J. J. McAnany. 3. J. E. Anderson. 1. 1. Department of Psychology. 2. Laboratory of
Integrative Neuroscience. 3. Department of Ophthalmology and Visual ...
M. W. Levine1,2 J. J. McAnany 3 J. E. Anderson1
1
Department of Psychology
2
Laboratory of Integrative Neuroscience
3
Department of Ophthalmology and Visual Sciences
University of Illinois at Chicago Chicago, Illinois
Abstract: The intersections of lighter alleys defining a grid of black squares display illusory effects (the Hermann grid and the scintillating grid). We have noted that a single light disk in an intersection at some remove from fixation is rendered less visible (the “vanishing disk”). Furthermore, contrast threshold for the vanishing disk becomes even higher when the alleys are curved instead of straight (Levine & McAnany, VSS 2006). We speculated that the effect of curvature is due to either an attention shift or inhibition from a “higher” center than the detection mechanism (Levine & McAnany, 2008). If this is correct, one would expect a measurable latency for the onset of this influence. To test for this temporal disparity, we presented vanishing disk stimuli in which half of each presentation featured straight alleys, and half featured curved alleys. That is, in the middle of a 250 msec presentation of a stimulus, straight alleys abruptly curved, or vice versa. Since these changes are equivalent, any difference in threshold for the disk between these conditions must be caused by the order of onset of curvature, and not be simply an effect of a transition. Thresholds for light disks were significantly higher when moderately curved alleys straightened than when straight alleys became curved to the same degree. We infer that the earlier curvature has time to initiate whatever disrupts visibility even though the alleys are straight in the latter part of the presentation; curvature later did not have time to exert its influence before detection was effected. (Note that had the decision been based on visual short-term memory, the stimuli that concluded with straight alleys would have evinced lower thresholds, not higher.) These and related results indicate a higher-level effect of curvature (complexity) upon detection. But other results presented here argue against this
conclusion – the effect is most likely due to the recruitment of cells with other preferred orientations at the earliest stages of cortical processing.
Introduction The intersections of the lighter alleys defining a grid of black squares show illusory effects (Schrauf et al., 1997; Spillmann, 1994): √ dark smudges (the Hermann grid) √ black spots on white disks placed in the intersections (the scintillating grid) We have noted that a single light disk in an intersection at some remove from fixation is hard to detect (the vanishing disk: McAnany & Levine, 2004). • In the other illusions, a non-existent stimulus is perceived. • In the vanishing disk, an actual stimulus is rendered less visible. Curving the alleys can eliminate an illusory non-existent stimulus (Geier et al., 2004) – but it enhances the vanishing disk (even less visible: Levine & McAnany, 2008).
Logic of these experiments: Curvature makes everything (Hermann smudges, scintillation, vanishing disk) less visible. Curvature even makes dark disks (which do not “vanish”) less visible. The effect of curvature is consistent with added complexity (Levine & McAnany, 2008). If complexity causes a feedback inhibition from “higher” centers (or changes attention), it would be expected to engage after a delay. We therefore designed a test in which the alleys switch from straight to curved (or vice versa) during a brief presentation.
HYPOTHESIS TESTED: If the effect of curvature takes time to develop, curved to straight should have a greater suppressive effect than straight to curved (because the effect of the curvature won’t have had time to develop if the curvature comes late).
Methods Stimuli: Arrays of “tetragrams” with curved alleys appear above and below fixation (see Fig. 1). A disk appears in one of 4 intersections. Subject selects an intersection by mouse click on response screen (Fig. 2). The staircase method was used to determine threshold. Two frames of equal durations; alleys are curved in one frame, straight in the other. Two possibilities: • Disk is present in both frames; interlaced staircases for upper visual field and for lower visual field. • Disk is present in only one frame; interlaced staircases for disk in first frame or disk in second frame, with a staircase for upper field and lower field in each (Fig. 2).
Tim e
Pr res evio po us ns e
Fixation screen 1st frame (straight) target
2nd frame (curved 5½ %) Response screen
{
Upper right Is correct
{
Figure 2
Wa
it f or
res po ns e
s
Fr
e am
r du
ion t a
50 Fix 0 m Un atio sec ti l n ne xt sti mu lu
e m a Fr
ra du
n tio
But cursor is in upper left
{
0 50
c se m
Experiment 1: The results of this experiment for subject ML are shown in Fig. 3. • Subject was tested with three levels of curvature. • Each curvature was tested with straight to curved (blue) and curved to straight (red) presentations. • Curved to straight consistently required a higher contrast for the disk to be visible. Averaged results from three subjects are shown in Fig. 4 (1 s.e.m. error bars). • Curved to straight required a higher contrast for the disk to be visible. • The effect was generally greater with greater curvature. This result seems to support the original hypothesis: the effect of curvature shows a delay before it develops fully.
Light disks
Figure 3
125 ms frames
Subject: ML
Light disks
Figure 4
125 ms frames
Mean of JA, ML, & JM
Experiment 2: We attempted to confirm our result by placing the disk in only one or the other frame. We expected that curvature would have its greatest effect in the Curved to straight condition with the disk in the second frame. But thresholds were highest for disks in the second frame, regardless of the order of curvature vs straight (Fig. 5). Thresholds were also higher in the second frame when both frames had half the curvature, but the second frame curvature was mirror-imaged (Fig. 5).
So the first experiment could be explained if the disk is detected at the start of the presentation (first frame), and is harder to detect when superimposed on curvature.
Light disks 125 ms frames Subject: ML
Figure 5
Figure 5: Thresholds for a disk in only one frame under various conditions. In each case, the disk was presented in one of the two frames. The light blue bars represent thresholds when the disk was presented in the first frame; magenta bars are for the disk presented during the second frame. • In the left set of bars, the alleys were curved in the first frame, straight in the second. In the next set of bars, the sequence was reversed. • The third set of bars represents trials in which the alleys were curved (at half the amplitude) in both frames, but switched to a mirror image in the second frame, as shown below.
• In the rightmost set of bars, the grid was the same in both frames (no change except for which frame contained the disk).
Experiment 3: Curvature also affects visibility of disks that are not obviously “vanishing”.
• Dark disks are less visible when alleys are curved.
• And dark disks are less “confounded” by Hermann “smudges”.
When dark disks are present throughout the presentation, Curved to straight requires a higher contrast (Fig. 6).
But when the disk is in only one frame: • Which frame shows higher threshold depends on which is curved (Fig. 7). • Second frame has a higher threshold when curve is mirrored; first frame if no change between frames.
• Higher threshold is for the disk in the curved frame (Figs. 7 & 8) (for all three subjects).
Dark disks
Figure 6
125 ms frames
Subject: ML
Dark disks
Figure 7
125 ms frames
Subject: JA
Dark disks
Figure 8
125 ms frames
Subject: JA
Summary: Curvature raises threshold for the vanishing disk more when it precedes straight alleys. The vanishing disk threshold is determined when the disk first appears. Dark disks are also affected more by curvature early in a presentation. But dark disks clearly show the
effects of curvature most strongly when the curvature is concurrent with disk presentation.
So: The answer is
Curvature raises threshold for
“details” in a way consistent with the added complexity (Levine & McAnany, 2008).
The effect of curvature does not
require a time delay to develop. It is thus most likely that curved alleys simply stimulate more cortical neurons than straight alleys (more orientations), making the disks (which have all orientations) harder to detect.
References Geier, J., Sera, L., and Bernath, L. (2004). Stopping the Hermann grid illusion by simple sine distortion. ECVP 2004 abstract
Levine, M. W. and McAnany, J. J. (2008). The effects of curvature on the grid illusions. Perception 37; 171-184. http://perceptionweb.com/perception/fulltext/p37/p5691.pdf
McAnany, J. J. and Levine, M. W. (2004). The blanking phenomenon: a novel form of visual disappearance. Vision Research 44; 993-1001.
Schrauf, M., Lingelbach, B., and Wist, E. R. (1997). The scintillating grid illusion. Vision Research 37; 1033-1038.
Spillmann, L., (1994). The Hermann grid illusion: a tool for studying human perspective field organization. Perception 23; 691-708.
