Sixteen healthy human subjects (age 22–29 years old, 7 males, 9 females) participated in this study. They were all right-handed subjects with normal or corrected to normal vision. We measured the eye dominance of each subject by 3 repeats of the “Hole-in-card” test (Miles, 1930), and the results were consistent for each individual subject. Among the 16 subjects, 5 subjects were left-eye dominant and 11 subjects were right-eye dominant. All subjects were naive to the experimental purpose. At the early stage of this study, 10 subjects (left-eye dominant: 5, right-eye dominant: 5) were recruited in the First Clinical College of Harbin Medical University. Since one subject had difficulty to keep fixation (the fixation break rate > 20%), data from this subject were excluded from further analysis. At the late stage of this study, the other 6 subjects (right-eye dominant: 6) were recruited in Beijing Normal University. The protocol of this study followed the ethical guidelines of the Declaration of Helsinki and was approved by the Institutional Review Board of the First Clinical College of Harbin Medical University and Beijing Normal University. All subjects gave written informed consent before participating in this study and received financial compensation for their participation.

To eliminate the possibility that the surrounding objects might serve as the allocentric referees, all experiments were conducted in a dark room. The computer screen was placed 57 cm in front of the subjects' eyes. A chin rest restricted any head movement of subjects. When setting up the experimental system, we carefully measured the distance between the chin rest and the screen to ensure that: the chin rest and the screen were parallel in both horizontal and vertical dimensions, and the vertical middle line of the chin rest was aligned with the vertical middle line of the screen. Thus, we did our best to ensure that the screen was centered and oriented parallel to the plane of the chin rest. In addition, we marked the midpoint of the chin rest and instructed the subjects to put their chins on this point before collecting data in each session. And the subjects were instructed to keep the same posture throughout each session.

At the early stage of this study, a computer keyboard was positioned in front of the subjects, with the up and down keys aligned with their body mid-sagittal plane. The left key and right key were positioned with equal distance from the body mid-sagittal plane. At the late stage of this study, to exclude the possibility that the leftward deviation of keyboard might cause the deviation of SSA, the keyboard was covered by two black boxes with the same size, only leaving the up, down, left and right keys uncovered. In addition, the up and down keys were aligned with the body mid-sagittal plane.

All visual stimuli were presented on a 21-inch CRT monitor (SUN X7149A, 1280 × 960 pixels, 100 Hz vertical refresh rate) and the luminance of the stimuli was measured with a photometer (LS-110; Konica Minolta). Before starting the experiments, we calibrated and linearized the screen with the photometer. We also presented a circle with a 15° radius that was centered at the center of the screen. When we measured the distances from multiple points on the circle to the center of the screen, we found that the distances were all 15°. Thus, there was no spatial distortion on the screen display.

We monitored and collected the eye position signal at a sample rate of 1 kHz by an infrared camera eye tracking system (EyeLink 1000 Desktop Mount; SR Research). We used MATLAB (version 2012a; The MathWorks) with Psychtoolbox (PTB-3; Brainard and Pelli, 2015) to control the presentation of the visual stimuli and collect the subjects' RT data.

In the main task, we excluded fixation break trials (early stage: 4.60%, 447 out of 9,720 trials; late stage: 4.35%, 188 out of 4,320 trials) and no response trials (early stage: 0.01%, 1 out of 9,720 trials; late stage: 0.14%, 6 out of 4,320 trials) from data analysis. In the control task, we first excluded fixation break trials (4.42%, 138 out of 3,120 trials). Then trials with RT differing more than 3 standard deviations from the mean RT of each day (in total, 1.48%, 44 out of 2,982 trials) were also excluded. The number of no response trials was zero.

To illustrate the subjects' egocentric judgment more clearly, we analyzed the percentage of rightward judgment at each visual cue location. To be noticed, the percentage of leftward judgment was a mirror image of the percentage of rightward judgment (100–the percentage of rightward judgment). The percentages of rightward judgment at each visual cue location in the main task are shown in Figure 2A. Data of each individual subject (same colored triangles) were fitted with a cumulative normal distribution function, shown as the same colored thin curve. The averaged data were denoted as black triangles and thick curve. As expected, the percentages of rightward judgment at the 3 most peripheral locations in the left VF (−15.5, −10.5, and −6.5°) were all 0% and those at the 3 most peripheral locations in the right VF (15.5, 10.5, and 6.5°) were all 100% among the subjects. When the visual cue got closer to the body mid-sagittal plane (from −3.5 to 3.5°), the percentage of rightward judgment gradually increased in the left VF, and it gradually decreased in the right VF (Figure 2B). Such results indicated that, when the visual cue was close to the body mid-sagittal plane, the egocentric judgment to the same location was ended with opposite outputs (leftward versus rightward, Table 2). Strikingly, the fitted psychometric curves of all subjects shifted to the left side of the body mid-sagittal plane, and the population point of subject equality (PSE) was at −0.31° ± 0.16° (mean ± SD). Compared with the body mid-sagittal plane, the leftward deviation of SSA was significant (p = 4.1135e-05, Wilcoxon test). These results showed that the visual SSA of all subjects deviated to the left side of the body mid-sagittal plane, regardless of the eye dominance.

We assessed the time of egocentric judgment by calculating the post-adjusted RTs of each individual subject in the main task (Figure 3A). As expected, the RTs gradually prolonged as the visual cue got closer to the body mid-sagittal plane. Notably, the RTs were asymmetric between left and right VFs: RTs were longer in the left VF than in the right VF. Two-way ANOVA showed that both eccentricity [F_{(5, 96)} = 10.92, p = 2.4673e-08, η_p² = 0.36] and laterality [F_{(1, 96)} = 4.12, p = 0.045, η_p² = 0.04] of the visual cue had significant influence on the RTs. But there was no significant interaction effect between the two factors [F_{(5, 96)} = 0.32, p = 0.901, η_p² = 0.02]. Moreover, RTs of all subjects but one were longest when the visual cue was closest to the SSA. For instance, the population RT was longest when the visual cue was at −0.5° (Figure 3B), consistent with the fact that the mean SSA was located at −0.31° (Figure 2B).

We further quantitatively analyzed the RT difference (left VF–right VF) between each mirror location for each individual subject. The results of the subtracted RTs of each subject are presented with different colored triangles in Figure 4. The solid symbols denote that the RT difference between left and right VFs reached the statistically significant level (p < 0.05, Wilcoxon test), whereas the dashed symbols do not. As shown, the RT difference between the 2 most center locations (−0.5° versus 0.5°) was greatest, with the RT difference of 6 out of 9 subjects reaching the statistically significant level. The RT difference gradually decreased as the eccentricity of visual cue increased. Eventually, at the peripheral locations, the RT difference of several subjects reversed.

The longer RT indicated that longer time was needed to judge the egocentric location of the visual cue. Our RT data showed that the most difficult egocentric judgment was around the SSA, rather than the body mid-sagittal plane.

At the late stage of this study, 6 subjects performed the main task with only 6 visual cue locations. The objective of the late stage was to assess the influence of eye position on the spatial judgment to a visual cue. We compared the eye position (averaged from visual cue onset to 250 ms later) between trials where subjects made two opposite judgments to the same visual cue.

First, the fitted psychometric curves of the rightward judgment percentages of 6 subjects shifted to the left side of the body mid-sagittal plane. The population PSE was at −0.18° ± 0.10° (mean ± SD) (Figure 5A). Compared with the body mid-sagittal plane, the leftward deviation of SSA was significant (p = 0.002, Wilcoxon test), which was consistent with the results of 9 subjects at the early stage.

The location of −0.5° was nearest to the location of SSA, thus the inconsistency of judgment (leftward versus rightward) was highest. We compared the eye positions between two groups of trials at −0.5°. We found that the distribution of the eye position was centered at the fixation point, regardless of the results of spatial judgment (Figures 5B,C). Furthermore, the horizontal eye positions between the two groups of trials were not significantly different both in the individual subject level (except one subject) and in the population level (Figure 5D, p = 0.496, Wilcoxon test). These results indicated that in the present study, the leftward deviation of SSA was not due to the influence of eye position.

The anatomical and physiological features of visual system of primates might explain our results—the leftward deviation of SSA and longer RT in the left VF. Previous studies have revealed that, while the lower visual areas in one hemisphere strictly process the visual information from contralateral VF (Tootell et al., 1982), the higher visual cortices process the visual information from both contralateral and ipsilateral VFs (Gross et al., 1969, 1972; Andersen et al., 1990; Raiguel et al., 1997; Ben Hamed et al., 2001). One reasonable explanation for our results is that the right hemisphere receives and processes more visual information from the ipsilateral VF than the left hemisphere does (Sheremata et al., 2010; Zhou et al., 2012; Sheremata and Silver, 2015). Therefore, compared with the left VF, the right VF might be overestimated, so that the SSA deviated to the left VF and the sensorimotor processing took less time in the right VF, i.e., shorter RTs. A strongly supportive evidence for this assumption is that the serious hemispatial neglect more frequently occurs after lesions of the right hemisphere but not after lesions of the left hemisphere (Beis et al., 2004; Ringman et al., 2004; Becker and Karnath, 2007; Kleinman et al., 2007).

It is well-known that the eye position strongly affects the perception of visual egocentric information (Barbeito and Simpson, 1991; Sridhar and Bedell, 2011, 2012). However, most previous studies did not monitor the eye position during experiments, thus the effect of eye position on visual egocentric judgment was ignored. The importance of eye position on visual egocentric judgment is caused by the fact that the retinal visual input combines with the eye position signal to build a head-centered reference frame. At the same time, the retinotopic position of an object is also transformed into the head-centered reference frame (Andersen et al., 1997). Indeed, the effect of eye position on the judgment of SSA in normal subjects has been found in previous studies (Morgan, 1978; Jeannerod and Biguer, 1989; Richard et al., 2005).

To exclude the effect of eye position on the judgment of SSA, we asked subjects to keep fixating straight ahead during the experiments and monitored the subjects' eye position. We found that the horizontal eye positions between the trials with opposite spatial judgments were not significantly different. Thus, the leftward deviation of SSA in the present study was not caused by the deviation of eye position.

The location of SSA is frequently measured in the studies of egocentric representation in healthy subjects and brain damage patients. SSA reflects the location of the subjective mid-sagittal plane, which subjectively separates the egocentric space into left and right halves. Thus, the comparison between the location of SSA and body mid-sagittal plane can provide useful information to help to understand the contribution of the two hemispheres to egocentric information processing. Based on the findings from previous studies, many factors can cause the deviation of SSA, which may share different neural mechanisms. First, the SSA will deviate toward the ipsilateral side of sensory stimulation if it is presented only in one side, including neck proprioception stimulation (Karnath et al., 1994, 2002), vestibular stimulation (Karnath et al., 1994) and acute experimental painful stimulation (Bouffard et al., 2013). The deviations of SSA under these conditions are mainly due to the post-training effect, which is very different from the mechanisms leading to the deviation of SSA in our present study. Second, patients with lateralized lesion of the peripheral nervous system or visual cortices also show the deviations of SSA, such as unilateral pathologic pain (Sumitani et al., 2007; Reinersmann et al., 2012), left vestibular loss (Saj et al., 2013) and homonymous hemianopia (Ferber and Karnath, 1999; Rousseaux et al., 2013) or quadrantanopia (Kuhn et al., 2010). Third, damage in cortical regions, in particular the parietal cortex, results in the symptom of hemispatial neglect and the ipsilesional deviation of SSA (Karnath, 1994; Farne et al., 1998; Ferber and Karnath, 1999; Schindler and Kerkhoff, 2004; Richard et al., 2004a,b, 2005; Saj et al., 2006; Rousseaux et al., 2013). These findings from clinical studies indicate that the unbalanced egocentric information processing between the two hemispheres causes the deviation of SSA. Here we report the leftward deviation of SSA in healthy human subjects when they fixate straight ahead, which might also reflect the asymmetric process of egocentric information between the two hemispheres.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.