Differential Effects of Continuous Flash Suppression at the Fovea and Periphery

Transcription

1 Differential Effects of Continuous Flash Suppression at the and Periphery Bobby Rohrkemper March 15, 4 Abstract Using a technique called Continuous Flash Suppression (CFS), it is possible to suppress a target image for an extended period of time. I found that both CFS and Binocular Rivalry (BR) behave differently in their ability to suppress target images in the fovea and the periphery. Plotting time of exclusive visibility and time of the color mondrian percept, each with a carefully designed control allows us to discuss mechanisms for CFS. Adaptation to flicker is supported as a possible mechanism. Contents 1 Introduction 2 2 Methods 3 3 Results 6 4 Discussion 5 Conclusion 11 A Appendix 11 A.1 Preliminary Studies A.2 Personal Benefit A.3 Relation to fmri Study A.4 Acknowledgements

2 1 Introduction In Binocular Rivalry (BR) (Blake and Logothetis, 2), competition occurs between two images. A different image is presented constantly to each eye, but the percept switches between the two images. It is also common to see pieces of both images. One possible modification of BR is to present one image constantly, but to change the other very rapidly. This allows for long durations of suppression of the target image. This type of stimulus is called Continuous Flash Suppression (CFS) (Tsuchiya and Koch, VSS, 4). If CFS were truly just a continuous version of flash suppression (which is very effective), then it would be possible to hide the target image indefinitely. However, after some time, the subjects do tend to see BR-like reversals. I wanted to know why this suppressed stimuli is not suppressed permanently. The most promising answer seems to be adaptation to the flickering stimulus (Wolfe, 1984). As a control, I used BR (Blake et al., 1992). He found that the time of exclusive visibility decreases in the periphery for BR (See Figure 1). The two main quantities I was interested in were time of exclusive visibility and time of the color mondrian percept. Exclusive visibility for the sixty second trials was calculated as sixty seconds minus the time during which the percept was transparent or piecemeal. The time of the color mondrian percept was defined to be the time during which the subject saw exclusively the color mondrians only. I hypothesize that adaptation may be the mechanism that causes a target stimulus to reappear in CFS. However, this hypothesis requires further testing. Another way to test this hypothesis would be to have two kinds of subjects. Those with slow adaptation to flicker would show long times of suppression for CFS. In contrast, those with fast adaptation to flicker would show short times of adaptation for CFS. Schieting s results show that adaptation to flicker occurs very easily in the periphery. He also found significant interocular transfer, which he says suggests that flicker adaptation occurs at a higher cortical level. Here, I propose that adaptation to the flickering color mondrians causes them to slowly disappear. It seems to be a cumulative effect that the subject is not consciously aware of until there is enough buildup. Then, the subject will begin to see pieces of the gabor image and may also see the gabor exclusively. In a study involving orthogonal circular gratings (See Figure 1), Blake showed that the total time, in which one of the two percepts was seen exclusively, correlates with retinal eccentricity. They did this for minute-long trials and showed that as the eccentricity approaches twelve degrees, the time of exclusive visibility approaches the length of the trial. This observation explains my hypothesis for the trials in which I plotted the time of exclusive visibility. Additionally, Blake found that exclusive visibility time increased with decreasing target size at a given eccentricity. Thus, he proposes that the receptive fields are composed of spatial zones of BR. Furthermore, Blake concludes that rivalry zones are larger in the periphery and that cortical neurons in the fovea have smaller receptive fields. Blake s data was not based on naïve subjects, but rather on the authors themselves. 2

3 Figure 1: Time of exclusive exclusive visibility (top) increases in the periphery for binocular rivalry (Blake et al., 1992). Different curves indicate different sizes of the target stimuli as indicated in the legend. 2 Methods Subjects were students from the Caltech community, aged years. All were male. Stimuli were presented in stereo using the Resonance Technology (RT) goggles at Caltech s Brain Imaging Center. Figures 3 and 2 show and describe the properties of BR and CFS. Subjects were given training sessions in which they were instructed to fixate the red cross at all times, and to develop a constant criteria which they will use throughout the experiment to distinguish between three possible percepts. Since rivalry rarely involves seeing a single percept dominate completely, it was very important for the subjects to establish this constant criteria. Through continuous monitoring of their percept, subjects were instructed to press one of three keys as indicated in Table 1. The color mondrians ( patterns) and gabors were generated prior to the experiment. The gabors were either horizontally or vertically oriented. As described in Section A.1, the size of the mondrian boxes were made large enough such that they could be discriminated at 9 degrees in the periphery. They were an average of 1 in length and 1 2 in width. Both the gabors and the color mondrians were used at maximum saturations. The spatial frequency of the gabors were 3 cycles degree. For each session, subjects repeated 4 runs, each 16 minutes in length. A run consisted of 8 minute-long trials, each separated by at least a minute of rest. All subjects were tested for normal stereo vision after the trials. Three tests were used, in a book produced by Stereo Optical Co., Inc. of Chicago, Illinois. Polarizing glasses were worn while viewing the images. The first was a random-dot-stereogram with a butterfly pop-out image. The second was a series of 9 forced choice pop-out tasks, which became successively more difficult towards the 9 th. The third test involved 3

4 Figure 2: Time of adaptation to flicker decreases in the periphery (Schieting and Spillmann, 1987). looking at images of animals and choosing the pop-out animal. Table 1: The subject s task. Percept Key press Gray gabor only 1 Transparent overlay or piecemeal rivalry 2 Color mondrian only 3 Table 2: The four trial types Constant gabors, constant color mondrians (BR) at the fovea Constant gabors, flickering color mondrians (CFS) at Hz in the fovea BR in periphery CFS in periphery 4

5 time time 9 degrees 4 degrees 15 degrees Figure 3: Sample images presented to each eye. Yellow markings indicate the stimuli size, but the diagram is not drawn to scale. See Table 2 for a description of the different trial types. The random dot frames aide in image stabilization between the two eyes. 5

6 3 Results All subjects were able to see pop-out in a random-dot stereogram using the provided 3D glasses. Additionally, they all scored greater than fifty percent on a 4-forced choice pop out test for stereo vision. The scores were as follows: AB - 6/9; BS - 5/9; ED - 9/9; RR - 7/9; and RW - 9/9. However, subject BS reported having an excessively dominant right eye, and subject AB was able to see pop-out in the stereogram, but he could not clearly identify the object. He also reported having trouble with random-dot stereograms in the past. Subject: ALL : CFS (RED) and BR (BLUE) Subject: ALL : CFS (RED) and BR (BLUE) 55 <.741 < < e 5 < < < Subject: All (AB & BS excluded) : CFS (RED) and BR (BLUE) Subject: All (AB & BS excluded) : CFS (RED) and BR (BLUE) 45 < <.1845 <.998 < e < < Figure 4: Averages across subjects. Left side:. Right side: Time of color mondrian percept. Top: All trials. Middle: First trials for each subject. Bottom: Second trials for each subject. Table 3 characterizes the significance of the interaction between BR and CFS. Here, I calculated the difference in means for each of the plots. Notice that when the subjects who do not have perfect stereo vision are excluded, the difference in means is much greater. This can be seen by comparing All with All (AB and BS excluded). Several subjects noted that it was very difficult to judge between the three percepts 6

7 Table 3: Difference in means between CFS and BR for each subject. Absolute values are reported. Subject Exclusive visibility Color perception ALL 5 6 All (AB and BS excluded) AB RR RR BS BS ED RW RW at the periphery. With training, this usually becomes easier. Subjectively, CFS is an effective way to suppress a target image for an extended period of time. The subjects noticed that when the color mondrians are flashed at Hz, they were much more likely to see the color mondrians only. One subject also said that for BR, the time of dominance of the color mondrians seems to depend on the random color at the fixation. Subject AB said that he saw the color more when there was red at the center. 7

8 Subject: RR 1 : CFS (RED) and BR (BLUE) 7 Subject: RR 2 : CFS (RED) and BR (BLUE) < e 6 < e <.6711 <.343 <.614 < Subject: RW 1 : CFS (RED) and BR (BLUE) 7 Subject: RW 2 : CFS (RED) and BR (BLUE) <.6345 < <.6723 < < e 5 < Subject: BS 1 : CFS (RED) and BR (BLUE) 7 Subject: BS 2 : CFS (RED) and BR (BLUE) < <.774 <.96 < < < Subject: ED 1 : CFS (RED) and BR (BLUE) 7 Subject: AB 1 : CFS (RED) and BR (BLUE) < < e 5 < e 5 < < < Figure 5: for eight sessions. Three subjects performed the experiment twice. 8

9 Subject: RR 1 : CFS (RED) and BR (BLUE) Subject: RR 2 : CFS (RED) and BR (BLUE) < < 6.585e 8 < <.96 < < Subject: RW 1 : CFS (RED) and BR (BLUE) Subject: RW 2 : CFS (RED) and BR (BLUE) < < < < 1.337e 6 < < e 5 Subject: BS 1 : CFS (RED) and BR (BLUE) Subject: BS 2 : CFS (RED) and BR (BLUE) < < e 6 <.547 < <.8571 <.9825 Subject: ED 1 : CFS (RED) and BR (BLUE) Subject: AB 1 : CFS (RED) and BR (BLUE) < < < 1.337e 6 < e 5 < < Figure 6: for eight sessions. Three subjects performed the experiment twice. 9

10 4 Discussion Figure 4 shows the results, averaged over all subjects. Based on the results given by Blake in Figure 1, we should see an increase in time of exclusive visibility for BR in the periphery(left side blue). I was able to replicate Blake s results. Also, based on results from Schieting in Figure 8, the time of adaptation to flicker decreases in the periphery. The fact that the time of the color mondrian percept also decreases in the periphery (right side red) suggests that the appearance of the gabor may be caused by adaptation to flicker. For the averaged plots, this trend is shown in all cases. However, it is not always statistically significant. For the average of all subjects, we find ISI <.18 for time of exclusive visibility and ISI <.27 for time of the color percept. However, if the subjects who reported having slightly abnormal stereo vision are removed, these values change to ISI <.5 and ISI <., respectively. The effect of ISI ( ) was always significant. This may be caused by large variability across subjects as can be seen in the plots for each individual subject. Figure 6 shows the individual subject plots for time of exclusive visibility. Figure 5 shows the individual subject plots for the time of the color percept. In both cases there are a few exceptions, which deserve discussion. In Figure 6, the line for BR (blue) should always have a positive slope. There are two exceptions to this, including the second session for subject BS, and the first for subject AB. BS reported difficulty with the task and also extreme difficulty in monitoring the percept. Generally, the subject s performance was more variable across trials the first time they performed the task, even with a small amount of prior training. This was also the case for AB. Note that the line for CFS (red) is plotted as a control. I used this control because the time of exclusive visibility for BR should decrease because there is greatly reduced color acuity in the periphery. Also note that there is no relationship ( <.97) for eccentricity. This is because the two slopes roughly average out to a flat line. In Figure 5, the line for CFS (red) should always have a negative slope. There was only one exception, and this was for the first session of subject RW. RW also reported being excessively tired. The second time he performed the task, his results were much more consistent with others. Note that the line for BR (blue) is plotted as a control. I decided to use this control because it allows me to rule out other factors besides adaptation to flicker. Specifically, it means that color adaptation is not a likely mechanism. If this control were not present, one could say that the gabor becomes visible because the cones quickly adapt to a color stimulus. I m concerned about several effects, which may be preventing me from inferring the mechanism responsible for the appearance of a target image in CFS. Some subjects noted that the target image faded out completely in the periphery. This could be due to Troxler s effect. This effect might cause the image area to be filled in with it s gray surroundings. In this case, a subject might think he sees the gabor. It should be noted that such fading out never occurred at the foveal presentations. Also, the effectiveness of a grating of a specific spatial frequency could vary with eccentricity.

11 Figure 7: Cortical magnification can be measured experimentally (Blake et al., 1992). They did this by using a similar forced choice task to mine, but instead used two keys instead of three. He then plotted the size needed for near-exclusive visibility at a given eccentricity. Figure 8: Accounting for cortical magnification allows the adaptation time to be independent of eccentricity (Schieting and Spillmann, 1987). 5 Conclusion In the future, this project could be improved through the use of the eye tracking device, which is incorporated into the RT goggles. It would be nice to at least use the device passively to alert subjects when they are no longer fixating within a small margin of error. Also, more using subjects would help to enhance the statistical power. This results of this project have suggested flicker adaptation as a mechanism for CFS. I have plotted time of exclusive visibility for BR and have used CFS as a control to show that BR is significantly different. More importantly, I have also plotted the time of the color percept for CFS and have used BR as a control to show that disappearance of the color mondrians is not simply due to color adaptation. A Appendix A.1 Preliminary Studies Fine tuning the CFS effect took considerable effort. I tested the following parameters: inter-stimulus interval, contrast power of both images, eccentricity, position of stimuli, 11

12 stimulus window size, and scale of gabor/mondrian. I also tested the effect of presenting stimuli in different quadrants in the periphery. There was not a significant effect. Contrast power is a significant factor in causing perceptual switches. The more salient image tends to dominate for longer. Also, it seems that if the salience of both images is raised, then the perceptual switches will occur more often, but the average total times for each percept will be roughly equal. I found that an ISI of.1 seconds (for CFS) is very effective for suppressing the target image. Shorter flickering intervals tend to fade together, and longer intervals tend to behave more similar to BR. A.2 Personal Benefit Being the first psychophysical experiment that I carried out from start to finish, this was quite a learning experience. I learned how to properly instruct a subject without biasing them. I found that training helped the subjects significantly even though the task seemed very easy. It was also the first time I had programmed stimuli. For this, I used Psychophysics Toolbox with MATLAB. I also learned about the importance of designing an experiment with careful controls. A.3 Relation to fmri Study CFS is a valuable technique for functional imaging studies as well. Here, I will include a portion of a proposal for a project currently underway. The effect of becoming conscious of visual stimuli on neuronal processing has been studied using fmri in binocular rivalry (BR) (Lumer et al., 1998; Tong et al., 1998; Lumer and Rees, 1999; Polonsky et al., ; Tong and Engel, 1), bistable stimuli (Kleinschmidt et al., 1998; Andrews et al., 2; Muckli et al., 2) in healthy subjects and bilateral stimulus presentation in a neglect and extinction patient (Vuilleumier et al., 2). In general, those studies revealed a) time-locked enhanced neuronal activity in the higher visual areas, such as Fusiform Face Area (FFA) and Parahippocampal Place Area (PPA), (but (Polonsky et al., ) and (Tong and Engel, 1) reported those in the lower visual areas) and b) time-locked transient activity that is related to the transition itself but not to the motor response in parietal and frontal lobes (Lumer and Rees, 1999). A problem with those studies, except for the extinction study, is that they could not control voluntary and top-down attention and expectation. Subjects may pay attention to one stimulus either when it is visible to sustain the visibility or when it is suppressed to overcome the invisibility. Controlling attention is a critical issue since it is known to modulate cortical activity, especially measured with fmri. For attentional effects on FFA, see (Wojciulik et al., 1998; O Craven et al., 1999; Vuilleumier et al., 1). For the importance of controlling attention in fmri experiments, see (Ress et al., ; Huk and Heeger,, 2). Therefore, the observed fmri modulation in lower visual areas or part of the modulation in higher areas could be due to attention. Here we propose Continuous Flash Suppression (CFS) to circumvent this issue. Using CFS, it is possible to control subjects attention and expectations very cleanly by incorporating several types of suppressed images on different trials (angry faces, neutral faces, places, and dummies without targets) and asking subjects to discriminate 12

13 the suppressed images by button presses. If these different trial types are randomly intermixed, the subject will not be able to predict what kind of stimulus (e.g. face vs. house vs. no target) is present until it actually becomes visible. Thus, visibility of a face would not be preceded by attention to the face as could be the case in BR. In addition to control of attention and expectation, CFS allows us to monitor fmri activity for unperceived stimuli for a long time (> 2 sec), which cannot be done with backward masking (Whalen et al., 1998). Because of this prolonged duration of invisibility, we can see if the activation in the higher cortical areas can be activated by invisible stimuli. Do invisible faces activate FFA? It should be noted that it is impossible to see activation from the baseline if we use BR, because the comparison must be made between suppressed perceived duration aligned at transition (instead of the baseline gray field) and suppressed duration. Therefore, the first and main hypothesis to test is 1) whether FFA or PPA activity is observed at all when subjects don t see the suppressed face or place stimuli, compared to the baseline, under the strict control of attention and expectation. We would not see any activation in these areas if they are totally correlated with awareness (Tong et al., 1998). Alternatively, we may see preceding build-up activation in those areas before subjects report visibility. Perhaps a threshold of activation needs to be exceeded for awareness of the suppressed image. We expect the build-up activation to be smaller than the full-blown activation observed when the suppressed stimulus becomes clearly visible after the report of visibility. In our CFS experiment, further hypotheses about the relationship between conscious awareness and neural activity will be tested. 2) Is amygdala activation independent of awareness (Whalen et al., 1998; Vuilleumier et al., 2)? This could be tested by comparing the responses to suppressed angry vs. neutral faces. Note the automatic detection of angry faces was not observed when (Pessoa et al., 2) it was tested with a very demanding concurrent task. 3) Is there any gradation of dependency on consciousness among our ROIs (FFA, Superior Temporal Sulcus (STS), Lateral Occipital Cortex (LOC), and amygdala)? Or, is the degree of activation in those areas during suppression always the same as in the consciously perceived trials or none as in the baseline period? 4) Does the activation in consciousness related areas ( e.g. FFA) follow the subjects reports or precede them? This question may not be able to be answered with the coarse time resolution of fmri, however, it would be possible if the rise time of activation starts 2 3 sec before the report or if we push the temporal sampling of slice acquisition (with T R = 1 sec, slice thickness 3mm, 16 slices, we could obtain all the above ROIs). 5) Do we see transition related activation in parieto-frontal areas (Lumer et al., 1998; Lumer and Rees, 1999)? If yes, the mechanisms of transition may be similar to that of BR. If no, adaptation to a series of changing patterns may be the key for a transition (Blake et al., 3; Muckli et al., 2). 13

14 A.4 Acknowledgements I would like to thank Nao Tsuchiya for assistance and advice throughout the project. Helpful comments also came from Shin Shimojo and Christof Koch. Equipment was provided by the Brain Imaging Center at Caltech and funding for subject payment was provided by the Koch Laboratory. References T. J. Andrews, D. Schluppeck, D. Homfray, P. Matthews, and C. Blakemore. Activity in the fusiform gyrus predicts conscious perception of rubin s vase-face illusion. Neuroimage, 17 (2):89 91, Clinical Trial Journal Article. R. Blake and N. K. Logothetis. Visual competition. Nat Rev Neurosci, 3(1):13 21, x Journal Article Review Review, Tutorial. R. Blake, R. P. O Shea, and T. J. Mueller. Spatial zones of binocular rivalry in central and peripheral vision. Vis Neurosci, 8(5):469 78, Journal Article. R. Blake, K. V. Sobel, and L. A. Gilroy. Visual motion retards alternations between conflicting perceptual interpretations. Neuron, 39(5):869 78, Journal Article. A. C. Huk and D. J. Heeger. Task-related modulation of visual cortex. J Neurophysiol, 83(6): , Clinical Trial Journal Article Randomized Controlled Trial. A. C. Huk and D. J. Heeger. Pattern-motion responses in human visual cortex. Nat Neurosci, 5(1):72 5, Journal Article. A. Kleinschmidt, C. Buchel, S. Zeki, and R. S. Frackowiak. Human brain activity during spontaneously reversing perception of ambiguous figures. Proc R Soc Lond B Biol Sci, 265 (1413): , Journal Article. E. D. Lumer, K. J. Friston, and G. Rees. Neural correlates of perceptual rivalry in the human brain. Science, 28(5371):19 4, Journal Article. E. D. Lumer and G. Rees. Covariation of activity in visual and prefrontal cortex associated with subjective visual perception. Proc Natl Acad Sci U S A, 96(4): , Journal Article. L. Muckli, N. Kriegeskorte, H. Lanfermann, F. E. Zanella, W. Singer, and R. Goebel. Apparent motion: event-related functional magnetic resonance imaging of perceptual switches and states. J Neurosci, 22(9):RC219, Clinical Trial Journal Article. K. M. O Craven, P. E. Downing, and N. Kanwisher. fmri evidence for objects as the units of attentional selection. Nature, 1(6753):584 7, Journal Article. L. Pessoa, M. McKenna, E. Gutierrez, and L. G. Ungerleider. Neural processing of emotional faces requires attention. Proc Natl Acad Sci U S A, 99(17): , 2. A. Polonsky, R. Blake, J. Braun, and D. J. Heeger. Neuronal activity in human primary visual cortex correlates with perception during binocular rivalry. Nat Neurosci, 3(11):1153 9, Journal Article. 14

15 D. Ress, B. T. Backus, and D. J. Heeger. Activity in primary visual cortex predicts performance in a visual detection task. Nat Neurosci, 3(9):9 5, Journal Article. S. Schieting and L. Spillmann. Flicker adaptation in the peripheral retina. Vision Res, 27(2): , Journal Article. F. Tong and S. A. Engel. Interocular rivalry revealed in the human cortical blind-spot representation. Nature, 411(6834):195 9, Journal Article. F. Tong, K. Nakayama, J. T. Vaughan, and N. Kanwisher. Binocular rivalry and visual awareness in human extrastriate cortex. Neuron, 21(4):753 9, Journal Article. P. Vuilleumier, J. L. Armony, K. Clarke, M. Husain, J. Driver, and R. J. Dolan. Neural response to emotional faces with and without awareness: event-related fmri in a parietal patient with visual extinction and spatial neglect. Neuropsychologia, (12): , Case Reports Clinical Trial Journal Article. P. Vuilleumier, J. L. Armony, J. Driver, and R. J. Dolan. Effects of attention and emotion on face processing in the human brain: an event-related fmri study. Neuron, (3):829 41, Journal Article. P. J. Whalen, S. L. Rauch, N. L. Etcoff, S. C. McInerney, M. B. Lee, and M. A. Jenike. Masked presentations of emotional facial expressions modulate amygdala activity without explicit knowledge. J Neurosci, 18(1):411 8, Journal Article. E. Wojciulik, N. Kanwisher, and J. Driver. Covert visual attention modulates face-specific activity in the human fusiform gyrus: fmri study. J Neurophysiol, 79(3):1574 8, Journal Article. J. M. Wolfe. Reversing ocular dominance and suppression in a single flash. Vision Res, 24(5): 471 8, Journal Article. 15