Individual differences in sensory eye dominance reflected in the dynamics of binocular rivalry

Vision Research xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Vision Research journal homepage: www.elsevier.com/locate/visres

Individual differences in sensory eye dominance reflected in the dynamics of binocular rivalry Kevin C. Dieter ⇑, Jocelyn L. Sy, Randolph Blake Vanderbilt Vision Research Center and Dept. of Psychology, Vanderbilt University, Nashville TN, 37240, USA

a r t i c l e

i n f o

Article history: Received 25 May 2016 Received in revised form 25 September 2016 Accepted 27 September 2016 Available online xxxx Keywords: Binocular rivalry Sensory eye dominance Binocular vision Individual differences

a b s t r a c t Normal binocular vision emerges from the combination of neural signals arising within separate monocular pathways. It is natural to wonder whether both eyes contribute equally to the unified cyclopean impression we ordinarily experience. Binocular rivalry, which occurs when the inputs to the two eyes are markedly different, affords a useful means for quantifying the balance of influence exerted by the eyes (called sensory eye dominance, SED) and for relating that degree of balance to other aspects of binocular visual function. However, the precise ways in which binocular rivalry dynamics change when the eyes are unbalanced remain uncharted. Relying on widespread individual variability in the relative predominance of the two eyes as demonstrated in previous studies, we found that an observer’s overall tendency to see one eye more than the other was driven both by differences in the relative duration and frequency of instances of that eye’s perceptual dominance. Specifically, larger imbalances between the eyes were associated with longer and more frequent periods of exclusive dominance for the stronger eye. Increases in occurrences of dominant eye percepts were mediated in part by a tendency to experience ‘‘return transitions” to the predominant eye – that is, observers often experienced sequential exclusive percepts of the dominant eye’s image with an intervening mixed percept. Together, these results indicate that the oftenobserved imbalances between the eyes during binocular rivalry reflect true differences in sensory processing, a finding that has implications for our understanding of the mechanisms underlying binocular vision in general. Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction Under ordinary viewing conditions, binocular visual perception belies little hint of its dual monocular origins – it feels as if we’re seeing the world through a single, cyclopean eye. Yet because the earliest stages of visual processing are patently monocular, it is feasible that the two eyes might contribute differentially to the perceptual experience, culminating from the processes promoting binocular combination. It is known, for example, that optical aberrations can differ between the two eyes of some individuals (Porter, Guirao, Cox, & Williams, 2001). Moreover, neural processing in the retina, in the thalamus, and in layer 4 of the primary visual cortex is accomplished largely by neurons that respond to inputs originating in one eye or the other but not both (Squire et al., 2003). Differential contributions from the two eyes might also arise within cortical neural mechanisms directly involved in binocular combination (Cumming & DeAngelis, 2001). It is under⇑ Corresponding author at: Dept. of Psychology, Vanderbilt University, PMB 407817, 2301 Vanderbilt Place, Nashville, TN 37240-7817, USA. E-mail address: [email protected] (K.C. Dieter).

standable, therefore, that some vision scientists are interested in establishing methods for determining the extent to which binocular vision is impacted by the level of balance between the contributions from the left eye (LE) and the right eye (RE). One broad category of methods used to assess the relative impact of the two eyes on binocular vision measures an observer’s reliance on one eye over the other for aligning targets in the environment, i.e., sighting dominance (e.g. Fink, 1938). Other measures focus on differences between the eyes in monocular acuity and contrast sensitivity (e.g. Suttle et al., 2009). By and large these methods have proven rather unreliable (e.g. Banks, Ghose, & Hillis, 2004; Khan & Crawford, 2001), and they tend to be unrelated to one another and/or to other binocular visual functions (Ehrenstein, Arnold-Schulz-Gahmen, & Jaschinski, 2005; Mapp, Ono, & Barbeito, 2003; Pointer, 2007; Rice, Leske, Smestad, & Holmes, 2008). A technique that has proven successful at both measuring differences between the eyes and relating these to the quality of binocular vision is binocular rivalry. The conditions instigating binocular rivalry are created by dichoptic presentation of conflicting visual stimuli to the two eyes, thereby provoking reciprocal,

http://dx.doi.org/10.1016/j.visres.2016.09.014 0042-6989/Ó 2016 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Dieter, K. C., et al. Individual differences in sensory eye dominance reflected in the dynamics of binocular rivalry. Vision Research (2016), http://dx.doi.org/10.1016/j.visres.2016.09.014

2

K.C. Dieter et al. / Vision Research xxx (2016) xxx–xxx

alternating periods of perceptual dominance and suppression between the two stimuli when they are viewed for an extended period of time (Alais, 2012; Blake & O’Shea, 2009). Importantly, the amount of time that one or the other monocular stimulus is seen depends on the relative salience of the two stimuli. Thus, for example, a high contrast stimulus will predominate over a lower contrast stimulus (Brascamp, van Ee, Noest, Jacobs, & van den Berg, 2006; Levelt, 1968) and a well-focused stimulus will predominate over a blurred one (Arnold, Grove, & Wallis, 2007). Similar biases in dominance are observed when disparities in monocular salience arise from intrinsic differences between the eyes, even when the stimuli being viewed by the LE and RE are equal in physical strength (Handa et al., 2004; Porac & Coren, 1978). One can systematically manipulate the contrast or the luminance of the stimulus viewed by the disadvantaged eye to achieve equal predominance during rivalry, thus acquiring a quantitative metric of sensory eye dominance (SED; Ooi & He, 2001). This metric is predictive of binocular visual performance in a stereoacuity task (Xu, He, & Ooi, 2011), and, when used to determine eye assignment for monovision correction (i.e., purposefully imbalanced refractive correction of the two eyes allowing distance vision by one eye and near vision for the other, sometimes favored by presbyopes or by patients following cataract surgery), this information can help improve binocular contrast sensitivity (Zheleznyak, Alarcon, Dieter, Tadin, & Yoon, 2015). These promising links between SED and measures of binocular visual function point to the potential utility of binocular rivalry for assessing individual differences in binocular function. However, it remains unknown exactly how imbalances in eye dominance affect the dynamics of ongoing binocular rivalry in order to promote increased predominance of the dominant eye. Specifically, some previous methods have relied primarily on brief presentations of dichoptic stimulation (Ooi & He, 2001; Xu, He, & Ooi, 2012; Xu et al., 2011). Such methods target onset rivalry, a brief period of time (
1 s) following the initial presentation of dichoptic stimuli that is characterized by an increased influence of factors such as attention and color on rivalry predominance (Carter & Cavanagh, 2007; Mitchell, Stoner, & Reynolds, 2004; Stanley, Forte, Cavanagh, & Carter, 2011). Others have focused on the overall predominance of one eye across an extended binocular rivalrytracking period (Handa et al., 2004; Porac & Coren, 1978; Zheleznyak et al., 2015). Importantly, there are a variety of ways that binocular rivalry dynamics could be altered that would yield a larger proportion of viewing time for one eye during an extended tracking block. These include changes in the relative frequency and/or duration of LE versus RE percepts (and relatedly, changes in alternation rate), and/or a reduction in the frequency or duration of periods of mixed perception. As such, while these methods manifestly provide a useful measure of the differences between the eyes, we sought to identify precisely how eye dominance alters binocular rivalry dynamics in order to give rise to the sometimes profound predominance of one or the other eye. Given the wide degree of individual variability in SED observed in previous studies (Al-Dossari, Blake, Brascamp, & Freeman, 2015; Ooi & He, 2001; Xu et al., 2011; Yang, Blake, & McDonald, 2010), we tested a large sample of observers and utilized an individual differences approach to identify signatures of SED in the dynamics of ongoing binocular rivalry. This allowed us to investigate both the variability in SED extent across our sample, as well as to relate each observer’s SED to particular characteristics of binocular rivalry dynamics. By focusing on signature changes in rivalry dynamics that are associated with SED, we hoped to gain insights into the nature of eye dominance’s impact on visual perception. For example, Levelt’s early work carefully described expected changes in dynamics when one of the rival images is raised or lowered in contrast (Levelt, 1968), and our

study offers the possibility of revealing important correlates of these patterns (also see Brascamp et al., 2006). In addition, quantifying the prevalence and extent of imbalances between the eyes during binocular rivalry across our sample of observers will be an important finding for those using binocular rivalry (and related methods involving interocular suppression, such as continuous flash suppression, Tsuchiya & Koch, 2005) to compare the visual properties of different classes of images. Researchers seek to attribute measured differences to the images themselves rather than to intrinsic differences between the eyes; yet, it is possible that SED could impact these results. 2. Method 2.1. Observers 89 observers (56 females, 33 males; median age 24 ± 11.4 yrs; range 18–68 yrs) participated in the study. These volunteers were recruited through advertisements posted in the Vanderbilt University Psychology Sign-Up System and the Vanderbilt Kennedy Center ‘‘Study Finder.” Two observers (KD & JS) are also authors. Based on prescreening conversations, all other observers had little or no prior experience viewing binocular rivalry, and they remained naïve as to the purpose of this study until after they completed it. Each observer reported normal or corrected to normal vision; however, three observers were excluded from analysis because laboratory measurements revealed monocular acuity in one eye that was worse than 40/20. One additional observer was excluded because he revealed after testing that he had monovision correction. All procedures were approved by the Vanderbilt University Human Research Protection Program, and each observer provided written informed consent prior to participation. This work was carried out in accordance with the Code of Ethics of the World Medical Association (Declaration of Helsinki). 2.2. Apparatus Stimuli were generated using the MATLAB Psychophysics Toolbox (Brainard, 1997). They were presented on a linearized Sony CPD-E540 monitor (1024  768 resolution) running at 100 Hz. Observers viewed stimuli through a mirror stereoscope that was mounted on a chin rest, fixed 80.5 cm from the display (viewing distance through the mirrors). 2.3. Stimuli To induce binocular rivalry, observers foveally viewed orthogonal sine wave gratings (± 45 degrees; 1.5 dva circular diameter; 30% contrast; 3 cyc/deg). The eye-to-stimulus pairing was alternated across ten 60-s tracking blocks. Rival targets were surrounded by identical fusion stimuli consisting of a clutter of overlapping circles presented in the background viewed by each eye, the presence of which promoted stable binocular eye alignment throughout the experiment. 2.4. Procedure Before beginning the binocular rivalry tracking experiment, observers completed a custom alignment procedure in which they used key presses to move LE and RE fusion stimuli on the screen (identical to those used in the tracking experiment, except without rival gratings) to positions where they remained aligned when using a version of the cover/uncover test. This process was completed three times, with the average LE and RE image coordinates then used to position the rival images on the screen during the

Please cite this article in press as: Dieter, K. C., et al. Individual differences in sensory eye dominance reflected in the dynamics of binocular rivalry. Vision Research (2016), http://dx.doi.org/10.1016/j.visres.2016.09.014

3

K.C. Dieter et al. / Vision Research xxx (2016) xxx–xxx

2.5. Data analysis During the tracking experiment, observers pressed (and held down) the left or right arrow key on the keyboard to indicate exclusive perceptual dominance of the left or right (respectively) tilted grating. When the observer experienced a mixed percept, he or she released both keys. For all analyses, only percepts (exclusive percepts and mixtures) longer than 300 ms were included, as shorter percept durations more likely reflect errant key presses than true visual percepts. To compute each observer’s average percept duration (Figs. 1 and 4), we took the median duration across all instances of perceptual dominance within each tracking block (separately for the LE and RE), and then calculated the mean value across the ten tracking blocks for each observer. To compute the normalized difference in number of percepts for each observer,

Proportion of observers

a 0.1

0 -5

-2.5

0

2.5

5

Difference between average RE and LE dominance durations (sec)

Proportion of observers

b 0.1

0 -50

-25

0

25

50

Difference in number of RE and LE dominance states (% relative to total)

c Difference in number of RE and LE dominance states (% relative to total)

tracking experiment. This procedure could be repeated at any time if an observer indicated that the stimuli appeared to diverge. Before beginning the main experiment, observers first tracked rivalry during one 60-s practice block. This was intended to familiarize each participant with the experience of binocular rivalry, and to give some practice using the response keys (observers had the option of repeating the practice if they still felt unfamiliar with the task). Observers were specifically instructed to press and hold down the left or right arrow key whenever and for as long as the left or right tilted (respectively) grating was perceived as exclusively dominant. Observers were also given instructions about mixed percepts – here, the experimenter described that the observer may at times perceive either a superimposed ‘‘mesh” of both images or simultaneously visible pieces of both gratings occupying different areas within the circle, forming a dynamic patchwork. Observers were instructed to press neither key when experiencing these kinds of mixtures. After 60 s of tracking elapsed, recording of key presses (and presentation of stimuli) continued until the observer’s next change in key press. This ensured that the duration of the final percept in each block was not truncated. Observers were instructed to maintain fixation on a small white dot at the center of the image (0.11 dva diameter) throughout each entire block. Prior to the rivalry experiment, each observer completed a brief fixation experiment consisting of two 60-s blocks. In these blocks, observers viewed stimuli binocularly without a stereoscope (a chin rest was used to maintain head position). During the first block, observers viewed a screen consisting simply of the fusion stimulus and fixation point. In the second block, a monocular rivalry stimulus was also presented (superimposed orthogonal gratings of different colors). In both blocks, observers were instructed to maintain fixation on the central fixation point and to move their eyes as little as possible. These blocks served as fixation practice, and eye-tracking results (recorded using an EyeLink 1000 desktop mount, recording at 1000 Hz) indicated good fixation (average standard deviation of eye position was 0.142 & 0.144 dva for block one and two, respectively). Most observers (70/85) also completed a test of sighting eye dominance. In this test, observers were seated approximately 180 cm directly opposite a white board that had a large dot drawn at about eye level. Observers were shown how to hold both arms out such that their hands were at eye level. Then, with both eyes open, they used their hands to form a small frame so that only the dot was visible through the opening. While alternately closing the LE and RE, observers reported which eye was closed when (and if) the dot appeared to move out of the frame. This eye was deemed the sighting dominant eye for a given observer (see Fig. 7). A few observers indicated that the dot remained within the frame regardless which eye was closed. These individuals (N = 4) likely did not perform the task correctly (e.g. made the frame too big), and thus were excluded from Fig. 7.

50

r = 0.53

25

0

-25

-50 -5

-2.5

0

2.5

5

Difference between average RE and LE dominance durations (sec) Fig. 1. Differences between the eyes in duration and frequency of perceptual dominance. We used kernel density estimation to examine the distribution of individual differences in LE and RE dominance duration (a) and in the relative frequency of instances of LE and RE exclusive dominance (b) across our sample of observers (N = 85). Results revealed that the median observer had roughly equal balance between the eyes, but that there was a wide range of individual variability. (c) Critically, we found that these two metrics were related (r = 0.53 ± 0.09, p < 106), suggesting that this individual variability did not arise by chance and instead providing support for true differences across observers in SED. N = 85; black line in (c) is the best-fit line to the data. Negative values indicate LE dominance; positive values indicate RE dominance.

we first separately totaled the unique instances of LE and RE dominance across all blocks for each observer. We then divided the difference between these values (RE  LE) by their sum (RE + LE), and multiplied by 100 to yield a percentage (Fig. 1b,c). A similar approach was used to calculate SED, our metric of eye dominance

Please cite this article in press as: Dieter, K. C., et al. Individual differences in sensory eye dominance reflected in the dynamics of binocular rivalry. Vision Research (2016), http://dx.doi.org/10.1016/j.visres.2016.09.014

4

K.C. Dieter et al. / Vision Research xxx (2016) xxx–xxx

b

0.1

Proportion of observers

a

100

SED (%)

50

0

-50

0 -100

-100 -50

0

50

100

15

30

45

60

Time (s)

c

50

Magnitude of SED (%)

SED (%)

25

0% -10 76

51

-75

%

% -50 26

1-2

5%

0

Quartile of SED Magnitude Fig. 2. Distribution of SED in a large sample of observers. We computed a metric of SED that indicates the relative proportion of exclusive percepts for an observer’s LE and RE. This metric ranges from 100% (complete LE dominance) to +100% (complete RE dominance). (a) The median observer had roughly equal balance between the two eyes (+2.5% SED), but there was a large degree of variability in this measure across observers (SD = 16.9%). (b) For each observer (individual line in plot), we estimated the stability of this metric by iteratively calculating SED in successive 15-s segments within each 60-s block. Results revealed that the mean slope was close to zero, and that there was no relationship between overall SED and SED drift (i.e. the slope of an individual’s function). The bolded line represents sample data for one observer with large, stable SED (+68.5%). In addition, the observer who began with almost complete RE dominance through the first 15-s was strongly RE dominant overall (+55.3% SED). (c) Dividing observers into quartiles based on their magnitude of SED revealed that a majority (i.e. the first three quartiles) of observers experienced relatively equal balance between the eyes, with a significant minority (i.e. the last quartile) experiencing a marked imbalance in the proportion of time that the images in their two eyes were perceived as perceptually dominant. N = 85; error bars in (c) represent SEM.

based on the proportion of total time in each block that the RE and LE stimuli were declared dominant exclusively (Figs. 2–7):

SED ¼ ðREprop  LEprop Þ=ðREprop þ LEprop Þ  100

ð1Þ

Note that the order of the difference (RE  LE) is arbitrary, and was chosen so that positive numbers (i.e. the right side of each plot) would reflect right eye dominance (for ease of interpretation). We generated kernel density estimates to examine the distribution of differences in duration and frequency of LE and RE percepts across observers (Figs. 1 and 5), as well as for the distribution of SED based on proportion data (Fig. 2). To form these distributions for Figs. 1 and 2, we evaluated a normal kernel function at 100 equally spaced points that spanned the range of data for each metric. This was implemented using the standard existing function ‘‘ksdensity” in the MATLAB statistics toolbox. Next, we investigated whether this estimate of an observer’s SED drifted within each tracking block – that is, whether SED was reliably stronger or weaker at particular times within a trial (see van Ee, 2005, for a similar analysis of alternation rate). To conduct this analysis, we iteratively calculated SED in 15-s segments (beginning 15 s into each tracking block), re-evaluating SED at 1 Hz (Fig. 2b). For each observer, we then averaged across the 10

tracking blocks and fit a line to the resulting drift function. The slope of this best fit line was taken as a measure of each observer’s SED drift. To estimate each observer’s alternation rate and proportion of mixed percepts (Fig. 3), we first calculated each measure separately within each tracking block, and then averaged across the ten blocks to get the final value. To compute alternation rate, we divided the number of alternations (i.e. the number of exclusive percepts minus 1) in a block by the total duration of that block (60 s plus any time added if the final percept extended beyond 60 s). The proportion of mixtures was simply the proportion of time in a given block in which an observer did not have an arrow key pressed down. To compute the fraction of return transitions (FRT), we examined the sequence of key presses made by each observer. A transition was defined as any instance where the observer’s exclusive percept changed from one to the other image (with or without an intervening mixed percept), or when perception changed from one image to a mixture, and then returned to that same image (note that these ‘‘return transitions” can only be identified in our data if an observer reports an intervening mixed percept). To calculate FRT, we divided the number of return transitions by the total

Please cite this article in press as: Dieter, K. C., et al. Individual differences in sensory eye dominance reflected in the dynamics of binocular rivalry. Vision Research (2016), http://dx.doi.org/10.1016/j.visres.2016.09.014

5

1

r = -0.35

0.5

b

0.5

0.25

d

1

r = 0.04 Proportion mixed percept

0.5

76

-10

0%

% -75 51

Quartile of SED Magnitude 0.5

0.25

76

-10

0%

% -75

Magnitude SED (%)

51

100

-50

50

5%

0

%

0

0

1-2

Proportion mixed percept

c

-50

Magnitude SED (%)

26

100

5%

50

1-2

0

%

0

0

26

Alternation rate (alt/sec)

a

Alternation rate (alt/sec)

K.C. Dieter et al. / Vision Research xxx (2016) xxx–xxx

Quartile of SED Magnitude Fig. 3. Relationship of SED to alternation rate and proportion of mixed percepts. (a) Alternation rate is negatively correlated with the magnitude of SED (r = 0.35 ± 0.09, p = 0.0009), a result indicating slower alternations between exclusive percepts for individuals with stronger imbalances between the two eyes. (b) However, a related analysis conducted after dividing individuals into quartiles based on their magnitude of SED revealed no significant relationship between these variables (F3,80 = 2.6, p = 0.059, n.s.). (c) No relationship was observed between the magnitude of SED and the proportion of time that an observer reported perceiving a mixture of the rival images (r = 0.04 ± 0.09, p = 0.74, n.s.). (d) Likewise, a quartile analysis revealed no relationship between these variables (F3,80 = 1.2, p = 0.32). N = 85; black lines in (a) & (c) are best-fit lines to data; error bars in (b) & (d) represent SEM.

number of transitions reported by each observer. Again, this was computed separately for each block, with these ten block measurements then averaged. We calculated FRT across all transitions, as well as separately for transitions to the LE and RE. We also established a combined metric in order to study the relative proportion of each observer’s total transitions that were made to his/her LE and RE. This metric was the difference in each eye’s FRT (RE  LE) divided by the total number of return transitions for that observer (Fig. 6). To perform the quartile analyses, we divided the 85 observers into 4 groups based on SED. Due to the odd number of observers, the median observer was excluded from the quartile analyses. Note that for some analyses this was based on the magnitude (absolute value) of SED (Figs. 2 and 3), while for others it was based on signed SED (Figs. 4–6). The relevant plots represent the group mean and standard error for each quartile of observers. To examine how signed SED quartile impacted the distributions of LE and RE percept durations (Fig. 5), we pooled percept data across all 21 observers in each quartile (separately for each eye). We then computed kernel density estimates for this pooled data as described above, here using 101 equally spaced durations between 0 and 10 s (inclusive). The plots in Fig. 5 represent the cumulative density functions (closely related to survival probability, which has been used as a metric for assessing rivalry dominance, Mamassian & Goutcher, 2005).

Statistical tests were performed using MATLAB and/or the VasserStats interface (vasserstats.net). To estimate the variability of the observed correlation coefficients and to ensure that results were not disproportionately influenced by outliers, we employed a resampling procedure. For each reported correlation, we drew 10,000 data samples by choosing 85 observers from our data set at random with replacement on each iteration. We then calculated the standard deviation of the 10,000 correlation coefficients (this value is reported in our results section for each correlation).

3. Results 3.1. Does individual variability arise from true differences between the eyes? We examined data from 85 observers who participated in a binocular rivalry tracking experiment. For all statistical analyses we report uncorrected p-values. To test for statistical significance, we compared these p-values to a Bonferoni corrected alpha value (for reported correlations, a = 0.0042; for reported ANOVAs, a = 0.01; p-values above these thresholds are indicated as n.s. throughout the results). We first sought to establish 1) whether observers in our sample indeed demonstrated differences between

Please cite this article in press as: Dieter, K. C., et al. Individual differences in sensory eye dominance reflected in the dynamics of binocular rivalry. Vision Research (2016), http://dx.doi.org/10.1016/j.visres.2016.09.014

6

K.C. Dieter et al. / Vision Research xxx (2016) xxx–xxx

8

b

r = -0.23

6

4

2

0 -100

8

RE percept duration (s)

LE percept duration (s)

a

-50

0

50

r = 0.43

6

4

2

0 -100

100

Avg. percept duration (s)

0

50

100

SED (%)

SED (%)

c

-50

LE percept duration

4

RE percept duration

2

-10

RE Dominance

76

51

-75

0%

%

% -50 26

LE Dominance

1-2

5%

0

Quartile of SED Fig. 4. Relationship of SED to LE & RE percept durations. SED is not correlated (r = 0.23 ± 0.11, p = 0.037, n.s.) with the average duration of an observer’s LE percepts (a), but is positively correlated (r = 0.43 ± 0.12, p < 104) with the average duration of an observer’s RE percepts (b). A quartile analysis revealed a significant interaction (F3,80 = 20.9, p < 104) between SED quartile and LE/RE percept durations. These results are consistent with longer percept durations of the dominant eye during binocular rivalry. Negative values of SED indicate LE dominance; positive values of SED indicate RE dominance. N = 85; black lines in (a) & (b) are best-fit lines to data; error bars in (c) are SEM.

their two eyes, and 2) whether those differences (if present) reflected reliable, genuine sensory differences rather than measurement variability. To make this determination, we first considered two ways in which sensory eye dominance might be expected to influence binocular rivalry dynamics – by altering the relative duration for which the LE and RE remained dominant on average (Fig. 1a), or by altering the relative frequency with which the LE and RE were perceived as perceptually dominant (Fig. 1b). The median observer on each metric demonstrated essentially no difference between the eyes (+0.06 s difference in average percept duration; +1% difference in relative percept frequency). However, there was a large degree of individual variability (SDs ±0.8 s and ±6.7%, respectively; see Fig. 1a,b), perhaps consistent with a subset of observers experiencing large differences between their two eyes. Indeed, 14 observers (16.4%) had a difference of more than 1 s between LE and RE average dominance durations, and 7 observers (8.2%) experienced a difference in percept frequency of greater than 10% (also see Fig. 2). While this range of individual variability is notable, it is critical to establish whether these patterns reflect a true difference between the eyes. To make this determination, we investigated whether the difference in LE and RE percept durations was related to their relative frequency. Specifically, if these two patterns (Fig. 1a,b) arise simply due to chance, no relationship between the measures should be expected – there is no a priori reason why seeing a given stimulus more often should inherently predict longer median dominance durations for that stimulus. In more

concrete terms, if the dominance duration distributions for the two eyes were in fact equivalent, we may observe (by chance) more right eye than left eye percepts (for example) without observing any difference in the median duration of those percepts. However, these two metrics were related in our data set (r = 0.53 ± 0.09, p < 106; Fig. 1c). In other words, the eye that was reported dominant more often was also reported to maintain its dominance for longer durations, on average. This finding is generally consistent with the concept of a dominant eye as the one associated with stronger sensory processing, and matches well with results demonstrating similar changes in rivalry dynamics when, for example, the contrast of one eye’s rival stimulus is higher than the contrast of the other eye’s stimulus (Brascamp et al., 2006; Levelt, 1968). We next explored the possibility that these results are driven largely by the first percept in each tracking block. Biases in the frequency of one eye’s dominance during onset rivalry have previously been used as a measure of sensory eye dominance (Ooi & He, 2001; Xu et al., 2011), and a tendency for an observer to report his/her LE or RE as dominant first in each tracking block could potentially account for the overall difference in percept frequency (Fig. 1b). Indeed, Xu et al. (2011) found that SED as measured using brief presentations of rivalry was correlated with predominance across 30-s of rivalry tracking. In our data set, the difference in average percept duration across eyes (Fig. 1a) was related to the proportion of blocks in which an observer’s first exclusive percept was his/her RE (r = 0.34 ± 0.1, p = 0.0014). However, even after

Please cite this article in press as: Dieter, K. C., et al. Individual differences in sensory eye dominance reflected in the dynamics of binocular rivalry. Vision Research (2016), http://dx.doi.org/10.1016/j.visres.2016.09.014

7

K.C. Dieter et al. / Vision Research xxx (2016) xxx–xxx

LE Percepts

RE Percepts

Cumulative probability

1

0.75

0.5

0.25 quartile 1

quartile 2

0 0

2

4

6

8

10

0

2

Percept duration (s)

4

6

8

10

Percept duration (s)

Cumulative probability

1

0.75

0.5

0.25 quartile 3

quartile 4

0 0

2

4

6

8

10

0

2

Percept duration (s)

4

6

8

10

Percept duration (s)

Fig. 5. LE and RE dominance duration distributions. SED impacted the relative duration of LE and RE percepts. For observers with the strongest LE dominance (quartile 1), there tended to be more long LE percepts (solid line) than RE percepts (dashed line; note the rightward shift of the LE function). When considering observers with the strongest RE dominance (quartile 4), this pattern was reversed. Notably, for observers with relatively equal balance between the eyes (quartiles 2 & 3), the functions are markedly similar for the two eyes.

b r = 0.53

50

0

-50

Return transitions to RE

0.1

76

-10

0%

% 51 -75

100

-50

SED (%)

50

26

0

%

0 -50

5%

-100 -100

Return transitions to LE

0.2

1-2

100

Fraction of return transitions

% diff. in FRT, RE-LE

a

Quartile of SED Fig. 6. Impact of SED on fraction of return transitions (FRT). We calculated the proportion of an observer’s total percept transitions that were returns to the same eye, separately for returns to the left and right eye (FRT). (a) A combined metric that indicates the difference in each observer’s frequency of returning to his/her right minus left eye, normalized to each observer’s total frequency of return transitions, was strongly related to SED (r = 0.53 ± 0.1, p < 106). (b) Dividing observers into quartiles based on signed SED likewise demonstrated that the frequency of an observer experiencing a return transition to his/her LE or RE was related to the strength and sign of SED. N = 85; black line in (a) represents the best-fit line to the data; error bars in (b) are SEM.

Please cite this article in press as: Dieter, K. C., et al. Individual differences in sensory eye dominance reflected in the dynamics of binocular rivalry. Vision Research (2016), http://dx.doi.org/10.1016/j.visres.2016.09.014

8

K.C. Dieter et al. / Vision Research xxx (2016) xxx–xxx

Sensory eye dominance (%)

100

50

0

-50

-100

LE sighting dominance RE sighting dominance

Fig. 7. Relationship between sensory and sighting eye dominance. In addition to measuring SED using binocular rivalry, observers also completed a test of sighting eye dominance. We found no relationship between these measures, with RE sighting dominant observers (for example) equally likely to demonstrate a wide range of sensory eye dominance values. Negative values of SED indicate LE dominance; positive values indicate RE dominance. Data are sorted by sighting, and then sensory, eye dominance to best illustrate the main result. Each bar represents data from one observer (N = 66).

accounting for the proportion of blocks in which the RE was the first exclusive percept, the relationship in Fig. 1c between dominance duration and incidence remains significant (partial r = 0.43 ± 0.14, p < 104).

3.2. How does sensory eye dominance impact the dynamics of binocular rivalry? Though differences between the eyes are frequently noted during binocular rivalry experiments, our quantitative analysis above firmly establishes that these reflect true differences and are not driven by measurement variability. Having now established real individual variability in LE and RE dynamics during binocular rivalry, we next sought to map out the precise manifestations of SED. To account for all possible means through which a dominant eye might gain disproportionate predominance (see Fig. 1), and for the sake of consistency with previous studies that measured SED in prolonged binocular rivalry tracking blocks (Handa et al., 2004; Porac & Coren, 1978; Zheleznyak et al., 2015), we compared various aspects of rivalry dynamics to a metric of SED based on overall differences in predominance between the LE and RE (see Section 2.5). This metric (which we call SED) ranges from 100% (complete LE predominance) to +100% (complete RE predominance), and reflects the difference in the proportion of exclusive visibility of the RE and LE. The SED metric provides a quantitative index for sorting individuals by the extent of their eye dominance, in turn making it possible to relate the overall strength and sign of eye dominance to precise signatures in rivalry dynamics. Of course, the SED index used here is just one of multiple ways to derive estimates of eye dominance using binocular rivalry (see, for example, the discrete sampling technique used to derive survival probability functions described by Mamassian & Goutcher, 2005). Because these various techniques all entail summary statements of rivalry temporal dynamics, there is no reason to expect that results derived from one method would not generalize to other approaches (also see Fig. 5). Similar to the results described earlier, the SED metric revealed a median observer with roughly equal balance between the eyes (+2.5% SED), but large variability across the sample (SD ± 16.9%; Fig. 2a,b). A quartile analysis based on the absolute value of SED (‘‘SED magnitude”) supports this result – a majority of observers have a relatively small difference between the two eyes (Q1-Q3), but a substantial minority (Q4) demonstrate a marked imbalance (Fig. 2c; One-Way ANOVA: F3,80 = 53.9,

p < 104; Tukey HSD tests for Q1 vs. Q4, Q2 vs. Q4, & Q3 vs. Q4, all p < 0.01). To determine the stability of SED over time, we calculated SED using sliding 15-s windows within each tracking block (see van Ee, 2005). We found that the average SED drift (i.e. slope of the best fit line) across observers did not differ from 0 (Fig. 2b; mean 0.06 ± 0.4 percent/s; single sample t84 = -0.5, p = 0.62, n.s.), and that there was no correlation between SED and slope (r = 0.16 ± 0.11, p = 0.13, n.s.). This result indicates that an individual’s SED was consistent during the entirety of each 60-s tracking block (see bolded line in Fig. 2b for data from one sample observer). We also investigated whether SED was impacted by practice effects across tracking blocks. Here, we compared SED in the first 5 to last 5 tracking blocks, and found no significant difference (t84 = 0.4, p = 0.7, n.s.). We next turned to analyses of how individual variability in SED (as quantified using the metric above) manifested in various quantitative metrics of binocular rivalry dynamics. First we examined alternation rate, which is a measure that reflects how quickly an observer switches between the competing exclusive percepts while viewing binocular rivalry. It is also often used as a proxy for measuring the strength of interocular suppression, with stronger suppression thought to be associated with slower alternations (e.g. van Loon et al., 2013; Wilson, 2007). We found that alternation rate was negatively correlated with the magnitude of SED (r = 0.35 ± 0.09, p = 0.0009) – that is, individuals with a stronger imbalance between the eyes tended to experience switches in perceptual dominance at a slower rate (Fig. 3a). Dividing observers into quartiles based on their magnitude of SED did not yield a significant relationship between quartile and alternation rate (OneWay ANOVA: F3,80 = 2.6, p = 0.059, n.s.; Fig. 3b). However, because the vast majority of the variability in SED magnitude seems to be carried in only the largest quartile (Fig. 2c), this analysis may underestimate the relationship between these variables. Next we looked at the relationship between the magnitude of SED and the proportion of time that an observer reported experiencing mixed perception. Observers were specifically instructed prior to the experiment that they might experience periods when they saw bits and pieces of both rival stimuli or periods when they saw both stimuli superimposed to form a grid-like percept. A decrease in mixed percepts is thought to be associated with stronger suppression (Klink, Brascamp, Blake, & van Wezel, 2010; van Loon et al., 2013), resulting in more frequent, clearly delineated periods of exclusive dominance. However, our data revealed no relationship between the magnitude of SED and the proportion of time that mixtures were perceived (r = 0.04 ± 0.09, p = 0.74, n.s.; Fig. 3c). Similar results are obtained from a quartile analysis (One-Way ANOVA: F3,80 = 1.2, p = 0.32, n.s.; Fig. 3d). This finding also provides some evidence that an increase in reports of one eye’s dominance was not due to bias. Specifically, one possibility is that observers with strong SED imbalance achieved this by using a more liberal reporting criterion for declaring exclusive dominance of that eye’s percepts. Such a result, however, would predict a negative correlation between the magnitude of SED and the proportion of mixed percepts. We next examined how LE and RE percept durations were related to SED. As previously described, differences in percept durations across the eyes were related to the difference in the frequency of each eye’s perceptual dominance (Fig. 1). However, understanding how the relationship between LE and RE percept durations changes with SED could help illuminate the mechanism by which SED impacts one’s perceptual experience during binocular rivalry. We found no significant relationship between LE percept durations and SED (r = -0.23 ± 0.11, p = 0.037, n.s.; Fig. 4a), but that RE percept durations were positively correlated with SED (r = 0.43 ± 0.12, p < 104; Fig. 4b). The stronger relationship

Please cite this article in press as: Dieter, K. C., et al. Individual differences in sensory eye dominance reflected in the dynamics of binocular rivalry. Vision Research (2016), http://dx.doi.org/10.1016/j.visres.2016.09.014

K.C. Dieter et al. / Vision Research xxx (2016) xxx–xxx

for RE durations is consistent with an overall prevalence of RE dominance in our dataset (see Section 3.4.). A quartile analysis (Fig. 4c) revealed an interaction between SED quartile and LE/RE percept duration (Two-Factor ANOVA, interaction: F3,80 = 20.9, p < 104) with no main effect of eye (F84,1 = 1.5, p = 0.23, n.s.) or SED quartile (F3,80 = 0.8, p = 0.49, n.s.) on percept durations. A similar result is seen when examining the cumulative densities of LE and RE percept durations across SED quartiles (Fig. 5). Over successive quartiles, one can clearly see the rightward shift of the dashed line (RE percept durations) relative to the solid line (LE percept durations), indicating a progressive increase in long RE durations (relative to LE) as the predominance of the RE increases across quartiles. Note that the SED quartiles in these analyses (Figs. 4 and 5) are based on signed SED, and thus differ from the SED quartiles based on magnitude plotted in Figs. 2 and 3. These results demonstrate that an imbalance between the eyes is associated with longer average percept durations in the dominant eye. One notable feature of the results in Fig. 4c is that average percept duration in the non-dominant eye does not seem to be impacted by SED. For example, RE percept durations in the quartile with strongest LE dominance (1–25% of SED) are roughly equivalent (on average) to those in the middle two quartiles (those with the most balanced SED). This pattern (which is similar for LE data) is notable because it suggests that unbalanced SED is associated with longer durations of perceptual dominance for the stronger eye, but no change in its suppression duration. This differs from cases where the contrast of one eye’s image is increased, where it would be expected to enjoy shorter suppression durations with no change to its dominance durations (Levelt, 1968). However, as this analysis is across observers, it is difficult to know precisely how it relates to the impact of contrast manipulations within observers. Finally, we sought to understand the finding (Fig. 1b) that observers demonstrated significant variability in how often one eye’s image was perceived as perceptually dominant. Binocular rivalry is typically thought of as a process characterized by alternating periods of LE and RE dominance – that is, an instance of LE dominance is expected to always be followed by an instance of RE dominance. However, such a process should result in a roughly equal number of LE and RE percepts, unless, perhaps, one is reliably biased toward seeing one eye as the first percept in a tracking block. We have already noted that while the strength of differences between the eyes does predict a tendency to see one’s dominant eye first, first percepts alone cannot fully explain SED. As a result, we considered the possibility that observers (particularly those with a large imbalance between their eyes) might experience consecutive percepts of the same eye’s image (i.e., return transitions). Thus, for example, the stimulus being viewed by the LE may be exclusively dominant for a few seconds, followed by a period of mixed perception, followed by a return to exclusive dominance of the LE stimulus. We computed the Fraction of Return Transitions (FRT, Brascamp et al., 2006), and related this to each observer’s measured SED. There was no relationship between the magnitude of an observer’s SED and his/her overall tendency to report return transitions (r = 0.24 ± 0.14, p = 0.0245, n.s.). Breaking this down, however, revealed that the pattern depended on eye – there was no significant relationship between signed SED and FRT to the LE (r = -0.16 ± 0.1, p = 0.14, n.s.), but there was a significant relationship between signed SED and FRT to the RE (r = 0.43 ± 0.13, p < 104). Given the increased prevalence of RE dominance in our data (see Section 3.4.), we also developed a combined metric that indicates the percentage difference in each observer’s total return transitions to the RE versus LE. This measure is strongly related to SED (r = 0.53 ± 0.1, p < 106, Fig. 6a). A quartile analysis (Fig. 6b) likewise revealed an interaction between SED quartile and LE/RE FRT (F3,80 = 11.6, p < 104). There was no main effect of quartile on FRT (F3,80 = 1.7, p = 0.18, n.s), but there

9

was a marginally significant effect of eye on FRT (F84,1 = 5.0, p = 0.028), likely due to the stronger relationship for the RE. Taken together, these results demonstrate that observers are more likely to experience return transitions for a stimulus imaged in their dominant eye compared to their non-dominant eye. 3.3. Is SED related to sighting eye dominance? Most observers in our study also completed a test of sighting eye dominance. Tests of sighting eye dominance are designed to establish which eye an observer prefers to use when directing their line of sight at a visual target (see Fink, 1938). Such tests are tapping into sensory-motor aspects of vision; they are not purely sensory in nature. Still, some previous studies using extended binocular rivalry tracking have found a link between sighting eye dominance and SED (Bosten et al., 2015; Handa et al., 2004; Porac & Coren, 1978; Zheleznyak et al., 2015). However, we found no such relationship between measures in our data (Fig. 7; also see Ooi & He, 2001; Xu et al., 2011). The average observer with LE sighting dominance had an SED of +5.4 ± 17.4%, while the average RE sighting dominant observer had an SED of +1.9 ± 14.7% (with no difference between the two, t63 = 0.9, p = 0.4, n.s.). 3.4. Is LE or RE dominance more common? Based on our test of sighting eye dominance, almost three out of every four of the participants in our large sample showed RE dominance (72.7% vs 27.3% – see Fig. 7). This is consistent with a body of previous findings indicating the prevalence of RE sighting dominance (see summary table in Lopes-Ferreira et al., 2013). We also found that RE SED (61.2% of observers) was more prevalent than LE SED (38.8% of observers; see Figs. 1 and 2), a pattern of individual differences noted in other studies of binocular rivalry (Al-Dossari et al., 2015; Zheleznyak et al., 2015). 4. Discussion Results from our large sample of observers confirm what others have noted – there exists considerable individual variability in the relative predominance of the two eyes during binocular rivalry (Figs. 1 and 2). While many observers witnessed the stereotypical back-and-forth of binocular rivalry alternations between the eyes, for a substantial number of observers one eye disproportionately dominated visibility. For these individuals, the stronger eye tended to experience both more frequent instances of perceptual dominance, as well as longer instances of perceptual dominance when visible. This difference between the eyes was unrelated to preferences in sighting eye dominance (Fig. 7), suggesting an independence of SED from its motor counterpart (Ooi & He, 2001). In our large sample of observers, all exhibited normal or corrected-to-normal vision, including good monocular and binocular visual acuity. This, coupled with the evidence described above linking SED to binocular visual function, would seem to suggest that monocular factors are not the sole source of SED. It may be noteworthy, however, that monocular acuity tends to develop with a unique trajectory in each eye, with stereopsis failing to emerge in infancy until the two eyes have reached relatively equal acuity (Birch, 1985). In cases of severe imbalance in monocular acuity or contrast sensitivity, then, differences in binocular rivalry dynamics could be driven by those facts. Indeed, for the three observers excluded from our data analysis because of large, uncorrectable acuity differences between the eyes, two had a strong imbalance in rivalry dynamics in favor of the eye with better acuity (also see Bosten et al., 2015). But, as pointed out above, the vast majority of our participants had uniformly good acuity. So what

Please cite this article in press as: Dieter, K. C., et al. Individual differences in sensory eye dominance reflected in the dynamics of binocular rivalry. Vision Research (2016), http://dx.doi.org/10.1016/j.visres.2016.09.014

10

K.C. Dieter et al. / Vision Research xxx (2016) xxx–xxx

can we surmise about why SED varies among individuals in our sample? Perhaps one underlying cause of SED is a chronic imbalance in the strength of interocular inhibition. According to one popular theory of binocular rivalry, the dominant eye drives suppressive forces onto mechanisms responding to the other eye (Blake, 1989). Following adaptation of this inhibitory drive (which gradually weakens suppression), an alternation will typically occur – however, in the case of a return transition, the suppression strength arising from responses to the dominant eye is stronger than that relating to the non-dominant eye, even after adaptation. Such a pattern could occur if the suppressive responses related to non-dominant eye stimulation started off weaker than those arising from dominant eye stimulation. For example, studies in which a stimulus in one eye is effectively erased by the onset of a stimulus in the other eye (termed ‘‘dichoptic masking;” Legge, 1979) have typically found that stronger magnitudes of dichoptic masking are associated with slower alternation rates (Baker & Graf, 2009; Bosten et al., 2015). If interocular inhibition were unbalanced, this could account, in part, for the negative correlation we found between alternation rates and SED (Fig. 3a). While weakened suppression is typically associated with an increase in the prevalence of mixed perception (van Loon et al., 2013), we found no relationship between the magnitude of SED and the proportion of time that mixed percepts were reported (Fig. 3c,d). However, an imbalance in interocular inhibition strength could arise solely from a decrease in one eye’s strength, or (in part or whole) from an increase in one eye’s strength, making the impact of SED on mixed perception unclear. This account postulating that SED arises from unbalanced interocular inhibition could also potentially explain why SED indexed using binocular rivalry is related to other aspects of binocular visual function. Divisive normalization has been proposed to play a critical role in interocular suppression (Li, Carrasco, & Heeger, 2016; Ling & Blake, 2012), and may also underlie binocular contrast normalization (Ding & Sperling, 2006; Moradi & Heeger, 2009). For example, imbalances between the two eyes as measured using binocular rivalry (SED) have been linked to impairments in stereoacuity (Xu et al., 2011), similar to impacts observed when the inputs to the two eyes differ in physical contrast (Halpern & Blake, 1988). Reducing this imbalance between the eyes improves stereoacuity (Xu, He, & Ooi, 2010; Xu et al., 2012). These results imply that binocular rivalry may rely on mechanisms also critical in mediating a wide range of binocular visual functions. As pointed out above, a substantial number of observers in our study (Fig. 2) exhibited a breakdown in the canonical back-andforth nature of binocular rivalry. Instead, they experienced successive returns of dominance to the same eye. How might this relate to models of rivalry in general and to SED in particular? To reiterate, perceptual dominance during binocular rivalry arises due to the currently dominant eye imposing inhibition onto the currently suppressed eye, with the strength of that inhibition waning over time due to adaptation. Eventually, this results in a perceptual switch to the other eye. Wilson (2007) has shown that this minimal model can adequately account for dominance durations in the absence of noise. Such an explanation, however, seemingly requires that perceptual dominance alternates between the two eyes, and thus, the presence of return transitions has been interpreted as evidence that additional components (e.g. noise) must be involved in rivalry transitions (Brascamp et al., 2006). In their study, Brascamp and colleagues found that return transitions occurred with substantial regularity under conditions where the contrast of one eye’s image was higher than that of the other’s image. Here, we demonstrate that return transitions can also arise fairly frequently when the images being viewed by the two eyes are equivalent in physical strength – this tendency is seen, however, primarily in people with relatively

pronounced SED (Fig. 6). The dynamical models proposed by Wilson (2007) and by Brascamp et al. (2006) both could easily account for our observed SED signatures by positing differential effective contrast of left- and right-eye inputs to an extent that varies across observers, or by altering the relative strength of reciprocal inhibition between the eyes. Moreover, the addition of noise within the rivalry process could contribute to eye dominance, too, depending on the relative strength and the extent of correlation of that noise within the two monocular channels. In terms of the potential contribution of noise, it is noteworthy that return transitions were common when a dominance duration experienced by the SED eye was fairly short relative to the observer’s median duration for that eye (e.g. short LE percept to mixture back to LE). After that short LE percept the LE representation might still remain relatively unadapted, which could lead to a switch attributable to a relatively strong noise excursion while the LE stimulus still retains a relatively strong neural representation capable of re-achieving competitive dominance. This line of reasoning has been applied by Pastukhov and Braun (2011) to account for cumulative history effects in rivalry dynamics. Whatever the bases for SED, our results have important implications for those using binocular rivalry as a tool to for investigating selective processing of different categories of visual stimuli. Binocular rivalry and its close cousin continuous flash suppression (Tsuchiya & Koch, 2005), have been widely employed to study the relation of visual awareness on aspects of neuro-cognitive processing associated with visual attention (see Dieter & Tadin, 2011; Paffen & Alais, 2011), emotion (Alpers & Pauli, 2006; Yang, Zald, & Blake, 2007), intuition (Lufityanto, Donkin, & Pearson, 2016), threat (Gayet, Paffen, Belopolsky, Theeuwes, & Stigchel, 2016), and multisensory interactions (van Ee, van Boxtel, Parker, & Alais, 2009), to list just a few. In studies like these, investigators attribute differences in rivalry strength between different classes of visual stimuli (e.g., neutral versus angry faces) to neural processes differentially responsive to features defining those classes (e.g., amygdala activation). A number of those studies purposefully image the to-besuppressed stimulus in the non-dominant eye. However, the impact of this procedure on the strength of interocular suppression will likely vary widely across individuals depending on their degree of eye dominance (Figs. 1 and 2). Thus, it is entirely possible that intrinsic differences between the two eyes may also impact the results for at least some observers. At minimum then, counterbalancing across eyes is critical, and investigators may want to quantitatively measure eye dominance and treat results separately across the two ocular configurations in extreme cases.

5. Conclusion Our results indicate that individuals vary greatly in how strongly their LE and RE contribute to binocular vision as measured in a binocular rivalry task. This variability had widespread impact on the dynamics of binocular rivalry, with the dominant eye experiencing longer and more frequent instances of perceptual dominance during binocular rivalry tracking. For observers with strong SED imbalance, binocular rivalry featured a greater proportion of return transitions to their stronger eye, thus deviating from the stereotypical back-and-forth description of binocular rivalry. These changes highlight the importance in considering the factor of eye when drawing inferences about the perceptual processing of images in paradigms utilizing interocular suppression. Acknowledgments Work supported by NIH NEI awards T32-EY 007135-19, F32-EY 025514, & R21-EY 022752. The authors thank Jan Brascamp for helpful discussions on these results.

Please cite this article in press as: Dieter, K. C., et al. Individual differences in sensory eye dominance reflected in the dynamics of binocular rivalry. Vision Research (2016), http://dx.doi.org/10.1016/j.visres.2016.09.014

K.C. Dieter et al. / Vision Research xxx (2016) xxx–xxx

References Al-Dossari, M., Blake, R., Brascamp, J. W., & Freeman, A. W. (2015). Chronic and acute biases in perceptual stabilization. Journal of Vision, 15(4), 1–11. Alais, D. (2012). Binocular rivalry: Competition and inhibition in visual perception. Wiley Interdisciplinary Reviews: Cognitive Science, 3, 87–103. Alpers, G. W., & Pauli, P. (2006). Emotional pictures predominate in binocular rivalry. Cognition Emotion, 20(5), 596–607. Arnold, D. H., Grove, P. M., & Wallis, T. S. A. (2007). Staying focused: A functional account of perceptual suppression during binocular rivalry. Journal of Vision, 7 (7), 1–8. Baker, D. H., & Graf, E. W. (2009). On the relation between dichoptic masking and binocular rivalry. Vision Research, 49, 451–459. Banks, M. S., Ghose, T., & Hillis, J. M. (2004). Relative image size, not eye position, determines eye dominance switches. Vision Research, 44(3), 229–234. Birch, E. E. (1985). Infant interocular acuity differences and binocular vision. Vision Research, 25(4), 571–576. Blake, R. (1989). A neural theory of binocular rivalry. Psychological Review, 96(1), 145. Blake, R., & O’Shea, R. P. (2009). Binocular Rivalry. In L. R. Squire (Ed.), Encyclopedia of neuroscience, 2 (pp. 179–187). Oxford: Academic Press. Bosten, J. M., Goodbourn, P. T., Lawrance-Own, A. J., Bargary, G., Hogg, R. E., & Mollon, J. D. (2015). A population study of binocular function. Vision Research, 110, 34–50. Brainard, D. H. (1997). The psychophysics toolbox. Spatial Vision, 10(4), 433–436. Brascamp, J. W., van Ee, R., Noest, A. J., Jacobs, R. H. A. H., & van den Berg, A. (2006). The time course of binocular rivalry reveals a fundamental role of noise. Journal of Vision, 6(11), 1244–1256. Carter, O., & Cavanagh, P. (2007). Onset rivalry: Brief presentation isolates an early independent phase of perceptual competition. PLoS ONE, 2(4), e343. Cumming, B. G., & DeAngelis, G. C. (2001). The physiology of stereopsis. Annual Review of Neuroscience, 24, 203–238. Dieter, K. C., & Tadin, D. (2011). Understanding attentional modulation of binocular rivalry: A framework based on biased competition. Frontiers in Human Neuroscience, 5, 5. Ding, J., & Sperling, G. (2006). A gain-control theory of binocular combination. Proceedings of the National academy of Sciences of the United States of America, 103(4), 1141–1146. Ehrenstein, W. H., Arnold-Schulz-Gahmen, B. E., & Jaschinski, W. (2005). Eye preference within the context of binocular functions. Graefe’s Archive for Clinical and Experimental Ophthalmology, 243(9), 926–932. Fink, W. H. (1938). The dominant eye: Its clinical significance. Archives of Ophthalmology, 19(4), 555–582. Gayet, S., Paffen, C. L., Belopolsky, A. V., Theeuwes, J., & Stigchel, S. V. d. (2016). Visual input signaling threat gains preferential access to awareness in a continuous flash suppression paradigm. Cognition, 149, 77–83. Halpern, D. L., & Blake, R. R. (1988). How contrast affects stereoacuity. Perception, 17 (4), 483–495. Handa, T., Mukuno, K., Uozato, H., Niida, T., Shoji, N., & Shimizu, K. (2004). Effects of dominant and nondominant eyes in binocular rivalry. Optometry Vision Science, 81(5), 377–383. Khan, A. Z., & Crawford, J. D. (2001). Ocular dominance reverses as a function of horizontal gaze angle. Vision Research, 41(14), 1743–1748. Klink, P. C., Brascamp, J. W., Blake, R., & van Wezel, R. J. (2010). Experience-driven plasticity in binocular vision. Current Biology, 20(16), 1464–1469. Legge, G. E. (1979). Spatial frequency masking in human vision: Binocular interactions. Journal of the Optical Society of America, 69(6), 838–847. Levelt, W. J. M. (1968). On binocular rivalry. Paris: Mouton. Li, H.-H., Carrasco, M., & Heeger, D. J. (2016). Deconstructing interocular suppression: Attention and divisive normalization. PLoS Computational Biology, 11(10), e1004510–e1004526. Ling, S., & Blake, R. (2012). Normalization regulates competition for visual awareness. Neuron, 75, 531–540. Lopes-Ferreira, D., Neves, H., Queiros, A., Faria-Ribeiro, M., Peixoto-de-Matos, S. C., & Gonzalez-Méijome, J. M. (2013). Ocular dominance and visual function testing. BioMed Research International, 2013(2), 1–7.

11

Lufityanto, G., Donkin, C., & Pearson, J. (2016). Measuring intuition: Nonconscious emotion information boosts decision accuracy and confidence. Psychological Science, 1–13. Mamassian, P., & Goutcher, R. (2005). Temporal dynamics in bistable perception. Journal of Vision, 5, 361–375. Mapp, A. P., Ono, H., & Barbeito, R. (2003). What does the dominant eye dominate? A brief and somewhat contentious review. Perception Psychophysics, 65(2), 310–317. Mitchell, J. F., Stoner, G. R., & Reynolds, J. H. (2004). Object-based attention determines dominance in binocular rivalry. Nature, 429(6990), 410–413. Moradi, F., & Heeger, D. J. (2009). Inter-ocular contrast normalization in human visual cortex. Journal of Vision, 9(3). Ooi, T. L., & He, Z. J. (2001). Sensory eye dominance. Optometry, 72(3), 168. Paffen, C. L., & Alais, D. (2011). Attentional modulation of binocular rivalry. Frontiers in Human Neuroscience, 5. Pastukhov, A., & Braun, J. (2011). Cumulative history quantifies the role of neural adaptation in multistable perception. Journal of Vision, 11(10). 12–12 http://doi. org/10.1167.11.10.12. Pointer, J. S. (2007). The absence of lateral congruency between sighting dominance and the eye with better visual acuity. Ophthalmic and Physiological Optics, 27(1), 106–110. Porac, C., & Coren, S. (1978). Sighting dominance and binocular rivalry. Optometry Vision Science, 55(3), 208–213. Porter, J., Guirao, A., Cox, I. G., & Williams, D. R. (2001). Monochromatic aberrations of the human eye in a large population. Journal of the Optical Society of America A: Optics, Image Science, and Vision, 18(8), 1793–1803. Rice, M. L., Leske, D. A., Smestad, C. E., & Holmes, J. M. (2008). Results of ocular dominance testing depend on assessment method. Journal of AAPOS, 12(4), 365–369. Squire, L. R., Bloom, F. E., McConnell, S. K., Roberts, J. L., Spitzer, N. C., & Zigmond, M. J. (2003). Fundamental neuroscience. San Diego, CA: Academic Press. Stanley, J., Forte, J., Cavanagh, P., & Carter, O. (2011). Onset rivalry: The initial dominance phase is independent of ongoing perceptual alternations. Frontiers in Human Neuroscience, 5, 1–9. Suttle, C., Alexander, J., Liu, M., Ng, S., Poon, J., & Tran, T. (2009). Sensory ocular dominance based on resolution acuity, contrast sensitivity and alignment sensitivity. Clinical Expert Optometry, 92(1), 2–8. Tsuchiya, N., & Koch, C. (2005). Continuous flash suppression reduces negative afterimages. Nature, 8(8), 1096–1101. van Ee, R. (2005). Dynamics of perceptual bi-stability for stereoscopic slant rivalry and a comparison with grating, house-face, and Necker cube rivalry. Vision Research, 45(1), 29–40. van Ee, R., van Boxtel, J. J. A., Parker, A. L., & Alais, D. (2009). Multisensory congruency as a mechanism for attentional control over perceptual selection. Journal of Neuroscience, 29(37), 11641–11649. van Loon, A. M., Knapen, T., Scholte, H. S., St. John-Saaltink, E., Donner, T. H., & Lamme, V. A. F. (2013). GABA shapes the dynamics of bistable perception. Current Biology, 23(9), 823–827. Wilson, H. (2007). Minimal physiological conditions for binocular rivalry and rivalry memory. Vision Research, 47(21), 2741–2750. Xu, J. P., He, Z. J., & Ooi, T. L. (2010). Effectively reducing sensory eye dominance with a push-pull perceptual learning protocol. Current Biology, 20(20), 1864–1868. Xu, J. P., He, Z. J., & Ooi, T. L. (2011). A binocular perimetry study of the causes and implications of sensory eye dominance. Vision Research, 51(23–24), 2386–2397. Xu, J. P., He, Z. J., & Ooi, T. L. (2012). Push–pull training reduces foveal sensory eye dominance within the early visual channels. Vision Research, 61, 48–59. Yang, E., Blake, R., & McDonald, J. E. II, (2010). A new interocular suppression technique for measuring sensory eye dominance. Investigative Ophthalmology & Visual Science, 51(1). 588-586.. Yang, E., Zald, D. H., & Blake, R. (2007). Fearful expressions gain preferential access to awareness during continuous flash suppression. Emotion, 7(4), 882–886. Zheleznyak, L., Alarcon, A., Dieter, K. C., Tadin, D., & Yoon, G. (2015). The role of sensory ocular dominance on through-focus visual performance in monovision presbyopia corrections. Journal of Vision, 15(6), 1–12.

Please cite this article in press as: Dieter, K. C., et al. Individual differences in sensory eye dominance reflected in the dynamics of binocular rivalry. Vision Research (2016), http://dx.doi.org/10.1016/j.visres.2016.09.014