Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2013 May 15.
Published in final edited form as: Vision Res. 2012 Jan 18;61:60–69. doi: 10.1016/j.visres.2012.01.003

Further support for the importance of the suppressive signal (pull) during the push-pull perceptual training

Jingping P Xu 1, Zijiang J He 1,*, Teng Leng Ooi 2,*
PMCID: PMC3342430  NIHMSID: NIHMS351190  PMID: 22273998

Abstract

We previously designed a push-pull perceptual training protocol that effectively reduces sensory eye dominance (SED) and enhances binocular depth detection in human adults (Xu et al, 2010a). During the training, an attention cue precedes a pair of binocular competitive stimulus to induce dominance of the weak eye and suppression of the strong eye. To verify that the success of the protocol is due to the suppression of the signals evoked by the stimulus in the strong eye, rather than to the attention cueing per se, we employed two new push-pull training protocols that did not involve attention cueing. Instead, we used the specific configurations of the boundary contours of the binocular competitive stimulus to render the strong eye suppressed. The first, MBC push-pull protocol has a half-image with grating feature but no boundary contour in the strong eye. The second, BBC push-pull protocol has a half-image with both grating feature and boundary contour in the strong eye. For both protocols, the weak eye receives a half-image with strong grating feature and boundary contour. These boundary contour configurations ensure that the weak eye remains dominant while the strong eye is suppressed during training. Each observer was trained with both protocols at two parafoveal (2 deg) retinal locations. We found that both protocols significantly reduce SED and binocular depth threshold. This confirms the basis of the push-pull protocol is the suppression of the strong eye, rather than the attention cueing per se. We further found that the learning effect (SED reduction) is more effective in the BBC push-pull protocol where the suppressed half-image in the strong eye carries both grating feature and boundary contour information, than in the MBC push-pull protocol where the boundary contour information is absent from the strong eye’s half-image. This suggests that the learning effect depends in part on the availability of the image attributes for processing (suppression) during the push-pull perceptual training.

Keywords: Adult cortical plasticity, Binocular boundary contour, Monocular boundary contour, Interocular inhibition, Push-pull, Sensory eye dominance, Stereopsis

1. Introduction

When the two eyes are stimulated with very different images at corresponding retinal points, only one image is perceived while the other image is suppressed by the interocular inhibitory mechanism. Typically, the choice of which image is selected for perception is based on the stimulus intensity or contrast, with the stronger image being perceived more frequently (higher predominance) (Fox, 1991; Levelt, 1965). There are some observers with clinically normal stereopsis, however, who consistently experience an image from one eye being dominant more frequently even when the two half-images have equal stimulus strength. This indicates such observers have the condition of sensory eye dominance (SED) with an intrinsic imbalance of interocular inhibition (Ooi & He, 2001; Porac & Coren, 1976; Xu et al, 2011c).

We have previously quantified SED by using a binocular rivalry stimulus with different intensity or contrast in the two half-images (e.g., Ooi & He, 2001; Xu et al, 2010a; 2011a). Figure 1a illustrates two pairs of orthogonal gratings for measuring SED at a local retinal location (Xu et al, 2010a). During the test, the observer first perceives the left pair of dichoptic stimulus. The contrast of the horizontal grating in the left eye (LE) is adjusted until he/she experiences an equal chance of perceiving the vertical and horizontal gratings. This measures the LE balance contrast. Then the observer is presented with the right pair of dichoptic stimulus, and with a similar procedure, his/her right eye (RE) balance contrast is obtained. Since the contrast of the vertical grating is kept the same while measuring the LE and RE balance contrast values, one can define their difference as the SED. The eye with the smaller balance contrast is the sensory dominant (strong) eye while the fellow eye is the non-dominant (weak) eye. In this paper, we refer to the SED measured in this way as the contrast-SED to distinguish it from its variant, which we call the boundary contour (BC) based SED (BC-SED).

Figure 1.

Figure 1

Open in a new tab

(a) Two pairs of orthogonal gratings for measuring contrast-SED. The left pair of dichoptic stimulus measures the LE balance contrast, which is obtained by adjusting the contrast of the horizontal grating in the LE while keeping the contrast of the RE’s vertical grating constant. The right pair of dichoptic stimulus is used to measure the RE balance contrast in a similar manner. The difference between the LE and RE balance contrast values defines the contrast-SED. (b) Two pairs of orthogonal gratings for measuring BC-SED. During the test, one keeps the contrast levels of the vertical and horizontal gratings constant while adjusting the relative phase-shift between the horizontal grating disc and its surrounding horizontal grating. In this way, the BC strength of the horizontal grating disc is varied to obtain the balance phase-shift for the LE (left pair of stimulus) and RE (right pair of stimulus). The difference between the LE and RE balance phase-shift values defines the BC-SED. (c) The MBC rivalry stimulus, which gives rise to a stable perception of the horizontal grating disc (lasting seconds). (d) and (e) depict pairs of orthogonal grating stimuli for measuring BC-SED. They differ from those in (b) in their grating orientations. In (d), the variable phase-shifted disc is vertical and both half-images have vertical grating background. In (e), the variable phase-shifted disc is oriented 135° and both half-images have 135° oriented grating background.

The BC-SED is measured with a different set of test stimuli, as illustrated in figure 1b. To obtain BC-SED, one keeps the contrast levels of the vertical and horizontal gratings constant while adjusting the relative phase-shift between the horizontal grating disc and its surrounding horizontal grating (Xu et al, 2010b, 2011a). In this way, the BC strength of the horizontal grating disc is varied until the observer experiences an equal chance of seeing the horizontal and vertical grating discs. The phase-shift at this point of equality is defined as the balance phase-shift. Thus, using a similar procedure as for measuring the contrast-SED, one can obtain the LE and RE balance phase-shift values, and define their difference as the BC-SED.

The contrast-SED and BC-SED measurements have a bias in characterizing the differential effects of interocular inhibition on two types of visual cortical properties for binocular surface representation, namely, the surface boundary contour (BC) and surface feature properties (Grossberg & Mingola, 1985; Nakayama et al, 1995; Ooi & He, 2006; Su et al, 2009; 2011a; van Bogaert et al, 2008; von der Heydt, 2003; Xu et al, 2010b; Zhou et al, 2000). When measuring contrast-SED (figure 1a), varying the grating contrast mainly changes the binocular compatibility of the interior region of the stimulus that affects its surface feature property. On the other hand, when measuring the BC-SED (figure 1b), adjusting the grating phase-shift varies the BC strength at the border of the horizontal grating disc, and thus mainly affects its boundary contour property.

In adults, the SED is modifiable through visual experience (perceptual learning), as with other forms of visual functions (Fahle, 1997; Karni & Sagi, 1991; Lu & Dosher, 2009; Paffen et al, 2008; Sasaki et al, 2010; Schoups et al, 1995; Seitz et al, 2009; Suzuki & Grabowecky, 2002, 2007; Watanabe et al, 2001; Xiao et al, 2008; Xu et al, 2010a, 2011a, b; Zhou et al, 2008). We have found that SED is effectively reduced with a push-pull training protocol (figure 2a) that targets the putative interocular inhibitory neural network (Xu et al, 2010a, 2011a, b). During the training trial, an attention cue (monocular frame) is presented to the weak eye for 100 msec, followed by a pair of orthogonal gratings (vertical/horizontal). The brief cue attracts transient (bottom-up) attention (e.g., Nakayama & Mackeben, 1992) to the weak eye, resulting in the vertical grating in the weak eye being perceived (push) while the horizontal grating in the strong eye is suppressed (pull) (Ooi & He, 1999). The cueing and orthogonal gratings presentation is repeated a second time, followed by a binocular mask to terminate the trial. The observer’s task is to discriminate the orientation of the two sequentially perceived gratings (vertical and near-vertical) in the weak eye. We found that observers who underwent such a training protocol over a 10-day period significantly reduced their SED. [Note: The role of the cue is to deploy transient (involuntary, bottom-up) attention to the weak eye to cause its stimulus to be perceived in dominance during binocular rivalry. Top-down (voluntary) attention can prolong the dominance duration of the image only when the image is already in the dominance state. In itself, top-down attention has little ability to cause the rivaling image to become dominant (Ooi & He, 1999).]

Figure 2.

Figure 2

Open in a new tab

(a) Sequence of stimulus presentation in the push-pull training protocol employed by Xu et al (2010a), using the attention cue to secure dominance of the weak eye. (b) and (c) show the stimulus presentation sequence for the MBC push-pull and BBC push-pull protocols, respectively. Instead of cueing for attention as in (a), the weak eye’s half-image is presented with a strong stimulus (horizontal grating disc). The strong eye’s half-image is defined by surface feature (vertical grating) in (b), and by both boundary contour and surface feature (vertical grating disc) in (c).

Equally important, we found the push-pull protocol is a more effective protocol than the push-only protocol, which is carried out in the same manner except that the strong eye is not stimulated with a grating disc (not shown). To explain the difference between the push-pull and push-only protocols, we have hypothesized that the suppression of the half-image in the strong eye during the push-pull training effectively shifts the balance of interocular inhibition between the two eyes. This is because with the push-pull protocol, repetitive stimulation of the strong eye while preventing its signals from reaching the higher level (thus failing to induce conscious perception) could effectively degrade the efficiency of the excitatory synaptic transmission within the strong eye’s channel and also depress the inhibition of the strong eye on the weak eye’s channel (Hebb, 1949; Xu et al, 2010a). This differs from the push-only training protocol, where the suppressed strong eye sees only a blank field, which presumably contributes little to the modulation of the interocular inhibition between the two eyes.

In this paper, we further evaluated our suppression hypothesis above by testing two proposals. First, we proposed that a push-pull training protocol without the attention cue prior to the binocular rivalry stimulus should still be effective, as long as the half-image presented to the strong eye is suppressed during the training. Second, we proposed that the learning efficacy is affected by the properties of the suppressed image in the strong eye during the push-pull training. Specifically, the learning effect on SED will be larger if the suppressed image in the strong eye carries boundary contour and surface feature information, rather than the surface feature information alone. This assumption is based on the reasoning that tilting the favor to the weak eye (i.e., learning) would be more effective if more information (modules) from the strong eye channel were recruited by the interocular inhibitory process (suppressed) during the training. In this paper, we investigated these two proposals by modifying the stimuli shown to the strong eye relative to those used in our original push-pull protocol (figure 2a). The new stimuli are displayed in figures 2b and 2c.

The MBC push-pull protocol (figure 2b) directly tests the first proposal. The stimulation in this protocol differs from the original one (figure 2a) because it capitalizes on the monocular boundary contour (MBC) rivalry display to cause the suppression of the image viewed by the strong eye. One can inspect the MBC rivalry display more closely by free fusing the second row of half-images in figure 2b (or figure 1c). With fusion, one perceives a relatively stable horizontal grating disc in front of the surrounding vertical grating square. This is because the visual system selects the horizontal grating disc with the boundary contour for perception as it suppresses the corresponding vertical grating patch in the fellow eye (Frisby & Mayhew, 1978; Ooi & He, 2005, 2006; Su et al, 2009, 2011a, b). Separately, our laboratory has found that observers can perceive the MBC grating disc when the stimulus duration is as short as 30 msec (Su et al, 2011b), indicating that the interocular inhibitory mechanism can quickly suppress the homogenous grating half-image. Accordingly, in the MBC push-pull protocol (figure 2b), presenting the MBC rivalry display to an observer for 500 msec should be sufficient to cause dominance of the MBC disc with horizontal grating viewed by the weak eye and suppression of the vertical grating viewed by the strong eye. Consequently, we predict that the MBC push-pull protocol can induce a significant learning effect (i.e., reduction of SED), even though it does not use the attention cue employed in the original push-pull protocol. [It should be pointed out that there is a possibility that the strong BC in the weak eye could attract top-down attention to prolong the dominance duration of the weak eye (Ooi and He, 1999). However, as pointed out above, top-down attention in itself has little ability to cause the weak eye to be dominant. Moreover, our specific goal is to test if the pre-cueing that brings bottom-up attention to the weak eye, and not top-down attention, is required to produce the learning effect.]

The BBC push-pull protocol (figure 2c) also tests the first proposal. Furthermore, a comparison between the outcomes of the MBC and BBC push-pull protocols allows us to test the second proposal. The stimulation in the BBC push-pull protocol is the same as in the MBC push-pull protocol, except for one modification. Here, the strong eye’s half-image has a vertical grating disc that is created by phase-shifting the area corresponding to the horizontal disc in the weak eye’s half-image by 180° relative to its surrounding vertical grating. Essentially, unlike the stimulation in the MBC push-pull protocol that lacks the BC information in the strong eye’s half-image, the stimulation in the BBC push-pull protocol has both BC and surface feature information in the strong eye’s half-image. Thus, the former stimulation only suppresses one type of information whereas the latter stimulation suppresses both types of information. We predict this will cause the BBC push-pull protocol to induce a larger learning effect than the MBC push-pull protocol. [Note: We refer to the BBC stimulus as such because each half-image has a disc that carries the boundary contour. Hence, when fused, a pair of binocular boundary contour (BBC) emerges.]

2. Material and methods

2.1 Apparatus

The stimuli were presented on a flat-screen CRT monitor (2048 × 1536 pixels @ 75Hz, except for the contrast-SED test with 1280 × 1024 pixels @ 100Hz) using a MacPro computer running MATLAB and Psychophysics Toolbox (Brainard, 1997; Pelli, 1997). The two half-images were viewed through a mirror haploscopic system attached to a chin-and-head rest that aided fusion from a viewing distance of 85 cm.

2.2 Observers

One author and six naïve observers (22–28 years old) with informed consent participated in the study. All observers had normal or corrected-to-normal visual acuity (at least 20/20), clinically acceptable fixation disparity (≤8.6 arc min), central stereopsis (≤40 arc sec), and passed the Keystone vision-screening test.

2.3 General stimuli and procedures

We first used the test stimuli similar to those in figure 1b to measure the BC-SED at eight concentric retinal locations (0°, 45°, 90°, 135°, 180°, 225°, 270°, and 315°) that are situated 2° in angular distance from the fovea. This allowed us to select two retinal locations with the largest BC-SED (~ 40–50° phase-shift), respectively, for training with the MBC push-pull and the BBC push-pull protocols (figure 2b and 2c). The two retinal locations selected were separated from one another by at least 2.83 deg. The two training protocols were implemented, one on each retinal location, on the same day over a 10-day duration (2 sessions/day). The training (learning) effect on BC-SED was monitored before and after each day’s training session using stimuli similar to those in figure 1b.

Additionally, we conducted several other tests at each training location before and after the training phase to assess the learning effect. These tests were: (1) BC-SED with three different grating stimulus orientations; (2) contrast-SED; (3) stereo threshold. These tests are described in detail in Section 2.5.

2.4 MBC push-pull and BBC push-pull protocols

The two retinal locations chosen for training were randomly assigned to the two protocols, which are the same in all aspects except for the design of the binocular rivalry stimuli employed (figures 2b & 2c). The stimulus design for the MBC push-pull protocol had a horizontal sinusoidal grating disc (3 cpd, 1.25°, 35 cd/m2, 1.8 log unit contrast) that was surrounded by a 7.5°×7.5° vertical grating background (3 cpd, 35 cd/m2, 1.2 log unit contrast) in one half-image. The other half-image had a homogeneous vertical grating (3 cpd, 35 cd/m2, 7.5°×7.5°, 1.2 log unit contrast). The binocular rivalry stimulus design in the BBC push-pull protocol also had one half-image with the horizontal sinusoidal grating disc and vertical background. However, the other half-image had a vertical sinusoidal grating disc at the location corresponding to the horizontal grating disc in the first half-image. The vertical grating disc was created by phase-shifting a circular region of the vertical grating surrounding it by 180°. All other parameters of the BBC stimulus design were the same as those for the MBC stimulus design.

The stimulation sequence for both protocols was identical. During the training, a trial began with fixation at the nonius target. Then, at the training location, the MBC or BBC stimulus was presented for 500 msec, and 400 msec later, a second MBC or BBC stimulus was presented for another 500 msec (figures 2b and 2c). The horizontal grating of the disc in the second presentation had a slightly different orientation from the horizontal grating in the first presentation. Four hundred msec after the second presentation, the onset of a binocular checkerboard sinusoidal grating mask terminated the trial (200 msec, 7.5°×7.5°, 3 cpd, 35 cd/m2, 1.8 log unit contrast). Notably, due to its higher contrast and stronger boundary contour, the horizontal grating disc in the weak eye was always perceived during each presentation interval because it successfully suppressed the corresponding vertical grating in the strong eye. The observer reported whether the first or second horizontal grating disc had a slight counterclockwise orientation, and an audio feedback was given. Fifty such training trials were run for each training block. We employed the QUEST procedure (Watson & Pelli, 1983) to obtain the orientation discrimination threshold at the end of the 50-trial block. Altogether, twelve blocks were performed during each training session/day, for each training protocol.

The two protocols were implemented on each training day, in an interleaved manner. Training on each protocol lasted for an hour, which was performed either in the morning or afternoon session.

2.5 Specific stimuli and procedures

2.5.1 Measuring BC-SED at 8 different retinal locations (to select two training locations)

The test stimulus comprised a pair of dichoptic vertical (1.2 log unit contrast) and horizontal (1.8 log unit contrast) sinusoidal grating discs (3 cpd, 1.25°, 35 cd/m2), each surrounded by a 7.5°×7.5° horizontal grating background (35 cd/m2, 3 cpd, 1.8 log unit contrast) (figure 1b). The horizontal grating of the disc had a variable phase-shift (0–180 degrees) relative to the larger horizontal grating background. During the test, a trial began with central fixation on the nonius target (0.45°×0.45°, line width=0.1°, 70 cd/m2) and the presentation of the dichoptic test stimulus (500 msec), followed by a 200 msec mask (7.5°×7.5°, 2-D sinusoidal grating, 3 cpd, 35 cd/m2, 1.8 log unit contrast). The observer’s task was to judge whether the disc was filled with more horizontal or vertical grating.

A staircase procedure was used to adjust the relative phase-shift of the horizontal grating disc after each trial with a step size of ~14.2° phase-shift (one pixel), until the observer obtained equal chance of seeing the vertical and horizontal gratings, i.e., the point of equality. Each block of trials (~50–60 trials) comprised 30 reversals, with the average of the last 26 reversals taken as the final balance phase-shift. When the horizontal grating disc was presented to the LE, we refer to its phase-shift at the point of equality as the LE balance phase-shift (left pair, figure 1b). Then the grating half-images were switched between the eyes to obtain the RE balance phase-shift (right pair, figure 1b). The difference in the balance phase-shift between the LE and RE is defined as the BC-SED. The measurement at each location was repeated twice.

2.5.2 Measuring BC-SED at the 2 training locations before and after the training phase

2.5.2.1 BC-SED with horizontal background orientation

The test stimuli were the same as the ones used in section 2.5.1 above (figure 1b). However, the method of constant stimuli, instead of the staircase method, was used to obtain the BC-SED. We tested with seven levels of relative phase-shifts of the horizontal grating (0°, 28.4°, 56.8°, 85.3°, 113.7°, 156.3°, 184.7°). Each relative phase-shift level was repeated 7 times/block over 6 blocks. As in the SED test in Section 2.5.1, the observer responded to seeing either the horizontal or vertical grating (predominant percept). In this, and the remaining SED tests below (including Section 2.5.3), we tested 4 stimulus combinations [2 locations (MBC/BBC) × 2 eyes (left/right)]. Each combination was repeated twice. The order of testing was randomized.

2.5.2.2 BC-SED with vertical background orientation

The test stimuli are depicted in figure 1d, where the vertical (1.8 log unit contrast) and horizontal (1.2 log unit contrast) sinusoidal grating discs (3 cpd, 1.25°, 35 cd/m2) are surrounded by a 7.5°×7.5° vertical grating background (35 cd/m2, 3 cpd, 1.8 log unit contrast). The test procedure (staircase) and observer’s task were the same as in section 2.5.1.

2.5.2.3 BC-SED with oblique background orientation

The test stimuli are shown in figure 1e. The dichoptic 45° (1.2 log unit contrast) and 135° (1.8 log unit contrast) grating discs (1.25°, 3 cpd, 35 cd/m2, 500 msec) are surrounded by a 7.5°×7.5° 135° grating background (3 cpd, 35 cd/m2, 1.8 log unit contrast). The test procedure and observer’s task were the same as in section 2.5.1.

2.5.3 Measuring contrast-SED at the 2 training-locations before and after the training phase

The stimulus comprised a pair of dichoptic vertical and horizontal sinusoidal grating discs (3 cpd, 1.25°, 35 cd/m2) (figure 1a). The contrast of the vertical grating was held constant (1.5 log unit) while the contrast of the horizontal grating was varied (0–1.99 log unit). A trial began with central fixation on the nonius target (0.45°×0.45°, line width=0.1°, 70 cd/m2), the presentation of the dichoptic orthogonal grating discs (500 msec), and terminated with a 200 msec mask (7.5°×7.5° checkerboard sinusoidal grating, 3 cpd, 35 cd/m2, 1.5 log unit contrast). The observer responded to his/her percept of seeing more vertical or horizontal grating orientation. The horizontal grating contrast was adjusted after each trial using the QUEST procedure (50 trials/block), until the observer obtained equal chance of seeing the vertical and horizontal gratings, i.e., the point of equality. When the horizontal grating was presented to the LE we refer to its contrast at the point of equality as the LE balance contrast. We then switched the grating discs between the eyes to obtain the RE balance contrast. The difference between the LE and RE balance contrast is defined as the contrast-SED.

2.5.4 Stereo threshold test with random dot stereogram at the 2 training-locations before and after the training phase

A 7.5°×7.5° random-dot stereogram (dot size=0.0132°, 35 cd/m2) with a variable crossed-disparity disc target (1.25°) was used (figure 6a). The contrast of the stereogram was individually selected for each observer, to make the stereo task moderately difficult and to avoid a possible ceiling-effect due to pixel-size constraint. With this criterion, the contrast levels were variously set for different observers (1.2 log unit: 1 observer, 1.3 log unit: 3 observers, 1.5 log unit: 1 observer; 1.7 log unit: 2 observers).

Figure 6.

Figure 6

Open in a new tab

(a) The random-dot stereogram used to measure binocular disparity threshold. (b) The average stereo thresholds measured with random dot stereogram at the MBC and BBC push-pull training locations. A significant reduction in binocular disparity threshold is found at both training locations.

We used the standard 2AFC method in combination with the staircase procedure to measure stereo disparity threshold. The temporal sequence of stimulus presentation was: fixation, interval-1 (200 msec), blank (400 msec), interval-2 (200 msec), blank (400 msec), and random-dot mask (200 msec, 7.5°×7.5°, 35cd/m2). The observer indicated whether the crossed-disparity disc was perceived in interval-1 or -2, and an audio feedback was given. Each block comprised 10 reversals (step size = 0.8 arc min, total ~50–60 trials), with the last 8 reversals taken as the average threshold. Each block was repeated 4 times, and measured over two days.

3. Results

3.1 Learning effect on BC-SED during the training phase

The average interocular balance phase-shift data obtained during the training-phase are shown in figures 3a and 3b, respectively, for the MBC push-pull and BBC push-pull protocols. The open symbols represent the results from the test stimuli where the test grating orientation (in the disc) was the same as the trained grating orientation (in the disc). For simplicity, we refer to these results as the “same” data. The closed symbols represent the results with test stimuli that were orthogonal to the trained orientation (in the disc). We refer to these results as the “orthogonal” data. [Note that the test (figure 1b) and training (figures 2b and 2c) stimuli had different background grating orientation.]

Figure 3.

Figure 3

Open in a new tab

The average interocular balance phase-shift data for (a) the MBC push-pull protocol and (b) BBC push-pull protocol. The interocular balance phase-shift data were obtained, respectively, with grating whose orientation was the same as, or orthogonal to, the grating used in the training, and measured before and after each day’s training. Generally, the interocular balance phase-shift reduces with days in training when tested with the same orientation grating. Gray symbols in both graphs plot the average interocular balance phase-shift data (for pre-training at day 0 and post-training at day 10) that were obtained using the method of constant stimuli. (c) BC-SED reduces with days in the training for both protocols. The gray symbols reveal the data obtained with the method of constant stimuli.

At the MBC training location (figure 3a), the same interocular balance phase-shift decreases as the training progressed when it was measured before (open black squares, slope=−3.915, R2=0.917, p<0.001) and after (open black circles, slope=−3.188, R2=0.943, p<0.001) each day’s training session. On the other hand, there is little change in the orthogonal interocular balance phase-shift measured before (filled black squares, slope=−0.153, R2=0.136, p=0.265) and after (filled black circles, slope=−0.697, R2=0.752, p=0.001) each day’s training session, indicating the learning effect does not transfer to a channel with an orthogonal orientation. Two-way ANOVA with repeated measures confirm that the slope in the orthogonal data is significantly shallower than that in the same data [interaction effect between testing stimuli (same/orthogonal) and training session: before, F(10, 60)=10.903, p<0.001; after, F(9, 54)=2.098, p=0.046]. This demonstrates the orientation specificity of the perceptual learning effect (Xu et al, 2010a). It is also clear that there is a significant difference in the interocular balance phase-shift measured before and after each day’s training session for the same data (main effect of the before-after: F(1,6)=91.176, p<0.001; interaction effect between the before-after and training session: F(9, 54)=0.653, p=0.747; 2-way ANOVA with repeated measures). However, there is no significant difference in the before and after interocular balance phase-shift for the orthogonal data [main effect of the before-after: F(1,6)=3.227, p=0.123; interaction effect between the before-after and training session: F(9, 54)=0.638, p=0.760; 2-way ANOVA with repeated measure]. This phenomenon of performance degradation immediately after the training has also been reported previously (Mednick et al, 2002, 2005; Ofen et al, 2007; Yotsumoto et al, 2009; Xu et al, 2010a, 2011b).

The interocular balance phase-shift data at the BBC-training location (figure 3b) show a similar trend. There is a clear learning effect in the same data [before: open black diamonds, slope=−3.193, R2=0.863, p<0.001; after: open black triangles, slope=−3.382, R2=0.817, p<0.001] but no reliable learning effect in the orthogonal data [before: filled black diamonds, slope=0.410, R2=0.357, p=0.052; after: filled black triangles, slope=0.250, R2=0.149, p=0.271]. A significant before-after difference in interocular balance phase-shift is only observed in the same data [main effect of the before-after: F(1,6)=65.113, p<0.001; interaction effect between the before-after and training session: F(9, 54)=1.431, p=0.198; 2-way ANOVA with repeated measure], and not in the orthogonal data [main effect of the before-after: F(1,6)=0.613, p=0.463; interaction effect between the before-after and training session: F(9, 54)=1.024, p=0.434; 2-way ANOVA with repeated measure].

Figure 3c plots the average BC-SED, which is defined as the difference between the same and orthogonal interocular balance phase-shift values of the corresponding conditions. BC-SED is significantly reduced, i.e., showing a learning effect, for all four sets of data [before/MBC: slope=−3.762, R2=0.898, p<0.001; after/MBC: slope=−2.490, R2=0.899, p<0.001; before/BBC: slope=−3.603, R2=0.911, p<0.001; after/BBC: slope=−3.631, R2=0.835, p<0.001]. The learning effect on BC-SED is similar at the two training locations with the MBC and BBC protocols, no matter whether they were measured before each day’s training session [main effect of training protocol (MBC/BBC): F(10,60)=11.792, p<0.001; interaction effect between the training protocol and training session: F(10,60)=1.611, p=0.125; 2-way ANOVA with repeated measures], or after the training session [main effect of training protocol (MBC/BBC): F(9,54)=4.562, p<0.001; interaction effect between the training protocol and training session: F(9,54)=0.481, p=0.881; 2-way ANOVA with repeated measures].

Altogether, the above observations during training with both MBC and BBC push-pull protocols demonstrate that learning to reduce BC-SED is possible without resorting to explicit attention cueing during the training, as performed in our previous studies (Xu et al, 2010a, 2011a, b). This finding reinforces the notion that suppression of the stimulus presented to the strong eye is the important factor triggering a significant plasticity within the interocular inhibitory network.

3.2 Assessing the learning effect on specific visual functions tested before and after the training

3.2.1 Learning effect on BC-SED with horizontal grating background

Besides using the staircase method, we also used the method of constant stimuli to measure the weak and strong eyes’ interocular balance phase-shift with the test stimuli in figure 1b (stimuli with horizontal grating background and similar to those used in Section 3.1 above). Figures 4a and 4b, respectively, plot the average data for the MBC and BBC push-pull training locations with fitted psychometric functions (cumulative normal distribution functions). Separately, we applied probit analysis to each observer’s data to estimate the interocular balance phase-shift (phase-shift at 50% performance level) and the standard deviation of the normal distribution of each psychometric function. The average interocular balance phase-shift data are plotted with gray symbols in figure 3 (pre-training at day 0 and post-training at day 10) and they show a similar trend as the ones measured with the staircase method during the training phase (section 3.1). At the MBC push-pull training location, there is a signification decrease in interocular balance phase-shift in the weak eye [same data: pre: 120.4±4.0 deg, post: 88.5±7.0 deg, t(6)=4.753, p=0.003], but little change in the strong eye [orthogonal data: pre: 69.7±10.5 deg, post: 75.736±8.557 deg, t(6)= −1.309, p=0.238]. A similar trend is found at the BBC push-pull training location [same interocular balance phase-shift: pre: 119.1±5.2 deg, post: 83.8±8.6 deg, t(6)=5.477, p=0.002; orthogonal interocular balance phase-shift: pre: 74.8±3.9 deg, post: 80.9±6.7 deg, t(6)= −0.967, p=0.371]. The pre- and post- BC-SED data are plotted in a bar graph in figure 5a. BC-SED is significantly reduced at both the MBC [pre: 50.7±8.5 deg, post: 12.8±10.9 deg, t(6)=3.887, p=0.008] and BBC push-pull [pre: 44.3±6.7 deg, post: 2.8±7.2 deg, t(6)=5.086, p=0.002] training locations. Although the BC-SED reduction at the MBC training location (37.9 deg) is smaller than that at the BBC training location (41.5 deg), their difference fails to reach the significant level [Main effect of training condition: F(1,6)=1.592, p=0.254; main effect of training session: F(1,6)=22.051, p=0.003; interaction effect between training condition and session: F(1,6)=0.326, p=0.589, 2-way ANOVA with repeated measures].

Figure 4.

Figure 4

Open in a new tab

(a) and (b), respectively, plot the average data for the MBC and BBC push-pull training locations with fitted psychometric functions (cumulative normal distribution functions). Generally, the psychometric functions for the same/after data are to the left of the same/before data, indicating a reduction in the weak eye’s balance contrast.

Figure 5.

Figure 5

Open in a new tab

(a) Average BC-SED data tested with the stimulus with horizontal grating background (figure 1b) using the method of constant stimuli. For training at both the MBC and BBC push-pull locations, BC-SED is significantly reduced after (post) the training. (b) Average BC-SED measured using the test stimuli with vertical grating background (figure 1d) in the pre and post training phase. BC-SED is significantly reduced at the MBC and BBC push-pull training locations. (c) Average BC-SED measured with the test stimuli with oblique grating background (figure 1e) before and after the training. A significant reduction of BC-SED is found at the BBC training location but not at the MBC training location. (d) Average contrast-SED measured before and after the training phase with the test stimuli shown in figure 1a. The reduction in BC-SED is significant at the BBC training location but not at the MBC training location.

3.2.2 Learning effect on BC-SED with vertical grating background

Figure 5b illustrates the average BC-SED measured using the test stimuli with vertical grating background (figure 1d) in the pre and post training phase. Clearly, the BC-SED data are reduced at both the MBC [t(6)=3.360, p=0.015] and BBC push-pull [t(6)=4.420, p=0.004] training locations. Although the BC-SED reduction at the MBC training location is smaller than that at the BBC training location, statistical analysis fails to reveal a significant difference between the two [Main effect of training session: F(1,6)=17.980, p=0.005; interaction effect between training location and session: F(1,6)=0.045, p=0.840, 2-way ANOVA with repeated measures].

3.2.3 Learning effect on BC-SED with oblique grating background

Figure 5c depicts the average BC-SED data measured with the test stimuli with oblique grating background (figure 1e) before and after the training. The reduction in BC-SED is significantly smaller at the MBC training location than at the BBC training location [Main effect of training session: F(1,6)=10.317, p=0.018; interaction effect between training location and session: F(1,6)=22.237, p=0.003, 2-way ANOVA with repeated measures]. Further analysis reveals a significant reduction at the BBC training location [t(6)=4.111, p=0.006] but not at the MBC training location [t(6)=1.652, p=0.150].

Overall, we found that both the MBC and BBC push-pull training protocols effectively reduce the BC-SED when the test stimuli (vertical/horizontal grating discs) are similar to the training stimuli (figures 4, 5a and 5b). We also found that the BBC push-pull protocol significantly reduces BC-SED when the training (vertical/horizontal) and test (oblique) stimuli have different orientations (figures 1e & 5c). This finding, that the learning to reduce BC-SED is transferable to a test stimulus whose orientation is 45° from the trained orientation, resembles our earlier finding using the push-pull protocol with attention cueing (Xu et al, 2011a). As discussed in that paper, we attribute the learning largely to the plasticity of interocular inhibition on the boundary contour mechanism. It should be emphasized that we can make this conclusion because the oblique test stimuli (figure 1e) reveal the operation of the surface BC mechanism more substantially than that of the surface feature mechanism. On the other hand, while test stimuli such as those in figure 1a may reveal a BC-SED learning effect, they do not allow us to differentiate the relative contributions of the surface boundary contour from the surface feature mechanisms.

More generally, we attach a significance to our current finding that learning transfers to test stimuli (figure 1e) that are oriented 45° from the training stimuli (figure 5c). This is in contrast to our earlier finding in Xu et al (2010a) that revealed a much reduced learning when the test stimuli were oriented 45° away from the training stimuli. Together, they argue for the notion that the surface BC mechanism has broader orientation tuning functions than the surface feature mechanism.

Also notable, is that the learning effect on BC-SED is larger at the BBC push-pull training location than at the MBC push-pull training location. In particular, the learning effect on the BC-SED with oblique grating background shown in figure 5c is significantly larger at the BBC than at the MBC training location. This tendency suggests that the suppression of the BC in the half-image presented to the strong eye during training more effectively reduces SED. The MBC push-pull protocol is not as effective because the half-image presented to the strong eye does not have a BC (i.e., there is no BC suppression but only surface-feature suppression in the strong eye).

3.3 Learning effect on contrast-SED

Figure 5d depicts the average contrast-SED measured before and after the training phase with the test stimuli shown in figure 1a. The reduction in contrast-SED is significantly larger at the BBC than the MBC push-pull training location [Main effect of training session: F(1,6)=5.274, p=0.061; Interaction effect between training condition and session: F(1,6)=10.328, p=0.018, 2-way ANOVA with repeated measures]. Further analysis reveals that the reduction is significant at the BBC training location [t(6)=3.625, p=0.011] but not at the MBC training location [t(6)=0.758, p=0.477]. This finding is consistent with the observations of the BC-SED with oblique grating background in figure 5c, where the BBC push-pull training results in a significantly stronger learning effect than the MBC push-pull training. Both findings together support the notion that binocular suppression of boundary contour in the strong eye during the training more effectively reduces sensory eye dominance.

3.4 Learning effect on stereo threshold

Figure 6b plots the average stereo thresholds measured with random dot stereograms (figure 6a) at the MBC and BBC push-pull training locations. A similar learning effect is found at both training locations [Main effect of the training session: F(1,6)=98.025, p<0.001; Interaction effect between training location and session: F(1,6)=1.655, p=0.246, 2-way ANOVA with repeated measures]. Further analysis reveals significant decreases of binocular disparity thresholds at both training locations [MBC: t(6)=9.191, p<0.001; BBC: t(6)=9.421, p<0.001]. This learning effect on binocular depth perception, a visual function that was not trained, is similar to those found in our previous studies that used the push-pull protocol with attention cueing (Xu et al, 2010a, 2011a, b).

4. Discussion

We have hypothesized that suppression of the image signals in the strong eye’s channel during the push-pull training is a critical factor for modifying the interocular inhibitory network in adults. Such a modification leads to reduced SED and improved stereopsis (Xu et al, 2010a). In the current study, we tested two proposals related to the suppression hypothesis to further our understanding. We employed two new stimulation strategies (MBC and BBC) in the push-pull protocol, which do not rely on attention cueing to suppress the strong eye as in Xu et al (2010a). Instead, both the MBC and BBC push-pull protocols rely on having a stronger image in the weak eye to suppress the image in the strong eye. As predicted, we were able to show that both protocols can significantly reduce SED and binocular depth threshold. This confirms that it is the suppression of the strong eye during training, rather than attention cueing per se, that is important for perceptual learning (first proposal).

Furthermore, when comparing the magnitude of the learning effect between the MBC and BBC push-pull protocols, we found that the latter produces a larger learning effect (second proposal). Notably, the stimulation in both protocols is the same in all aspects, except for the property of the suppressed image in the strong eye. The difference between the two protocols rests on the composition of the suppressed image; in the MBC protocol the suppressed image only has surface feature information whereas in the BBC protocol the suppressed image has both surface feature and surface BC information. This difference underscores not only the significant role of the BC information in interocular inhibition, but also supports the notion that the types of signals suppressed from the strong eye are critical during perceptual learning. The reason for this might be due to the fact that increasing the number of stimulus properties in the image causes the interocular inhibitory network to interact with more modular (local filter) processes in the visual cortex, thereby, strengthening the inhibition of the weak eye onto the strong eye. It follows from the assumption that these modular processes contribute to binocular competition, hence SED, relatively independently and in a cumulative manner.

Our hypothesis above is drawn from the general learning rule that when neural signals at the pre-synaptic level fail to proceed to the post-synaptic level, the efficiency of the synaptic transmission will be reduced (Hebb, 1949). A way to influence synaptic plasticity is to rely on the inhibitory connections to the synapses that selectively modulate the flow of neural signals across synapses (Adini et al, 2002; Harauzov et al, 2010; Hensch et al, 1998; Huang et al, 1999; Huang et al, 2010; Karmarkar & Dan, 2006; Xu et al, 2010a). In view of this, we recognize that the interocular inhibitory network underlying binocular suppression offers an ideal model for conducting psychophysical studies to demonstrate the role of the putative inhibitory neurons in modulating neural plasticity. This is largely because inputs (stimuli) to the two eyes can be independently manipulated and the perceptual consequence of inhibition (binocular suppression) can be measured with psychophysical methods.

Finally, on a speculative note, we consider the possible neural loci along the visual pathway where the perceptual learning effect to reduce SED might occur. This is by no means straightforward since the measured SED, like most psychophysical performance, is contributed by multiple neural representations along the visual pathway. Nevertheless, although neither our current study nor previous studies (Xu et al, 2010a; 2011a, b) were specifically designed to address this issue, they do provide some clues. There are reasons to deduce the primary visual cortex (V1) contributes significantly to the learning effect. First, the perceptual learning likely occurs in the neural network that carries the eye-of-origin information in area V1. This is because SED, which measures the imbalance of interocular inhibition, involves the eye-of-origin information. The eye-of-origin information is explicitly coded by the monocular neurons that are mostly found in area V1, and not beyond. Our studies have shown that the learning effect on SED reduction is eye specific (Xu et al, 2010a, 2011a). For example, both in figure 3a and 3b and also in figure 3a of Xu et al (2010a), the push-pull training produces little learning effect on interocular balance when the test and trained stimuli were presented in opposite eyes (orthogonal conditions, represented by filled symbols in the figures).

Second, the learning effect on SED reduction cannot be observed outside the push-pull training location. For example, Xu et al (2010a) found little transfer of learning from the push-pull training location to the adjacent untrained locations. Equally significant, we found that the learning effect fails to transfer to the push-only training location. (In that study, we trained two retinal locations in an interleave manner, respectively, with the push-pull and push-only protocols. The observers performed the same grating orientation discrimination task in the weak eye.)

Third, the finding in Xu et al (2011a) that top-down attention is not required for the learning effect to occur is consistent with the notion that top-down attention only exerts an indirect, and relatively modest influence on the early-level visual processes (e.g., V1), presumably through a feedback network from the extrastriate cortices (Kastner & Ungerleider, 2000; Yoshor et al, 2007). Yet, interestingly, we also found that directing top-down attention to the push-pull training location significantly augments the learning effect (SED reduction). This suggests that the learning effect on SED involves plasticity of the higher visual level that provides top-down attention modulation to the early visual cortex. Altogether, it is likely that the learning effect (SED reduction) is contributed by a distributed neural plasticity along the visual pathway that recognizes the eye-of-origin information (Fahle, 1997; Karni & Sagi, 1991; Mollon & Danilova, 1996; Liu & Weinshall, 2000; Schoups et al, 1995; Seitz et al, 2009; Watanabe et al, 2002; Xiao et al, 2008; Zhang et al, 2010).

5. General conclusion

We have shown that an important reason for the strong learning effect with the push-pull training protocol is the suppression of the image signals in the strong eye’s channel (processes) during the perceptual training. This confirms that the critical role of the transient (bottom-up) attention cue in the push-pull training protocol used in Xu et al (2010) is to render the half-image presented to the strong eye suppressed. It also reinforces the notion that transient attention paves the way for dominance during binocular competition (Ooi & He, 1999). Furthermore, our current study reveals that the push-pull perceptual training protocol can modify the surface feature and surface boundary contour mechanisms.

Highlights.

  • A push-pull training protocol with an attention cue reduces sensory eye dominance

  • The present study uses two new push-pull training protocols without attention cue

  • Both protocols have strong boundary contour and feature contents in the weak eye

  • Sensory eye dominance and stereo threshold are effectively reduced after training

  • Suppressive signals to the strong eye during the push-pull training causes learning

Acknowledgments

This study was supported by a grant from the National Institutes of Health (EY015804) to TLO and ZJH.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Adini Y, Sagi D, Tsodyks M. Context-enabled learning in the human visual system. Nature. 2002;415:790–793. doi: 10.1038/415790a. [DOI] [PubMed] [Google Scholar]
  2. Brainard DH. The Psychophysics Toolbox. Spat Vis. 1997;10:433–436. [PubMed] [Google Scholar]
  3. Fahle M. Specificity of learning curvature, orientation, and vernier discriminations. Vision Res. 1997;37:1885–95. doi: 10.1016/s0042-6989(96)00308-2. [DOI] [PubMed] [Google Scholar]
  4. Fox R. Binocular rivalry. In: Regan DM, Cronly-Dillon JR, editors. Binocular Vision Vision and Visual Dysfunction. Vol. 9. Macmillan; London: 1991. pp. 93–110. [Google Scholar]
  5. Frisby JP, Mayhew JEW. The relationship between apparent depth and disparity in rivalrous-texture stereogram. Perception. 1978;7(6):661–678. doi: 10.1068/p070661. [DOI] [PubMed] [Google Scholar]
  6. Grossberg S, Mingolla E. Neural dynamics of form perception: Boundary completion, illusory figures, and neon color spreading. Psychol Rev. 1985;92:173–211. [PubMed] [Google Scholar]
  7. Harauzov A, Spolidoro M, DiCristo G, De Pasquale R, Cancedda L, Pizzorusso T, Viegi A, Berardi N, Maffei L. Reducing intracortical inhibition in the adult visual cortex promotes ocular dominance plasticity. J Neurosci. 2010;30:361–371. doi: 10.1523/JNEUROSCI.2233-09.2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Hebb DO. The organization of behavior. Wiley; New York: 1949. [DOI] [PubMed] [Google Scholar]
  9. Hensch TK, Fagiolini M, Mataga N, Stryker MP, Baekkeskov S, Kash SF. Local GABA circuit control of experience-dependent plasticity in developing visual cortex. Science. 1998;282:1504–1508. doi: 10.1126/science.282.5393.1504. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Huang JZ, Kirkwood A, Morales B, Pizzorusso T, Porciatti V, Bear MF, Maffei L, Tonegawa S. BDNF is a key regulator of the maturation of inhibition and critical period of mouse visual cortex. Cell. 1999;98:739–755. doi: 10.1016/s0092-8674(00)81509-3. [DOI] [PubMed] [Google Scholar]
  11. Huang S, Gu Y, Quinlan EM, Kirkwood A. A refractory period for rejuvenating GABAergic synaptic transmission and ocular dominance plasticity with dark exposure. J Neurosci. 2010;30:16636–16642. doi: 10.1523/JNEUROSCI.4384-10.2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Karmarkar UR, Dan Y. Experience-dependent plasticity in adult visual cortex. Neuron. 2006;52:577–585. doi: 10.1016/j.neuron.2006.11.001. [DOI] [PubMed] [Google Scholar]
  13. Kastner S, Ungerleider LG. Mechanisms of visual attention in the human cortex. Annu Rev Neurosci. 2000;23:315–41. doi: 10.1146/annurev.neuro.23.1.315. [DOI] [PubMed] [Google Scholar]
  14. Karni A, Sagi D. Where practice makes perfect in texture discrimination: evidence for primary visual cortex plasticity. Proc Natl Acad Sci U S A. 1991;88:4966–4970. doi: 10.1073/pnas.88.11.4966. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Levelt W. On binocular rivalry. Royal Van Gorcum; Assen, The Netherlands: 1965. [Google Scholar]
  16. Liu Z, Weinshall D. Mechanisms of generalization in perceptual learning. Vision research. 2000;40(1):97–109. doi: 10.1016/s0042-6989(99)00140-6. [DOI] [PubMed] [Google Scholar]
  17. Lu ZL, Dosher B. Mechanisms of Perceptual Learning. Learning and Perception. 2009;1:19–36. doi: 10.1556/LP.1.2009.1.3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Mednick SC, Nakayama K, Cantero JL, Atienza M, Levin AA, Pathak N, Stickgold R. The restorative effect of naps on perceptual deterioration. Nat Neurosci. 2002;5:677–681. doi: 10.1038/nn864. [DOI] [PubMed] [Google Scholar]
  19. Mednick SC, Arman AC, Boynton GM. The time course and specificity of perceptual deterioration. Proc Natl Acad Sci USA. 2005;102:3881–3885. doi: 10.1073/pnas.0407866102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Mollon JD, Danilova MV. Three remarks on perceptual learning. Spatial Vision. 1996;10:51–58. doi: 10.1163/156856896x00051. [DOI] [PubMed] [Google Scholar]
  21. Nakayama K, Mackeben M. Sustained and transient components of focal visual attention. Vision Res. 1992;29:1631–1647. doi: 10.1016/0042-6989(89)90144-2. [DOI] [PubMed] [Google Scholar]
  22. Nakayama K, He ZJ, Shimojo S. Visual surface representation: a critical link between lower-level and higher level vision. In: Kosslyn SM, Osherson DN, editors. Vision: An Invitation to Cognitive Science. MIT Press; Cambridge, MA: 1995. pp. 1–70. [Google Scholar]
  23. Ofen N, Moran A, Sagi D. Effects of trial repetition in texture discrimination. Vision Res. 2007;47:1094–1102. doi: 10.1016/j.visres.2007.01.023. [DOI] [PubMed] [Google Scholar]
  24. Ooi TL, He ZJ. Binocular rivalry and visual awareness: the role of attention. Perception. 1999;28:551–574. doi: 10.1068/p2923. [DOI] [PubMed] [Google Scholar]
  25. Ooi TL, He ZJ. Sensory eye dominance. Optometry. 2001;72:168–178. [PubMed] [Google Scholar]
  26. Ooi TL, He ZJ. Surface representation and attention modulation mechanisms in binocular rivalry. In: Alais D, Blake R, editors. Binocular Rivalry. MIT Press; Cambridge, MA: 2005. pp. 117–135. [Google Scholar]
  27. Ooi TL, He ZJ. Binocular rivalry and surface-boundary processing. Perception. 2006;35:581–603. doi: 10.1068/p5489. [DOI] [PubMed] [Google Scholar]
  28. Paffen CL, Verstraten FA, Vidnyánszky Z. Attention-based perceptual learning increases binocular rivalry suppression of irrelevant visual features. J Vis. 2008;8:25.1–11. doi: 10.1167/8.4.25. [DOI] [PubMed] [Google Scholar]
  29. Pelli DG. The VideoToolbox software for visual psychophysics: transforming numbers into movies. Spat Vis. 1997;10:437–442. [PubMed] [Google Scholar]
  30. Porac C, Coren S. The dominant eye. Psychol Bull. 1976;83:880–897. [PubMed] [Google Scholar]
  31. Sasaki Y, Nanez JE, Watanabe T. Advances in visual perceptual learning and plasticity. Nat Rev Neurosci. 2010;11:53–60. doi: 10.1038/nrn2737. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Schoups AA, Vogels R, Orban GA. Human perceptual learning in identifying the oblique orientation: retinotopy, orientation specificity and monocularity. J Physiol. 1995;483:797–810. doi: 10.1113/jphysiol.1995.sp020623. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Seitz AR, Kim D, Watanabe T. Rewards evoke learning of unconsciously processed visual stimuli in adult humans. Neuron. 2009;61:700–707. doi: 10.1016/j.neuron.2009.01.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Su Y, He ZJ, Ooi TL. Coexistence of binocular integration and suppression determined by surface border information. Proc Natl Acad Sci U S A. 2009;106:15990–15995. doi: 10.1073/pnas.0903697106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Su RY, He ZJ, Ooi TL. Revealing boundary-contour based surface representation through the time course of binocular rivalry. Vision Res. 2011a;51:1288–1296. doi: 10.1016/j.visres.2011.04.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Su RY, He ZJ, Ooi TL. Seeing grating-textured surface begins at the border. J Vis. 2011b;11:14. doi: 10.1167/11.1.14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Suzuki S, Grabowecky M. Evidence for perceptual “trapping” and adaptation in multistable binocular rivalry. Neuron. 2002;36:143–157. doi: 10.1016/s0896-6273(02)00934-0. [DOI] [PubMed] [Google Scholar]
  38. Suzuki S, Grabowecky M. Long-term speeding in perceptual switches mediated by attention-dependent plasticity in cortical visual processing. Neuron. 2007;56:741–753. doi: 10.1016/j.neuron.2007.09.028. [DOI] [PubMed] [Google Scholar]
  39. van Bogaert EA, Ooi TL, He ZJ. The monocular-boundary-contour mechanism in binocular surface representation and suppression. Perception. 2008;37:1197–1215. doi: 10.1068/p5986. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. von der Heydt R. Image parsing mechanisms of the visual cortex. In: Werner JS, Chalupa LM, editors. The Visual Neurosciences. MIT press; Cambridge, MA: 2003. pp. 1139–1150. [Google Scholar]
  41. Watanabe T, Náñez JE, Sasaki Y. Perceptual learning without perception. Nature. 2001;413:844–848. doi: 10.1038/35101601. [DOI] [PubMed] [Google Scholar]
  42. Watanabe T, Nanez JE, Koyama S, Mukai I, Liederman J, Sasaki Y. Greater plasticity in lower-level than higher-level visual motion processing in a passive perceptual learning task. Nat Neurosci. 2002;5:1003–1009. doi: 10.1038/nn915. [DOI] [PubMed] [Google Scholar]
  43. Watson AB, Pelli DG. QUEST: a Bayesian adaptive psychometric method. Percept Psychophys. 1983;33:113–120. doi: 10.3758/bf03202828. [DOI] [PubMed] [Google Scholar]
  44. Xiao LQ, Zhang JY, Wang R, Klein SA, Levi DM, Yu C. Complete transfer of perceptual learning across retinal locations enabled by double training. Curr Biol. 2008;18:1922–1926. doi: 10.1016/j.cub.2008.10.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Xu JP, He ZJ, Ooi TL. Effectively reducing sensory eye dominance with a push-pull perceptual learning protocol. Curr Biol. 2010a;20:1864–1868. doi: 10.1016/j.cub.2010.09.043. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Xu JP, He ZJ, Ooi TL. Surface boundary contour strengthens image dominance in binocular competition. Vision Res. 2010b;50:155–170. doi: 10.1016/j.visres.2009.11.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Xu JP, He ZJ, Ooi TL. Perceptual learning to reduce sensory eye dominance beyond the focus of top-down visual attention. Vision Res. 2011a doi: 10.1016/j.visres.2011.05.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Xu JP, He ZJ, Ooi TL. Push-pull training reduces foveal sensory eye dominance within the early visual channels. Vision Res. 2011b doi: 10.1016/j.visres.2011.06.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Xu JP, He ZJ, Ooi TL. A binocular perimetry study of the causes and implications of sensory eye dominance. Vision Research. 2011c;51:2386–2397. doi: 10.1016/j.visres.2011.09.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Yoshor D, Ghose GM, Bosking WH, Sun P, Maunsell JH. Spatial attention does not strongly modulate neuronal responses in early human visual cortex. J Neurosci. 2007;27:13205–9. doi: 10.1523/JNEUROSCI.2944-07.2007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Yotsumoto Y, Chang L, Watanabe T, Sasaki Y. Interfereqnce and feature specificity in visual perceptual learning. Vision Res. 2009;49:2611–2623. doi: 10.1016/j.visres.2009.08.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Zhang T, Xiao LQ, Klein SA, Levi DM, Yu C. Decoupling location specificity from perceptual learning of orientation discrimination. Vision Research. 2010;50:368–374. doi: 10.1016/j.visres.2009.08.024. [DOI] [PubMed] [Google Scholar]
  53. Zhou Y, Huang C, Xu P, Tao L, Qiu Z, Li X, Lu ZL. Perceptual learning improves contrast sensitivity and visual acuity in adults with anisometropic amblyopia. Vision Res. 2008;46:739–750. doi: 10.1016/j.visres.2005.07.031. [DOI] [PubMed] [Google Scholar]
  54. Zhou H, Friedman HS, von der Heydt R. Coding of border ownership in monkey visual cortex. J Neurosci. 2000;20:6594–611. doi: 10.1523/JNEUROSCI.20-17-06594.2000. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES