Perceptual entrainment of individuallyunambiguous motions

School of Psychology, The University of Sydney,NSW, Australia, &

Australian Centre of Excellence in Vision Science, AustraliaIsabelle Mareschal

School of Psychology, The University of Sydney,NSW, Australia, &

Australian Centre of Excellence in Vision Science, AustraliaColin W. G. Clifford

When two bistable spheres defined by dots oscillating around a common axis (coaxial condition) are placed near eachother, their motions become entrained such that they appear to rotate in the same direction. When the dots in the twospheres oscillate around parallel axes (non-coaxial condition), entrainment is much reduced, suggesting that thisphenomenon is driven by interactions between the global stimulus representations rather than between the locallyambiguous motion signals. In order to examine where these interactions may be arising, we created highly unambiguousspheres by introducing size and contrast cues on the dots. We find that coupling is strong for the two nearly unambiguousspheres rotating in opposite directions of motion in the coaxial condition but, importantly, that this effect is subject toattentional control. Forcing observers to withdraw their attention from the rotating spheres by performing a demanding taskat fixation significantly reduces the amount of coupling, arguing against mediation by low-level interactions between thespheres. These results reveal a disjunct between the local motion signals that are unambiguous and the global percept,implicating a process that requires active suppression of reliable visual information. We propose that this is a dynamicprocess that requires feedback from higher levels to bind opposite motions into a unified directional percept.

Keywords: perceptual entrainment, unambiguous motion, attention

Citation: Mareschal, I., & Clifford, C. W. G. (2012). Perceptual entrainment of individually unambiguous motions. Journal ofVision, 12(1):24, 1–8, http://www.journalofvision.org/content/12/1/24, doi:10.1167/12.1.24.

Introduction

In bistable displays, the perception of a stimulusalternates between two competing perceptual interpreta-tions even though its physical properties remainunchanged. A common example is structure from motion(SFM) where the oscillation of dots around a single axisyields the perception of a three-dimensional rotatingsphere. The motion of this stimulus is inherently ambig-uous and the perceived direction of rotation switchesfrequently between the two alternatives (Bradley, Chang,& Andersen, 1998; Nawrot & Blake, 1989; Treue, Husain,& Andersen, 1991). The neural processes underlyingbistability have been proposed to result both from low-level adaptation mechanisms as well as high-level group-ing and attentional processes. Support for the idea thatadaptation underlies the bistability comes from evidencethat the rate of switches between the two alternativepercepts can be decreased or abolished by presenting thestimulus in ways that minimize adaptation. This has beenachieved either by producing a wobble in the axis ofrotation (Blake, Sobel, & Gilroy, 2003), by perceptuallyremoving the object from view (Leopold, Wilke, Maier, &Logothetis, 2002), or by adapting out one of the directions

of rotation with an unambiguously rotating sphere(Nawrot & Blake, 1989).Evidence for higher level influences on the alternation

rate comes from experiments showing that the duration ofa dominant percept can be increased by attentionalmanipulations. Nakatani and van Leeuwen (2005) findthat an observer’s level of attention can account forindividual differences in switch rates for the Necker cube,and Brouwer and van Ee (2006) report that the duration ofperception of the direction of the SFM is under attentionalcontrol. Using a quartet of dots whose direction of motioncan be perceived as either horizontal or vertical, Kohler,Haddad, Singer, and Muckli (2008) report that observerscan switch the direction of motion by attention. There is,however, some debate as to the overall role of attentionin bistable perception with Pastukhov and Braun (2007)reporting that some phenomenal switches occur withoutany prompting by attention.High-level perceptual grouping also influences the

interpretation of ambiguous visual stimuli (e.g., Gillam,1972, 1976; Ramachandran & Anstis, 1983). In binocularrivalry, it has been shown that the brain reassemblespatches of images presented in different eyes into coherentpercepts (Kovacs, Papathomas, Yang, & Feher, 1996).Using bistable stimuli like the SFM, Grossman and

Journal of Vision (2012) 12(1):24, 1–8 http://www.journalofvision.org/content/12/1/24 1

doi: 10 .1167 /12 .1 .24 Received August 1, 2011; published January 26, 2012 ISSN 1534-7362 * ARVO

Dobbins (2003) find that coupling between the twospheres is strong when both spheres are ambiguous butreduced when one is made unambiguous. They argue thatthe lack of coupling between the ambiguous and unam-biguous stimuli is not due to the dissimilarity between thestimuli but rather due to global feedback, which isproportional to the amount of ambiguity. Freeman andDriver (2006) replicate this result when the unambiguity isdetermined by opacity. However, when luminance anddepth cues are introduced, they report strong couplingbetween stimuli that differ in the strength of the direc-tional ambiguity of their motion.A critical factor in elucidating the nature of the under-

lying mechanisms of bistability is whether high-level cuesare relevant even in cases where low-level cues aresufficient to disambiguate the stimuli. By some computa-tional accounts, visual competition is a feedforward,hierarchical process in which the higher levels are engagedonly when resolution is not possible at the previous level(Freeman, 2005; Wilson, 2003). However, alternativeframeworks emphasize the top-down influence of priorknowledge in a more complex interplay of feedforwardand feedback signals (Hohwy, Roepstorff, & Friston,2008; Watson, Pearson, & Clifford, 2004). In order todetermine if high-level effects influence the perception ofbistable stimuli when low-level signals are disambiguated,we examined coupling using two highly unambiguousspheres. Since the direction of either sphere presented inisolation is stable, any coupling that arises between the twospheres presented together will not simply be the passiveresult of random, single sphere dynamics but represent anactive process of the visual system to override the physicalcharacteristics of one of the spheres. Given that higherlevel grouping principles are believed to exert theirinfluence on the visual competition process through atten-tionally modulated feedback (Carmel, Walsh, Lavie, &Rees, 2010; Kanai, Bahrami, & Rees, 2010; Kanai,Carmel, Bahrami, & Rees, 2011), we also postulated thatcoupling would be reduced when attention is divertedfrom the SFM stimulus.

Methods

Observers

One of the authors and 5 naive observers served assubjects. All wore optical correction as necessary.

Apparatus and stimuli

ADell Optiplex computer runningMATLAB (MathWorks)was used for stimulus generation, experiment control, andrecording subjects’ responses. The programs controlling

the experiment incorporated elements of the PsychToolbox(Brainard, 1997). Stimuli were displayed on a DiamondDigital monitor (resolution = 1024 * 768 pixels, refreshrate = 85 Hz, background luminance = 50 cd/m2)driven by the computer’s built-in Radeon graphics card.The display was calibrated using a photometer and line-arized using look-up tables in software. At the viewingdistance of 57 cm, 1 pixel subtended 2.1 arcmin.

Stimulus

A set of 200 dots whose position oscillates around eithera vertical (for left/right motion) or a horizontal axis (forUp/Down motion) was used to create the illusion of aspherical object rotating in space. When the dots are equalin size and contrast, the motion is ambiguous and flipsbetween the two alternative percepts with durations ineither direction following a gamma distribution (Leopold& Logothetis, 1999). We created highly unambiguousdirectional spheres by introducing size and contrast cuesinto the stimulus resulting in an “imposed” direction ofmotion. This was achieved by making the dots in onetrajectory (e.g., “up” if the imposed direction was upward)more visible/salient by increasing their size and contrastcompared to those in the other trajectory (e.g., “down”).The larger dots were black and had a diameter of 7 arcmin,while the smaller dots had a diameter of 3 arcmin and aluminance of 32 cd/m2. The dots followed simpleharmonic motion with a peak speed of 2.1 deg/s. Eachillusory sphere subtended 4.2 deg in diameter and itscenter was separated from the fixation cross by 3.3 deg.

ProcedureExperiment 1: No attentional load—Constantmonitoring of two directions of motion

Schematic example of one frame of the stimulus isshown in Figure 1a. Observers fixated a cross that waspresent throughout the stimulus duration and wererequired to monitor the direction of motion of two nearlyunambiguous spheres presented on either side of fixationthat always rotated in opposite directions. Two conditionswere tested: a coaxial condition, whereby the spheresrotated either upward or downward and shared a commonaxis of motion (though the motions were always inopposite directions), and a non-coaxial condition, wherebythe spheres rotated either to the left or right of a verticalaxis and had parallel axes of motion. Observers completeda minimum of 10 blocks, each lasting 240 s. During ablock, 8 regular directional switches were imposed on thestimulus every 30 s (plus a temporal jitter of T400 ms) toensure a balanced representation of the two motions oneither side of fixation, resulting in eight full cycles of dualmotion. The observer had to indicate the direction of

Journal of Vision (2012) 12(1):24, 1–8 Mareschal & Clifford 2

Journal of Vision (2012) 12(1):24, 1–8 Mareschal & Clifford 3

motion using response keys (“a” and “s” in the coaxialcondition or “a” and “z” in the non-coaxial condition) forthe stimulus on the left of fixation and (“k” and “l”(coaxial) or “k” and “m” (non-coaxial)) for the stimuluson the right of fixation (Figure 1a). Data from the 8 cyclesof a block were pooled together and averaged acrossblocks (for a duration of 29.6 s because of the temporaljitter). We were not interested in biases due to spatiallocation and combined the four “up” (or “right”) con-ditions on the left of fixation with the four “up” (or“right”) conditions on the right of fixation. Only ObserverIM ran the conditions interleaved.A single sphere condition was also run on each observer

to measure a “baseline” level from which to calculate theamount of coupling that would be predicted from switchesbased on the single sphere dynamics. Observers com-pleted a minimum of 4 blocks in two conditions: (a) whenthe direction of rotation of the single sphere was around ahorizontal axis (termed “coaxial,” as the single spheremotion is the same as one of the spheres in the coaxialcondition) and (b) when the direction of rotation wasaround a vertical axis (termed non-coaxial). In half theblocks, the sphere was on the left side of fixation, and inthe other half, it was on the right side of fixation toaccount for any biases due to spatial location. Theproportion of coupling predicted from the single spheredynamics was calculated as

Predicted proportion coupling ¼ ð1j pÞ * qþ ð1j qÞ * p;ð1Þ

where each sphere in isolation had the probability of beingseen to rotate in its imposed direction of motion, p (e.g.,“up”) and q (e.g., “down”).In the few instances where no keys were pressed or the

two keys for the same sphere were pressed (between0.01% and 0.5% of all frames), responses for thecorresponding frames were coded as not indicating thecorrect direction of motion.

Experiment 2: Attentional load—Regular sampling oftwo directions of motion in the coaxial condition

High attentional load task: The stimuli were similar tothose in Experiment 1, except that a string of numbers waspresented at fixation and observers were required to countthe number of times a white “2” appeared over the 240-speriod. The nine distracter numbers (4 white and 5 black)and the one target had an equal probability of appearing inthe string. While observers performed the counting task atfixation, they were also required to do a directional taskon both spheres at regular intervals throughout the trial.Starting from 3 s after the stimulus onset, and occurringevery 5 s thereafter, a cross would appear at fixation insteadof a number. This cued observers to report the direction ofmotion of the two spheres within a response window usingthe same response keys as in Experiment 1. The durationsof the response window, the fixation cross, and the targetand non-target numbers were always equal and wereindividually set for each observer based on practice runsto ensure performance above 70% (between 300 and500 ms). At the end of each of the eight cycles, the twospheres would disappear and a number would appear atfixation corresponding to the total number of times thetarget was present in the cycle. Using a key press, observersindicated whether this total was correct or not, where anincorrect total was produced by adding 1 to the final counton 50% of the trials. The total was present for 5 s, followedby a gray screen for 2 s before the next cycle commenced.Observers practiced on the task prior to data collectionand performed a minimum of 10 blocks of 240 s each.Low attentional load task: The stimulus presentation

was the same to avoid any potential confounds caused bythe string of numbers and observers were instructed to onlyperform the directional task. At the end of each 240-msperiod, the sum answer that subjects were instructed toignore was presented for 2 s followed by a blank screen for2 s. Observers performed a minimum of 10 blocks.

Results

Experiment 1

Observers monitored the perceived directions of twospheres over time. Although the spheres were always

Figure 1. SFM stimulus procedure and data for one observer.(a) Schematic of one frame in Experiment 1. Fixation is a crossthat is present throughout the stimulus duration. Dashed line andarrows illustrate the directions of motion but were not present inthe experiment. Observers were tested in blocks that lasted 240 sconsisting of eight 30-s periods of unambiguous motion (only2 periods shown). In each period, the stimuli on the left and right offixation rotated in opposite directions (illustrated by arrows) whileobservers reported both directions of motion with two key presses.(b) The perceived direction of motion (red lines) of the spheres didnot always match their physical directions, leading to coupling(shaded gray areas). (c) Directional data in coaxial and non-coaxial conditions for Observer IM. The leftmost column plots thetwo directions of motion separately in the (top) coaxial and(bottom) non-coaxial conditions. The blue curve plots the propor-tion of times the observer said “up” (or “right”) for an upward (orrightward) rotating sphere over the 30-s time period. The red curveis the proportion of times the observer classified a downward (orleftward) motion as going “up” (or “right”). Error bars contain T1standard error. The middle column plots data in the same formatfor single sphere stimuli. The rightmost column plots in green theproportion of coupling as a function of time. The black curvesrepresent the proportion of random coupling predicted from thesingle sphere dynamics. Error bars are 95% confidence intervals.

Journal of Vision (2012) 12(1):24, 1–8 Mareschal & Clifford 4

rotating in opposite directions (Figure 1a depicting coaxialcondition), observers would often perceive them as rotatingin the same direction (shaded areas in Figure 1b). Figure 1cplots the proportion of time Observer IM reported that thespheres were moving in one of the imposed directions as afunction of time (top shows coaxial condition, whereasbottom shows non-coaxial). If neither sphere had appearedto switch from its imposed direction of motion, this valueshould always equal 1 for one of the directions and 0 forthe other. In both conditions, the curves start to convergetoward the center (0.5), slightly earlier in the coaxialcondition than the non-coaxial condition.In the single sphere data, fluctuations in the directions

of motion do not start to appear until after approximately13 s. The rightmost column plots coupling that reflects theproportion of times in each frame the observer indicatedthat the two spheres were rotating in the same direction.Note that for Observer IM coupling diverges significantly

from the prediction based on the single sphere data(Equation 1) after roughly 2 s in the coaxial condition.In the non-coaxial condition, the coupling is weaker andonly diverges from the single sphere prediction afterapproximately 22 s.Figure 2 plots coupling in the coaxial and non-coaxial

conditions for the five naive observers. Although there issome intersubject variability regarding the onset andstrength of coupling, all observers display significantdepartures from the single sphere dynamics in the coaxialcondition. The non-coaxial condition leads to moreintersubject variability with LB perceiving the veridicaldirections of the two spheres throughout the trials,whereas the other observers display various degrees ofcoupling in the non-coaxial condition.The time-averaged proportion of coupling in the two

dual sphere conditions is plotted against the predictedcoupling in the single sphere conditions for the six

Figure 2. Coupling in (left) coaxial and (right) non-coaxial conditions for five naive observers. The black curves represent the proportion ofcoupling predicted from the single sphere dynamics. Error bars are 95% confidence intervals. Bottom right plots are the proportion ofmeasured coupling (dual sphere conditions) and predicted coupling (single sphere conditions) averaged across time in the coaxialarrangement (left) and in the non-coaxial arrangement (right). Error bars are 95% confidence intervals and solid line depicts equalproportions of actual and predicted coupling.

Journal of Vision (2012) 12(1):24, 1–8 Mareschal & Clifford 5

observers, in the bottom right of Figure 2. The left-handplot shows data in the coaxial condition with all datapoints well above the unity line: The coupling in the dualaxis condition is significantly different from that predictedfrom the single sphere dynamics (within-subjectsANOVA, F1,20 = 41.7, p G 0.0001). The right-hand plotshows data in the non-coaxial condition with the datapoints clustering above the line of equality and the amountof coupling also significantly different from the singlesphere condition (F1,20 = 10.1, p = 0.0049). There is also asignificant interaction (F1,20 = 12.5, p = 0.0021) betweenthe condition (coaxial or non-coaxial) and the configu-ration (dual or single) indicating that coupling is signifi-cant in the dual coaxial condition.

Experiment 2

In order to determine if diverting attention away fromthe spheres would reduce coupling, observers performed acounting task at fixation while reporting the direction ofmotion of the spheres only when cued. Figure 3 plots theeffects of attention on the amount of coupling between therotating spheres in the coaxial condition. When observersperformed the high attentional load task at fixation (graysymbols), there was a significant decrease in the propor-tion of coupling compared to the low attentional load task(t5 = 3.35, p G 0.05), most markedly for AA, IM, and LB.For all observers, the amount of coupling measured inExperiment 2 with the low attentional load task (respond

Figure 3. Proportion of coupling in the dual sphere conditions sampled at 6 discrete points within the 30-s period. Gray symbols representthe high attentional load condition; black symbols are the low load condition. Error bars are T1 standard errors. Percent correct scoresrepresent performance on the secondary task.

Journal of Vision (2012) 12(1):24, 1–8 Mareschal & Clifford 6

upon presentation of fixation cross only: black symbols)was similar to the constant monitoring technique inExperiment 1 (as can be seen by comparing black symbolsin Figure 3 with the green curves from Figures 1 and 2).

Discussion

We find rapid and reliable coupling between twounambiguous spheres when they rotate around a commonaxis, supporting the perceptual construct of a single,unidirectional, object. When attention is diverted fromthe structure-from-motion stimuli, we find that the percep-tual grouping of their directions of rotation is significantlyreduced. This study differs fundamentally from previousexperiments using bistable stimuli where the visual systemreceived conflicting or ambiguous information from thesame spatial location. Here, we have the oppositesituation: Faced with locally unambiguous signals, thevisual system overrides reliable information to generate apercept that is inconsistent with one-half of the stimulus.It has been reported previously that the alignment of

local and global cues can enhance performance on manybasic, low-level tasks. Using a contour detection task,Field, Hayes, and Hess (1993) report that performancedecreases as the degree of collinearity between the localGabor patches and the global contour axis is reduced. Asimilar dependence on the coalignment between the localGabor patches and the global axis of the stimulusunderlies contrast facilitation and discrimination tasks(e.g., Bonneh & Sagi, 1998; Cannon & Fullenkamp,1991; Kapadia, Ito, Gilbert, & Westheimer, 1995). Morerecently, the relative importance of both local and globalcues has been examined in more higher level processessuch as binocular rivalry. Silver and Logothetis (2004)found that the perceptual grouping of dot arrays withoutany change in low-level stimulus features could alterdominance periods in rivalry. Using biological motion toinduce perceptual grouping, Watson et al. (2004) reportrivalry between point-light walkers that occurs due tohigh-level perceptual grouping rather than low-level cues,since at the local level these stimuli would fail to initiaterivalry. Consistent with this finding of global dominance,Andrews and Lotto (2004) report that rivalry can occurbetween physically identical monocular stimuli if theyelicit different higher level percepts. They also show thatphysically different stimuli that would otherwise causerivalry fail to do so if they are perceived as belonging to thesame object. However, Carlson and He (2004) examinedthis issue using bar and grating stimuli and report thatrivalry was due to the local interactions, not globalpercepts. Our results using bistable stimuli support theformer findings whereby global cues in the stimulus alterthe stability of one sphere’s direction of rotation. In order toachieve a unidirectional motion percept, the visual systemoverrules local motion signals in half of the stimulus.

The ability of the visual system to impose coupling onthe rotating spheres is severely hampered when attentionis diverted away from them. Attention is known tomodulate the size of visual phenomena such as the motionaftereffect (Chaudhuri, 1990; Lankheet & Verstraten,1995; Rees, Frith, & Lavie, 1997), the tilt aftereffect(Spivey & Spirn, 2000), and contrast facilitation (Freeman,Driver, Sagi, & Zhaoping, 2003), as well as the amount ofadaptation to illusory contours (Montaser-Kouhsari &Rajimer, 2004). Our results are consistent with thesereports and we find that observers must attend to thecoaxial spheres for coupling to occur. We suggest thatcoupling is the result of a feedback mechanism thatsuppresses local motion signals and that attention modu-lates the strength of this feedback.There is growing interest in determining the anatomical

areas involved in the perception of bistable stimuli(Carmel et al., 2010; Kanai et al., 2010, 2011). In theirmost recent paper, Kanai et al. (2011) report a structuraland functional segmentation of the superior parietal lobule(SPL) that receives feedback from higher cortical areas tomodulate bistable perception. Given that our behavioralresult implicates higher level mechanisms, it would be ofinterest to determine whether the SPL might also regulatethe entrainment of highly unambiguous stimuli.

Acknowledgments

C.W.G. Clifford was funded by an Australian ResearchCouncil Future Fellowship.

Commercial relationships: none.Corresponding author: Isabelle Mareschal.Email: isabelle.mareschal@sydney.edu.au.Address: School of Psychology, The University ofSydney, Room 508, Griffith Taylor Building, NSW2006, Australia.

References

Andrews, T. J., & Lotto, B. R. (2004). Fusion and rivalryare dependent on the perceptual meaning of visualstimuli. Current Biology, 14, 418–423.

Blake, R., Sobel, K. V., & Gilroy, L. A. (2003). Visualmotion retards alternations between conflicting per-ceptual interpretations. Neuron, 39, 869–878.

Bonneh, Y., & Sagi, D. (1998). Effects of spatialconfiguration on contrast detection. Vision Research,38, 3541–553.

Bradley, D. C., Chang, G. C., & Andersen, R. A. (1998).Encoding of three-dimensional structure-from-motionby primate are MT neurons. Nature, 392, 714–717.

Brainard, D. (1997). The psychophysics toolbox. SpatialVision, 39, 217–281.

Journal of Vision (2012) 12(1):24, 1–8 Mareschal & Clifford 7

Brouwer, G. J., & van Ee, R. (2006). Endogenousinfluences on perceptual bistability depend on exog-enous stimulus characteristics. Vision Research, 46,3393–3402.

Cannon, M. W., & Fullenkamp, S. C. (1991). Spatialinteractions in apparent contrast: Inhibitory effectsamong grating patterns of different spatial frequen-cies, spatial positions and orientations. VisualResearch, 31, 1985–1998.

Carlson, T. A., & He, S. (2004). Competing globalrepresentations fail to initiate binocular rivalry.Neuron, 43, 907–914.

Carmel, D., Walsh, V., Lavie, N., & Rees, G. (2010).Right parietal TMS shortens dominance in durationsin binocular rivalry. Current Biology, 20, 799–800.

Chaudhuri, A. (1990). Modulation of the motion after-effect by selective attention. Nature, 344, 60–62.

Field, D. J., Hayes, A., & Hess, R. F. (1993). Contourintegration by the human visual system: Evidence for alocal “association field”.Vision Research, 33, 173–193.

Freeman, A. W. (2005). Multistage model for binocularrivalry. Journal of Neurophysiology, 94, 4412–4420.

Freeman, E., & Driver, J. (2006). Subjective appearanceof ambiguous structure-from-motion can be driven byobjective switches of a separate less ambiguouscontext. Vision Research, 46, 4007–4023.

Freeman, E., Driver, J., Sagi, D., & Zhaoping, L. (2003).Top-down modulation of lateral interactions in earlyvision: Does attention affect integration of the wholeor just perception of the parts? Current Biology, 13,985–989.

Gillam, B. (1972). Perceived common rotary motion ofambiguous stimuli as a criterion for perceptualgrouping. Perception & Psychophysics, 11, 99–101.

Gillam, B. (1976). Grouping of multiple ambiguouscontours: Towards an understanding of surfaceperception. Perception, 5, 203–209.

Grossman, J. K., & Dobbins, A. C. (2003). Differentialambiguity reduces grouping of metastable objects.Vision Research, 43, 359–369.

Hohwy, J., Roepstorff, A., & Friston, K. (2008). Predic-tive coding explains binocular rivalry: An epistemo-logical review. Cognition, 108, 687–701.

Kanai, R., Bahrami, B., & Rees, G. (2010). Human parietalcortex structure predicts individual differences inperceptual rivalry. Current Biology, 20, 1626–1630.

Kanai, R., Carmel, D., Bahrami, B., & Rees, G. (2011).Structural and functional fractionation of right superiorparietal cortex in bistable perception. Current Biology,21, 106.

Kapadia, M. K., Ito, M., Gilbert, C. D., & Westheimer, G.(1995). Improvement in visual sensitivity by changesin local context: Parallel studies in human observersand V1 of alert monkeys. Neuron, 15, 843–856.

Kohler, A., Haddad, L., Singer, W., & Muckli, L. (2008).Deciding what to see: The role of intention andattention in the perception of apparent motion. VisionResearch, 48, 1096–1106.

Kovacs, I., Papathomas, T. V., Yang, M., & Feher, A.(1996). When the brain changes its mind: Interoculargrouping during binocular rivalry. Proceedings of theNational Academy of Sciences of the United States ofAmerica, 93, 15508–15511.

Lankheet, M. J., & Verstraten, F. (1995). Attentionalmodulation of adaptation to two-component trans-parent motion. Vision Research, 35, 1401–1412.

Leopold, D. A., & Logothetis, N. K. (1999). Multistablephenomena: Changing views in perception. Trends inCognitive Sciences, 3, 254–264.

Leopold, D. A., Wilke, M., Maier, A., & Logothetis, N. K.(2002). Stable perception of visually ambiguouspatterns. Nature Neuroscience, 5, 605–609.

Montaser-Kouhsari, L., & Rajimehr, R. (2004). Attentionalmodulation of adaptation to illusory lines. Journal ofVision, 4(6):3, 434–444, http://www.journalofvision.org/content/4/6/3, doi:10.1167/4.6.3. [PubMed] [Article]

Nakatani, H., & van Leeuwen, C. (2005). Individualdifferences in perceptual switching rates: The roleof occipital alpha and frontal theta band activity.Biological Cybernetics, 93, 343–354.

Nawrot, M., & Blake, R. (1989). Neural integration ofinformation specifying structure from stereopsis andmotion. Science, 244, 716–718.

Pastukhov, A., & Braun, J. (2007). Perceptual reversals needno prompting by attention. Journal of Vision, 7(10):5,1–17, http://www.journalofvision.org/content/7/10/5,doi:10.1167/7.10.5. [PubMed] [Article]

Ramachandran, V. S., & Anstis, S. M. (1983). Perceptualorganization in moving patterns. Nature, 304, 529–531.

Rees, G., Frith, C., & Lavie, N. (1997). Modulatingirrelevant motion perception by varying attentionalload in an unrelated task. Science, 278, 1616–1619.

Silver, M. A., & Logothetis, N. K. (2004). Grouping andsegmentation in binocular rivalry. Visual Research,44, 1675–1692.

Spivey, M. J., & Spirn, M. J. (2000). Selective visualattention modulates the direct tilt aftereffect. Percep-tion & Psychophysics, 62, 1525–1533.

Treue, S., Husain, M., & Andersen, R. A. (1991). Humanperception of structure from motion. Visual Research,31, 59–75.

Watson, T. L., Pearson, J., & Clifford, C. W. G. (2004).Perceptual grouping of biological motion promotesbinocular rivalry. Current Biology, 14, 1670–1674.

Wilson, H. R. (2003). Computational evidence for a rivalryhierarchy in vision. Proceedings of the NationalAcademy of Sciences of the United States of America,100, 14499–14503.

Journal of Vision (2012) 12(1):24, 1–8 Mareschal & Clifford 8