1937). There is considerable evidence that orientation contrast and tilt after-effect are both caused by .... and WK, respectively, froma serie of measurements where A was fixed at 24 deg .... Heidelberg New York, pp 3-38. Campbell FW, Maffei ...
Ex mentai BranResearch
Exp Brain Res (1981) 43:193-198
9 Springer-Verlag 1981
Inhibition, Disinhibition, and Summation Among Orientation Detectors in Human Vision* W. Kurtenbach and S. Magnussen Neurologische Universitfitsklinik mit Abteilung fiir Neurophysiologie, Hansastr. 9, D-7800 Freiburg, Federal Republic of Germany
Summary. A vertical test line viewed together with an oblique inducing line appears to be tilted in the opposite direction (orientation contrast); a similar effect results from adapting to an oblique line before presenting the test line (tilt after-effect). The error in perceived orientation caused by a clockwise tilted inducing line may be partially cancelled by a second clockwise tilted inducing line; likewise, adapting to a configuration of two clockwise orientations gives weaker tilt after-effects than adapting to the more effective of the two orientations alone. The angular functions of these "disinhibitory" effects mirror the angular functions of the principal effects. However, combining tilt after-effect and simultaneous contrast in a disinhibition design by having the subject first adapt to a clockwise orientation and then presenting the test line together with a clockwise inducing line results in partial summation rather than disinhibition. The results are consistent with Carpenter's and Blakemore's (1973) hypothesis that orientation contrast and the tilt after-effect are manifestations of cortical inhibition in a network of inhibitory, recurrent lateral connections between cortical orientation detectors. After-effects must be due to prolonged inhibition, probably caused by a sensitivity change in previously inhibited detectors. Key words: Human vision - Orientation detectorsLateral inhibition - Psychophysics
Most cells in the visual cortex of the cat and monkey respond to bars and edges over a restricted range of * Parts of this paper were presented at the annual meeting of the Association for Research in Vision and Ophthalmology, Sarasota, Florida, 30 April to 5 May, 1979, and the 3rd European Conference on Visual Perception, Brighton, England, 3 to 6 September, 1980 Offprint requests to: Dr. S. Magnussen, Institute of Psychology, University of Oslo, Box 1094 Blindern, Oslo 3, Norway
orientations and with a fairly well-defined optimal orientation (Hubel and Wiesel 1962, 1968; Campbell et al. 1968; Schiller et al. 1976). It has been suggested that lateral inhibition between such orientation detecting cells (Blakemore and Tobin 1972; Nelson and Frost 1978; Sillito 1979) forms the basis of various psychophysical interactions between line or grating stimuli that differ in orientation (Wallace 1969; Blakemore et al. 1970, 1971, 1973; Campbell et al. 1973; Thomas and Shimamura 1975; Braddick et al. 1978). The present paper examines two widely studied phenomena, both of which involve changes in perceived orientation. Orientation contrast or tilt illusion (TI) describes the case when the perceived orientation of a test line (or grating) is changed by an adjacent inducing line (grating). In the tilt after-effect (TAE) the test line is presented after the subject has adapted for some time to another orientation (Hofmann and Bielschowsky 1909; Gibson and Radner 1937). There is considerable evidence that orientation contrast and tilt after-effect are both caused by orientation-specific lateral inhibition (Carpenter and Blakemore 1973; Sekuler and Littlejohn 1974; Tolhurst and Thompson 1975; Virsu and Taskinen 1975; Magnussen and Kurtenbach 1980a, b). In particular, both effects are subject to'~ Carpenter and Blakemore (1973) found that for the TI, the orientation error induced by a line tilted 1 0 2 0 deg clockwise away from the test line was partially cancelled by a second clockwise inducing line. Likewise, Magnussen and Kurtenbach (1980b) found that when the subject adapted to two clockwise adapting lines, one of which formed a 12-deg angle with the test line, weaker tilt after-effects were obtained than by adapting to the 12-deg line alone. Since, in this experiment, the second adapting line was within the angular range of the tilt after-effect, the results are difficult to reconcile with the rivaling
0014-4819/81/0043/0193/$ 1.20
194
W. Kurtenbach and S. Magnussen: Lateral Inhibition of Orientation Detectors in Human Vision
"fatigue from excitation" hypothesis of visual aftereffects (Sutherland 1961; Coltheart 1971; Klein et al. 1974). In the present experiments we investigated the interaction between orientation contrast and tilt after-effect in some detail. Knowing that disinhibition occurs within the simultaneous contrast and within adaptation conditions, we asked whether disinhibition would also occur between them. For example, would an after-effect caused by adapting to a clockwise orientation be reduced by presenting the test line together with a clockwise inducing line?
Methods Method and procedure were described in detail in a previous paper (Magnussen and Kurtenbach 1980a). Briefly, the stimuli (Fig. 1, insets) were black lines, 0.03 deg broad and 1.3 deg long, viewed against a background of 40 cd/m 2. The contrast of the lines relative to the background was about 0.3. The stimuli were presented to both eyes using a modified three-channel tachistoscope (Scientific Prototype, Model N-1000). We used a parallel-setting technique in combination with a method of adjustment. The subject's task was to align the variable comparison line (C) with a physically vertical test line (T) below. T h i s was done by issuing "left-right" instructions to the experimenter who varied the orientation of C. C was controlled by a micrometer and its position was read to the nearest 0.07 deg. To generate an after-effect the subject viewed an adapting line (A) for 2 min, moving his eyes back and forth along a horizontal fixation bar to avoid after-images. The initial adaptation period was followed by a sequence of 1.5-s presentations of the test pattern alternating with 10-s readaptation periods; this sequence was continued until at least five settings were obtained. To measure orientation contrast, an inducing line (I) was presented together with test line (T) as seen in Fig. 1, inset. The 1.5-s presentations of the test pattern were cycled with 5-s "blanks" containing only a fixation point, until a least five settings were obtained. In most of the experiments we used more than one adapting or inducing line, the details of these experiments are described in the text. Each experimental session opened with five parallel settings of T and C in the absence of A or I. This "null" position was rechecked before adapting to each new orientation. Only small day-to-day fluctuations were observed. The authors served as subjects.
Results Control Measurements
Figure 1 presents some reference data for the main experiment. Figure la shows a set of TI and TAE measurements; open circles plot counterclockwise shifts in the perceived orientation of T for various clockwise orientations of I, solid circles show the effect of adapting to various clockwise orientations of A. In this and the following figures clockwise orienta-
(0)
(b)
BLARe; TEST
W~
o EONIRASI
BLANK
l~/l
[]
CONTRAST ~
ADAPT TEST
~%
9 AFTEREFFECT
~-Z
-
t-a-J z
-
AI (])
r
m
0
~
ADAPT lEST
~~
9 AFTEREFFECT
~'-3
TES~
I
I
I
I
I
I
I
I
I
i
I
0 I0 20 30 40 50 rio 0 I0 ZO 30 40 50 60 ORIENTATIONOF A AND l,deg ORIENTATION OF A2 AND g,deg Fig. 1. a Empty circles show perceived counterclockwise tilt of a vertical test line, T, for various clockwise orientations of a simultaneously presented inducing line, I; solid circles show perceived tilt of T after 2 min inspection of a clockwise adapting line, A. Stimulus patterns are diagrammed in insets, C is the comparison line. b Empty squares show the effect of a second inducing line, D, on the perceived orientation of T, for various clockwise orientations of D with I fixed at 12 deg clockwise tilt. Solid squares show the effect of adapting to two clockwise adapting lines one of which, A1, was fixed at 12 deg clockwise tilt and the other, A2, presented at various clockwise orientations. Arrow indicates D = I and A2 = A1, respectively, and the dashed line indicates the orientation error induced by t alone, or by adapting to A1 alone. Bars denote half standard error of the mean ($7). Where no bar is shown, ST falls within the symbol size. The two set of measurements are adjusted to a common baseline. The smooth curves in a) and b) are both fitted by eye. Results for subject WK
tions from the vertical are expressed in positive values, counterclockwise orientations in negative values. The two procedures yield very similar results, confirming the more detailed measurements of Magnussen and Kurtenbach (1980a). Figure 1 illustrates disinhibition. To demonstrate disinhibition in simultaneous contrast, a third line D was added to the display (see upper inset Fig. lb). The orientation of I was fixed at 12 deg clockwise tilt and D was varied between 12 deg (D and I superimposed) and 60 deg clockwise tilt. The results are shown as open squares. The dashed horizontal line indicates the shift in perceived orientation of T induced by I in the absence of D. For all angles tested D reduces the orientation error except in the obvious case of D = I (indicated by arrow). This confirms the results of Carpenter and Blakemore (1973). Solid squares in Fig. lb shows disinhibition in the after-effect condition. In this experiment the subject adapted to two lines, A z and A2 (see lower inset, Fig. lb). The orientation of A1 was fixed at 12 deg clockwise tilt and A2 was varied between 6 and 60 deg clockwise tilt. The dashed horizontal line
W. Kurtenbach and S. Magnussen: Lateral Inhibition of Orientation Detectors in Human Vision
195
-4
-4 A
SM
A
WK -3
z < l)
03 -E3
i
-so -40
'~ -,,o/
I
I0
I
I
20 30 ADAPT
I
I
I
40 50 60 IEST
i.-..%, -
,
-60-50
~-zo -I0 /
-40 - 3 0 " ~
,
'/
Io
[
I
20
30
I
I
40 50 60
z COUNTERCLOCKWISE
3
CLOCKWISE
ORIENTATION OF ], deg Fig. 2. Interaction between orientation contrast and tilt aftereffect. The data points show the perceived orientation of T for various orientations of I after adapting to A at 24 deg clockwise tilt (indicated by arrow). Adapting and test patterns are shown as insets. Heavy and dashed lines are theoretical curves representing, respectively, linear summation and independence of orientation contrast and tilt after-effect. Bars denote half standard error of the mean (S'i'). Where no bar is shown, Sx falls within the symbol size. Results for SM
again indicates the after-effect of adapting to A1 alone, this baseline value is higher than the values obtained for all combinations of A 1 and A2. Essentially similar results were found with A t fixed at 24 deg and A 2 presented at larger clockwise angles. The two sets of data in Fig. lb have been slightly corrected for a day-to-day shift in baseline. Again the adaptation and simultaneous contrast procedures produce orientation shifts of comparable magnitude and a single curve has been fitted by eye.
Interaction Between Adaptation and Simultaneous Contrast One way of deciding whether the processes underlying orientation contrast and the tilt after-effect are fundamentally the same is to study how the two phenomena interact. Magnussen and Kurtenbach (1980a) already showed that the effects of adaptation and simultaneous contrast sum linearly, when presented to opposite but equal angles about the test orientation. Do they also disinhibit each other? The inset to Fig. 2 shows the design of the experiment. The subject first adapted to a particular orientation of A, then T was presented together with I. Figures 2 and 3 show the results for subjects SM
COUNTERCLOCKWISE
3 ORiENIATION OF [
CLOCKWISE
deg
Fig. 3. Same as Fig. 2, but for subject WK
and WK, respectively, f r o m a serie of measurements where A was fixed at 24 deg clockwise tilt (indicated by arrow), and I varied from 53 deg counterclockwise to 60 deg clockwise tilt. The continuous lines in the graphs are theoretical curves representing linear summation of orientation contrast and after-effect. The angular function of orientation contrast for clockwise inducing angles were determined for both subjects, and smooth curves were fitted by eye to the individual results. This function is assumed to be symmetric about the test orientation (Campbell and Mallei 1971; Carpenter and Blakemore 1973; Magnussen and Kurtenbach 1980a), and the whole curve is shifted along the ordinate by a factor which corresponds to the orientation error caused by adaptation to 24 deg clockwise tilt alone(zero value of I). Consider first the left part of the graphs, showing results for counterclockwise orientations of I. This is essentially an extension of the experiment by Magnussen and Kurtenbach (1980a) to asymmetric orientations of A and I. It shows that an after-effect of clockwise adaptation may be reduced, cancelled or even counteracted by a simultaneous contrast effect of variable strength from the opposite side of the test line. The results in this part of the figure are from a single run (n = 5) and there is some variability among the data points, in particular for WK. But to a first approximation they are well described by the theoretical curve. Consider next the right part of the graphs, showing results for clockwise orientations of I. This part averages results from two runs (n -- 10). The dashed lines are theoretical curves indicating results
196
W. Kurtenbach and S. Magnussen: Lateral Inhibition of Orientation Detectors in Human Vision
,I::D
orientation error induced by adaptation, whereas combining two lines simultaneously or adapting to two lines may decrease the effect by as much as 0.75 deg. At inducing angles of 36 deg and beyond the shift in the apparent orientation of T is determined by adaptation alone.
I
A
oWK oSM
-3-
J
I
i'(~0..
SUMMATION
LI
Discussion IiI
OlSINHIBITION I
z
O __
I
I
I
I
I
0 I0 ZO 3O 40 50 60 CLOCKWISE ORIENTATION OFI,deg Fig. 4. Interaction between orientation contrast and tilt aftereffect. In this experiment the subject adapted to A at 12 deg clockwise tilt (indicated by arrow), and only clockwise orientations of I were tested. The dashed line defines the border between disinhibition and partial summation (i.e., independence of orientation contrast and tilt after-effect). In this case this is identical to the effect of A or I alone. The smooth curve is reproduced from Fig.lb. Results for both subjects, adjusted to a common baseline. Vertical bars indicate half standard error of the mean
predicted by independence of the two effects. In this case the shift in apparent orientation of T would be determined by whichever effect is strongest. Data points falling below this curve would thus indicate disinhibition. For both subjects the measured points lie between the limits of complete summation and independence. In other words, combining adaptation and simultaneous contrast in a disinhibition design does not result in disinhibition but in partial summation. We have tested this finding several times, always with the same result. Figure 4 shows the results of an experiment where A was fixed at 12 deg clockwise tilt and I presented at various clockwise angles. The dashed horizontal line indicates the effect of adapting to A alone, defining the border between disinhibition and partial summation. Results are shown for both subjects, those for SM having been slightly adjusted along the ordinate to form a common baseline with WK. This experiment is analogous to the disinhibition experiments in Fig. lb, and for comparision we have reproduced the smooth curve from this figure. The results, however, are very different. At inducing angles below 36 deg, the data points fall clearly above the demarcation line and show that simultaneous contrast may add as much as 0.75 deg to the
Blakemore et al. (1970) proposed that when two stimuli form an acute angle, inhibition occurs between cortical units which code these neighbouring orientations. The resulting distribution of excitation in the population of orientation detectors is skewed. The excitation peaks are shifted to slightly more extreme orientations. Assuming that the brain identifies stimulus orientation on the basis of its most active detectors, the perceived orientation of such lines should be distorted, compared to when seen alone. Carpenter and Blakemore (1973) further showed that a simple network of recurrent inhibitory connections between orientation specific cells may explain the perceptual distortion. The same inhibition hypothesis can be used to explain the tilt after-effect on the additional assumption that it is an after-effect of prolonged inhibition (Blakemore et al. 1971; Magnussen and Kurtenbach 1980a). Some of the basic results which support the inhibition hypothesis are replicated in Fig. 1. It is noteworthy that our measurements not only confirm the presence of disinhibition in both simultaneous contrast and tilt after-effect, but that the curves also look like compressed mirror images of the parental effects, as expected if the effect is caused by inhibition. The present experiments were restricted to angles where only direct (or repulsion) effects are found: If the angle between the two lines is between 0 and 60 deg, the lines appear tilted away from each other. Between 60 and 90 deg indirect (or attraction) effects are frequently observed (Hofmann and Bielschowsky 1909; Gibson and Radner 1937): the two lines appear tilted towards each other. Such indirect effects are possibly manifestations of disinhibition (O'Toole and Wenderoth 1977; O'Toole 1979) as well. In agreement with the inhibition hypothesis, Magnussen and Kurtenbach (1980a) found that aftereffect and orientation contrast summed linearly if the subject first adapted to a line tilted clockwise from the test line and then viewed the testline together with an equally tilted counter-clockwise inducing line. The present experiments show linear summation throughout the tested range, when the adapting and inducing orientations are also asymmetrical
w. Kurtenbach and S. Magnussen: Lateral Inhibition of Orientation Detectors in Human Vision about the test orientation. The perceived orientation error is quite accurately predicted by an additivity rule (left parts of Figs. 2 and 3). We thus confirm and extend our previous finding that orientation contrast and tilt after-effects of opposite signs sum linearly, giving strong support to the idea of a c o m m o n inhibitory mechanism. The results from the main experiment, in which the adapting and inducing lines were both clockwise with respect to the test line (right part of Figs. 2, 3, and 4), may at first be m o r e difficult to reconcile with the inhibition hypothesis. The adding of a third line to the display produced smaller effects when tested within either the adaptation or simultaneous condition (Fig. 1). However, the effect increased when the two procedures were combined. W h e n the subject first adapted to a clockwise orientation and then viewed the test line together with a clockwise inducing line, the test line looked even m o r e tilted than by adaptation or simultaneous stimulation alone. In other words, once established, a T A E m a y be cancelled by a competing T I of opposite orientation, but it cannot be r e m o v e d by a T I induced on the same side as the test orientation. H o w can inhibition persist over several minutes to generate after-effects? According to Carpenter and B l a k e m o r e (1973) the simultaneous contrast is caused by active lateral inhibition, transmitted from stimulated detectors to its neighbours. Because of the similarity of both effects, the after-effect might simply be explained by prolonged active inhibition. This suggests a continued active transmission of inhibitory signals from previously excited detectors onto their neighbouring detectors. Such a prolonged active inhibition would share the main properties of conventional inhibition and would not be subject to disinhibition if it operates through interneurons and with recurrent, presynaptic connections. There is another possibility that appear biologically more plausible: Detectors which received inhibition from stimulated detectors during the adaptation period may b e c o m e less sensitive, i.e., they adapt, according to the amount of inhibition received. They therefore act for a certain time after the stimulus has been switched off, as if they were still inhibited. Such a synaptical depression would of course not be subject to disinhibition but its inhibitory effect must sum with inhibitory signals. Partial and not complete summation might be found, because a maximally adapted neuron is no longer able to sum active inhibition and synaptical depression linearly. Our psychophysical data cannot decide between the cellular mechanisms behind the prolonged inhibition, but well-known inhibitory mechanisms can
197
account for the disinhibition and summation found. This is also supported by neurophysiological data (Creutzfeldt and Heggelund 1974; Maffei et al. 1973; Movshon and Lennie 1979; Vautin and Berkley 1977). Finally, the hypothesis of inter-channel inhibition is not unique to the orientation domain but has been invoked to account for other psychophysical interaction p h e n o m e n a within spatial frequency, movement, and retinal disparity as well (Braddick et al. 1978; Sekuler et al. 1978; Nelson 1977; Stromeyer et al. 1977). Whilst it is likely that a c o m m o n principle is involved, it remains to be seen whether the interactions between simultaneous contrast and after-effects demonstrated here for orientation have counter-parts along other spatial stimulus dimensions.
Acknowledgements. S.M. thanks the Alexander-von-Humboldt Foundation (Bonn-Bad Godesberg, FRG) and NAVF (Norway) for generous support during a sabbatical leave in Freiburg. W.K. received support through SFB-70, Teilprojekt A6 (Spillmann), We thank Prof. L. Spillmann for providing laboratory facilities and for comments on an earlier draft, T. G/inter for modifying the tachistoscope, and J. Humburger for preparing the figures.
References Blakemore C, Tobin EA (1972) Lateral inhibition between orientation detectors in the cat's visual cortex. Exp Brain Res 15:43%440 Blakemore C, Carpenter RHS, Georgeson MA (1970) Lateral inhibition between orientation detectors in the human visual system. Nature 228:37-39 Blakemore C, Carpenter RHS, Georgeson MA (1971) Lateral thinking about lateral inhibition. Nature 234:418-419 Blakemore C, Muncey JP, Ridley RM (1973) Stimulus specificity in the human visual system. Vision Res 13:1915-1931 Braddick O, Campbell FW, Atkinson J (1978) Channels in vision. Basic aspects. In: Held R, Leibowitz HW, Teuber HL (eds) Handbook of sensory physiology, vol VIII. Springer, Berlin Heidelberg New York, pp 3-38 Campbell FW, Maffei L (1971) The tilt after-effect. A fresh look. Vision Res 11:833-840 Campbell FW, Cleland BG, Cooper GF, Enroth-Cugell C (1968) The angular selectivity of visual cortical cells to moving gratings. J Physiol (Lond) 198:237-250 Campbell FW, Gilinsky AS, Howell ER, Riggs LA (1973) The dependence of monocular rivalry on orientation. Perception 2:123-125 Carpenter RHS, Blakemore C (1973) Interaction between orientations in human vision. Exp Brain Res 18:287-303 Coltheart M (1971) Visual feature-analyzers and the after-effects of tilt and curvature. Psychol Rev. 78:114-121 Creutzfeldt OD, Heggelund P (1974) Neural plasticity in visual cortex of adult cats after exposure to visual patterns. Science 188:1025-1027 Gibson JJ Radner M (1937) Adaptation, after-effect and contrast in the perception of tilted lines. I. Quantitative studies. J Exp Psychol 20:452-467 Hofmann FB, Biclschowsky A (1909) Ober die Einstellung der scheinbaren Horizontalen und Vertikalen bei Betrachtung
t98
W. Kurtenbach and S. Magnnssen: Lateral Inhibition of Orientation Detectors in Human Vision
eines von schr/~gen Konturen erffillten Gesichtsfeldes. Pflfigers Arch 126:453-475 Hubel, DH, Wiesel TN (1962) Receptive fields, binocular interaction and functional architecture in the cat's visual cortex. J Physiol (Lond) 160:106-154 Hubel DH, Wiesel TN (1968) Receptive fields and functional architecture of monkey visual cortex. J Physiol (Lond) 195: 215-243 Klein S, Stromeyer CF III, Ganz L (1974) The simultaneous spatial frequency shift. A dissociation between the detection and perception of gratings. Vision Res 14:1421-1432 Mallei L, Fiorentini A, Bisti S (1973) Neural correlate of perceptual adaptation to gratings. Science 182:1036-1038 Magnussen S, Kurtenbach W (1980a) Linear summation of tilt illusion and tilt aftereffect. Vision Res 20:39-42 Magnussen S, Kurtenbach W (1980b) Adapting to two orientations. Disinhibition in a visual aftereffect. Science 207: 908-909 Movshon JA, Lennie P (1979) Pattern-selective adaptation in Visual cortical neurones. Nature 278:850-852 Nelson JI (1977) The plasticity of correspondence. After-effects, illusions and horopter shifts in depth perception J Theor Biol 66:203-266 Nelson JI, Frost BJ (1978) Orientation selective inhibition from beyond the classic visual receptive field. Brain Res 139: 35%365 O'Toole BI (t979) Exposure-time and spatial-frequency effects in the tilt illusion. Perception 8:557-564 O'Toole B, Wenderoth PM (1977) The tilt illusion. Repulsion and attraction effects in the oblique meridian. Vision Res 17: 367-374 Schiller PH, Finlay BL, Volman SF (1976) Quantitative studies of
single-cell properties in monkey striate cortex. II. Orientation specificity and ocular dominance. J Neurophysiol 39: 1320-1333 Sekuler R, Littlejobn J (1974) Tilt after-effect following very brief exposures. Vision Res 14:151-152 Sekuler R, Pantle A, Leviuson E (1978) Physiological basis of motion perception. In: Held R, Leibowitz HW, Teuber HL (eds) Handbook of sensory physiology, vol VIII. Springer, Berlin Heidelberg New York, pp 67-96 Sillito AM (1979) Inhibitory mechanisms influencing complex cell orientation selectivity and their modification at high resting discharge levels. J Physiol (Lond) 289:33-53 Stromeyer III CF, Klein S, Sternheim CE (1977) Is spatial adaptation caused by prolonged inhibition? Vision Res 17: 603-606 Sutherland NS (1961) Figural after-effects and apparent size. Q J Exp Psychol 13:222-228 Thomas JP, Shimamura KK (1975) Inhibitory interaction between visual pathways tuned to different orientations. Vision Res 15:1373-1380 Tolhurst D, Thompson PG (1975) Orientaton illusions and aftereffects. Inhibition between channels. Vision Res 15:96%972 Vautin RG, Berkley MA (1977) Responses of single cells in cat visual cortex to prolonged stimulus movement. Neural correlates of visual aftereffects. J Neurophysiol 40:1051-1065 Virsu V, Taskinen H (1975) Central inhibitory interactions in human vision. Exp Brain Res 23:65-74 Wallace GK (1969) The critical distance of interaction in the Z611ner illusion. Percept Psychophys 5:261-264
Received May 19, 1980