .
1.2
ifi &j 0.8 I
0.8 1.0
3.0
1.5
Spatial
8.0
1.0
12.0
Frequency (&kg)
and phase angle at the second harmonic (15 Hz) of the stimulus frequency. The second harmonic was analyzed because it is the primary component of the pattern-reversal steady-state VEP at the moderate pattern reversal frequency used in this study (Regan, 1989). A moving I-see DFT analysis window advanced in 500-msec steps so that the data in each 1-set window overlapped 50% with the previous one, resulting in steady-state VEP amplitude and phase values for every 500-msec data bin. A 50% cosine taper was applied (Harris, 1978) yielding a Tukey window with a - 3 dB bandwidth of 1. I Hz for the 1-set window. RESULTS
Steady-state VEP amplitudes were scalar averaged for each individual at every SF presented in the adapted and nonadapted conditions. Figure 1 shows mean VEP amplitudes across the test SFs for the nonadapted and adapted conditions for one observer adapted to 6 e/deg. As is typical of adult observers, nonadapted VEP amplitude peaks in the region of 4-6c/deg and then decreases linearly with increasing SF. Adapted VEP amplitude shows a region of attenuated responding about 1.0 octave in width (all reported bandwidths are full 1
T
I. 1.0
I
1.5
3.0
1
4.0
8.0
120
Spatial Frequency (c/de&
Fig. 1. Mean adapted (0) and nonadapted (0) VEP amplitudes across test grating SF for one observer adapted to 6.0c/deg, indicated by arrow. Error bars are f 1 SEM.
1.31
20
I
I 6.0
12.0
Spatial Frequency (e/deg) Fig. 2. Grouped data showing mean adapted/nonadapted VEP amplitude ratios across test grating SF for observers (n = 8) adapted to 6.0 c/deg, indicated by arrow. Error bars are f 1 SEM.
Fig. 3. Grouped data showing mean adapted/nonadapted VEP amplitude ratios across test grating SF for observers (a = 3) adapted to 4.Oc/deg, indicated by arrow. Mean SEM = 0.135.
width/half-amplitude) near the adapting SF. A region of VEP amplitude enhancement is displaced about 1J-2.0 octaves below the adapting SF. There is no symmetrical enhancement at higher SFs. The remaining data are presented as adapted/nonadapted VEP amplitude ratios for ease of communication. Local ratios less than 1.0 indicate attenuation of VEP amplitude by SF adaptation, while ratios greater than 1.0 indicate enhan~ment of VEP amplitude by SF adaptation. Figure 2 shows the grouped data for the observers adapted to 6c/deg. VEP amplitude was attenuated by adaptation across about 1.5 octaves near the adapting SF. The maximum attenuation effect was about 40% of the nonadapted VEP amplitude. Attenuation was maximum about 0.5 octaves below the adapting SF. There was a region of VEP enhancement displaced about 2.0 octaves below the adapting SF; maximum VEP enhancement was about 20%, about half the maximum VEP attenuation. The grouped data for the observers adapted to 4c/deg are shown in Fig. 3. The adaptation effects at 4 c/deg parallel the 6 c/deg data with comparable regions of VEP attenuation and enhancement. One noticeable difference is that the maximum VEP amplitude attenuaton for 4c/deg adaptation is not shifted below the adapting SF. Results for every observer are presented in Fig. 4 for 6 c/deg adaptation and Fig. 5 for 4c/deg adaptation. Inspection of these data reveal that 10 of the I1 individual sessions showed the VEP enhancement effect in the SF region 1.0-2.0 octaves below the adapting SF (P < 0.01). Corresponding enhancement of VEP amplitude at SFs above the adapting SF was not obtained consistently. Two ANOVAs were conducted, one on the grouped data at each adapting SF using
Sm
Smut et al. 1.3.
-
ngaLwsrs+vae$s
-
LwmmphbFS*rUp
7.0
Spatial Frequency (cldeg) Fig. 6. Mean adapted/nona&pt~ VEP amplitude ratios across test grating SF for one occiput adapted to 7 cjdeg, with test SFs swept from low to high SF (0) vs from high to low SF (0). Adapting SF indicated by arrow.
Spatial Frequency (c/deg)
Fig. 4. Mean adaptedinonadapted VEP amplitude ratios across test grating SF for individual observers adapted to 6cjdeg. Adapting SF indicated by arrow.
the following factors: condition (adapted/ nonadapted), trials (IO/condition), test SF (19/trial) and observers (8 for 6 c/deg adaptation; 3 for 4c/deg adaptation). SF-tuned adaptation effects were confirmed by significant condition by test SF interactions for 1296) = 24.48, adaptation, 6 c/deg F(l8, P < 0.001, and 4 c/deg adaptation, F( 18,486) = 3.55, P
evaluate each treatment effect. This made it possible to examine effects involving observers, or individual differences, in comparison to the pooled trials effect. The question asked here was whether the differences among observers were greater than would be expected from the differences among trials within observers. Consistent indi~dual differences in overall VEP amplitude were obtained within both adaptation groups, F(7,72) = 173.15, P -e0.001 for 6 c/deg adaptation, and F(2, 27) = 23.91, P < 0.001 for 4 cjdeg adaptation. The magnitude of SF adaptation effects differed among observers within both adaptation groups as indicated by a condition by observers interaction, F(7, 72) = 65.47, P < 0.001 for 6c/deg adaptation and F(2, 27) = 6.42, P < 0.01for 4 c/deg adaptation. Finally, the spatial location of SF adaptation effects differed among individuals adapted to Gc/deg as shown by a condition by SF by observers interaction, F( 126, 1296) = 7.15, P < 0.001. The VEP suppression and enhancement effects were not dependent on the order of presentation of the SFs, that is, there was no hysteresis. As shown in Fig. 6, VEP adaptation effects were similar for low to high and high to low SF sweeps for an individual subject studied in both conditions.
DISCUSSION
*0
2.0
4.0
8.0
120
Spatial Frequency (c!deg)
Fig. 5. Mean adapted/nonadapted VEP amplitude ratios across test grating SF for individual participants adapted to 4c/deg. Adapting SF indicated by arrow.
There were three major findings: (a) VEP amplitude was attenuated by SF adaptation in a region near the adapting SF; (b) the maximum VEP attenuation effect for 6c/deg adaptation was not found at the adapting SF, but was displaced lower by about 0.5 octaves; and (c) there was a “remote VEP facilitation” effect, a region of increased VEP amplitude following SF adaptation located about 2.0 octaves below the
VEP attenuation and enhancement
adapting SF. These are discused in the following sections. VEP attenuation near the adapting SF Steady-state VEP amplitude was attenuated within an SF region of about 1.0 octave near the adapting SF. Amplitude attenuation averaged about 30% at the adapting SF, but was greater than 50% for several subjects. These adaptation effects are comparable in magnitude and bandwidth to the elevations in psychophysical contrast sensitivity (e.g. Blakemore & Campbell, 1969; Williams & Wilson, 1983) and VEP attenuation for steadystate (Blakemore & Campbell, 1969; Mecacci & Spinelli, 1976) and transient VEPs (Manahilov & Vassilev, 1986; Smith & Jeffreys, 1978) that have been reported in earlier studies. Note that the bandwidth of VEP attenuation following SF adaptation is not necessarily equivalent to the bandwidth of an underlying SF-tuned mechanism or channel (Dealy & Tolhurst, 1974) since, as discussed below, a mechanism’s tuning may be determined by both its excitatory tuning and tuning of inhibitory effects from mechanisms tuned to other SFs. Although VEP attenuation near the adapting SF is a substantial and consistent effect under the conditions used here, the effect seems to depend upon several critical variables: (a) high contrast adapting gratings (40% in the present study)-perhaps related to this, Legge and Foley (1980), using low contrast masking gratings, found lowering of contrast thresholds at the masking SF; (b) low contrast test gratings (10% in the present study)-Bach, Greenlee and Btihler (1988), using high contrast test gratings, found enhancement of VEP amplitude at the adapting SF; (c) relatively low adapting SF (4 and 6 c/deg here)-with low contrast test gratings, steadystate VEP amplitude descends to its noise level in EEG by 10-15 c/deg; (d) not collecting nonadapted VEP data for an extended period of time following SF adaptation-adaptation effects may persist for several hours (e.g. Mecacci & Spinelli, 1976); and (e) using monocular presentations, horizontal gratings, or screening for fixation disparity-persons with horizontal fixation disparity tend not to show VEP amplitude attenuation after exposure to vertical adapting gratings (Suter et al., 1990a).
Maximum VEP attenuation adapting SF
1171
displaced from
The maximum VEP amplitude attenuation for 6 c/deg SF adaptation was displaced about 0.5 octaves from the adapting SF. Analogous psychophysical effects have been reported. For example, Williams, Wilson and Cowan (1982) obtained maximum depression in psychophysical contrast sensitivity about 0.5 octaves above their adapting SFs of 1.4 and 2.0 c/deg. Wilson, McFarlane and Phillips (1983) studied the effects of high contrast sinusoidal masking gratings on psychophysical thresholds for difference-of-Gaussian test stimuli, resembling 1.5 cycles of sinusoidal gratings, of various SF. Small displacements in the maximum threshold elevation were obtained for masking gratings ranging from 1.O to 11.3 c/deg. Nearby masking frequencies (e.g. 1.0 and 1.4, 1.4 and 2.0, 8.0 and 11.3 c/deg) produced nearly identical threshold elevation curves. For example, 8.0 and 11.3 c/deg masks produced peak threshold elevations at 8.0 c/deg, while 1.4 and 2.0 c/deg masks both had peak effects at 2.0 c/deg. It was suggested that the identical curves were caused by a single SF-tuned neural mechanism having detected both test stimuli within a pair. For narrow, 0.75 deg masking and test fields, the psychophysical threshold for a 2.0 c/deg grating was elevated greatest for high contrast grating masks of 2.8 and 4.0c/deg (Legge & Foley, 1980). As others have observed from psychophysical displacements (Williams et al., 1982), the effects just reviewed are quite consistent with models of spatial vision that postulate a small number of SF-tuned mechanisms underlying the range of SFs to which the human visual system is sensitive. Maximum attenuation should occur at the adapting SF only when the adapting SF happens to correspond to the peak sensitivity of one of the underlying SF-tuned mechanisms. Otherwise, maximum attenuation would be displaced. Displaced maximum attenuation specifically argues against a continuum, or indefinite number of SF-tuned mechanisms. If there were an indefinite number of SF-tuned mechanisms then there would be a neural mechanism tuned to every possible adapting SF. The maximum attenuation effect should never be displaced. The present electrophysiological data support the hypothesis of a limited number of neural subunits tuned to different center SFs, such as the six subunits of the modified line
1172
Smw Sum
element model (Wilson BEGelb, 1984), or the four subunits of Watt and Morgan’s (1985) MIRAGE model, rather than an indefinite number of SF-tuned mechanisms. Maximum response attenuation following SF adaptation has often been reported at the adapting SF in previous psychophysical (e.g. Blakemore & Campbell, 1969) and VEP (e.g. Mecacci & Spinelli, 1976) studies. There are at least three possible explanations. First, adapting and/or test contrasts higher than those used in the present study (4O%C and lO%C, respectively) may stimulate adjacent subunits tuned to different SFs which would tend to obscure the contribution of the subunit whose center SF is closest to the adapting SF. A second possible explanation is based on the finding that SF tuning varies with eccentricity in the visual field (Swanson dc Wilson, 1985). As field size increases, subunits with differing tuning characteristics would make an increasing contribution to the VEP. The responses of these mechanisms could obscure an otherwise orderly relationship between SF adaptation effects and SF for the subunits supporting central vision. Third, in early studies, the intention in using the SF adaptation paradigm was primarily to demonstrate the presence of SF-tuned mechanisms rather than the properties of the proposed mechanisms. As a consequence, correspondences of adaptation effects with adapting SFs might have been unintentionally emphasized over small deviations. For example, the maximum VEP attenuation following high contrast 4c/deg SF adaptation appears to be shifted downward by about 0.5 octaves in one VEP SF adaptation study (Mecacci & Spinelli, 1976, Fig. 2, p. 478), but this displacement is not noted in text. Related to this last observation, in studies in which extensive data are collected from only several or a single observer, it is possible that observers who yield data with small displacements of VEP adaptation effects are rejected as observers under the assumption that the deviations represent noisy data. Remote VEP amplitude facilitation by SF adap tation There are several psychophysical findings that resemble the remote VEP amplitude facilitation reported here. Tolhurst and Barfield (1978) studied the effects of 4.25 c/deg high contrast adapting gratings on contrast sensitivity. Sensitivity to SFs near the adapting gratings was diminished, while sensitivity was enhanced at
et al.
1.5 octaves above the adapting grating. Adaptation effects were not examined below the adapting SF. Similarly, De Valois (1977)’ measured contrast sensitivity following adaptation to different high contrast gratings ranging in SF from 0.84 to 13.45 c/deg. In general, contrast sensitivity was diminished near the adapting SF, but enhanced in SF regions about 2.0-4.0 octaves from the adapting SF. Williams and Wilson (1983) studied contrast sensitivities to gratings following adaptation to a high contrast 4.0 c/deg grating. Sensitivity was diminished across about 1.Ooctaves near the adapting SF. There was significant enhancement of sensitivity about 1.5 octaves below the adapting SF. Remote enhancement of contrast sensitivity has also been reported using difference-of-Gaussian stimuli at 2 deg eccentricity following sinusoidal grating adaptation at 5.7 and 8.0c/deg (Williams et al., 1982). These psychophysical findings have been interpreted to indicate that the SF-tuned mechanisms processing patterned stimuli are not completely independent, but may be coinhibitory. Specifically, mechanisms with neighboring center SFs may inhibit one another to an extent that is proportional to their excitatory output. If this were the case, SF adaptation of one mechanism (M, ), reducing its excitatory output, would produce a “release from inhibition” effect on another mechanism (Mr), temporarily increasing the responsivity of M2 (De Valois, 1977; Tolhurst, 1972). Therefore, SF adaptation near M, should lower contrast thresholds and increase VEP amplitude to SFs near MZ. This effect requires that while SFs are being presented near M2, M, is simultaneously responding below its normal level. According to this reasoning, unlike the SF adaptation paradigm, the SF masking paradigm would not release adjacent mechanisms from inhibition. In masking, the added high-contrast grating is simultaneously present with the test grating, but at a different temporal frequency. The response spectrum of a neural mechanism, M,, sensitive to the masking grating, will show increased activity at the fundamental and harmonics of the temporal frequency of the masking grating, and at multiple discrete cross-modulation frequencies of the test and mask temporal frequencies to the extent that M, is also sensitive to the test grating SF (Regan & Regan, 1988). While SF adaptation causes M, to respond “less” during testing, masking causes M, to respond “differently”. In masking, since M, is not
VEP attenuation and enhancement
itsnormal ievel, it would not release M, from inhibition. Electrophysiological evidence appears to be consistent with this reasoning, Fiorentini et al. (1983) and Regan (1983) searched for, but found no facilitation ei%cts on steady-state VEP amplitudes using masking paradigms that produced marked VEP amplitude attenuation when test and masking gratings were similar in SF. However, there is at least one report of increased psychophysical sensitivity at SFs distant from a high contrast masking grating (Tolhurst & Barfield, 1978). Interaction between SF-tuned mechanisms is supported by psychophysical studies exploring the effects of concurrent adaptation to several high contrast gratings differing in SF. The net e&M on contrast sensitivity varies as a function of the difference between the two adapting SFs. Within a range of about 1.0 octave, SF adaptation effects are additive; however, for adapting gratings differing in SF by I&2.0 octaves, the effects are subtractive as would be predicted if the two gratings were stimulating coinhibitory SF-tuned neural mechanisms (Greenlee & Magnussen, 1988). Tolhurst (1972) has reported similar effects of composite adapting stimuli on psychophysical thresholds. The appearance of illusory gratings higher and lower than the adapting SF following SF adaptation is also suggestive of disinhibition following SF adaptation (Georgeson, 1976b, 1980). The remote VEP facilitation effects reported here may be objective evidence for interacting SF-tuned mechanisms in humans previously ~y~thes~d on the basis of the psychophysical findings just reviewed. Accumulated psychophysical and electrophysiological evidence suggests a half-amplitude bandwidth of about 1.O octave for SF-tuned mechanisms (De Valois h De Valois, 1988). Since the areas of VEP facilitation are displaced by about 2.0 octaves from the adapting SF they would seem to represent an interaction between mechanisms. Additional electrophysiological evidence is provided by single cell studies in cat striate cortex (De Valois & Tootell, 1983). Most cells investigated were strongly inhibited by a narrowly tuned range of SFs from 1.0 to 2.0 octaves removed from their optimum SF. It was reasoned that the inhibitory effects represented inputs from interacting SF-tuned cehs, and that these interactions sharpen SF sefectivity of individual cells. Remote enhancement of contrast sensitivity following SF adaptation has been reconciled responding below
1173
with the modified line element model of spatial vision (Wilson $ Gelb, 1984) by allowing a gain increase in an adjacent SF-tuned mechanism following SF adaptation (Williams & Wilson, 1983; Williams et al., 1982), but without inco~rating coinhibition between subunits. If, as suggested above, this effect is a result of disinhibition from a neighboring SFtuned neural mechanism, the line element model might be modified to incorporate interaction effects between mechanisms, specifically tonic coinhibition. A second interpretation of the remote VEP facilitation effect is based on possible interactions between transient (movement) and sustained (pattern) neural mechanisms (Ikeda & Wright, 1974). In the present study SF adaptation led to increased VEP amplitude at relatively low SFs. Both transient and sustained mechanisms contribute to the VEP response to contrast-reversing gratings with the transient component pr~o~nating at SFs below about 3 c/deg (Kulikowski, 1977). If sustained mechanisms are selectively adapted by a stationary grating (Kulikowski, 19771, and the two mechanism types are in a coinhibitory relationship (Georgeson, 1976a), then stationary grating adaptation at moderate SFs should disinhibit transient mechanisms, resulting in increased VEP amplitude at low SFs. The transient-sustained coinhibition hypothesis cannot explain remote facilitation above the adapting SF, since effects of any transient mechanism d&inhibition by stationary SF adaptation would be confined to low SFs. Remote facilitation above 4.25 c/deg (Tolhurst & Barfield, 1978) and 1.19 c/deg SF adaptation has been reported for psychophysical data, but the effect has not been shown for VEP adaptation. Remote VEP facilitation in SF regions above the adapting SF was not found consistently in the present study. Although this may argue against SF-tuned coinhibition, it is also consistent with electrophysiolo~~l evidence suggesti~ a possible asymmetry in inhibitory effects between SF-tuned mechanisms. For cat striate cortex there are stronger inhibitory effects of high SFs on lower-tuned cells than for low SFs on higher-tuned cells, a phenomenon that would selectively sharpen the SF tuning of mechanisms tuned to higher SFs (De Valois & Tootell, 1983). The present findings must be viewed cautiously since our test SFs did not extend to 2.0 octaves above the adapting SFs, the region in which De Valois (1977) found
STEW Surea et al.
1174
psychophysical facilitation. Steady-state VEP amplitude approaches EEG noise by about 12 c/deg for the lO%C gratings used as test stimuli in the present study. An alternative approach would be to study the effects of SF adaptation on higher SFs by using contrast sweeps at fixed SFs (e.g. Allen et al., 1986) to establish VEP contrast thresholds. Acknowledgements-We gratefully acknowledge the research assistance of Michael S. Lutge and Brenda L. Bass, the library support of David Kosakowski and Loma Frost, and the technical support of Tom Gsbom and Michael Fleming. Some of these data were reported at the Meeting of the Association for Research in Vision and Ophthalmology,
May 1989, Sarasota, Florida. This research was supported by the California State University Bakersfield Foundation.
REFERENCES Allen, D., Norcia, A. M. & Tyler, C. W. (1986). Comparative study of electrophysiological and psychophysical measurement of the contrast sensitivity function in humans. American Journal of Optometry and Physiological Optics, 63, 442449.
Bach, M., Greenlee, M. W. & Biihler, B. (1988). Contrast adaptation can increase visually evoked potential amplitude. Clinical Vision Sciences, 3, 185-194. Blakemore, C. & Campbell, F. W. (1969). On the existence of neurones in the human visual system selectively sensitive to the orientation and size of retinal images. Journal of Physiology, London, 203, 237-260.
Blakemore, C., Muncey, J. P. J. 8r Ridley, R. M. (1971). Perceptual fading of a stabilized cortical image. Nature, London, 233, 204-205.
Bowker, D. 0. & Tulunay-Keesey, U. (1983). Sensitivity to countermodulating gratings following spatiotemporal adaptation. Journal of the Optical Society of America, 73, 427435.
Campbell, F. W. & Robson, J. G. (1968). Application of Fourier analysis to the visibility of gratings. Journal of Physiology, London, 197, 551-566.
Dealy, R. S. & Tolhurst, D. J. (1974). Is spatial adaptation an after effect of prolonged inhibition? Journal of Physiology, London, 241, 261-270.
De Valois, K. K. (1977). Spatial frequency adaptation can enhance contrast sensitivity. Vision Research, 17, 1057-1065.
De Valois, R. L. & De Valois, K. K. (1988). Spatial oision. New York: Oxford University Press. De Valois, K. K. & Tootell, R. B. H. (1983). Spatialfrequency-specific inhibition in cat striate cortex cells. Journal of Physiology, 336, 359-376.
Fiorentini, A., Pirchio, M. & Spinelli, D. (1983). Electrophysiological evidence for spatial frequency selective mechanisms in adults and infants. Vision Research, 23, 119-127.
Georgeson, M. A. (1976a). Antagonism between channels for pattern and movement in human vision. Nature, London, 259, 413415.
Georgeson, M. A. (1976b). Psychophysical hallucinations of orientation and spatial frequency. Perception, 5, 99-111.
Gcorgeson, M. A. (1980). The perceived spatial frequency, contrast and orientation of illusory gratings. Perception, 9, 695-712. Greenlee, M. W. & Magnussen,
among adapted Harris, F. analysis
S. (1988). Interactions spatial frequency and orientation channels concurrently. Vision Research, 28, 1303-1310. J. (1978). On the use of windows for harmonic with the discrete Fourier transform. Proceedings
of the IEEE, 66, 51-83. Ikeda, H. C Wright, M. J. (1974). Evidence for “sustained” and “transient” neurones in the cat’s visual cortex. Vision Research, 14, 133-136.
Klein, S. A. & Levi, D. M. (1985). Hyperacuity thresholds of 1 set: Theoretical predictions and empirical validation. Journal of the Optical Society of America, 2, 1170-I 190. Kulikowski, J. (1977). Separation of occipital potentials related to the detection of pattern and movement. In Desmedt, J. E. (Ed.), Visual evoked potentials in man: New developments (pp. 184196). Oxford: Clarendon Press. Legge, G. E. & Foley, J. M. (1980). Contrast masking of human vision. Journal of the Optical Society of America, 70, 1458-1471.
Maffei, L., Fiorentini, A. & Bisti, S. (1973). Neural correlate of perceptual adaptation to gratings. Science, 182, 10361038. Manahilov, V. & Vassilev, A. (1986). The effect of standing contrast and pattern adaptation on the human visually evoked potential. Electroencephalography and Clinical Neurophysiology, 64, 536-545.
Mecacci, L. & Spinelli, D. (1976). The effects of spatial frequency adaptation on human evoked potentials. Vision Research, 16, 477-479.
Norcia, A. M. & Tyler, C. W. (1985). Spatial frequency sweep VEP: Visual acuity during the first year of life. Vision Research, 25, 1399-1408. Regan, D. (1982). Visual information channeling in normal and disordered vision. Psychological Review, 89, 407-444.
Regan, D. (1983). Spatial frequency mechanisms in human vision investigated by evoked potential recording. Vision Research, 23, 1401-1407.
Regan, D. (1989). Human brain electrophysiology. New York: Elsevier. Regan, M. P. & Regan, D. (1988). A frequency domain technique for characterizing nonlinearities in biological systems. Journal of Theoretical Biology, 133, 293-3 17. Sachs, M. B., Nachmias, J. & Robson, J. G. (1971). Spatial-frequency channels in human vision. Journal of the Optical Society of America, 61, 11761186. Sclar, G., Lennie, P. & DePriest, D. D. (1989). Contrast adaptation in striate cortex of macaque. Vision Research, 29, 747-755.
Smith, A. T. & Jeffreys, D. A. (1978). Size and orientation specificity of transient visual evoked potentials in man. Vision Research, 18, 651-655.
Suter, P. S., Suter, S., Armstrong, C. A., Powers, J. C. & Bass, B. L. (1990a). Binocular VEP amplitude and lateral associated phoria. Investigative Ophthalmology and Visual Science, 31(4/Suppl.),
92.
Suter, S., Suter, P. S. % Deegan, J. F. II. (1990b). Steadystate VEP phase stability and acuity in adults and infants. Clinical Vision Sciences, 5, 71-80.
Swanson, W. H. & Wilson, H. R. (1985). Eccentricity dependence of contrast matching and oblique masking. Vision Research, 25, 1285-1295.
VI% att~ua~on Tolhurst, D. J. (1972). Adaptation to square-wave gratings: Inhibition between spatial frequency channels in the human visual system. Journal ofPhysiology, London, 226, 23 l-248. Tolhurst, D. J. & Barfield, L. P. (1978). Interactions between spatial frequency channels. Vision Research, lg, 951-958. Tyler, C. W. & Apkarian, P. A. (1985). I%cts of contrast, orientation and binocularity in the pattern evoked potential. Vlrion Research, 25, 755-766. Tyler, C. W., Apkarian, P. & Nakayama, K. (1978). Multiple spatial-frequency tuning of electrical responses from human visual cortex. ~~pe~~~a~ &a& Research, 33, 535-550. Watson, A. B. (1987). The cortex transform: Rapid computation of simufated neural images. Computer Vi&n, Graphics and Image Processing, 39, 31 l-327.
and enhancement
1175
Watt, R. J. & Morgan, M. J. (1985) A theory of the primitive spatial code in human vision. Vision Research, 25, 1661-1674. Williams, D. W. St Wilson, H. R. (1983). Spatial-frequency adaptation affects spatial-probability summation. Journal of the Optical Socieiy of America, 73, 13674371. Williams, D. W., Wilson, H. R. & Cowan, J. D. (1982). Localized effects of spatial-frequency adaptation. Journal of the Optical Sociefy of America, 72, 818--W. Wilson, H. R. & Gelb, D. J. (1984). Modified line-element theory for spatial-frequency and width dis&mination. .Journal oj r&e Opfieal S&eiy of America A, 1, 124-13I. Wilson, H. R,, McFarlane, D. K. Br PM&p, G. C. (1983). Spatial frequency tuning of orientation selective units estimated by oblique masking. Y&&r Research, 23, 873-882.
0042-6989/91 S3.00 + 0.00 Copyright 0 1991 Pcrgamon Prm pk
SPATIAL-FREQUENCY-TUNED ATTENUATION AND ENHANCEMENT OF THE STEADY-STATE VEP BY GRATING ADAPTATION S-I-EVE SUTER,’COLINA. ARMSTR~NG,~ PENELOPE S. SUTER’ and JUS~INA C. Powsas3 ‘Department of Psychology, California State University, Bakersfield, CA 93311, ‘San Diego State University/University of California San Diego Joint Doctoral Program in Clinical Psychology, San Diego, CA 92182 and Department of Psychology, University of California, Riverside, CA 92521, U.S.A. (Received 14 September 1989; in revised form
7
August 1990)
Abstract-Steady-state visual evoked potentials (VEPs) were recorded from adults using lO%C fast spatial frequency (SF) sweeps of horizontal gratings under two conditions: (a) after exposure to a 4O%C grating of 6 or 4 c/deg, and (b) after exposure to a blank screen equalling the adapting gratings in space-averaged luminance. SF adaptation attenuated VEP amplitude near the adapting SF, but maximum attenuation was displaced from the adapting SF for 6 c/deg adaptation. Small displacements in maximum attenuation would be expected if underlying neural subunits are tuned to a small number of different center SFs. In addition, SF adaptation caused amplitude enhancement 1.O-2.0 octaves below the adapting SF, providing electrophysiological evidence in humans for coinhibitory relationships among neural mechanisms that have been postulated on the basis of analogous psychophysical findings. The results are consistent with coinhibition between SF-tuned subunits and between transient and sustained mechanisms. VEP
Adaptation
Spatial frequency
INTRODUCTION
Several lines of evidence suggest that the visual world of humans and other mammals may be constructed from spatial frequency (SF) information extracted from the retinal image by a set of SF-tuned channels (Campbell & Robson, 1968). These neural mechanisms are thought to function as bandpass filters differing in center SF. The overall contrast sensitivity function is the envelope of the underlying channel sensitivities. This view is primarily supported by psychophysical studies using SF adaptation, masking and subthreshold summation paradigms (reviewed by De Valois & De Valois, 1988; Regan, 1982). Effects that are limited to the region of the SF spectrum near the adapting, masking or summation stimuli are suggestive of SF-tuned subunits. Psychophysical studies indicate that: (a) SF adaptation diminishes contrast sensitivity (e.g. Blakemore & Campbell, 1969) and the apparent contrast of suprathreshold stimuli (Blakemore, Muncey & Ridley, 1971) over a region of about 1.0 octave (half-amplitude bandwidth) near the adapting SF stimulus; (b) masking effects are confined to a similarly narrow range of SF differences between mask-
ing and test stimuli (e.g. Legge & Foley, 1980); and (c) subthreshold summation effects are usually seen only when the subthreshold stimuli are within a narrow SF region (e.g. Sachs, Nachmias & Robson, 1971). Based on such evidence, several quantitative models have been proposed that incorporate sets of SF-tuned neural subunits with particular performance characteristics (e.g. Klein & Levi, 1985; Legge & Foley, 1980; Watson, 1987; Watt & Morgan, 1985; Wilson & Gelb, 1984). Electrophysiological evidence bearing on the hypothesized neural mechanisms is quite limited, but critical since the psychophysical data themself cannot provide direct evidence about neural mechanisms (Regan, 1983). However, SF-tuned effects of SF adaptation have been noted for visual evoked potential (VEP) amplitudes in humans (e.g. Mecacci & Spinelli, 1976) and single cortical cells of monkeys (e.g. Sclar, Lennie & DePriest, 1989) and cats (e.g. MafIei, Fiorentini & Bisti, 1973). Masking paradigms yield SF-tuned VEP attenuation (e.g. Fiorentini, Pirchio & Spinelli, 1983). In summary, psychophysical and neurophysiological findings are consistent in a general sense as they relate to the multiple SF-tuned channel hypothesis.
1167
SlwE SumR et al.
1168
In the present study a VEP fast-SF sweep technique was used to explore for adaptation effects at many SFs in the absence of state changes. This allowed detailed SF-resolution of adaptation effects near the adapting SF while also making it possible to search for SF adaptation effects several octaves removed from the adapting SF. METHOD
Participants
Data were collected from 8 adult observers from the university community, ages 19-45 yr, with no overt strabismus, and normal corrected acuity. Observers wore their best refractive corrections for the viewing distance used. Design
Steady state VEPs were obtained under two binocular viewing conditions-after preexposure to: (a) a stationary adapting sinusoidal grating of 6 c/deg (n = 8 participants) or 4 c/deg (n = 3 participants)--the adapted condition; and (b) a uniform field matching the adapting grating in space-averaged luminance-the nonadapted condition. There were 10 trials per condition. Stationary gratings were used as adapting stimuli in order to avoid adaptation of movement processing mechanisms which may be associated with adaptation to contrast reversing gratings (Kulikowski, 1977). However, Bowker and Tulunay-Keesey (1983) have shown that psychophysical SF adaptation effects are insensitive to the temporal characteristics of the adapting grating for SFs of 3.0 c/deg and higher. Stimulus presentation
Gratings were vertically-generated on a Sony Model PVM-122 video monitor with 1000 lines of horizontal resolution, rotated 90 deg so that all gratings were horizontal. Horizontal gratings were used because persons with horizontal fixation disparity do not show VEP amplitude attenuation with vertical adapting and test gratings (Suter, Suter, Armstrong, Powers & Bass, 1990a). Each grating subtended 11 by 11 deg at the viewing distance of 95 cm with space averaged luminance of 38 cd/m2. Adapting sinusoidal gratings were presented at 4O%C, test gratings at lO%C. Pilot studies indicated that VEP adaptation effects are not obtained consistently at test grating contrasts above
lO%C, which is consistent with other published research (e.g. Mecacci & Spinelli, 1976). Depending upon the condition, observers viewed either the blank screen or grating for 5 min initially, and then for 30 set between the lo-set trials. Accurate accommodation and fixation were assisted by a small fixation target in the center of the screen. Participants were instructed and reminded to keep fixation moving one stripe-width above and below the fixation target during these periods in order to avoid formation of grating afterimages. During each trial, test SFs were swept from 2-3 octaves below, to approx. 1 octave above the adapting SF, increasing in equal log SF steps every 500 msec. The purpose of using the swept-SF technique was to obtain VEP data for many SFs under constant state conditions. In addition, it was reasoned that by presenting each SF only briefly during test trials, possible SF adaptation effects during VEP testing would be minimized. Test gratings were counterphase reversed at 7.5 Hz. This reversal rate yields a robust second harmonic EEG response in adults (Allen, Norcia & Tyler, 1986; Suter, Suter & Deegan, 1990b) in a region of the EEG temporal frequency spectrum that is uncontaminated by EEG alpha. Since there are spatio-temporal interactions in the steady-state VEP (Regan, 1983; Tyler, Apkarian & Nakayama, 1978) it should be noted that the present results may not hold for extremely slow or fast stimulation rates. VEP recording and pre -analysis
EEG was recorded from a bipolar montage of gold cup electrodes placed 3 cm above the inion and 6 cm laterally to the right, grounded over the left ear. This placement optimizes recording of pattern-evoked steady-state VEPs to gratings (Tyler & Apkarian, 1985). The raw EEG was monitored on an oscilloscope during all trials in order to detect and repeat trials containing gross deviations in EEG amplitude or temporal frequency which were considered artifacts. Altogether, 7% of test trials were repeated because of EEG artifacts, equipment failure, or EEG amplitude beyond the digitization limit. After the EEG had been pre-processed by a Coulbourn S75-01 amplifier, evoked responses were extracted from noise by a specialized microprocessor-based steady-state VEP system (Norcia & Tyler, 1985). Following 8-bit digitization at 200 HZ, a discrete Fourier transform (DFT) algorithm computed the EEG amplitude
VEP attenuation and enhancement
m
II“3;;.
20
3 % I. ,8
5
1169
L
,.(‘...__..___... .
.._....
~_..__._....__..____~_.._.
. .. .
1.2
ifi &j 0.8 I
0.8 1.0
3.0
1.5
Spatial
8.0
1.0
12.0
Frequency (&kg)
and phase angle at the second harmonic (15 Hz) of the stimulus frequency. The second harmonic was analyzed because it is the primary component of the pattern-reversal steady-state VEP at the moderate pattern reversal frequency used in this study (Regan, 1989). A moving I-see DFT analysis window advanced in 500-msec steps so that the data in each 1-set window overlapped 50% with the previous one, resulting in steady-state VEP amplitude and phase values for every 500-msec data bin. A 50% cosine taper was applied (Harris, 1978) yielding a Tukey window with a - 3 dB bandwidth of 1. I Hz for the 1-set window. RESULTS
Steady-state VEP amplitudes were scalar averaged for each individual at every SF presented in the adapted and nonadapted conditions. Figure 1 shows mean VEP amplitudes across the test SFs for the nonadapted and adapted conditions for one observer adapted to 6 e/deg. As is typical of adult observers, nonadapted VEP amplitude peaks in the region of 4-6c/deg and then decreases linearly with increasing SF. Adapted VEP amplitude shows a region of attenuated responding about 1.0 octave in width (all reported bandwidths are full 1
T
I. 1.0
I
1.5
3.0
1
4.0
8.0
120
Spatial Frequency (c/de&
Fig. 1. Mean adapted (0) and nonadapted (0) VEP amplitudes across test grating SF for one observer adapted to 6.0c/deg, indicated by arrow. Error bars are f 1 SEM.
1.31
20
I
I 6.0
12.0
Spatial Frequency (e/deg) Fig. 2. Grouped data showing mean adapted/nonadapted VEP amplitude ratios across test grating SF for observers (n = 8) adapted to 6.0 c/deg, indicated by arrow. Error bars are f 1 SEM.
Fig. 3. Grouped data showing mean adapted/nonadapted VEP amplitude ratios across test grating SF for observers (a = 3) adapted to 4.Oc/deg, indicated by arrow. Mean SEM = 0.135.
width/half-amplitude) near the adapting SF. A region of VEP amplitude enhancement is displaced about 1J-2.0 octaves below the adapting SF. There is no symmetrical enhancement at higher SFs. The remaining data are presented as adapted/nonadapted VEP amplitude ratios for ease of communication. Local ratios less than 1.0 indicate attenuation of VEP amplitude by SF adaptation, while ratios greater than 1.0 indicate enhan~ment of VEP amplitude by SF adaptation. Figure 2 shows the grouped data for the observers adapted to 6c/deg. VEP amplitude was attenuated by adaptation across about 1.5 octaves near the adapting SF. The maximum attenuation effect was about 40% of the nonadapted VEP amplitude. Attenuation was maximum about 0.5 octaves below the adapting SF. There was a region of VEP enhancement displaced about 2.0 octaves below the adapting SF; maximum VEP enhancement was about 20%, about half the maximum VEP attenuation. The grouped data for the observers adapted to 4c/deg are shown in Fig. 3. The adaptation effects at 4 c/deg parallel the 6 c/deg data with comparable regions of VEP attenuation and enhancement. One noticeable difference is that the maximum VEP amplitude attenuaton for 4c/deg adaptation is not shifted below the adapting SF. Results for every observer are presented in Fig. 4 for 6 c/deg adaptation and Fig. 5 for 4c/deg adaptation. Inspection of these data reveal that 10 of the I1 individual sessions showed the VEP enhancement effect in the SF region 1.0-2.0 octaves below the adapting SF (P < 0.01). Corresponding enhancement of VEP amplitude at SFs above the adapting SF was not obtained consistently. Two ANOVAs were conducted, one on the grouped data at each adapting SF using
Sm
Smut et al. 1.3.
-
ngaLwsrs+vae$s
-
LwmmphbFS*rUp
7.0
Spatial Frequency (cldeg) Fig. 6. Mean adapted/nona&pt~ VEP amplitude ratios across test grating SF for one occiput adapted to 7 cjdeg, with test SFs swept from low to high SF (0) vs from high to low SF (0). Adapting SF indicated by arrow.
Spatial Frequency (c/deg)
Fig. 4. Mean adaptedinonadapted VEP amplitude ratios across test grating SF for individual observers adapted to 6cjdeg. Adapting SF indicated by arrow.
the following factors: condition (adapted/ nonadapted), trials (IO/condition), test SF (19/trial) and observers (8 for 6 c/deg adaptation; 3 for 4c/deg adaptation). SF-tuned adaptation effects were confirmed by significant condition by test SF interactions for 1296) = 24.48, adaptation, 6 c/deg F(l8, P < 0.001, and 4 c/deg adaptation, F( 18,486) = 3.55, P
evaluate each treatment effect. This made it possible to examine effects involving observers, or individual differences, in comparison to the pooled trials effect. The question asked here was whether the differences among observers were greater than would be expected from the differences among trials within observers. Consistent indi~dual differences in overall VEP amplitude were obtained within both adaptation groups, F(7,72) = 173.15, P -e0.001 for 6 c/deg adaptation, and F(2, 27) = 23.91, P < 0.001 for 4 cjdeg adaptation. The magnitude of SF adaptation effects differed among observers within both adaptation groups as indicated by a condition by observers interaction, F(7, 72) = 65.47, P < 0.001 for 6c/deg adaptation and F(2, 27) = 6.42, P < 0.01for 4 c/deg adaptation. Finally, the spatial location of SF adaptation effects differed among individuals adapted to Gc/deg as shown by a condition by SF by observers interaction, F( 126, 1296) = 7.15, P < 0.001. The VEP suppression and enhancement effects were not dependent on the order of presentation of the SFs, that is, there was no hysteresis. As shown in Fig. 6, VEP adaptation effects were similar for low to high and high to low SF sweeps for an individual subject studied in both conditions.
DISCUSSION
*0
2.0
4.0
8.0
120
Spatial Frequency (c!deg)
Fig. 5. Mean adapted/nonadapted VEP amplitude ratios across test grating SF for individual participants adapted to 4c/deg. Adapting SF indicated by arrow.
There were three major findings: (a) VEP amplitude was attenuated by SF adaptation in a region near the adapting SF; (b) the maximum VEP attenuation effect for 6c/deg adaptation was not found at the adapting SF, but was displaced lower by about 0.5 octaves; and (c) there was a “remote VEP facilitation” effect, a region of increased VEP amplitude following SF adaptation located about 2.0 octaves below the
VEP attenuation and enhancement
adapting SF. These are discused in the following sections. VEP attenuation near the adapting SF Steady-state VEP amplitude was attenuated within an SF region of about 1.0 octave near the adapting SF. Amplitude attenuation averaged about 30% at the adapting SF, but was greater than 50% for several subjects. These adaptation effects are comparable in magnitude and bandwidth to the elevations in psychophysical contrast sensitivity (e.g. Blakemore & Campbell, 1969; Williams & Wilson, 1983) and VEP attenuation for steadystate (Blakemore & Campbell, 1969; Mecacci & Spinelli, 1976) and transient VEPs (Manahilov & Vassilev, 1986; Smith & Jeffreys, 1978) that have been reported in earlier studies. Note that the bandwidth of VEP attenuation following SF adaptation is not necessarily equivalent to the bandwidth of an underlying SF-tuned mechanism or channel (Dealy & Tolhurst, 1974) since, as discussed below, a mechanism’s tuning may be determined by both its excitatory tuning and tuning of inhibitory effects from mechanisms tuned to other SFs. Although VEP attenuation near the adapting SF is a substantial and consistent effect under the conditions used here, the effect seems to depend upon several critical variables: (a) high contrast adapting gratings (40% in the present study)-perhaps related to this, Legge and Foley (1980), using low contrast masking gratings, found lowering of contrast thresholds at the masking SF; (b) low contrast test gratings (10% in the present study)-Bach, Greenlee and Btihler (1988), using high contrast test gratings, found enhancement of VEP amplitude at the adapting SF; (c) relatively low adapting SF (4 and 6 c/deg here)-with low contrast test gratings, steadystate VEP amplitude descends to its noise level in EEG by 10-15 c/deg; (d) not collecting nonadapted VEP data for an extended period of time following SF adaptation-adaptation effects may persist for several hours (e.g. Mecacci & Spinelli, 1976); and (e) using monocular presentations, horizontal gratings, or screening for fixation disparity-persons with horizontal fixation disparity tend not to show VEP amplitude attenuation after exposure to vertical adapting gratings (Suter et al., 1990a).
Maximum VEP attenuation adapting SF
1171
displaced from
The maximum VEP amplitude attenuation for 6 c/deg SF adaptation was displaced about 0.5 octaves from the adapting SF. Analogous psychophysical effects have been reported. For example, Williams, Wilson and Cowan (1982) obtained maximum depression in psychophysical contrast sensitivity about 0.5 octaves above their adapting SFs of 1.4 and 2.0 c/deg. Wilson, McFarlane and Phillips (1983) studied the effects of high contrast sinusoidal masking gratings on psychophysical thresholds for difference-of-Gaussian test stimuli, resembling 1.5 cycles of sinusoidal gratings, of various SF. Small displacements in the maximum threshold elevation were obtained for masking gratings ranging from 1.O to 11.3 c/deg. Nearby masking frequencies (e.g. 1.0 and 1.4, 1.4 and 2.0, 8.0 and 11.3 c/deg) produced nearly identical threshold elevation curves. For example, 8.0 and 11.3 c/deg masks produced peak threshold elevations at 8.0 c/deg, while 1.4 and 2.0 c/deg masks both had peak effects at 2.0 c/deg. It was suggested that the identical curves were caused by a single SF-tuned neural mechanism having detected both test stimuli within a pair. For narrow, 0.75 deg masking and test fields, the psychophysical threshold for a 2.0 c/deg grating was elevated greatest for high contrast grating masks of 2.8 and 4.0c/deg (Legge & Foley, 1980). As others have observed from psychophysical displacements (Williams et al., 1982), the effects just reviewed are quite consistent with models of spatial vision that postulate a small number of SF-tuned mechanisms underlying the range of SFs to which the human visual system is sensitive. Maximum attenuation should occur at the adapting SF only when the adapting SF happens to correspond to the peak sensitivity of one of the underlying SF-tuned mechanisms. Otherwise, maximum attenuation would be displaced. Displaced maximum attenuation specifically argues against a continuum, or indefinite number of SF-tuned mechanisms. If there were an indefinite number of SF-tuned mechanisms then there would be a neural mechanism tuned to every possible adapting SF. The maximum attenuation effect should never be displaced. The present electrophysiological data support the hypothesis of a limited number of neural subunits tuned to different center SFs, such as the six subunits of the modified line
1172
Smw Sum
element model (Wilson BEGelb, 1984), or the four subunits of Watt and Morgan’s (1985) MIRAGE model, rather than an indefinite number of SF-tuned mechanisms. Maximum response attenuation following SF adaptation has often been reported at the adapting SF in previous psychophysical (e.g. Blakemore & Campbell, 1969) and VEP (e.g. Mecacci & Spinelli, 1976) studies. There are at least three possible explanations. First, adapting and/or test contrasts higher than those used in the present study (4O%C and lO%C, respectively) may stimulate adjacent subunits tuned to different SFs which would tend to obscure the contribution of the subunit whose center SF is closest to the adapting SF. A second possible explanation is based on the finding that SF tuning varies with eccentricity in the visual field (Swanson dc Wilson, 1985). As field size increases, subunits with differing tuning characteristics would make an increasing contribution to the VEP. The responses of these mechanisms could obscure an otherwise orderly relationship between SF adaptation effects and SF for the subunits supporting central vision. Third, in early studies, the intention in using the SF adaptation paradigm was primarily to demonstrate the presence of SF-tuned mechanisms rather than the properties of the proposed mechanisms. As a consequence, correspondences of adaptation effects with adapting SFs might have been unintentionally emphasized over small deviations. For example, the maximum VEP attenuation following high contrast 4c/deg SF adaptation appears to be shifted downward by about 0.5 octaves in one VEP SF adaptation study (Mecacci & Spinelli, 1976, Fig. 2, p. 478), but this displacement is not noted in text. Related to this last observation, in studies in which extensive data are collected from only several or a single observer, it is possible that observers who yield data with small displacements of VEP adaptation effects are rejected as observers under the assumption that the deviations represent noisy data. Remote VEP amplitude facilitation by SF adap tation There are several psychophysical findings that resemble the remote VEP amplitude facilitation reported here. Tolhurst and Barfield (1978) studied the effects of 4.25 c/deg high contrast adapting gratings on contrast sensitivity. Sensitivity to SFs near the adapting gratings was diminished, while sensitivity was enhanced at
et al.
1.5 octaves above the adapting grating. Adaptation effects were not examined below the adapting SF. Similarly, De Valois (1977)’ measured contrast sensitivity following adaptation to different high contrast gratings ranging in SF from 0.84 to 13.45 c/deg. In general, contrast sensitivity was diminished near the adapting SF, but enhanced in SF regions about 2.0-4.0 octaves from the adapting SF. Williams and Wilson (1983) studied contrast sensitivities to gratings following adaptation to a high contrast 4.0 c/deg grating. Sensitivity was diminished across about 1.Ooctaves near the adapting SF. There was significant enhancement of sensitivity about 1.5 octaves below the adapting SF. Remote enhancement of contrast sensitivity has also been reported using difference-of-Gaussian stimuli at 2 deg eccentricity following sinusoidal grating adaptation at 5.7 and 8.0c/deg (Williams et al., 1982). These psychophysical findings have been interpreted to indicate that the SF-tuned mechanisms processing patterned stimuli are not completely independent, but may be coinhibitory. Specifically, mechanisms with neighboring center SFs may inhibit one another to an extent that is proportional to their excitatory output. If this were the case, SF adaptation of one mechanism (M, ), reducing its excitatory output, would produce a “release from inhibition” effect on another mechanism (Mr), temporarily increasing the responsivity of M2 (De Valois, 1977; Tolhurst, 1972). Therefore, SF adaptation near M, should lower contrast thresholds and increase VEP amplitude to SFs near MZ. This effect requires that while SFs are being presented near M2, M, is simultaneously responding below its normal level. According to this reasoning, unlike the SF adaptation paradigm, the SF masking paradigm would not release adjacent mechanisms from inhibition. In masking, the added high-contrast grating is simultaneously present with the test grating, but at a different temporal frequency. The response spectrum of a neural mechanism, M,, sensitive to the masking grating, will show increased activity at the fundamental and harmonics of the temporal frequency of the masking grating, and at multiple discrete cross-modulation frequencies of the test and mask temporal frequencies to the extent that M, is also sensitive to the test grating SF (Regan & Regan, 1988). While SF adaptation causes M, to respond “less” during testing, masking causes M, to respond “differently”. In masking, since M, is not
VEP attenuation and enhancement
itsnormal ievel, it would not release M, from inhibition. Electrophysiological evidence appears to be consistent with this reasoning, Fiorentini et al. (1983) and Regan (1983) searched for, but found no facilitation ei%cts on steady-state VEP amplitudes using masking paradigms that produced marked VEP amplitude attenuation when test and masking gratings were similar in SF. However, there is at least one report of increased psychophysical sensitivity at SFs distant from a high contrast masking grating (Tolhurst & Barfield, 1978). Interaction between SF-tuned mechanisms is supported by psychophysical studies exploring the effects of concurrent adaptation to several high contrast gratings differing in SF. The net e&M on contrast sensitivity varies as a function of the difference between the two adapting SFs. Within a range of about 1.0 octave, SF adaptation effects are additive; however, for adapting gratings differing in SF by I&2.0 octaves, the effects are subtractive as would be predicted if the two gratings were stimulating coinhibitory SF-tuned neural mechanisms (Greenlee & Magnussen, 1988). Tolhurst (1972) has reported similar effects of composite adapting stimuli on psychophysical thresholds. The appearance of illusory gratings higher and lower than the adapting SF following SF adaptation is also suggestive of disinhibition following SF adaptation (Georgeson, 1976b, 1980). The remote VEP facilitation effects reported here may be objective evidence for interacting SF-tuned mechanisms in humans previously ~y~thes~d on the basis of the psychophysical findings just reviewed. Accumulated psychophysical and electrophysiological evidence suggests a half-amplitude bandwidth of about 1.O octave for SF-tuned mechanisms (De Valois h De Valois, 1988). Since the areas of VEP facilitation are displaced by about 2.0 octaves from the adapting SF they would seem to represent an interaction between mechanisms. Additional electrophysiological evidence is provided by single cell studies in cat striate cortex (De Valois & Tootell, 1983). Most cells investigated were strongly inhibited by a narrowly tuned range of SFs from 1.0 to 2.0 octaves removed from their optimum SF. It was reasoned that the inhibitory effects represented inputs from interacting SF-tuned cehs, and that these interactions sharpen SF sefectivity of individual cells. Remote enhancement of contrast sensitivity following SF adaptation has been reconciled responding below
1173
with the modified line element model of spatial vision (Wilson $ Gelb, 1984) by allowing a gain increase in an adjacent SF-tuned mechanism following SF adaptation (Williams & Wilson, 1983; Williams et al., 1982), but without inco~rating coinhibition between subunits. If, as suggested above, this effect is a result of disinhibition from a neighboring SFtuned neural mechanism, the line element model might be modified to incorporate interaction effects between mechanisms, specifically tonic coinhibition. A second interpretation of the remote VEP facilitation effect is based on possible interactions between transient (movement) and sustained (pattern) neural mechanisms (Ikeda & Wright, 1974). In the present study SF adaptation led to increased VEP amplitude at relatively low SFs. Both transient and sustained mechanisms contribute to the VEP response to contrast-reversing gratings with the transient component pr~o~nating at SFs below about 3 c/deg (Kulikowski, 1977). If sustained mechanisms are selectively adapted by a stationary grating (Kulikowski, 19771, and the two mechanism types are in a coinhibitory relationship (Georgeson, 1976a), then stationary grating adaptation at moderate SFs should disinhibit transient mechanisms, resulting in increased VEP amplitude at low SFs. The transient-sustained coinhibition hypothesis cannot explain remote facilitation above the adapting SF, since effects of any transient mechanism d&inhibition by stationary SF adaptation would be confined to low SFs. Remote facilitation above 4.25 c/deg (Tolhurst & Barfield, 1978) and 1.19 c/deg SF adaptation has been reported for psychophysical data, but the effect has not been shown for VEP adaptation. Remote VEP facilitation in SF regions above the adapting SF was not found consistently in the present study. Although this may argue against SF-tuned coinhibition, it is also consistent with electrophysiolo~~l evidence suggesti~ a possible asymmetry in inhibitory effects between SF-tuned mechanisms. For cat striate cortex there are stronger inhibitory effects of high SFs on lower-tuned cells than for low SFs on higher-tuned cells, a phenomenon that would selectively sharpen the SF tuning of mechanisms tuned to higher SFs (De Valois & Tootell, 1983). The present findings must be viewed cautiously since our test SFs did not extend to 2.0 octaves above the adapting SFs, the region in which De Valois (1977) found
STEW Surea et al.
1174
psychophysical facilitation. Steady-state VEP amplitude approaches EEG noise by about 12 c/deg for the lO%C gratings used as test stimuli in the present study. An alternative approach would be to study the effects of SF adaptation on higher SFs by using contrast sweeps at fixed SFs (e.g. Allen et al., 1986) to establish VEP contrast thresholds. Acknowledgements-We gratefully acknowledge the research assistance of Michael S. Lutge and Brenda L. Bass, the library support of David Kosakowski and Loma Frost, and the technical support of Tom Gsbom and Michael Fleming. Some of these data were reported at the Meeting of the Association for Research in Vision and Ophthalmology,
May 1989, Sarasota, Florida. This research was supported by the California State University Bakersfield Foundation.
REFERENCES Allen, D., Norcia, A. M. & Tyler, C. W. (1986). Comparative study of electrophysiological and psychophysical measurement of the contrast sensitivity function in humans. American Journal of Optometry and Physiological Optics, 63, 442449.
Bach, M., Greenlee, M. W. & Biihler, B. (1988). Contrast adaptation can increase visually evoked potential amplitude. Clinical Vision Sciences, 3, 185-194. Blakemore, C. & Campbell, F. W. (1969). On the existence of neurones in the human visual system selectively sensitive to the orientation and size of retinal images. Journal of Physiology, London, 203, 237-260.
Blakemore, C., Muncey, J. P. J. 8r Ridley, R. M. (1971). Perceptual fading of a stabilized cortical image. Nature, London, 233, 204-205.
Bowker, D. 0. & Tulunay-Keesey, U. (1983). Sensitivity to countermodulating gratings following spatiotemporal adaptation. Journal of the Optical Society of America, 73, 427435.
Campbell, F. W. & Robson, J. G. (1968). Application of Fourier analysis to the visibility of gratings. Journal of Physiology, London, 197, 551-566.
Dealy, R. S. & Tolhurst, D. J. (1974). Is spatial adaptation an after effect of prolonged inhibition? Journal of Physiology, London, 241, 261-270.
De Valois, K. K. (1977). Spatial frequency adaptation can enhance contrast sensitivity. Vision Research, 17, 1057-1065.
De Valois, R. L. & De Valois, K. K. (1988). Spatial oision. New York: Oxford University Press. De Valois, K. K. & Tootell, R. B. H. (1983). Spatialfrequency-specific inhibition in cat striate cortex cells. Journal of Physiology, 336, 359-376.
Fiorentini, A., Pirchio, M. & Spinelli, D. (1983). Electrophysiological evidence for spatial frequency selective mechanisms in adults and infants. Vision Research, 23, 119-127.
Georgeson, M. A. (1976a). Antagonism between channels for pattern and movement in human vision. Nature, London, 259, 413415.
Georgeson, M. A. (1976b). Psychophysical hallucinations of orientation and spatial frequency. Perception, 5, 99-111.
Gcorgeson, M. A. (1980). The perceived spatial frequency, contrast and orientation of illusory gratings. Perception, 9, 695-712. Greenlee, M. W. & Magnussen,
among adapted Harris, F. analysis
S. (1988). Interactions spatial frequency and orientation channels concurrently. Vision Research, 28, 1303-1310. J. (1978). On the use of windows for harmonic with the discrete Fourier transform. Proceedings
of the IEEE, 66, 51-83. Ikeda, H. C Wright, M. J. (1974). Evidence for “sustained” and “transient” neurones in the cat’s visual cortex. Vision Research, 14, 133-136.
Klein, S. A. & Levi, D. M. (1985). Hyperacuity thresholds of 1 set: Theoretical predictions and empirical validation. Journal of the Optical Society of America, 2, 1170-I 190. Kulikowski, J. (1977). Separation of occipital potentials related to the detection of pattern and movement. In Desmedt, J. E. (Ed.), Visual evoked potentials in man: New developments (pp. 184196). Oxford: Clarendon Press. Legge, G. E. & Foley, J. M. (1980). Contrast masking of human vision. Journal of the Optical Society of America, 70, 1458-1471.
Maffei, L., Fiorentini, A. & Bisti, S. (1973). Neural correlate of perceptual adaptation to gratings. Science, 182, 10361038. Manahilov, V. & Vassilev, A. (1986). The effect of standing contrast and pattern adaptation on the human visually evoked potential. Electroencephalography and Clinical Neurophysiology, 64, 536-545.
Mecacci, L. & Spinelli, D. (1976). The effects of spatial frequency adaptation on human evoked potentials. Vision Research, 16, 477-479.
Norcia, A. M. & Tyler, C. W. (1985). Spatial frequency sweep VEP: Visual acuity during the first year of life. Vision Research, 25, 1399-1408. Regan, D. (1982). Visual information channeling in normal and disordered vision. Psychological Review, 89, 407-444.
Regan, D. (1983). Spatial frequency mechanisms in human vision investigated by evoked potential recording. Vision Research, 23, 1401-1407.
Regan, D. (1989). Human brain electrophysiology. New York: Elsevier. Regan, M. P. & Regan, D. (1988). A frequency domain technique for characterizing nonlinearities in biological systems. Journal of Theoretical Biology, 133, 293-3 17. Sachs, M. B., Nachmias, J. & Robson, J. G. (1971). Spatial-frequency channels in human vision. Journal of the Optical Society of America, 61, 11761186. Sclar, G., Lennie, P. & DePriest, D. D. (1989). Contrast adaptation in striate cortex of macaque. Vision Research, 29, 747-755.
Smith, A. T. & Jeffreys, D. A. (1978). Size and orientation specificity of transient visual evoked potentials in man. Vision Research, 18, 651-655.
Suter, P. S., Suter, S., Armstrong, C. A., Powers, J. C. & Bass, B. L. (1990a). Binocular VEP amplitude and lateral associated phoria. Investigative Ophthalmology and Visual Science, 31(4/Suppl.),
92.
Suter, S., Suter, P. S. % Deegan, J. F. II. (1990b). Steadystate VEP phase stability and acuity in adults and infants. Clinical Vision Sciences, 5, 71-80.
Swanson, W. H. & Wilson, H. R. (1985). Eccentricity dependence of contrast matching and oblique masking. Vision Research, 25, 1285-1295.
VI% att~ua~on Tolhurst, D. J. (1972). Adaptation to square-wave gratings: Inhibition between spatial frequency channels in the human visual system. Journal ofPhysiology, London, 226, 23 l-248. Tolhurst, D. J. & Barfield, L. P. (1978). Interactions between spatial frequency channels. Vision Research, lg, 951-958. Tyler, C. W. & Apkarian, P. A. (1985). I%cts of contrast, orientation and binocularity in the pattern evoked potential. Vlrion Research, 25, 755-766. Tyler, C. W., Apkarian, P. & Nakayama, K. (1978). Multiple spatial-frequency tuning of electrical responses from human visual cortex. ~~pe~~~a~ &a& Research, 33, 535-550. Watson, A. B. (1987). The cortex transform: Rapid computation of simufated neural images. Computer Vi&n, Graphics and Image Processing, 39, 31 l-327.
and enhancement
1175
Watt, R. J. & Morgan, M. J. (1985) A theory of the primitive spatial code in human vision. Vision Research, 25, 1661-1674. Williams, D. W. St Wilson, H. R. (1983). Spatial-frequency adaptation affects spatial-probability summation. Journal of the Optical Socieiy of America, 73, 13674371. Williams, D. W., Wilson, H. R. & Cowan, J. D. (1982). Localized effects of spatial-frequency adaptation. Journal of the Optical Sociefy of America, 72, 818--W. Wilson, H. R. & Gelb, D. J. (1984). Modified line-element theory for spatial-frequency and width dis&mination. .Journal oj r&e Opfieal S&eiy of America A, 1, 124-13I. Wilson, H. R,, McFarlane, D. K. Br PM&p, G. C. (1983). Spatial frequency tuning of orientation selective units estimated by oblique masking. Y&&r Research, 23, 873-882.