Psycho gicd Research PsychologischeForschung
Psychol Res (1992) 54:240-245
© Springer-Verlag 1992
Motion aftereffects with rotating ellipses Paola Bressanl, Lucia Tomatl, and Giorgio Vallortigara2* 1 Dipartimento di Psicologia Generale, Universit~ di Padova, Piazza Capitaniato 3, 1-35139 Padova, Italy 2 Istituto di Filosofia, Pedagogia, Didattica delle Lingue Moderne, Universit~t di Udine, Via Antonini 8, I-33100 Udine, Italy Received February 17, 1992/Accepted May 15, 1992
Summary. The perceptual outcome and the motionaftereffect duration generated by the rotation on the frontal plane of an ellipse with a bar depend on whether the bar is placed along the major or the minor axis. When the bar is placed along the minor axis, a stereokinetic transformation occurs, and the pattern looks like a tilting ring with a perpendicular bar moving rigidly with it. Placing the bar along the major axis prevents the stereokinetic transformation: subjects report deformations and relative motion of the bar with respect to the ellipse. We found that motion aftereffects last longer when the bar is placed along the minor rather than along the major axis. A series of experiments was carried out to investigate whether differences in aftereffect duration are related to the stereokinetic transformation. Results seem to suggest that they are not.
Introduction After prolonged observation of a stimulus moving in one direction, a stationary stimulus appears to move in the opposite direction. This phenomenon is called motion aftereffect (MAE) and has been widely investigated (for a review see Anstis, 1986). It is usually explained in terms of Sutherland's (1961) ratio model, according to which the perceived direction of motion of a stimulus depends on the ratio of the firing rates of motion detectors tuned to opposite directions. The model claims that in order for movement to be seen in one particular direction, the firing rate of detectors tuned to that direction must exceed, by a certain minimum ratio, the firing rate of detectors tuned to the opposite direction. After adaptation to one motion direction, the corresponding detectors will fire less, allowing those tuned to the opposite direction to exceed the critical * The authors' names are in alphabetical order. Correspondence to: P. Bressan or G. Vallortigara
ratio and to produce, as a consequence, erroneous motion signals in the presence of a stationary stimulus. MAEs have been studied with both translatory and rotatory motion (see Holland, 1965). As of yet, however, little attention has been paid to the fact that with some figures rotation in the frontoparallel plane produces the impression of rigid or nonrigid objects moving in 3-D space. A paradigmatic case of "stereokinetic effect" (STKE) can be easily observed when a contour ellipse is slowly rotated (10-30 rpm) on the frontal plane (see Musatti, 1924; Bressan & Vallortigara, 1986; Vallortigara, Bressan, & Zanforlin, 1986). After brief inspection, one usually perceives first a deforming amoeba-like object and then a rigid circular ring tilting in 3-D space. It has been reported that after long observation of some of these STK patterns (e. g., Benussi's cone) compelling impressions of depth following cessation of rotation occur, and 3-D motion aftereffects can also be observed (Piggins, Robinson, & Wilson, 1984). Although this might be due to the stimulation of motion-in-depth neural channels (see Regan & Beverley, 1978), it may also be that some sort of cognitive effect is at work in these cases, since a long exposition to an object moving in 3-D space may favour interpretation of the ordinary MAE as an MAE in depth. Apart from the depth effect, STK patterns exhibit another interesting feature. Perception of 3-D objects is usually preceded by perception of relative motion between parts of the 2-D pattern (deformations). One wonders whether these relative-motion percepts may somehow affect MAE. There are grounds for thinking that this is a reasonable question, because of the psychophysical evidence for adaptation mechanisms that are selectively specific to changing size (Beverley & Regan, 1979), motion in depth (Regan & Beverley, 1978), and rotation in depth (Petersik, Shepard, & Malsh, 1984). Unfortunately, perceptual changes associated with the inspection of STK patterns usually occur at different times for different subjects; this makes the assessment of the effect associated with the STK transformation on MAE experimentally impossible. However, Zanforlin (1991) has recently shown that the orientation of certain features
241
o
O
Q
b
(D
a
Fig. la, b. Patterns used in Experiment 1 (top) and Experiment2 (bottom). The bar was placed either along the major axis (1 a) or along the minor axis (1 b) of the ellipse
3O
•
test with bar
responding pattern of proximal stimulation consists of alternate expansion and contraction on the same regions of the retina, a reduction of the usual counterrotation aftereffect may be observed. Second, apart from the depth effect, when the bar is placed along the minor axis, the whole pattern appears to move rigidly, whereas when the bar is placed along the major axis, the bar and the ellipse appear to move independently of each other. Since there is evidence that MAEs can be influenced by high-level, cognitive-like factors (e. g., attention, Chauduri, 1990), one may then argue that these alternative interpretations of the pattern could somehow affect MAE. Both hypotheses, therefore, predict either qualitatively different aftereffects (e. g., depth aftereffects when the bar is placed along the minor axis) or different aftereffect duration (e. g., longer duration, depending on whether the bar is placed along the major or the minor axis).
[ ] test without bar
Experiment 1 Method
2O
Subjects. Thirty university students, naive in relation to the purpose of the experiment, served as subjects.
15
Procedure. Stimuli were created with an Apple Macintosh IIx computer 10
0
i
along minor axis
along major axis
orientation of the bar
Fig. 2. MAE duration(mean ± SE) as a functionof the orientationof the bar and of the stimulusin the test phase (Experiment 1: long-bar condition) within otherwise identical STK figures determines how fast the depth effect occurs. An ellipse with a bar placed along the minor axis (see Figure 1 b) rapidly gives rise to an STK transformation (the pattern appears like a sort of rigid umbrella tilting in 3-D space). If the same bar is placed along the major axis (see Figure 1 a), on the other hand, no rigid motion of the whole is normally perceived: subjects report deformations of the ellipse and relative motion of the bar with respect to the ellipse. Although 3-D percepts do occasionally occur, in naive subjects the pattern needs to be observed for a very long time. Moreover, even when the ellipse looks like a tilting ring, the bar always appears to move at a different speed, so that no solid motion of the whole ever occurs for this pattern in either 2-D or 3-D space. Do these different perceptual outcomes affect the quality or the duration of MAE? There are at least two reasons for supposing that they might. First, perception of motion in depth during adaptation might lead to a qualitatively different aftereffect, i.e., to an MAE in depth (Regan, Beverley, & Cynader, 1979). Alternatively, since the cor-
and displayed on its 13-in screen. Subjects were divided into two groups of 15. In the adaptation phase (2 rain) subjects from both groups were presented with a pattern rotating at a speed of 27 rpm consisting of a contour ellipse (major axis 77 mm, minor axis 43 ram, contour thickness 4 mm) and a bar (length 77 ram, thickness 2 ram) placed along either the major or the minor axis of the ellipse (see Figure 1, top row). Each subject saw both patterns (in the first group, 6 subjects with the order minor-major and 9 with the order major-minor: in the second group, 8 subjects with the order minor-major and 7 subjects with the order major-minor). In the test phase, the first group was presented with a stationary pattern identical to that of the adaptation phase (i. e., ellipse plus bar), the second group with the ellipse alone. Since none of our subjects was familiar with motion aftereffects, before the experiment all observers were individually shown the phenomenon. They were presented with a rotating outline square (side length 53 mm, contour thickness 4 mm, speed 14 rpm) for 2 rain (adaptation phase), and then with the same stimulus at rest (test phase). All subjects reported a motion aftereffect. Subjects were tested individually. They sat at a distance of 156 cm from the screen. Vision was monocular. After the moving stimulus had been replaced by the stationary test pattern, observers pressed a key when an impression of movement began and pressed it again when the impression ceased (cumulative recordings were possible). Each trial was followed by an interval of about 2 min during which subjects were asked to report what they had seen during the adaptation phase, and in particular whether rigid motion in 3-D space or relative motion between the parts of the pattern had been experienced.
Results Results are shown in Figure 2. The analysis of variance revealed a significant main effect of the Orientation of the bar within the ellipse, F(1,26) = 6.579, p = .015, and of the Presence of the bar in the test phase, F(1,26)= 5.968, p = .020. Neither the main effect of Order of presentation, F(1,26) = 0.543, nor any of the interactions reached significance (Presence of the bar x Order of presentation,
242 30
[ ] test with bar [ ] test without bar
25
20
15
10
O
i
along minor axis
Experiment 2 It may be objected that the two adapting patterns of Experiment 1 are not comparable, in that the bar stopped exactly at the boundaries of the ellipse when it was oriented along the major axis, whereas it crossed the boundaries of the ellipse when it was oriented along the minor axis. One may observe that the presence of this crossing could have been crucial in producing the differences observed in MAE. Since the ellipse approximates a circle and therefore a condition of absence of optical stimulation for rotation, the crossing might have provided stronger local cues for perceiving a sliding of the bar with respect to the ellipse. This issue was addressed directly by a control experiment.
Method along major axis
orientation of the bar
Fig. 3. MAE duration (mean _+ SE) as a function of the orientation of the bar and of the stimulus in the test phase (Experiment 2: short-bar condition)
F(1,26) = 0.001; Orientation of the bar x Presence of the bar, F(1,26) = 0.558; Order of presentation x Orientation of the bar, F(1,26) = 0.016; Presence × Orientation × Order, F(1,26) = 0.003). The main findings were as follows. First, MAE was longer after adaptation to the pattern when the bar was placed along the minor rather than along the major axis of the ellipse. This occurred irrespective of the presence or absence of the bar during the test phase. Second, MAE was longer when the adaptation and test stimuli were identical (i. e., ellipse plus bar during both the adaptation and the test phases) than when they were different (i. e., ellipse plus bar during the adaptation and ellipse alone during the test phase). This occurred irrespective of the orientation of the bar during the adaptation phase. Out of 30 subjects, 23 reported seeing a tilting ring with a perpendicular bar (the "umbrella" percept) when viewing the pattern with the bar placed along the minor axis of the ellipse. This percept was remarkably stable and the whole configuration looked rigid. None of these subjects reported a similar impression when the bar was placed along the major axis of the ellipse. In this case the ellipse underwent deformations and distortions and the bar seemed to move at a different speed. Two subjects reported seeing the tilting ring with both bar orientations, but experienced rigid movement of the whole only when the bar was placed along the minor axis. Finally, 5 subjects were unable to obtain any stereokinetic effects, and therefore reported distortions and relative movements with both patterns. Differences in MAE durations showed the same trend even in these subjects, though the difference was not significant (minor axis 20.6+ 12.3 s; major axis 10.6+3.7 s). To sum up, it seems that, contrary to our expectations, no qualitatively different aftereffects were produced by the two patterns. Moreover, there were indeed differences in MAE duration, but in the opposite direction from that expected.
Subjects. Twenty-eight university subjects served as subjects. None had taken part in the previous experiment.
Procedure. Apparatus, procedure, and design were exactly the same as in Experiment 1. The only difference was the length of the bar placed inside the ellipse, which this time was shorter (14 mm rather than 77 ram) so that, irrespective of its orientation, it did not cross the contour of the ellipse (see Figure 1, bottom row).
Resul~ Results are shown in Figure 3. The ANOVA revealed a significant effect of the Orientation of the bar, F(1,23) = 15.103, p = .001, and a close-to-significant effect of the Presence of the bar during the test phase, F(1,23) = 3.553, p = .069. Neither the main effect of Order of presentation, F(1,23) = 0.165, nor any of the interactions reached significance (Presence of the bar x Order of presentation, F(1,23) = 0.339; Presence of the bar x Orientation of the bar, F(1,23)= 2.478; Order of Presentation x Orientation, F (1,23) = 0.618; Presence x Order x Orientation, F(1,23)= 0.098). The results were therefore very similar to those obtained in the previous experiment. The only difference was that the effect of the presence of the bar in the test phase was clearly significant in Experiment 1, and only close to significance in Experiment 2. To understand whether this reflected a substantial difference or simply chance variations, we carried out a general ANOVA putting together the sets of data from the two experiments (and taking the bar length as a factor). The ANOVA revealed two significant main effects, that is, Orientation of the bar with respect to the ellipse, F(1,53) = 22.417,p
Psychol Res (1992) 54:240-245
© Springer-Verlag 1992
Motion aftereffects with rotating ellipses Paola Bressanl, Lucia Tomatl, and Giorgio Vallortigara2* 1 Dipartimento di Psicologia Generale, Universit~ di Padova, Piazza Capitaniato 3, 1-35139 Padova, Italy 2 Istituto di Filosofia, Pedagogia, Didattica delle Lingue Moderne, Universit~t di Udine, Via Antonini 8, I-33100 Udine, Italy Received February 17, 1992/Accepted May 15, 1992
Summary. The perceptual outcome and the motionaftereffect duration generated by the rotation on the frontal plane of an ellipse with a bar depend on whether the bar is placed along the major or the minor axis. When the bar is placed along the minor axis, a stereokinetic transformation occurs, and the pattern looks like a tilting ring with a perpendicular bar moving rigidly with it. Placing the bar along the major axis prevents the stereokinetic transformation: subjects report deformations and relative motion of the bar with respect to the ellipse. We found that motion aftereffects last longer when the bar is placed along the minor rather than along the major axis. A series of experiments was carried out to investigate whether differences in aftereffect duration are related to the stereokinetic transformation. Results seem to suggest that they are not.
Introduction After prolonged observation of a stimulus moving in one direction, a stationary stimulus appears to move in the opposite direction. This phenomenon is called motion aftereffect (MAE) and has been widely investigated (for a review see Anstis, 1986). It is usually explained in terms of Sutherland's (1961) ratio model, according to which the perceived direction of motion of a stimulus depends on the ratio of the firing rates of motion detectors tuned to opposite directions. The model claims that in order for movement to be seen in one particular direction, the firing rate of detectors tuned to that direction must exceed, by a certain minimum ratio, the firing rate of detectors tuned to the opposite direction. After adaptation to one motion direction, the corresponding detectors will fire less, allowing those tuned to the opposite direction to exceed the critical * The authors' names are in alphabetical order. Correspondence to: P. Bressan or G. Vallortigara
ratio and to produce, as a consequence, erroneous motion signals in the presence of a stationary stimulus. MAEs have been studied with both translatory and rotatory motion (see Holland, 1965). As of yet, however, little attention has been paid to the fact that with some figures rotation in the frontoparallel plane produces the impression of rigid or nonrigid objects moving in 3-D space. A paradigmatic case of "stereokinetic effect" (STKE) can be easily observed when a contour ellipse is slowly rotated (10-30 rpm) on the frontal plane (see Musatti, 1924; Bressan & Vallortigara, 1986; Vallortigara, Bressan, & Zanforlin, 1986). After brief inspection, one usually perceives first a deforming amoeba-like object and then a rigid circular ring tilting in 3-D space. It has been reported that after long observation of some of these STK patterns (e. g., Benussi's cone) compelling impressions of depth following cessation of rotation occur, and 3-D motion aftereffects can also be observed (Piggins, Robinson, & Wilson, 1984). Although this might be due to the stimulation of motion-in-depth neural channels (see Regan & Beverley, 1978), it may also be that some sort of cognitive effect is at work in these cases, since a long exposition to an object moving in 3-D space may favour interpretation of the ordinary MAE as an MAE in depth. Apart from the depth effect, STK patterns exhibit another interesting feature. Perception of 3-D objects is usually preceded by perception of relative motion between parts of the 2-D pattern (deformations). One wonders whether these relative-motion percepts may somehow affect MAE. There are grounds for thinking that this is a reasonable question, because of the psychophysical evidence for adaptation mechanisms that are selectively specific to changing size (Beverley & Regan, 1979), motion in depth (Regan & Beverley, 1978), and rotation in depth (Petersik, Shepard, & Malsh, 1984). Unfortunately, perceptual changes associated with the inspection of STK patterns usually occur at different times for different subjects; this makes the assessment of the effect associated with the STK transformation on MAE experimentally impossible. However, Zanforlin (1991) has recently shown that the orientation of certain features
241
o
O
Q
b
(D
a
Fig. la, b. Patterns used in Experiment 1 (top) and Experiment2 (bottom). The bar was placed either along the major axis (1 a) or along the minor axis (1 b) of the ellipse
3O
•
test with bar
responding pattern of proximal stimulation consists of alternate expansion and contraction on the same regions of the retina, a reduction of the usual counterrotation aftereffect may be observed. Second, apart from the depth effect, when the bar is placed along the minor axis, the whole pattern appears to move rigidly, whereas when the bar is placed along the major axis, the bar and the ellipse appear to move independently of each other. Since there is evidence that MAEs can be influenced by high-level, cognitive-like factors (e. g., attention, Chauduri, 1990), one may then argue that these alternative interpretations of the pattern could somehow affect MAE. Both hypotheses, therefore, predict either qualitatively different aftereffects (e. g., depth aftereffects when the bar is placed along the minor axis) or different aftereffect duration (e. g., longer duration, depending on whether the bar is placed along the major or the minor axis).
[ ] test without bar
Experiment 1 Method
2O
Subjects. Thirty university students, naive in relation to the purpose of the experiment, served as subjects.
15
Procedure. Stimuli were created with an Apple Macintosh IIx computer 10
0
i
along minor axis
along major axis
orientation of the bar
Fig. 2. MAE duration(mean ± SE) as a functionof the orientationof the bar and of the stimulusin the test phase (Experiment 1: long-bar condition) within otherwise identical STK figures determines how fast the depth effect occurs. An ellipse with a bar placed along the minor axis (see Figure 1 b) rapidly gives rise to an STK transformation (the pattern appears like a sort of rigid umbrella tilting in 3-D space). If the same bar is placed along the major axis (see Figure 1 a), on the other hand, no rigid motion of the whole is normally perceived: subjects report deformations of the ellipse and relative motion of the bar with respect to the ellipse. Although 3-D percepts do occasionally occur, in naive subjects the pattern needs to be observed for a very long time. Moreover, even when the ellipse looks like a tilting ring, the bar always appears to move at a different speed, so that no solid motion of the whole ever occurs for this pattern in either 2-D or 3-D space. Do these different perceptual outcomes affect the quality or the duration of MAE? There are at least two reasons for supposing that they might. First, perception of motion in depth during adaptation might lead to a qualitatively different aftereffect, i.e., to an MAE in depth (Regan, Beverley, & Cynader, 1979). Alternatively, since the cor-
and displayed on its 13-in screen. Subjects were divided into two groups of 15. In the adaptation phase (2 rain) subjects from both groups were presented with a pattern rotating at a speed of 27 rpm consisting of a contour ellipse (major axis 77 mm, minor axis 43 ram, contour thickness 4 mm) and a bar (length 77 ram, thickness 2 ram) placed along either the major or the minor axis of the ellipse (see Figure 1, top row). Each subject saw both patterns (in the first group, 6 subjects with the order minor-major and 9 with the order major-minor: in the second group, 8 subjects with the order minor-major and 7 subjects with the order major-minor). In the test phase, the first group was presented with a stationary pattern identical to that of the adaptation phase (i. e., ellipse plus bar), the second group with the ellipse alone. Since none of our subjects was familiar with motion aftereffects, before the experiment all observers were individually shown the phenomenon. They were presented with a rotating outline square (side length 53 mm, contour thickness 4 mm, speed 14 rpm) for 2 rain (adaptation phase), and then with the same stimulus at rest (test phase). All subjects reported a motion aftereffect. Subjects were tested individually. They sat at a distance of 156 cm from the screen. Vision was monocular. After the moving stimulus had been replaced by the stationary test pattern, observers pressed a key when an impression of movement began and pressed it again when the impression ceased (cumulative recordings were possible). Each trial was followed by an interval of about 2 min during which subjects were asked to report what they had seen during the adaptation phase, and in particular whether rigid motion in 3-D space or relative motion between the parts of the pattern had been experienced.
Results Results are shown in Figure 2. The analysis of variance revealed a significant main effect of the Orientation of the bar within the ellipse, F(1,26) = 6.579, p = .015, and of the Presence of the bar in the test phase, F(1,26)= 5.968, p = .020. Neither the main effect of Order of presentation, F(1,26) = 0.543, nor any of the interactions reached significance (Presence of the bar x Order of presentation,
242 30
[ ] test with bar [ ] test without bar
25
20
15
10
O
i
along minor axis
Experiment 2 It may be objected that the two adapting patterns of Experiment 1 are not comparable, in that the bar stopped exactly at the boundaries of the ellipse when it was oriented along the major axis, whereas it crossed the boundaries of the ellipse when it was oriented along the minor axis. One may observe that the presence of this crossing could have been crucial in producing the differences observed in MAE. Since the ellipse approximates a circle and therefore a condition of absence of optical stimulation for rotation, the crossing might have provided stronger local cues for perceiving a sliding of the bar with respect to the ellipse. This issue was addressed directly by a control experiment.
Method along major axis
orientation of the bar
Fig. 3. MAE duration (mean _+ SE) as a function of the orientation of the bar and of the stimulus in the test phase (Experiment 2: short-bar condition)
F(1,26) = 0.001; Orientation of the bar x Presence of the bar, F(1,26) = 0.558; Order of presentation x Orientation of the bar, F(1,26) = 0.016; Presence × Orientation × Order, F(1,26) = 0.003). The main findings were as follows. First, MAE was longer after adaptation to the pattern when the bar was placed along the minor rather than along the major axis of the ellipse. This occurred irrespective of the presence or absence of the bar during the test phase. Second, MAE was longer when the adaptation and test stimuli were identical (i. e., ellipse plus bar during both the adaptation and the test phases) than when they were different (i. e., ellipse plus bar during the adaptation and ellipse alone during the test phase). This occurred irrespective of the orientation of the bar during the adaptation phase. Out of 30 subjects, 23 reported seeing a tilting ring with a perpendicular bar (the "umbrella" percept) when viewing the pattern with the bar placed along the minor axis of the ellipse. This percept was remarkably stable and the whole configuration looked rigid. None of these subjects reported a similar impression when the bar was placed along the major axis of the ellipse. In this case the ellipse underwent deformations and distortions and the bar seemed to move at a different speed. Two subjects reported seeing the tilting ring with both bar orientations, but experienced rigid movement of the whole only when the bar was placed along the minor axis. Finally, 5 subjects were unable to obtain any stereokinetic effects, and therefore reported distortions and relative movements with both patterns. Differences in MAE durations showed the same trend even in these subjects, though the difference was not significant (minor axis 20.6+ 12.3 s; major axis 10.6+3.7 s). To sum up, it seems that, contrary to our expectations, no qualitatively different aftereffects were produced by the two patterns. Moreover, there were indeed differences in MAE duration, but in the opposite direction from that expected.
Subjects. Twenty-eight university subjects served as subjects. None had taken part in the previous experiment.
Procedure. Apparatus, procedure, and design were exactly the same as in Experiment 1. The only difference was the length of the bar placed inside the ellipse, which this time was shorter (14 mm rather than 77 ram) so that, irrespective of its orientation, it did not cross the contour of the ellipse (see Figure 1, bottom row).
Resul~ Results are shown in Figure 3. The ANOVA revealed a significant effect of the Orientation of the bar, F(1,23) = 15.103, p = .001, and a close-to-significant effect of the Presence of the bar during the test phase, F(1,23) = 3.553, p = .069. Neither the main effect of Order of presentation, F(1,23) = 0.165, nor any of the interactions reached significance (Presence of the bar x Order of presentation, F(1,23) = 0.339; Presence of the bar x Orientation of the bar, F(1,23)= 2.478; Order of Presentation x Orientation, F (1,23) = 0.618; Presence x Order x Orientation, F(1,23)= 0.098). The results were therefore very similar to those obtained in the previous experiment. The only difference was that the effect of the presence of the bar in the test phase was clearly significant in Experiment 1, and only close to significance in Experiment 2. To understand whether this reflected a substantial difference or simply chance variations, we carried out a general ANOVA putting together the sets of data from the two experiments (and taking the bar length as a factor). The ANOVA revealed two significant main effects, that is, Orientation of the bar with respect to the ellipse, F(1,53) = 22.417,p
