Adaptation to invisible motion results in low-level but not high-level aftereffects

Abstract

After prolonged exposure to moving stimuli, illusory motion is perceived in stimuli that do not contain consistent motion, a phenomenon termed the motion aftereffect (MAE). In this study, we tested MAEs under binocular suppression that renders the motion adaptor invisible for the entire adaptation period. We developed a variant of the continuous flash suppression method to reliably suppress target motion stimuli for durations longer than several tens of seconds. Here, we ask whether motion systems are functional in the absence of perception by measuring the MAE, a question difficult to address using binocular rivalry that accompanies a switch of percept between visible and invisible. Results show that both the MAEs with static and dynamic tests are attenuated with an invisible adaptor when the adaptor and the test stimulus are presented to the same eye. In contrast, when the test pattern was presented to the other eye, the dynamic MAE was observed in invisible adaptor conditions. These results indicate that low-level adaptation survives under total binocular suppression, a finding predicted by previous studies. In contrast, disappearance of interocular transfer in the dynamic MAE suggests that a high-level motion detector does not operate when the motion adaptor is rendered invisible.

Introduction

The mechanism underlying conscious experience of the visual world, the neural substrate of visual awareness, has been a subject of substantial interest. Psychophysical studies have addressed this question by investigating whether aftereffects persist for adaptors perceptually rendered invisible (Blake & Fox, 1974; He & MacLeod, 2001; Lehmkuhle & Fox, 1975; O'shea & Crassini, 1981; Wade & Wenderoth, 1978; Wiesenfelder & Blake, 1990). Here, aftereffects resulting from adaptation are considered evidence for the operation of detectors tuned to the physical attributes of the adaptor stimulus. Therefore, the combination of measuring aftereffects and applying techniques that mask perception of the adaptor (e.g., binocular rivalry, crowding) is a powerful method for studying the neural substrate of visual awareness. 

In this study, we focused on the motion aftereffect (MAE) where prolonged viewing of a motion stimulus subsequently evokes the perception of illusory motion in the opposite direction in a stationary target. Some classical studies have reported that binocular suppression does not interfere with the formation of MAEs (Lehmkuhle & Fox, 1975; O'shea & Crassini, 1981). Recently, Blake, Tadin, Sobel, Raissian, and Chong (2006) suggested that the apparent insusceptibility of MAE to binocular suppression results from a compressive non-linearity in the MAE strength at high and intermediate adapting contrast. They demonstrate that rivalry suppression indeed interferes with the formation of MAEs under low contrast adaptor conditions (Blake et al., 2006). However, it is contentious whether motion detectors are completely suppressed or only attenuated during periods where the adaptor is not perceived. This question arises because the observed MAE reflects the net adaptation induced by visible and invisible adaptors due to perceptual alternation between the adapting and the rival stimuli during binocular rivalry. Moreover, non-linearity in temporal accumulation of adaptation and various noises in measurement make it difficult to extrapolate the results obtained by binocular rivalry to the point of total perceptual elimination of the adaptor (Figure 1a). One way to address this question is to render the adaptor invisible during the entire adaptation period. 

Figure 1

 

Motion aftereffect under binocular suppression. (a) Aftereffect under conventional binocular rivalry. The upper row colored in red denotes suppressed phases of adaptor, whereas green denotes dominant phases of adaptor. The lower row schematizes the extent of adaptation in each phase. Here, α denotes the extent of adaptation in the adaptor dominant phase and β denotes the extent of adaptation in the suppressed phase. (b) Two models for MAE with complete suppression of motion adaptor. If there were no accumulation of adaptation during the perceptually suppressed phase of the adaptor, i.e., β_i = 0 in panel a, MAE would disappear when the adaptor is completely suppressed from awareness (case “a”). In contrast, if the adaptation is attenuated but not lost, the MAE would be observed even in the case of complete suppression (case “b”). Note that the results with partial suppression observed with conventional binocular rivalry (shown in the bold line) cannot dissociate these two possibilities due to the non-linearity between the dominance duration of the adaptor and the magnitude of the aftereffect.

To completely mask motion stimulus from visual awareness, we modified a recently developed method, continuous flash suppression (CFS; Tsuchiya & Koch, 2005). This technique can suppress stationary stimuli for several tens of seconds. In the original CFS method, observers viewed a stationary target stimulus with one eye, while multiple rectangles of various size, contrast, and position (Mondrian stimulus) flickered at 10 Hz in front of the other eye. We replaced the static rectangles with drifting sinusoidal gratings (dynamic Mondrian stimulus) perpendicular to the target motion stimuli (Figure 1b). This method, dynamic continuous flash suppression (dCFS), suppresses visual awareness of the target motion grating during the entire period of adaptation. 

One challenging and interesting aspect of MAE is that a single motion adaptor evokes adaptation in multiple processing levels. In relation to our question of study, whether any difference exists in contribution to visual awareness between early and higher visual cortex is a subject of intense investigation and dispute (Logothetis & Schall, 1989; Lumer, Friston, & Geraing, 1998; Polonsky, Blake, Brown, & Heeger, 2000; Sheinberg & Logothetis, 1997; Tong & Engel, 2001; for an overview, see Blake & Logothetis, 2002; Tong, Meng, & Blake, 2006). Although a direct exploration of this question is beyond the scope of this study, this problem raises the possibility that susceptibility of motion detectors to the effect of binocular suppression is different between processing levels. 

In the literature on MAE, it is well established that different levels of motion adaptation can be examined with test stimuli that are either static or dynamic (the static or dynamic test). The motion aftereffect with a stationary test pattern (static test) shows position specificity (Anstis & Gregory, 1965), spatial frequency selectivity (Cameron, Baker, & Boulton, 1992), and partial interocular transfer (Wade, Swanston, & de Weert, 1993). These characteristics reveal that MAE evoked with a static test (static MAE/sMAE) reflects adaptation in the motion system located at the early level. In contrast, MAE with a dynamic test, where a flickering or counter-phasing grating is presented as a test pattern, shows weaker position specificity and spatial frequency selectivity (Ashida & Osaka, 1994; Culham, Verstraten, Ashida, & Cavanagh, 2000) and shows robust interocular transfer (Nishida, Ashida, & Sato, 1994; Raymond, 1993). Therefore, the MAE evoked with a dynamic test (dynamic MAE/dMAE) reflects high-level adaptation (Nishida & Ashida, 2000; Nishida et al., 1994; Nishida & Sato, 1995). We employed different types of test stimulus and interocular transfer measurements to examine binocular suppression in lower and higher processing levels separately. 

General methods

Apparatus

Stimuli were generated using a PC/AT compatible personal computer with a C++ library and were presented on two CRT screens (Eizo FlexScan T550). An eight-bit gray scale with a gamma correction was provided by an nVidia video card. Average luminance was 35.5 cd/m ². The screens were aligned horizontally. Observers viewed one of two screens with each eye using a mirror stereoscope with the aid of a chin rest. Left- and right-eye images were presented against a gray, uniform background. To promote stable binocular alignment, each image was surrounded by a square frame (5.6 × 5.6 arc deg) composed of static random dot patterns (“fuse frame”) with a red fixation cross at the center. The refresh rate was 60 Hz. The viewing distance was 80 cm and each pixel subtended 1.1 × 1.1 arc min. 

Stimuli

Figure 2 shows a schematic drawing of the dichoptic stimulus setup using a mirror-stereoscope together with the experimental procedure. Observers viewed the adaptation stimulus for 30 sec with their non-dominant eye. The adaptation stimuli consisted of three sinusoidal gratings aligned vertically with gaps of 0.29 arc deg. Each grating subtended 2 (H) × 0.53 (V) arc deg. Spatial frequency of the sinusoidal grating was 0.62 c/deg during the adaptation phase. Adaptation motion was generated by shifting the grating every 100 ms, with a displacement of a 90-deg phase angle. This corresponds to 4.0 deg/sec. The direction of motion was leftward for the center grating and rightward for other gratings. The gratings above and below the center grating were introduced to magnify the motion aftereffect (cf. Wade et al., 1993). In the invisible adaptor condition with dCFS, a dynamic Mondrian stimulus was presented to the dominant eye. Here the dominant eye was determined for each observer with the “Miles” and “Porta” test (Crovits & Zener, 1962; Miles, 1930). Each Mondrian was a rectangle subtended 1.4 × 1.4 arc deg. This rectangle comprised horizontal, sinusoidal gratings moving vertically between vertical bars for which the contrast was reversed (black/white) every 33.3 ms. The spatial frequency of the sinusoidal grating was 3.0 c/deg. This grating was shifted every 16.6 ms with a displacement of 36 deg phase angle to generate motion. The direction of motion (up/down) was reversed with random timing to avoid motion adaptation by these stimuli. The arrangement of these Mondrians was changed every 100 ms. In the visible condition, only the fixation and “fuse frame” were presented on a gray background. 

Figure 2

 

A diagram of the stimulus and procedure. (a) A diagram of stimulus. (b) A component of the dynamic CFS display. (c) Basic procedure of the main experiment (same-eye test). Note that in the other-eye test, the test stimuli were presented to the eye other than that with which observers viewed adaptors.

After a 30-sec presentation of adaptation stimuli, a uniform gray field and reference frames were displayed for 1 sec. Subsequently, observers viewed stationary (static test) or flickering (dynamic test) sinusoidal gratings for 30 sec. The gratings were also arranged in three rows and had a spatial frequency of 1.86 c/deg. The difference in spatial frequency between the test and adaptation patterns was introduced to attenuate the influence of early level adaptation in the other eye condition for the dynamic test (see Discussion for details). In the static test condition, the gratings were a static pattern. In the dynamic test condition, the gratings flickered every 500 ms (2.0 Hz); thus, contrast reversal occurred every 250 ms. The test stimulus was presented either to the same eye to which the adaptor was presented (same-eye test) or to the other eye (other-eye test). 

Procedure

Same-eye experiment

The adaptation and test stimuli were presented successively for 30 sec with a gap of 1 sec ( Figure 2c). Observers judged the direction of perceived motion for the center grating continuously by pressing one of the two keys assigned for rightward and leftward motion. Two keys were necessary due to the perceptual bi-stability of the dynamics test. Moreover, observers were instructed to refrain from pressing any key when no motion was perceived. The response was recorded at 60 Hz. Observers performed 10 trials for both the static test and the dynamic test in a randomized order. As a control, they also judged the direction of MAE induced by adaptors without viewing dynamic Mondrians for 10 trials for each dynamic and static tests, respectively. Five observers including one author participated in this experiment. 

Other-eye experiment

The basic procedure was the same as for the same-eye test except that the eyes for which the adaptors and test stimuli were presented to were swapped. Either dynamic or static test was presented during one session of the experiment. There were 10 trials for each condition and 20 for the control condition. Six observers participated in this experiment. Three of the six observers participated in both the static test and the dynamic test sessions while three others performed only the dynamic test condition. 

Storage effect experiment

The stimuli and procedures were the same as for the other-eye test except that dynamic Mondrians were presented during the gap between the adaptation and stimulus phases. Observers viewed the dynamic Mondrians and test stimuli with the same eye. The adaptation stimuli were presented to the other eye. The duration of the gap was 30 sec. Four observers participated in this experiment. Three of the four observers also participated in the other-eye test. 

Results

We measured dynamic and static MAEs during complete suppression of visual awareness of the motion adaptor using dCFS as well as during visible control conditions. 

Data analysis

In this experiment, complete suppression of the visual awareness of adaptors was essential. To eliminate trials in which suppression was in any way incomplete, observers were instructed to terminate the trial whenever they perceived the horizontal motion of the adaptors. Seven out of nine observers did not report that they viewed the adaptors' motion at any time during the experiment. One of the other two remaining observers reported motion perception of the adaptors in all conditions and we omitted data collected from this observer. The other observer reported motion perception in a few trials of the other-eye test. We omitted data collected from this observer for the other-eye test only. 

For each video refresh frame (16.7 ms), we sampled responses from observers both for the same direction as the adaptor and the direction opposite to the adaptor. We calculated the response ratios of motion in the same and opposite direction of the adaptor for each refresh frame (forward/reversed ratio) from these data. Subtraction of the forward ratio from the reversed ratio at each refresh frame was employed as an index for the magnitude of aftereffect at that frame (MAE response). We averaged the MAE response across observers and plotted it as a function of the time elapsed from the onset of the test stimuli. 

In addition, we calculated the total of MAE responses for all frames (30 sec = 1800 frames) and divided the total in the invisible adaptor condition by the total in the visible adaptor condition. We used this ratio between visible and invisible adaptor conditions as an index of suppression for MAE with dCFS (“suppression index”). 

Same eye experiment

For the static MAE in the visible adaptor condition (control) ( Figure 3a), the period in which the reversed ratio was significantly larger than the forward ratio (MAE period) lasted approximately 20 sec from the onset of the test stimulus. In the invisible adaptor condition, the static MAE (sMAE) survived but the magnitude was decreased. The duration in the invisible condition was significantly shorter than in the visible condition (approximately 15 sec) and MAE response ratios were smaller. The mean and standard error of the suppression index were 0.33 ± 0.12. For the dynamic MAE (dMAE), the overall trend was similar to that of the sMAE: the aftereffect was attenuated but did not abolish under complete perceptual suppression of the motion adaptor ( Figure 3b). The mean and standard error of suppression index were 0.46 ± 0.10. Here, no significant difference was found in the suppression index between the dynamic and static MAE condition ( t(4) = 0.67, p = 0.54). In summary, in the same-eye test, robust sMAE and dMAE were observed in both the visible and invisible conditions while the magnitudes of both MAEs were decreased for invisible motion adaptors. 

Figure 3

 

Same-eye experiment. (a) Results with static test stimuli. In the upper two panels, mean response ratios in control (left) and suppressed (right) conditions are shown. The red line indicates mean response ratios for “reversed“ motion (reversed ratio). The black line represents mean response ratios for “forward” motion (forward ratio). The black and red shaded areas are the 95% confidence intervals in each time frame. Lower panel shows the MAE response, a subtraction of mean reversed ratio and forward ratio. The results in the control condition are in blue and the results in suppression condition are in red. The blue and red areas are the 95% confident intervals in each time frame. (b) Dynamic test stimuli. The response ratios in the control and suppressed conditions are in the upper two panels. The lower panel is the MAE response in each condition calculated by the forward and reversed response ratios described above. The blue and red areas are the 95% confident intervals for each time frame in the control and suppression conditions.

Other-eye experiment

In the static test, all observers did not perceive any MAE with both visible and invisible adaptors ( Figure 4). We did not calculate suppression index for this condition. In contrast, robust inter-ocular transfer was observed in the visible condition in the dynamic test ( Figure 4). Interestingly, the inter-ocular transfer component of dMAE disappeared in the invisible condition. The mean suppression index was virtually zero (−0.06 ± 0.14) for the dMAE in this condition. 

Figure 4

 

Other-eye experiment. (a) Results with static test stimuli. In the upper two panels, mean response ratio in control (left) and suppressed (right) conditions are shown. Lower panel, MAE response in control (blue) and suppression (red) conditions are shown. (b) Results with dynamic test stimuli. Upper panels show mean response ratios. Lower panels show the MAE responses in control and suppression conditions.

Storage effects with dynamic continuous flash suppression

One concern is that dynamic Mondrians may affect the perception of subsequent flickering motion and counteract high-level adaptation effects. To eliminate this possibility, we tested the storage effect of dMAE. Both static and dynamic MAEs survive post-adaptation blank intervals for time periods equivalent to the duration of the MAE, a phenomenon termed the storage effect (Griffith & Spitz, 1959; Spigel, 1960, 1962a, 1962b, 1964; Wohlgemuth, 1911). 

The stimuli and the procedure for examining the storage effect were essentially the same as in the visible other-eye dMAE experiment, except that 30-sec presentations of dynamic Mondrians in the same eye as the test were inserted between the adaptation phase and the test phase ( Figure 5a). 

Figure 5

 

Storage effect experiment. (a) A diagram of the storage effect experiment. (b) Mean response ratios for reversed (upper) and forward (lower) ratios are shown. Results of other eye test with dynamic test stimuli in control and suppressed conditions are presented for comparison. (c) MAE responses calculated from data shown in the panel b. Results of other-eye test with dynamic test stimuli in control and suppressed conditions are presented for comparison.

Results show that the dynamic MAE was also perceived robustly in this experiment ( Figure 5b). That is, the storage effect was observed in cases where dynamic Mondrians were presented during the waiting period. Thus, it is unlikely that dynamic Mondrians counteracted the percept of flickering motion that canceled high-level adaptation in the invisible condition of the other eye experiment. These results support our conclusion that suppression of visual awareness of motion adaptors leads to complete suppression of high-level adaptation. 

Discussion

In this study, we report the magnitude of MAE with an adaptor rendered completely invisible in awareness. The main purpose of this study is to show that whether motion systems are functional under the strong binocular suppression, i.e., dCFS. As we summarized in Figure 6, results show that both the static and dynamic MAEs were attenuated with invisible adaptors with the same-eye test. In contrast, in the other-eye test condition, the dMAE is not observed with an invisible adaptor. These results lead to two conclusions. First, the early motion system can operate under the strong binocular suppression. Second, at least part of the higher-order motion system does not operate under strong binocular suppression. In the following discussion, we will further elucidate these conclusions. Finally, we will speculate why the dMAE was abolished in the other-eye test condition. This speculative discussion will clarify the factors that limit the interpretation of our results. 

Figure 6

 

Summary of the results. Mean suppression index for each condition. Note that the larger suppression index denotes the smaller suppression effect, i.e., when the suppression index is 1.0, it means completely no suppression on the MAE was observed and 0.0 means the complete abolishment of MAE. Error bars show the 95% confident intervals.

The results of sMAE in the same eye condition decisively demonstrate that low-level adaptation can occur without the motion adaptor reaching awareness. We first discuss the implications of this result in relation to previous literature. Several studies demonstrated that MAE under binocular rivalry is attenuated (Blake et al., 2006; Wiesenfelder & Blake, 1990). However, as we discussed in the Introduction, it is difficult to determine whether adaptation accumulated across periods where the adaptor was not perceived. We circumvented this problem by compressing the duration of perceived motion to zero. As a result, sMAE was attenuated to less than half of what was observed in the control condition. It is widely accepted that sMAE reflects adaptation in early-level motion processing. Therefore, the present results for sMAE strongly support the hypothesis that low-level motion detectors operate, at least partly, even when the motion adaptor is invisible. These results and interpretation parallel with the presence of the “positive motion aftereffect” where the sMAE is observed in the same direction as the perceived direction of adaptor whose first- and second-order structures move in opposite directions (Nishida & Sato, 1995). 

Interestingly, the attenuation of dMAE was similar to that of the sMAE in the same eye condition. The mean suppression indices for the dMAE and the sMAE were 0.46 and 0.33, respectively, and no significant difference was found between them. Similarly, Whitney and Bressler (2007) reported that the dMAE from luminance- (first-order) and contrast-defined (second-order) motion persists under crowding conditions. Although the method for repeated suppression was different from that of the present study and they did not provide a comparison with the sMAE, both studies demonstrate that the dMAE persists without the awareness of adaptors in the same-eye condition. 

In contrast with the results in the same-eye condition, dMAE with invisible adaptors completely disappeared in the other eye condition. The critical features that contribute to the interpretation of these results is that we used the condition where no inter-ocular transfer of sMAE was observed. dMAE is known to reflect not only high-level but also low-level adaptation (Nishida & Ashida, 2001; see Figure 7). However, low-level adaptation, if any, should have been revealed in the interocular transfer of sMAE as well as that of dMAE. Given that no significant interocular transfer was observed with the static test, it is highly unlikely that the interocular transfer of dMAE included low-level components of adaptation with the current setting. We can therefore conclude that disappearance of dMAE in the other-eye condition reveals that the adaptation of higher-level system does not occur with the invisible motion adaptor. 

Figure 7

 

A diagram illustrating the levels of adaptation and the magnitude of observed MAE. Red hexagrams denote adaptation at the early level (Me, Me′), whereas the green rectangle denotes higher level adaptation (Mh). The dMAE reflects both adaptations at the early and higher level, whereas the sMAE reflects only adaptation at the early level (cf. Nishida & Ashida, 2000, 2001). Given that we observed no interocular transfer in the sMAE (Me′ = 0), and no dMAE in the other-eye condition (f(Mh, Me′) = 0), we conclude that higher level adaptation does not occur when the motion adaptor is rendered invisible.

We identified conditions where the inter-ocular transfer of sMAE is not observed to address high-level adaptation. Here, we explain why the current setting was effective in removing interocular transfer of sMAE. We believe that the critical factor was the difference in spatial frequencies: 0.62 c/deg for adaptation gratings and 1.86 c/deg for test gratings. It is known that the adaptation of early motion detectors displays spatial frequency selectivity (Thompson, 1998). The bandwidth of the selectivity is around 1 octave, which is smaller than the difference in spatial frequency between our adaptation and test stimulus. Meanwhile, the robust interocular transfer component of dMAE with the visible adaptor (control condition) suggests that high-level adaptation is indeed less selective to spatial frequency, thus assuring that if there was any high-level adaptation in the invisible condition, it should have been observed within our experimental setup. 

Furthermore, to support the conclusion on the total cancellation of high-level adaptation with invisible adaptors, we confirmed that the dynamic Mondrians themselves did not drastically affect the subsequent perception of the dynamic test. This was shown by a robust storage effect of dMAE where 30 sec of dynamic Mondrians were inserted between motion adaptation and the dynamic test. The storage effect we observed is in line with Verstraten, Fredericksen, Grüsser, and van de Grind (1994), where they showed that the aftereffect of the motion adaptor remains potent in cases where a second motion stimulus orthogonal to the adaptor is presented after adaptation. Together with the directional orthogonality of the dynamic Mondrians to the adaptor, the very fact that they consisted of multiple motion patches seemed to have worked in favor of intact storage. This is in contrast to another study by Verstraten, Fredericksen, Van Wezel, Lankheet, and Van de Grind (1996), where they observed that the dynamic MAE is not stored when dynamic testing is preceded by a static test pattern. 

As discussed above, processing in the early motion system survives suppression, albeit weakened, while the higher-level motion system does not operate when the motion adaptor is rendered invisible. Although it is difficult to provide decisive answers on the mechanism underlying suppression of the higher-level motion system, we discuss the matter in relation to previous psychophysical and physiological findings. Here we consider two possible explanations. One straightforward interpretation is that the signal of the invisible adaptor does not arrive at the site of high-level adaptation. This model seems to be consistent with complete suppression of face adaptation for unseen faces by Moradi, Koch, and Shimojo (2005). At higher levels in the ventral visual pathway such as the inferior temporal cortex (IT), the proposed sites for face adaptation, the majority of neurons are perceptually modulated during binocular rivalry (Sheinberg & Logothetis, 1997; Tong, Nakayama, Vaughan, & Kanwisher, 1998). Although a dorsal pathway analogue of IT, where majority of neurons correlate to the perception of motion, is yet to be found, the first interpretation assumes the existence of such a cortical area and assigns the site of high-level motion adaptation to this cortical level. 

However, there is another possibility that the signal of the suppressed motion adaptor reaches the cortical level of the higher-level motion system but has no effect on the percept of the following dynamic test stimulus as well as on the online perception of the motion adaptor itself. We bring up this possibility because approximately 60% of neurons in the middle temporal (MT) area or V5, a higher-level cortical area for motion processing, respond to perceptually suppressed stimuli during binocular rivalry, though they are mostly binocular neurons (Logothetis & Schall, 1989). Numerous other studies indicate a difference between the dorsal and ventral stream pathway in the relationship with visual awareness or binocular rivalry (Fang & He, 2005; He, Carlson, & Chen, 2005; Lamme, 2006; Lamme & Roelfsema, 2000; Milner & Goodale, 1995). Assuming that the results of Logothetis and Schall (1989) hold for human observers at the site of high-level adaptation, neurons which have preference for the adaptor (horizontal translational motion) would still respond even when it is rendered invisible. That is, neural representations of the perceived dynamic Mondrians (perceptual network) and the non-perceived motion adaptor would co-exist. Under these assumptions, the second model claims that the perceptual network is independent from the neural activity that represents the non-perceived motion adaptor, thus leading to the total suppression of high-level adaptation. In relation to our view of independent perceptual networks, a study by Maier, Logothetis, and Leopold (2007) showed that whether middle temporal (MT) neurons are perceptually modulated during flash suppression depended upon the rivaling monocular stimulus, indicating that the perceptual network is not fixed to a certain population of neurons. The implication of our hypothesis of independent circuitry for high-level adaptation and their view of a flexible perceptual network are interesting and calls for further investigation. 

The two models are still possible due to the lack of knowledge on the precise neural substrates of the higher-level motion detector and where rivalry between two stimuli is finally resolved. Moreover, the relationship between normal rivalry suppression and CFS is unclear. The mechanism of CFS is believed to be different from that of binocular rivalry (Tsuchiya, Koch, Gilroy, & Blake, 2006). As the dCFS method can be considered a subtype of CFS, it is possible that input of the non-perceived motion stimulus may not have penetrated into higher cortical levels such as the middle temporal (MT) or middle superior temporal area (MST). At this point, it is difficult to relate our psychophysical findings with certain brain areas owing to the simple lack of knowledge on motion adaptation and cortical circuitry. Nonetheless, exploration of the neural substrate of our psychophysical findings holds out a new dimension to the search for the neural mechanism of visual awareness. 

Acknowledgments

We thank Randolph Blake, Min-Suk Kang, Eunice Yang, Ryota Kanai, and Junji Watanabe for comments on an early version of the manuscript. This research was supported by Grant-in-Aids for Scientific Research on Priority Areas-Higher Order Brain Functions—from the Ministry of Education, Culture, Sports, Science and Technology (17022015) for MW. KM is supported by the Japan Society for the Promotion of Science. 

Commercial relationships: none. 

Corresponding author: Kazushi Maruya. 

Email: [e-mail]. 

Address: 301 Wilson Hall, 111 21st Avenue South, Nashville, Tennessee, 37203, USA.