Unfortunately , Story does not break down her data for monocular viewing according to whether T- and I-figures were on the same or opposite sides as the eye used so that this prediction would only apply to half the trials she reports .sx Nevertheless , there is no sign of this trend in her results for monocular viewing .sx ( =3 ) The effects to be expected due to the different spatial positions of the two eyes should be even more striking when the distance between shape and eye is less than in Story's experiment , and when the I-figure is shown to one eye and the T to the other :sx although these conditions have often been used in experiments on FAE no effects of this sort have been reported .sx ( It might , however , be worth looking for them in future experiments .sx ) ( =4 ) Finally , although Story suggests that the different visual angles subtended by the figures at the retinae might be the explanation of the effects obtained under binocular viewing , she does not show in detail how these effects would be predicted by the geometry of the situation , and it is difficult to see how the effects found could in fact be produced in this way .sx Nevertheless , the suggestion is an interesting one and could be followed up by experiments in which the figures are placed closer to the eye and conditions of alternating monocular viewing are employed .sx It is possible that the reason why the A-effect is obtained only when both eyes are used is that binocular vision itself provides a cue to the distance of the figures and thus to their relative apparent sizes ( v. below) :sx thus , the fact that the effect only occurs with binocular viewing does not necessarily conflict with the hypothesis that under some conditions the FAE may be determined by apparent size , and indeed can be interpreted within the framework of this hypothesis .sx Size of circles .sx If smaller circles than those used by Sutherland are employed , the A-effect does not occur ( Day and Logan , 1961 ; Terwilliger , 1961 ; McEwen 1959 ; Oyama , 1956) :sx the usual result under these conditions is that the T-circle looks smaller than C whether I is nearer or further away .sx ( It should be noted that Terwilliger did not obtain this result :sx when the retinal size of T and I was the same , he found no change in the apparent size of T. ) This effect is also found when T and I shapes are the same distance away as one another ( Day and Logan ( 1961 ) , cf .sx also " hler and Wallach ( 1944)) .sx Day and Logan make the interesting suggestion that this shrinkage may resemble a time error effect though they do not discuss the details of how this might occur .sx Unfortunately , from what is known about time errors , one might expect the opposite effect with small circles .sx When a series of stimuli are being judged , there is usually a point in the middle of the series where ( after practice ) there is no constant error :sx above this point , time errors tend to be negative , below it , positive .sx We shall call this point the " adaptation point .sx " Subjects will have an adaptation point at the start of an experiment and it will usually be shifted in the course of the experiment :sx now when a small circle is shown as I-figure this should shift the adaptation point downwards .sx If it shifts it downwards further for that part of the visual field on which the I-figure is shown than for other parts , we would expect the T-figure to be judged larger than the C-figure :sx the T-figure is less far away from the adaptation point at that part of the visual field than is the C-figure from the adaptation point at its part of the visual field .sx Day and Logan obtained exactly the opposite result to this .sx Thus , there is some difficulty in applying this type of explanation , though the correspondence between the change in direction of the FAE with different sized circles ( found by Day and Logan ) and the change in direction of TE ( found by Watson , 1957 ) is very suggestive .sx Nevertheless , Day and Logan's work does make it difficult to interpret the A-effect as due to differences in apparent size because of their finding that when large circles are used and both are far away , the T-circle appears larger than the C. Outline and filled-in circles .sx Day and Logan show that the A-effect occurs with outline circles but not with filled-in circles :sx it is hard to see what explanation could be offered for this at present .sx Further discussion .sx One very ingenious recent experiment has demonstrated in a most convincing way that an FAE determined wholly by apparent size does occur under certain conditions :sx Gregory ( personal communication ) has shown that if the apparent size of a figure is made to shrink continuously while the retinal size remains the same , when the shrinkage in apparent size is stopped suddenly there is a dramatic increase in the apparent size of the figure .sx This phenomenon is very striking and is seen by all observers .sx Since this shows that a FAE determined by continuous change in apparent size can occur , the question arises of why it is so difficult to demonstrate the effect with static figures .sx There are three possible answers to this .sx ( 1 ) It may be that just as with FAE due to retinal size , the effect through apparent size only occurs if the difference between the apparent sizes of the T- and I-figures is optimal ( cf .sx the distance paradox) .sx If this is correct , we would only expect to obtain a FAE due to apparent size under limited conditions .sx This suggestion could be tested experimentally by keeping one circle a constant size and distance and varying the size and distance of the other keeping retinal size equal .sx We would expect an effect due to apparent size to occur only within a limited range of size and distance of the other figure .sx In Gregory's experiment , because the apparent size of the inspection figure changes continuously , these changes are bound to straddle the point which would be optimal for producing the effect .sx ( 2 ) The conditions of the experiments performed with static figures are such that there may be a temptation to judge in terms of retinal size :sx it is known that when two shapes of different real size are aligned side by side , subjects tend to make judgements in terms of retinal size ( Joynson and Kirk , 1960) .sx It would be interesting to test for the occurrence of the A-effect , using for T- and C-figures two shapes of the same physical size but different retinal sizes at different distances away from the observer and not aligned opposite one another .sx The T-circle could be kept the same retinal size as the I , and the C-circle would be a different retinal size :sx subjects would be asked to compare the real size of T- and C-figures .sx These experimental conditions should tend to favour judgements in terms of apparent physical size rather than apparent retinal size .sx ( 3 ) It may be that apparent size only influences FAE when the apparent size has changed continuously , i.e. where there has been an apparent movement effect :sx if established this would be an important finding since it would reveal a difference in the mechanisms underlying apparent movement and judgements of apparent size ( v. below) .sx This could only be established by a thorough investigation of the static A-effect along the lines set out in ( 1 ) and ( 2 ) above .sx THEORETICAL CONSIDERATIONS .sx The work of Hubel and Wiesel ( 1959 ) suggests a new theoretical approach to FAE problems .sx In order to see the experiments described above in perspective , it may be worth setting out briefly what this approach is :sx it has suggested itself independently to a number of workers in the field , and Papert is currently engaged on testing some of its implications .sx It must be stressed that a new approach is necessary since the sort of theory espoused by " hler and Wallach ( 1944 ) and by Osgood and Heyer ( 1952 ) is unable to account for many of the phenomena of FAE .sx They both assume that inspection of a contour results in any contour subsequently falling near the second contour being seen as displaced away from it :sx the amount it is displaced is said to depend upon the distance separating the two contours on the retina , and there will be a point at which displacement is maximal .sx Three instances of well attested phenomena which this theory is unable to explain will be quoted .sx ( 1 ) In Figure 1 , if the I-line is fixated , the T-line should appear as shown ( P) :sx displacement should be small where I and T lie near together gradually increasing to a maximum and then decreasing .sx In fact T is seen occupying the position of line A. ( 2 ) Similarly when a curved line is shown , and a straight line used as I-figure , the straight line should appear like line P in Figure 1 ( b ) but in fact appears like line A. ( 3 ) The theories are unable to account for the after effect of seen motion .sx Both theories under discussion assume that the FAE occurs before any analysis of the stimuli is undertaken .sx Hubel and Wiesel have demonstrated by recording from single cells that in the cat considerable analysis of the stimulus on the retina occurs at or before the level of the striate cortex .sx In particular they present evidence to show that in the striate cortex there are cells whose response is determined by the orientation of lines on a given part of the retina ; i.e. the orientation of lines is coded in separate fibres at this level of the cat visual system .sx If we assume that there are cells with similar receptive fields in human beings we have a very simple explanation of the effect shown in Figure 1 ( a) :sx inspection of a line in one orientation will result in heavy firing of the cells maximally responsive to lines in this orientation , and to some firing of cells maximally responsive to lines in neighbouring orientations .sx If any adaptation occurs in these cells as a result of prolonged firing , when a T-contour in a slightly different orientation to the I line is exposed on the same part of the retina , the cells fired maximally by it will be ones which are normally maximally responsive to contours in orientations lying further away from the orientation of the I-figure .sx It is reasonable to suppose that the orientation in which a contour is seen will depend upon the balance of firing in cells representing contour orientation :sx the firing in any one cell will be determined partly by the contrast of the contour with its background , etc. , but such effects would be balanced out if the ratio of firing in all cells sensitive to orientation in a given region of the retina were computed .sx If there are also cells sensitive to curvature of a line a similar mechanism would explain the sort of finding depicted in Figure 1 ( b) .sx As yet there is no physiological demonstration of the existence of such cells .sx Hubel and Wiesel have , however , found cells which respond differentially according to the direction in which a stimulus is moved across the retina .sx If direction of movement is coded in single cells in human beings , adaptation in these cells might clearly underly [SIC] the after-effect of movement .sx Once again the direction in which something is seen to move might depend upon the ratios of firing in cells sensitive to movement in different directions , and after prolonged movement in one direction a stationary image would produce less firing in the cells which had just been stimulated than normally , hence apparent movement in the opposite direction would be seen to occur .sx This explanation of FAE is based on sound physiological evidence and is so simple that it seems highly convincing .sx It does not , however , explain mere displacements in apparent spatial position occurring as a FAE :sx for this phenomenon , the Osgood and Heyer type of explanation appears reasonably plausible .sx This explanation in fact fits well with the explanation outlined above since Osgood and Heyer argue that the position at which a contour is seen itself depends upon ratios of firing in different cells .sx