THE INFLUENCE OF COLOUR AND CONTOUR RIVALRY ON THE MAGNITUDE OF THE TILT AFTER-EFFECT’ NICHOLAS
apartment
J. WADE
of Psychology, University of Dundee, Dundee DDI 4HN. Scotland and
PETER WENDEKITH Department of Psychoiogy. University of Sydney, Sydney 2006, Australia (Received
in revised
form
17 October
1977)
~~act-Tijt
after-effects were generated by inspection of gratings inclined 10 or 15” from the vertical in six experiments. The resuIts indicated that the magnitude of the tilt after-effect: was not influenced by the cotour of the inspection and test _m?ltings (Experiments I-IV]; was not affected by binocular rivalry suppression (Experiment V); and was the same under various conditions of monocular and binoc~tar ‘ikspection .anh testing (Experiment VI).
at which a line appears vertical can be modified by prior exposure to a near-vertical line. This tilt after-effect (TAE) involves a shift of the apparent vertical in the direction of the previously tilted line. The maximum TAE is produced by inspecting a line tilted between IO and 20” from the vertical (Gibson and Radner, 1937; Mitchell and Muir, 1976; Muir and Over, 1970). Experimental interest in after-effects has increased considerably of late, because they might furnish information concerning the operation of hypothetical line detectors in the human visual system. For example, the effect of adapting one eye and testing the other (interocular transfer) has been taken to reflect the complement of binocular cells in the human visual cortex (Hohmann and Creutzfeldt, 1975). Second, the influence of cotour on the magnitude of after-effects has led to the suggestion that some line detectors are eolourcoded (Broerse, Over and Lovegrove, 1975: Lovegrove and Over, 1973; Over and Wenderoth. 1974). An additional approach has been to examine the infiuence of binocular rivalry suppression on the magnitude of after-effects (Blake and Fox, 1974; Kavadellas and Held, 1977; Lehmkuhle and Fox, 1975a). From these last studies the site for spatial after-effects is assumed to be cortical, and that for rivalry would appear to be more central still, as the after-effect magnitude was not influenced by phenomenal suppression of the inspection figure. A series of experiments, which are addressed to these aspects of TAE magnitude, is reported. In the first three experiments the effect of colour was exam-
The orientation
ined; monocular
after-effects
were generated
using in-
spection and test gratings that were either the same or different colours. The effects of cofour, interocular transfer and binocular rivalry were investigated in the fourth experiment, and the fifth was also concerned with colour and rivalry. The sixth experiment corn-’ Requests for reprints should be addressed to the first author.
pared the magnitudes of after-effects following monocular and binocular adaptation and testing. This last experiment was conducted to differentiate between two models of after-effect ma~itude. The results of the experiments indicated that the magnitude of the TAE: (a) was not influenced by the colour of the adapting and test figures. (b) was not affected by binocular rivalry suppression, and (cc)was the same under various conditions of monocular and binocular adaptation and testing.
The TAEs were generated by inspection of a 10” clockwise grating that was either red or green and tested with a grating of the same or different colour. Both inspection and test gratings were presented to the right eye. Lovegrove and Over (1973) reported after-effects in the order of 1.3” following inspection of similarly coloured gratings presented to one eye, but no after-effect when the colours differed. In a second report Broerse er al. (1975) found larger monocular TAEs under both conditions, but those in which the inspection and test gratings differed in colour were half the size of the same-cofour after-effects (1.1 * refative to 2.2“). This experiment was designed to examine the dependence of monocular TAEs upon the colour of the inspection and test stimuli. Spatial after-effects can be generated at their maximum magnitude following very brief exposure (Sagara and Oyama, 1957; SekuIer and Littlejohn. 1974), although their dissipation is very rapid; longer inspeo tion reduces the rate of dissipation. In order to reduce the possibility of appreciable dissipation in this, and in the following experiments the test grating was presented immediately after the inspection grating, and the subject’s task was to report whether the test grating appeared tilted with its top to the right or left (i.e. clockwise or counter-clockwise, respectively). This sequence was then repeated, with the inspection grating immediateiy repfacing the test grating. The inspec-
838
NICHOLASJ. WADE and PETERWENDEROTU
tion period was IOsec and the test duration was 0.5 sec. so that a cycle lasted 10.5 sec. A sequential estimation procedure was employed, in which the orientation for the test grating was dependent upon the subject’s previous report. If the test grating was judged clockwise it would be shifted in a counterciockwise direction for the next presentation. The size of the shifts decreased throughout a trial. following the simplest version of the PEST procedure devised by Taylor and Creefman (1967): in a trial the test grating was presented initially at + 10’ (clockwise) or - IO” (counter-clockwise); following a subject’s response the grating was moved 16’ in the opposite direction (e.g. from + lo” to -6’): the step size was halved each time a response differed from the one immediately preceding it (e.g. if - 6’ was judged “left” then the test would be shifted 8’ clockwise to -t2”); a trial terminated when the next step size would have been 0.25”. and the apparent vertical was taken as the orientation that would have been presented with the 0.25” shift. METHOD Apparatus
The stimuli were presented in a three-field tachistoscope IScientific Prototvoe. Model GB\ which had been modified io allow presentaiion of stimuli continuously variable in orientation and pivoted about an axis at the centre of an aperture subtending 3.3’ at the observer’s eye. Additional modifications enabled control of both the colour of the stimulus via Kodak Wtatten (gelatin) tilters (No. 29. red: and No. 61. green). The stimulus fields contained either blank cards or square wave gratings. both with spaceaveraged luminance of 7.5 cd/m’. The Michelson contrast CCL,, - L,,,)ilL.
f
LJI
was 0.8. The spatial frequency of both inspection and test gratings was 3.5 c,‘dcg Subjects Sixteen subjects participated in the experiment. All had normal or corrected-to-normal vision. Procrdure Subjects used their right eye throughout the experiment and the left eye was occluded. The sequential estimation proeedure mentioned above was employed to determine the apparent vertical. Four estimates were derived for each condition, two each from initial positions of +lO* and - IO’. The orientation of the briefly presented test grating could be changed. according to the rules of the PEST procedure. during the IO-see inspection period. The four measures were taken successively. The four combinations of inspection and test colour were examined. namely red-red, red-green, greed-red and green-green. For each combination there were four pre-test and four post-test estimates of the apparent vertical. The only difference between the pre- and post-test procedures was that a blank coloured fieid was inspected in the pre-
test rather than a coloured grating as in the post-rest. _t rest of at least 5 min was given between rht termination of a post-test series and the commencement of rhr ncyt
pre-test condition. Thus. four after-effects were measured under each colour combination within a session. and each subject was tested in two sessions.
RESULTS
AND
DISCLSSIOZ
The measure of the TAE was the difference between the pre- and post-test apparent verticals, and the means for the four colour combinations are given in fable 1, together with their standard errors. The
values were all in the range of 2.1-2.6’. There was little suggestion of their dependence upon the colour
combination, and this conclusion was supported by the statistical analysis of the data. Three planned contrasts between the means were examined, and the critical value in each case was F0,gs(lrJ5) = 4.06: there were no differences between the after-effects following both red or green inspection and testing (F = 0.58): inspection of a red grating did not affect the magnitude of after-effect tested with a red or green grating (F = 1.46); and a green inspection grating did not influence the after-effect with a green or red test grating (F = 0.58). These results are clearly at variance with those of Lovegrove and Over (1973) and Broerse et al. (1975), but the reason for this is not readily apparent. The magnitudes of our same-coloured TAEs were commensurate with theirs, but in our experiment the inspection-test colour difference did not diminish the TAE; Lovegrove and Over reported no TAE in this condition and Broerse et nI. found one of i.l”--both studies employing a 15” inspection grating. Thus. their results have not been consistent in this regard. although different psychophysical procedures were employed in their two studies: Lovegrove and Over used an adjustment method and Broerse et al. employed a staircase procedure. Moreover, the Wratten filters used in our experiment (Nos. 29 and 61) differed slightly from the ones they used (Nos. 26 and 55). In order to attempt to resolw this discrepancy a further experiment was conducted. EXPERlMEFiT
II
This experiment was essentially similar to the previous one, except that Wratten filters 26 and 55 were employed for the cofoured gratings. IMETHOD
AppWCltUS
The stimuli were presented in a Gerbrands tachistoscope and they subtended 3.75” at the eye. Coiour filters (Kodak Wratten Nos. 26 and 55) could be inserted
(Model I-4A).
Table 1. Means (deg) and standard errors (S.E.) for the TAE in Experiment I Inspection colour : Test colour: Mean S.E.
Red
Red Red
Green Red
Green Green
Green
+2.12 0.56
+ 2.30
+2.21
+2.51
0.67
0.50
0.49
829
Magnitude of the tilt after-effect Table 2. Means (deg) and standard errors (S.E.) for the TAE in Experiment II Inspection colour : Test colour:
Red Red
Green Red
Green Green
Red Green
Mean S.E.
3.50 0.76
4.11 0.50
3.25 0.82
2.78 0.28
in the fields. which had a space-averaged luminance of 0.7cd/m2. The spatial frequency of the gratings was 5.6c/deg. and the contrast was 0.7. Subjects Ten subjects took part in the experiment, all with normal or coK~ted-t~normai vision. Procedure The procedure was essentially similar to that of Experiment I. All subjects used the right eye alone for observation. and an identical PEST sequence was employed-with a cycling IO set inspection and 0.5 set tests. Pre-tests were measured without any filters in the fields. and the mean of two sequences (starting from f IO”) was used for determining all the after-effects. Four after-effect coIour combinations were examined, as in Experiment 1. The inspection grating was IO” from the vertical (half the subjects were given - IO” and the other half + IO” inspection orientations). Two measures were taken for each condition.
METHOD
Apparatus The same apparatus was used as in Experiment II. Subjecrs Five subjects took part in the experiment. normal or corrected vision.
all having
Proceduie The PEST procedure described above was used, but the timing sequence followed that reported by Broerse et al.: the inspection grating was inclined IS” from the vertical and it was presented for I sec. followed immediately by the test grating. which was exposed for IS0 msec. There was an intertrial interval of 8 sec. during which the subject reported the direction of tilt of the test grating. and the experimenter made the appropriate adjustment for the next trial. Otherwise the procedure was the same as in Experiment II.
RESULTS AND DISCUSSION The mean TAEs for the four conditions are given in Table 2. together with their standard errors. (Since the after-effects for subjects inspecting - 10” gratings were in the opposite direction to those viewing + 10” gratings, they were combined according to their expected direction, i.e. the visual vertical was shifted in the direction of the inspection orientation. This was the case for all subjects under all conditions.) The TAEs were larger than in the previous experiment, but no differences were evident statistically between the colour combinations of inspection and test figures. That is, inspection of a red (green) grating yieMs the same magnitude of tilt after-effect when tested with a red (green) OT greed (red) grating. These results support those of Experiment I, and remove the possibility that the colour filters employed could have led to the discrepancy with the data of Lovegrove and Over (1973) and Broerse et al. (1975). EXPERKMENT fft
Broerse et al. (1975) used a different timing sequence for their sequential testing and this was employed in the third experiment, which again examined the effects of inspection and test tiolour on the magnitude of the TAE.
RESULTS AND DISCUSStOFU The mean after-effects for the four inspection-test coIour combinations are given in Table 3. The aftereffects were all in the expected direction. and were not differentiated statistically. Thus, the results of three experiments have been consistent in failing to find any evidence of colour selectivity for the T’AE. The reason for the discrepancy between our results and those of Over and his colleagues remains unresolved, but it seems unlikely to be associated with apparatus or procedural factors.
EXPERIMENT
IV
The effect of colour on the TAE was examined yet again in this experiment, but the test grating was always green. However, rather than inducing the T’AE in one eye atone and occluding the other, all conditions involved binocular inspection and then monocutar testing, The binocuiar inspection figures could be homogeneous &ours (red or green), a coloured grating in one eye and a homogeneously coloured disc in the other, or oppositely tilted coloured gratings
Table 3. Means (deg) and standard errors (SE) for the TAE in Experiment III Inspection colour: Test colour: Mean SE.
Red
Red
Green Red
Green Green
2.95 I.21
3.35 0.98
2.00
2.65
0.45
0.78
Red Green
830
NICHOLAS J. WADE
h-ddt-g orteat(ri&eye Blank
I
1!Ygrating
I
I
and
PETER WE~DEROTH
Kavadellas and Held (1977) have reported that the magnitude of a TAE induced by alternating coloured gratings. was not influenced by rivalry during inspection. Accordingly. it was predicted that the conventional TAE measured here would not be affected bq rivalry during inspection. In that event testing the right eye after RIV inspection might result in a net TAE which is the simple sum of the directionallyopposite MON and IOT effects induced in the two eyes independently.
Fig. I. The four spatial conditions in Experiment trol (CON). Monocular (MON). Interocular (IOT) and Rivalry (RIV).
IV: Con-
Transfer
in each eye. These combinations resulted in sixteen different conditions. which are summarized below. I. Confrol (CON) trials. Both the test (right) and other (left) ey-e were presented with homogeneous (z&o contrast) fields. 2. Monocrrlar (MON) trials. The left eye viewed a homogeneous field. the test eye viewed a grating tilted i 15’ (clockwise relative to the observer). 3. Intrrocular tmnsfb (rOTI trials. The test eye viewed a homogeneous field, the other eye viewed a grating tilted - 15” (counter-clockwise). 4. Ricalry (RI E’) frids. The combination of MON and IOT. in which the left eye viewed a - 15’ grating and the right eye viewed a + 15’ grating. Under each of these inspection conditions. four colour conditions were examined and these were: I. RR: 2. GG: 3. RG: 4. GR:
Both fields red. ’ Both fields green. Left eye field red, right eye field green. Left eye tield green. right eye field red,
There were thus sixteen conditions, with a green test grating presented to the right eye in each one. Homogeneous fields induce no TAEs so that the control conditions were the baselines (pre-tests) against which the TAE magnitudes were measured. However, in case unexpected differences did occur between the four controls. CON,, was the intended pre-test condition for MONRR and RIV,,; and each of the other CON colour conditions was intended as control for the corresponding MON. IOT and RIV treatments. As it turned out. there were no differences between the CON conditions and they were combined. After-effects measured via interocular transfer typically are smaller than those derived monocularly (e.g. Movshon. Chambers and Blakemore, 1972; Mitchell and Ware, 1974). In the IOT condition the inspection grating was - 15’ and so the TAE would be in a counter-clockwise direction. In the RIV condition the oppositely oriented gratings would be expected to vary in phenomenal clarity, so that the gratings would either alternate in visibility. or different parts of them would be visible simultaneously. Blake and FOX (1974) and Lehmkuhle and Fox (1975a) have shown that phenomenal suppression of targets in rivalry does not influence the magnitude of a number of spatial. monocularly measured after-effects. Additionally,
The apparatus was the same as that used in Experiment I. The eye to which the stimulus was presented was controlled by means of polarizing filters. The space-averaged luminance of the stimulus fields was 2.2cd/m’. and the contrast was 0.8. and the spatial frequency of the gratings was 3.5 c’deg.
Su&cts Ten subjects took part in the experiment. All had normal or corrected-to-normal vision. Eight subjects were right eye sighting dominant and two left eye dominant. as determined by aligning a small aperture in a card, held by both hands. with a distant dot. Procrd1cre Each subject was tested under all sixteen conditions, which were presented in a different random order to each one. The procedure for locating the apparent vertical was the same as that described for Experiment I. RESULTS
ASD
DtSCLSSlON
The means for the four control conditions were very similar. with an overall value of -0.35’ (this slight counter-clockwise bias was due largely to one subject whose apparent vertical was reliably close to -4”). The TAEs were derived by subtracting the average of the four CON measures for each subject from all their other values (including the individual CON measures), and the means are given in Table 4. together with their associated standard errors. The overall mean monocular TAE was + 1.52’ and the effect was very similar whether the colours in the two eyes during inspection were the same (+ 1.46”) or different (+ 1.57’) as Fig. 2(a) shows. In addition. whether the colour of the inducing stimulus was the same as that of the test stimulus or not also made little difference and this is shown in Fig. 2(b). The mean MON effect when both inspection and test stimuli were green was + 1.31’ while that obtained when the inspection figure was red and the test figure was green was + 1.72”. Although this obtained difference, of about 0.4’. was not significant (see below) it was in the direction opposite to that predicted from the results of Lovegrove and Over (1973) and Broerse et a/. (1975). The overall mean interocularly-transferred TAE was - 1.27’. which was 840/, of the MON effect. Although this degree of transfer is rather higher than the 70% or so which others have reported (e.g. MOW shon et al.. 1972; Mitchell and Ware, 1974) in the latter cases IOT was measured with one eye occluded during inspection. Lehmkuhle and Fox (1976) have shown recently that IOT effects are increased by up to 25”i, when the nonadapted eye. rather than being
831
Magnitude of the tilt after-effect Table 4. Means (deg) and standard errors (S.E.) of TAEs under control (CON). monocular (MON). interocular transfer (IOT) and rivalry (RIV) conditions Mean
Condition CON:
RR RG GR GG
MON:
RR RG GR GG
IOT:
RR RG GR GG
RIV:
RR RG GR GG
-0.15 +0.14 +0.09 - 0.08 0.00 Mean CON + 1.91 + 1.62 +1.53 + 1.10 Mean MON +I.52 - 1.37 - 1.38 - 1.12 - 1.22 Mean IOT -1.27 + 0.48 +0.15 - 0.06 + 0.49 Mean RIV +0.27
S.E. 0.17 0.39 0.21 0.14 0.31 0.47 0.48 0.26 0.35 0.37 0.42 0.39 0.” 0% 0.41 0.42
occluded, views a homogeneous field during inspection. However, while this fact might account for our obtained degree of transfer it must be noted that Broerse et al. (1975). who presented white light in the non-adapted eye, obtained transfer of only 65%. The degree of transfer in our experiment is also unexpectedly high when it is considered that 8 of the 10 subjects had right eye dominance so that for these subjects. transfer was from the nondominant to the dominant eye. Several reports have suggested that IOT is greater in the reverse direction, from the dominant to the nondominant eye (Movshon er al., 1972; Mitchell and Ware, 1974; Mitchell, Reardon and Muir, 1975). Finally, the rivalry after-effect was small. clockwise but not significantly different from zero (see below): its mean value was +0.27’. This value is, however, very close to that predicted from the sum of the MON and IOT effects: algebraic summation of MON and IOT, the two components involved in the RIV effect, gives + 1.52” + (- 1.27’) = 0.25”. Moulden (1974) found that a similar relationship applied to the movement after-effect duration, but not for TAE magnitude. All of the above statements generally were supported in the statistical analysis of the data. Fifteen planned contrasts were carried out amongst the sixteen conditions and the critical value in each case was F,.,,(1,135) = 3.91. Three contrasts showed that there were no differences between the CON effects: they tested the difference between the RR and GR means; the GG and RG means; and the average of the former pair minus the average of the latter pair. In these three tests, F = 0.30, 0.24 and 0.04. respectively. Similar tests showed
that there were no differ-
’ Rivalry was not directly monitored in any of these experiments. However, apart from the fact that there is independent evidence that + 15’ gratings phenomenally alternate both authors noted vigorous rivalry when acting as subjects in Experiment IV. The 30” disparity between the two stimuli was sufficiently great to prevent binocular fusion.
Fig. 2. (a) Mean TAEs in Experiment IV as a function of colour of left-eye inspection field in relation to right-eye inspection field for conditions CON (0). MON (0). IOT (a) and RIV (Cl). (b) Mean TAEs as a function of dour of inspection stimulus relative to test stimulus under CON (0). MON (0). IOT (H) and RIV (0. the MON effects (F = 0.74. 1.87 and 1.69, respectively), the IOT effects (F = 0.32, 0.13 and 0.03. respectively) or the RIV effects (F = 1.45, 0.57 and 0.13, respectively). That is, the left eye-right eye colour combination during inspection had no significant effect upon the magnitudes of TAEs under the four conditions, when tested with a green grating in the right eye. The remaining three contrasts showed: that the MON mean (+ I .52”) was significantly different from the IOT mean (- 1.27”) with f = 155.43; that the CON and RIV means were not different (F = 1.41): and that the algebraic sum of MON and IOT. + 1.52” + ( - 1.27”) = 0.25’ was not significantly different from the sum of RIV and CON. +0.27” + (0.00) = +0.27’. Since the CON mean was normalized to zero this last test provides statistical support for the additivity hypothesis. However, it must also be noted that, in a further test. the mean RIV effect was not significantly different from zero (F = 2.82) although the two component TAEs, MON and IOT. were greater than zero (F = 79.39 and 56.07, respectively, P < 0.0001). Perhaps a more powerful indication of the degree to which the data supported the additivity of the MON and IOT effects is given by an individual subject’s analysis: when, for each subject. a predicted MON effect was calculated by subtracting IOT from RIV the Pearson productmoment correlation between the obtained and predicted values was +0.73, a reasonably high value which was significantly different from zero (t = 3.06. df = 9, P < 0.02). In summary, the results of this experiment supported those of Experiments I-III: the magnitude of the TAE is unaffected by the colour characteristics of the inspection and test figures. In addition. the results suggested that oppositely directed TAEs in each eye might combine additively.
ences between
EXPERIMEW v
The results of Experiment IV did not provide conclusive evidence concerning the effect of binocular rivalry on the magnitude of the TAE despite the considerable phenomenal alternation which occurred during inspection’. The reason is that the experimen-
832
N~c~or~s J.
tal paradigm
in fact tested the hypothesis
(MON
WADE
and
PETERWENDEROTH
that:
Test Grating Left
- x) - (IOT - x) = RI’?.
That is. the magnitude of the RIV effect in the above equation would be unchanged if phenomenal rivalry of the inducing stimuli had no effect on the magnitudes of the MON and IOT effects (i.e. x = 0). or if it reduced both by a fixed amount (i.e. x > 0). In order to test the effect of rivalry on the TAE directly it is necessary to devise conditions analogous to those used with other after-effects by Blake and Fox (1974) and Lehmkuhle and Fox (1975a). such that the rivalrous stimulus in the other eye produces phenomenal rivalry without itself inducing any aftereffect. Thus, Fox and his colleagues compared aftereffects in which the nonadapted eye was presented with a blank field (equivalent to MON in Experiment IV) with those in which the nonadapted eye viewed a rivalrous stimulus which itself induced no aftereffect. Since they found that the after-effects were the same magnitudes in the two cases they inferred that it is the time For which the adapting stimulus is physically present, not phenomenally present, which detetmines after-effect magnitude. In this experiment, we examined an equivalent situation but using the TAE. As in Experiment IV, the test stimulus was always a green grating. The control (pre-test) conditions involved inspection of homogeneous green fields in both eyes. In the two monocular conditions, which were compared to assess the effect of rivalry. the adapted eye viewed a + 15” green grating and the other eye viewed either a red or green stimulus. The only difference between the two monocular conditions was that in one case the nonadapted eye viewed a homogeneous field (condition MON) while in the other condition it viewed a horizontal grating (condition RIVH). The rationale for the last condition is that althou~ phenomenal rivalry between a horizontal grating and a + 15” grating should be as vigorous as that between + 15” and - 15” gratings, nevertheless a horizontal grating induces no TAE (e-g. Mitchell and Ware, 1974). The latter is such a welf-documented finding that we did not consider it necessary to include an IOT control condition with the horizontal grating in the other eye during inspection. An additional factor of interest in this experiment was eye dominance: half of the observers were left-eye sighting dominant, half were right-eye dominant and each subject was tested with each eye. The complete experimental design (IO conditions in all) is shown in Fig 3. The CON apparent verticals were baseline pre-test measures which were subtracted from the other apparent verticals to provide TAE measures. If rivalry had no effect on the TAE magnitude, then MON effects and RIVH effects were expected to be equivalent. There was no reason to expect column (test eye) differences, although it was of interest to see whether there was any interaction between test eye and eye dominance. IMETHOD
Apparatus
The apparatus was identical to that used in the last experiment, with the addition of polarizing filters to allow testing of the left eye.
eye
Right eye
00
COI
Mar
G
0
RIVt
G
G
Fig. 3. Conditions
G
G
R
G
G
in Experiment V (see text for de&is).
Stthjrcrs Twenty subjects participated, includinn both authors. All had normal or corrected-to-normai &ion. Ten subjects had right-eye sighting dominance and the other half had left-eye dominanoz. Procrdure
The procedures were the same as those in Experiment IV and the TAG were. once again. averages of four measures taken from + IO” starting positions.
The mean TAEs obtained in the four colourcontour conditions are given in Table 5 where separate means are shown for the effects obtained in the dominant and nondominant eyes. The overall mean TAE in the four MON conditions was +2.16” and that in the RIVH conditions was +2X)2”. Thus, phenomenal suppression of the adapting grating during part of the inspection period had little effect upon the TAE generated by it. The RIVH effects were also similar whether the colour of the horizontal grating in the other eye was the same as the colour of the adapting grating (+ 1.99’) or different from it ( + 2.06’). This is consistent with the results of Experiment IV, where colour had no influence, and it is also of some interest since it indicates that the RIVH effects were the same whether rivalry was induced by contour alone or by contour and colour, whereas phenomenal alternation rate tends to be greater under the latter than the former condition (Wade, 1975). Lastly, the overall mean TAEs were of similar magnitude in the dominant eye (+ 1.97’) and the nondominant eye (+2.22”), a result consistent with previous data on the TAE and the motion aftereffect (Mitchell and Ware, 1974: Wade. 1976). The statistical analysis supported these points. A two-way analysis of variance Finer, 1962, p. 306) showed that neither the between-subjects main effect, namely eye dominance. nor the eye dominance by treatments interaction, were significant (F = 0.47. 1.01, respectivefy, P > 0.05). The remaining withinsubjects effects were tested by seven planned contrasts
a33
Magnitude of the tilt after-effect Table 5. Means (deg) and standard errors (SE.) of TAEs under monocular (MON) and rivalry (RIVH) conditions with a green inducing and test grating and the colour in the other eye either the same (green) or different (red). Results for the sighting dominant and nondominant eyes are shown separately Inducing condition: Colour in other eye relative to test eye: Dominant eye tested: Nondominant eye tested :
for which f,,,,(1,126) colour in
the critical = 3.92. The
value
RWH
MON Same
Different
2.04 0.48 2.43 0.45
2.18 0.48
Mean S.E. Mean SE.
in
contrasts
each
case
indicated
was that
the nonadapted eye affected neither the MON effects in the left eye (F = 0.01) or the right eye (F = 0.55). nor the RIVH effects in the left fF = 0.00) or right eye (F = 0.19). Also, the MON and RIVH aftereffects did not differ from each other in the left eye (F = 2.48) or in the right eye (F = 0.24). The sole significant difference was that the overall TAE in the right eye (+2.35”) exceeded that in the left eye (+ 1.85’) with F = 7.47, P c 0.01. Spatial after-effects can he generated at their maximum magnitude following very brief exposure (Sagara and Oyama, 1957; Sekuler and Littlejohn, 1974), although their dissipation is very rapid. Longer inspection reduces the rate of dissipation. It is possible that binocular rivalry suppression influences the duration of after-effects rather than their magnitude. The present experiment has shown that TAE magnitude is not affected by rivalry. However, Lehmkuhle and Fox (1975a) found that the duration of the movement after-effect was unaffected by rivalry suppression. Thus. it would appear that both the magnitude and duration of spatial aftereffects are dependent upon the physical presence of the inspection stimulus. Lehmkuhle and Fox (1975a) did find that the movement after-effect was smaller in a rivalry mimic condition, in which the inspection figure is physically absent during intervals yoked to rivalry suppression. This might have reflected the absence of an inspection stimulus immediately prior to the test stimulus on some proportion of the mimic trials.
1.65
1.99 0.45
Same
0.46 2.32 0.49
The results of Experiment IV suggested that binocular rivalry suppression either has no effect on TAE magnitude or that it subtracts a common amount from both the monocular and interocuiarly-transferred components of a rivalrous after-effect. Experiment V indicated that the first alternative is correct. Taken together, the last two experiments have shown that TAEs, like other after-effects studied by Fox et al., can be generated at fun strength during binocular rivafry. Even so, the additive expression with which these results are consistent, namely MON + IOT = RIV, does not address itself directly to the mechanism of the TAE; at least two models of TAE magnitude have been proposed which comply with this formulation.
2.00 0.45 2.12 0.39
First, Moulden (1974) postulated that the magnitudes of spatial after-effects depend upon the ratio of the.num~r of neurones which are both adapted and tested to the total number of neurones tested. On the other hand, Lehmkuhle and Fox (1975b) proposed that after-effect magnitude depends simply upon the numerator of this ratio, that is, upon the total number of cells both adapted and tested. Each of these models is consonant with the colour, interocular transfer and rivahy data presented so far. There is, however, one issue which might differentiate between the two models, namely whether monocular after-effects differ in magnitude from binocular after-effects. Both models assume that the generation of after-effects involves three independent populations of visual cortical neurones: binocular (B) cells which can be fully stimulated by either eye alone or by both eyes; and two groups of monocular cells each of which requires to be stimulated by either the left eye (ML) or the right eye (Ma). According to the Lehmkuhle and Fox model, the total number of cells adapted and tested determines after-effect magnitude so that: Binocular TAE = B + ML + MR Monocular TAE = I3 + ML or B + Ma_ Hence, if ML = MR, it follows that binocular aftereffects must exceed monocular after-effects. According to Moufden’s model, however, the magnitude of the after-effect depends upon the ratio of cells adapted and tested to ceils tested, so that: Binocular TAE =
EXPERIMENT VI
Different
B+Mt.+Ma = 1o B + M,, + MR .
B + ML Monocular TAE = B+M,
or
B+M, Gz&=
IO -*
That is, on Moulden’s model, binocular and monocular after-effects should be equal in magnitude. Lehmkuhle and Fox 61975b) present data which are consistent with their model while Moulden’s (1974) data favour his formulation. Our final experiment measured the TAE using achromatic gratings under conditions of monocular adaptation and testing (MM) as well as under binocular adaptation and testing (BB). In order to render the test of the two modefs more powerful two additional conditions were run: adapting one eye but testing both (MB) and adapting both eyes while test-
834
NICHQLASJ. WADE
and
ing only one (BM). after the manner of Moulden (1974). The two models make different predictions about the inequalities in these four after-effect conditions. Specifically for the mixed cases. the Lehmkuhle and Fox model predicts that: MB = B + MiL or B + Ma and BM - B + ML or B + MR so that. according to that model the predicted rank ordering of the four after-effects is BB>MM=MB=BM.
(1)
According to Moulden. however: MB=
B + ML
B + Ma
B + ML e MR Or B + M,_ + Ma
< 1.0
and BM =
B + M,, Bt
Or
B + Ma -= B + MR
1.0
so that. according to Moulden’s model, the predicted rank ordering of the four after-effects is BB = MM = BM > MB.
(2)
The aim of Experiment VI was to gather data which would bear upon these different predictions with the purpose of falsifying at least one of the two hypotheses. Moulden’s (1974) data for both the movement after-effect and the TAE conformed to expression (2).
SIETHODS The apparatus was identical to that used in Experiment V except that no colour filters were used so that the spaceaveraged luminance of the achromatic gratings was 13.7 cd/m’ and the contrast was approximately 0.9. In the monocular inducing and test conditions, the other eye viewed a grey homogeneous field of space averaged luminance equal to that of the test and inducing gratings. Subjects
There were 24 subjects. half of whom were left-eye sighting dominant and the remainder had right-eye dominance. Procedure All subjects were tested under each of the BB. MM. BM and MB conditions. Independent pre-tests (equivalent to CON trials of previous experiments. but using grey fields) were given prior to the after-effect trials for each condition. Within both the left and right eye dominant groups half the subjects were adapted or tested with the
PETERWE~DEROTH
left eye under M conditions.
and half were adapied
same as in Experiment V. RESULTS ASD DISCLSS~ON
The overall mean TAEs obtained in the four treatment conditions are given in Table 6. A three-way analysis of variance (Wirier. 1962, p. 337) indicated that there were no significant effects in the experiment. Overall. subjects who used the left eye in the M conditions obtained TAEs not significantly different from those measured in subjects who used the right eye (A 1.41” and + 1.80’. respectively). Further. after-effects obtained from left-eye dominant (+2.05’) and right-eye dominant subjects ( + 1.16”) were not significantly different. In these two cases, the obtained means look different but the larger variance associated with between subjects effects resulted in the f values in these cases (0.37 and 1.95. respectively) falling short of the critical value [F0,9s(l,20) = 3.491. The main effect within subjects. that of the combination of monocular/binocular adaptation or testing. was also not significant: the critical F value for all within subjects effects was F0.ss(3,60) = 2.76 and in this case, F = 0.20. All the interactions of M/B combination with test eye (F = 1.03). dominant eye (f = 1.30) and both test and dominant eye (F = 2.66) were nonsignificant too. It would be inappropriate to place too much emphasis on any of the above tests with the exception of the M;B combination effect which was based upon all 24 subjects. The other effects derive from small samples and are sometimes between subject effects which were included more for purposes of counterbalancing rather than hypothesis testing. Thus. the essential tinding of Experiment VI is that the data do not support either the Lehmkuhle and Fox or the Moulden model. On the other hand, it could be argued that the number of B cells was considerably greater than the M cells, so that in the case of the Moulden model the ratio (B + M,)/(B + ML c Ma) is only marginally less than 1.0. In that event, the two models would be virtually indistinguishable because it would also be true in the Lehmkuhle and Fox model that B + ML + MR would be only marginally greater than, for example. B + ML. The most hkeiy possibility is that both of these models are oversimplified. in that they class cells as either entirely monocular or entirety binocular, whereas recent neurophysiologica1 evidence stresses that cells vary in degrees of ocular dominance (e.g. Wubel and Wiesel. 1974a, b). The other oversimplification in these models is that they assume that binocular neurones can be equally well stimulated binocularly or monocularly. an assumption which is not supported by
Table 6. Means (deg) and standard errors (S.E.) of TAEs obtained under conditions of binocular induction and test (BB), monocular induction and test (MM), monocular induction but binocular test (MB) and binocular induction with monocular test (BM). -
Condition: Mean S.E.
or
tested with the right eqs. Otherwise, procedures were th?
BB
MM
>fB
BM
+ I.53 0.28
+ 1.60
7 I.56
0.12
0.39
t1.71 0.33
835
Magnitude of the tilt after-effect
neurophysiological data (e.g. Joshua 1970; Vidvasagar, 1976).
and
Bishop,
CONCLUS1ON
There probably are a number of ways in which the visual system could be wired in order to amount both for the results of our experiments and for those of others to which we have referred One requirement in Formulating modeis is that they should be consistent with known neurophysiotogy. We are therefore not only wary of formulations which refer only to extreme classes of solely monocular or binocular cells, but we also wonder about the rote of binocular colour-selective cells in vision (e.g. Zeki, 1973). since current models seem to stress the monocularity of colour-selective neurones (e.g. Broerse rr al.. 1975; Coitheart, 1973), In summary. the experiments which are reported here add the tilt after-effect to the class of effects which are unaffected in magnitude by phenomenal rivafry suppression. Furthermore, it appears that monocular and interocularfy transferred TAEs can be used to predict the size of rivalrous effects: simple linear summation seems to work. As far as inducingand test-figure colour is concerned. we have obtained no evidence to suggest that colour makes any difference to TAE magnitude so that there is a need to identify specific Finally,
the precise conditions under which colourTAEs, which others have reported. occur. although we derived different predictions
about the relative magnitudes of binocular and monocular TAEs From two alternate and simple models we have been unable to confirm either model. Acknowledgemenrs-Experiments I. IV. V and VI were sup ported by Grant No. G974/9fl/N from the Medical Research Council to NJW. and Experiments 11 and III were supported by Grant No. A74/15177 from the Australian Research Grants Committce to PW in 1977. Pw’s visit to Dundee was made possible by the University of Sydney, from which ht was on sabbatical leave. We thank Linda Wilson for her assistance in data collection. REFERENCES Blake R. and Fox R. (1974) Adaptation to invisible gratings and site of binocular rivalry suppression. Nurure. toed. 249, 488-490. Broerse J., Over R. and Lovegrove W. (1975) Loss of wavelength selectivity in contour masking and aftereffect following dichoptic adaptation. Percept. ~~~~~0~~~s. 17. 333-336, Coltheart M. (1973) Colour-specificity and monocularity in the visual cortex, Vision Res. 13, 2595-2598. Gibson J. J. and Radner M. (1937) Adaptation, after-effect and contrast in the perception of tilted lines. 1. Quantitative studier J. exp. Psycho/. 20, 453-467. Hohmann A. and Creutzfeldt 0. D. (1975) Squint and development of binocularity in humans. Nature, tend. 2S4, 613-614.
Hubel D. H. and Wiesel T. N. (1974a) Sequence regularity and geometry of orientation columns in the monkey striate cortex. J. Camp. Neural., 15% 267-294, Hubel D. H. and Wieset T. N. (1974b) Uniformity of monkey striate cortex: a parallel relationship between field size. scatter and magnification factor. J. Camp. Neural. l§s. 295-30s.
Joshua 13. E. and Bishop P. 0. (19701 Binocular single vision and depth discrimination. Receptive field disparities for central and peripheral vision and binocular interaction on peripheral single units in cat striate cortex. Expl Brain Res. IO, 389-410. Kavadellas A. and Held R. (1977) Monocularity of colourcontingent tilt aftereFfects. Percept. Psychophys. 21. 12-14. LehmkuhIe S. W. and Fox R. (1975a) Effect of binocular rivalry suppression on the motion after-effect. Vision Res. 15. 855-859. Lehmkuhle S. W. and Fox R. (1975b) Binocular interaction of the motion uzerelfecr; a simple model. Paper presented to the Association for Research in Vision and Ophthalmology. Sarasota. Lehmkuhle S. W. and Fox R. (1976) On measuring interocular transfer. Vision Res. 16. 428-430. Lovegrove W. J. and Over R. (1973) Colour selectivity in orientation masking and after-effect. Vision Rrs. 13. 895-90 I. Mitchell D. E. and Muir D. W. (1976) Does the tilt after-
effect occur in the oblique meridian?
Vision
Res.
16.
609-613.
Mitchell D. E., Reardon 1. and Muir D. W. (1975) Interocular transfer of the motion after-effect in normal and
stereoblind observers. Expi Brain Res. 22, 16h-173. Mitchell D. E. and Ware C. (1974) Interocular transfer of a visual after-effect in normal and stereoblind humans. J.
Physiol..
Lond.
236. 707-72 1.
Moulden B. (1974)After-effects,or a treatise of the con%quences of adaptation to movement and tilt. Unpub tished Ph.D. thesis. Universitv of Readine Movshon 1. A., Chambers B. E. i. and Blakekore C. (1972). Interocular transfer in normal humans and those who lack stereopsis. Perception 1. 483-490. Muir D. W. and Over R. (1970) Tilt after-effects in central and peripheral vision. J. exp. Psychol. 85, 165-170. Over R. and Wenderoth P. (1974) Is spatial masking selective to wavelength? Vision Res. 14, 1.57-l 58. Sagara M. and Oyama T. (1957) Experimental studies of figural after-effects in Japan. Psycho/. Bull. 54. 327-336. Sekuler R. and Littlejohn J. (1974) Tilt after-effect following very brief exposures. Vision Res. 14, 151-152. Taylor M. M. and Creelman C. D. (1967) PEST: efficient estimates on probability functions. J. ucousr. Sot. Am. 41, 782-787. Vidvasagar T. R. (1976) Orientation specific colour adaptation at a binocular site. Eiarure. Land. 261. 39-40. Wade N. J. (1975) Monocular and binocular rivalry between contours. Perception 4. 85-95. Wade N. J. (1976) On interocular transfer of the movement after-effect in individuals with and without normal binocular vision. Perception 5, 113-I 18. Winer B. 1. (1962) Statistical Principles in Experimental Design. McGraw-Hill. New York. Zcki S. M. (1973) Colour coding in rhesus monkey prestriate cortex. Brain Res. 53, 422-427.