Introduction

Parkinson’s disease (PD) is associated with visual and visuocognitive abnormalities that have a significant impact on quality of life over and above the disease’s hallmark motor symptoms1,2,3,4,5,6,7. Persons with PD (PwPD) whose motor symptoms are more severe on the left side of their bodies, or whose motor symptoms began on the left side (LPD), seem to demonstrate unique spatial and attentional processing changes relative to those with right motor onset (RPD) (e.g8,9,10,11). This may stem from the fact that the right hemisphere is dominant for spatial processing, encoding both left and right visual hemifields, whereas the left hemisphere encodes mainly the right half of space9,12,13. PD is usually asymmetrical in its onset, and the brain pathology on the side contralateral to the initial motor symptoms remains more severe throughout the illness14. Hence, in LPD, the primary processor of left-space has been compromised, with attendant perceptual consequences including, perhaps, hemispatial neglect.

For LPD, the following results are suggestive of neglect:1 On line bisection tasks, this PD subgroup estimates the center of a horizontal line to be to the right of the actual center8,15 for men;16 or more rightward than RPD and HC at certain field positions17,2 The initial direction of eye scanning when exploring a novel stimulus is biased toward the right18,3 Objects presented on the left appear smaller than those presented on the right19. There is also evidence that this syndrome may affect daily living in these individuals, since LPD report bumping into doorways more often on the left than right side15, which could contribute to falls or driving difficulties17,20.

Several studies found no evidence of an LPD-specific effect in terms of neglect, however. Laudate et al.17 showed no difference between controls and LPD on line bisection midpoints in eight out of nine conditions (which differed only in their position on the screen). In addition, two studies by Norton and colleagues attempting to “explain” hemineglect in LPD at a functional level found no evidence of a left-hemifield deficit. The first measured whether space is perceptually compressed in the left hemifield in LPD21 by having subjects compare the perceived thickness of bands on vertically oriented Gabor patches in the left and right hemifield, as well as measuring the perceived contrast of Gabor patches in the left and right hemifield. A second study measured sustained attention using a multiple object tracking task where the targets were systematically restricted to the left or right hemifield22. In both, performance of LPD did not differ from that of control participants. In addition, Salazar and colleagues23 found no evidence of a rightward bias for LPD on a computer-based line bisection task with a sample of 42 RPD and 37 LPD. In that study, both LPD and RPD actually showed a rightward bias, but only when beginning their adjustment of a movable center hatch on the left of the line. In sum, these results raise significant doubts about whether a true hemineglect effect exists on line bisection tasks in LPD.

To address this question, we used a novel psychophysical paradigm to assess line bisection. This paradigm offered two advantages over previous studies. First, it used a two alternative forced choice procedure that assessed the perception of line length apart from possible noise generated by biases and motor behaviors necessary to move a hatch-mark manually on a computer screen or draw a line on a piece of paper. Second, a very brief stimulus presentation allowed us to remove the role that biased visual exploration of the stimulus might play. In addition, we tested a subset of participants using a modified version of this paradigm that contained a fixation cross and employed eye tracking, to investigate whether any neglect effects we might find would persist when gaze was assuredly matched between groups.

Experiment 1

Methods

Participants

Participants were 49 non-demented PwPD, including 21 LPD (12 men, 9 women, mean age = 64.2, standard deviation [SD] = 7.3; see table one), 28 RPD (15 men, 13 women, mean age = 65.5SD = 6.8) and 29 healthy control adults (HC, 11 men, 18 women, mean age = 67.6, SD = 8.7). The groups did not differ significantly in terms of age, F(2,75) = 1.27, p = .29, or years of education, F(2,75) = 0.93, p = .40. Participants were excluded from the study on the basis of having neurological conditions (other than PD, for the PD group), coexisting serious chronic medical illnesses (including psychiatric illness), use of psychoactive medication besides antidepressants and anxiolytics in the PD group, history of intracranial surgery (e.g., deep brain stimulation or other invasive PD treatments), traumatic brain injury, alcohol dependence or recent substance abuse. All participants received a detailed neuro-ophthalmological examination to rule out visual disorders including significant glaucoma, cataracts, or macular degeneration. The minimum Mini-Mental State Examination (MMSE24; score for inclusion in the study was a 26 out of a possible total score of 30. The minimum score for the sample included in this study was 27. Both PD groups had a median Hoehn and Yahr (H&Y) score of 2 out of a total possible score of 5, indicating mild bilateral motor stage of disease. The range of scores for LPDs was between 1 and 4. There was only one LPD participant with a H&Y score of 4, and none with 3 or 3.5. The range of scores for RPDs was between 1 and 3. LPD and RPD did not differ significantly on their H&Y scores (Kolmogorov-Smirnov, p = .49). Scores were unavailable for 2 LPD and 3 RPD. Unified Parkinson Disease Rating Scale (UPDRS) scores were available for 17 LPD, who had a mean motor score of 19.4 (SD 8.1), and for 19 RPD, who had a mean motor score of 19.1 (SD 9.0). Because we had scores for more participants on H&Y than on UPDRS, we report H&Y scores for the participants in each experiment. All participants were taking medications as prescribed, and were therefore in an “on” state during the study.

Stimulus

The stimulus was a long horizontal line (Fig. 1), 15.9 degrees in length and 7 arc min in width, with a perpendicular hatchmark 4.0 degrees in length and 7 arc min in width. The bisected line and bisecting hatchmark were both black and presented on a white background. The bisected line was placed in the center of the screen. The hatchmark was positioned along the horizontal line in one of 12 locations—0.5, 1, 2.1, 4.2, 8.3 or 16.7% offset to the left or right of the bisected line’s center. For example, a 100% offset would indicate the hatchmark being positioned on the endpoint of the line. In order to preclude saccades during the stimulus presentation, the presentation time was set at a brief duration, 83.3 msec, whereas stimulus-driven saccades are on the order of 200 msec25. For the vertical condition, the stimulus was identical, except that it was rotated 90 degrees, and there were 10 hatchmark positions instead of 12. The reason for the difference was that piloting, individuals achieved chance performance in the vertical condition at the 1% position, whereas in the horizontal condition, and extra, more difficult position was required to reach a floor of chance accuracy. Vertical and horizontal stimuli were presented in separate blocks; within those blocks, hatch position conditions were presented in a random order. Each hatch position was presented 8 times. Stimuli for Experiments 1 and 2 were programmed using Psychophysics Toolbox and MatLab26 and were presented on a 21” CRT monitor (Hewlitt Packard FP2141SB) running at 120 Hz. In both experiments, participants’ heads were positioned approximately 20” from the screen, using a chinrest.

Fig. 1
figure 1

Experiment 1 line bisection task. The task begins with a blank screen that is present for 1000 msec. A horizontal line with a vertical hatchmark to the left or right of the horizontal line’s midpoint is then presented for 83 msec. The observer is instructed to verbally respond “left” or “right” after the bisected line disappears. For the vertical condition, the hatchmark and horizontal line were rotated 90 degrees, and the task for subjects became indicating whether the hatchmark was above or below the vertical line’s midpoint. Observers responded verbally with “above” or “below.”

Full size image

Procedures

The task was to indicate orally whether the hatchmark was to the left or the right of the longer line’s midpoint (or above or below the midpoint for the vertical condition; Fig. 1). The hypothesis was that abnormal saccades drive neglect in LPD, and therefore eliminating saccades would also eliminate neglect. That is, LPD would not show a neglect-like pattern on this task and would perform similarly to RPD and HC.

Data analysis

The main dependent variable was the proportion of trials in which participants reported that the hatchmark was to the right of center for each hatch-position. From the proportion right data, a perceived midpoint (PM) was extracted using a Weibull function of the form \({y={\text{1}} - {e^{{{(\frac{{ - x}}{a})}^b}}}}\) where y is the proportion of trials for which the observer reports that the hatchmark is to right of center, x is the hatch position, and a and b are adjustable curve-fitting parameters27. Neglect would manifest itself as a rightward shift in PM, and a decrease in the proportion of trials viewed as right of the midpoint, since neglect tends to shift the PM of the line to the right. Outliers were identified as those who showed a PM more than two standard deviations from the mean in their subgroup (e.g., LPD).

Results

Horizontal condition

Results for proportion-right data are shown in Fig. 2a. LPD reported “right [of midpoint]” less frequently than did RPD and HC, especially at the hatch positions near the center of the line (i.e., where the stimulus was difficult to judge and uncertainty was high). However, statistical analyses did not support this being a reliable difference. An ANOVA was performed on proportion-right data with group as a between-subjects variable and hatch position as a within-subjects repeated measure. The main effect for group was not significant, F(2,72 ) = 1.66, p = .20, η2 = 0.04 There was a main effect for hatch position, F(4.2,300)) = 338.2, p < .001, η2 = 0.82. The interaction between group and hatch position was not significant F(8.3,300) = 0.73, p = .67, η2 = 0.02 (Hyunh-Feldt correction applied for violation of sphericity assumption). Adding handedness as a factor in the model did not substantially change the results. There was still no effect for group, F(2,66) = 0.61, p = .55, η2 = 0.02, and no interaction between group and hatch position, F(7.8,258.8 = 0.86, p = .55, η2 = 0.03).

Fig. 2
figure 2

Experiment 1 results: Negative values on x axis refer to left offset. LPD responded “rightward of center” less often than RPD and HCat hatch-positions near the midpoint of the horizontal line. RPD and HCshowed a similar trend for all hatchmark positions. b) Vertical condition results from experiment 1. c) Perceived midpoints for horizontal line bisection. d) Perceived midpoints for vertical line bisection.

Full size image

Vertical condition

Results for vertical line bisection are shown in Fig. 2b. An ANOVA was performed on proportion-above data with group as a between-subjects variable and hatch position as a within-subjects repeated measure. Including all subjects, the main effect for group was not significant, F(2,72) = 2.48, p = .09, η2 = 0.07. There was a main effect for hatch position, F(4.2, 303) = 493.7, p < .001, η2 = 0.87. The interaction between group and hatch position was significant F(8.4,303) = 2.00, p = .043, η2 = 0.05. When adding handedness to the model, the main effect for group became significant, F(2,66) = 7.37, p = .001, η2 = 0.18, the main effect for hatch position remained significant, F(4.3,281.7) = 195.4, p < .001, η2 = 0.75, and the interaction between group and hatch position remained significant F(8.5,281.7) = 3.22, p = .001, η2 = 0.09 Table 1.

Experiment 2

Methods

Participants

Participants were a subgroup of those from Experiment 1. They were 36 non-demented PwPD, including 17 LPD (10 men, 7 women, mean age = 64.2, SD = 6.9), 19 RPD (11 men, 8 women, mean age = 64.6, SD = 6.4) and 17 HC (6 men, 11 women, mean age = 64.6, SD = 9.1). The groups did not differ significantly in terms of age, F(2) = 0.03, p = .97 or male: female ratio χ22 = 2.5, p = .29. LPD had a Hoehn and Yahr range of 1–4, with a median of 2, and RPD had a range of 1–3 and a median of 2; the two groups did not differ significantly, Kolmogorov-Smirnov, p = .21.

Stimulus and procedure

The stimuli and procedures, shown in Fig. 3, were generally similar to those described for Experiment 1. A fixation cross was used to ensure that the stimulus was processed by the same part of the retina for all participants. The stimulus was a black horizontal line, averaging 15.9 degrees in length and 7 arc min in width, with a perpendicular hatchmark 3.0 degrees in length and 7 arc min in width, both presented on a gray background. The fixation cross was positioned in the center of the screen, and participants were instructed to look directly at it each time it appeared. After 1000 msec, the horizontal section of the fixation cross was overlaid with a long horizontal line offset from center, such that the left or the right half of the line was longer than its opposite half. The presentation time of this stimulus was 100 msec (slightly longer than in experiment 1, in an effort to provider more reliable eye data) There were 8 different left-lengths (with right length being held constant; four decrements [-13.5% of total line length, -6.8%, -3.2%, -1.5%] and four increments [13.5%, 6.8%, 3.2% and 1.5%] for the left half of the line), and 8 different right-lengths used. The task was to determine whether the fixated hatchmark was to the left or the right of the long horizontal line’s center. All conditions were presented in a random order, and each was repeated 8 times. Data were averaged across sides such that, for example, trials where the line length was decreased by 13.5% by shifting the left endpoint of the line towards the center were pooled with trials where the line length was increased by 13.5% by shifting the right endpoint of the line away from the center.

Fig. 3
figure 3

Experiment 2 line bisection task. The task began with a fixation cross presented in the center of the screen. After 1000 msec, a horizontal line was overlaid on the fixation cross so that the vertical portion of the cross formed a vertical hatchmark, offset to the left or right of the horizontal line’s midpoint. The horizontal line remained onscreen for 100 msec, after which participants were instructed to verbally respond “left” or “right” if the vertical hatchmark bisected the line left or right of the horizontal line’s perceived center.

Full size image

Eye-tracking apparatus

Eye-tracking and recording were performed using an Applied Science Laboratories (ASL) eye-tracking system. A model D6 camera array was located underneath the stimulus monitor, and used infrared light to discern the participant’s pupil position and corneal reflection. These reflection points were monitored with Eye-Trac software to locate the position of the participant’s eye, and sampled at a rate of 120 Hz, with a maximum accuracy of 0.5 degrees of visual angle. Although the participant used binocular vision for the experiment, only their left eye was tracked. A 9-point calibration sequence was administered at the beginning of the session and as needed during testing to ensure that the equipment was accurately calculating the participant’s gaze relative to the display for the duration of the experiment. Eye-tracking data were processed with MATLAB. Trials where the participant’s gaze was more than two degrees of visual angle leftward or rightward of the fixation cross, or where a saccade was made during the stimulus presentation, were eliminated.

Results

Response as a function of line-end position

The results for Experiment 2 are shown in Fig. 4. For five subjects (1 LPD, 2 HC, 2 RPD), eye tracking was completely unsuccessful due to failed calibration. For the remaining subjects, the mean proportion of trials fixated was 72.2% (SD = 23.3%) for LPD, 80.1% (24.0%) for HC, and 65.7% (33.9% )for RPD. Groups did not differ on the proportion of trials where fixation was maintained, F(2,42) = 1.08, p = .35, η2 = 0.05.

Fig. 4
figure 4

Experiment 2 results a. Line bisection on trials for which observers maintained fixation. b. Perceived midpoints in each of the three participant groups.

Full size image

An ANOVA was performed on the proportion-right data with group as a between-subjects variable and hatch position as a within-subjects repeated measure. This analysis showed a significant effect for line-end position, F(3.7, 157.2) = 322.9, p < .001, η2 = 0.80, but not for group, F(2,42) = 1.13, p = .33, η2 = 05. The interaction between group and hatch position was not significant, F(7.5, 157.2) = 0.26, p = .77. Using the PM data, there were no group differences, F(2, 49), = 0.38, p = .69, η2 = 0.02.

Discussion

The present study used a novel psychophysical paradigm to examine line bisection performance in LPD. Across two experiments, both of which likely precluded any strategic eye movements to scan the target, we found no evidence of hemineglect as assessed by horizontal line bisection. Instead, we found that perception of line length was biased in LPD for vertical lines, where a significant interaction between hatch position and group was driven by LPD participants’ reduced proportions of trials judged as having the hatch mark above the midpoint, relative to the other groups (Fig. 2b).

Hemineglect in LPD

No direct evidence was found for hemineglect in LPD in the present study. A horizontal bias was present in LPD in the same direction as that seen in hemineglect patients with right parietal damage, but this bias was not statistically distinct from that seen in healthy control or RPD participants. The present result was obtained using a psychophysical paradigm that eliminated non-perceptual bias with its use of a two alternative forced choice procedure. Experiment one also prevented strategic eye movements to scan the target by using a brief presentation time. In Experiment two, when gaze was also controlled using a fixation cross and eye tracking, the difference in line bisection performance between HC and LPD diminished further and was still non-significant.

This study was also the first to remove exploratory eye movements as a possible factor in the results of a line bisection task in PD. Since stimulus presentation was limited to 83 msec, it is unlikely that observers were able to generate any exploratory saccades during stimulus presentation. This removes reduced saccadic amplitude as a factor in the results, which is important since reduced saccadic amplitude has been demonstrated in PD28. The results of Experiment 1 indicate instead that a perceptual bias in vertical but not horizontal line bisection exists in LPD, but it is independent of abnormalities in saccadic functioning.

In Experiment 2 an additional aspect of oculomotor functioning, gaze bias, was also precluded as a possible factor in the results. Although it is unlikely that participants were able to explore the stimulus with their eyes for more than one fixation, it is possible that PD participants had more variability or a directional bias in their gaze position when the stimulus appeared on the screen, and this could have affected their perception. Gaze bias was controlled using a fixation cross and eye tracking; trials on which fixation was not maintained were removed. Here, there was no difference between LPD and HC, and the non-significant difference was smaller than that seen for Experiment 1. The results of Experiment 2 are similar to other studies on PD that utilized psychophysical paradigms with eye tracking and a fixation cross, finding little to no group differences between LPD, RPD, and HC. When examining LPD participants’ perceptions of spatial compression, no biases were found in the data that would support the hypothesis of left hemifield neglect21. Additionally, no evidence was found to support a left-hemifield deficit in sustained attention to moving objects22. This pattern of results across experiments suggests that when the possibility of gaze bias and biased exploratory eye movements are eliminated using a fixation cross and eye tracking, no LPD differences in the processing of left-space emerge relative to processing by RPD or control participants.

A point of potential importance for future research is that we focused on LPD and found that some of these participants showed notable neglect-like biases, but so did some with RPD, as well as some HC (see PM scatter plots, Fig. 2). Side of motor symptom onset is likely only one factor accounting for the variance in line bisection bias in PD. Individual differences in spatial biases that occur in healthy individuals and those with PD are a potentially informative topic that to date has received little attention. Another point is that since all patients were taking dopamine replacement medications as prescribed, it is possible that medication obscured potential group difference that may exist in the absence of these drugs.

Vertical line bisection

A bias in LPD performance occurred with respect to the upper and lower visual fields: LPD viewed the hatchmark as above the center less frequently than RPD and HC. Although the vertical condition was mostly included in the present study as a control condition for the neglect-sensitive horizontal condition, it appears that the results of the vertical condition are reflective of a true difference in the LPD group. Vertical-axis bias has been shown previously in LPD, with conflicting results16,17. Lee and colleagues16found that LPD estimated a vertical line’s midpoint to be below its physical center, whereas control adults and RPD did the opposite, but Laudate and colleagues17 did not find a difference between LPD and a healthy control group on vertical line bisection. It is unclear why the two previous studies found conflicting results, though both were based on small samples. The present data are consistent with those of Laudate and colleagues, and in the opposite direction of Lee and colleagues. Of relevance to the present discussion, in both previous studies eye movements were free to be made during an unlimited stimulus presentation time. The results of these studies are therefore ambiguous as to why LPD performed as they did (bisecting lines below or above their true vertical center), since they could be explained by eye movement differences, or by altered perception of the upper visual field of space. In the present study, LPD perceived the vertical line’s midpoint as being below its physical position, and since scanning of the target was not possible, an abnormality in processing of up- or down-space appears to exist in LPD independently of saccadic abnormalities.

Possible neural correlates of bias in vertical length perception in PD

The present results may be related to multiple forms of neuropathology that have been demonstrated in PD. Parietal neuropathology in PD has been shown in the form of decreased gray matter density29,30 and cortical thinning31,32. In addition, some individuals with PD, namely those with mild cognitive impairment, show hypometabolism in parietal cortex33,34. Because the parietal lobe is important for attention and representation of visual space12,13,35,36,37, this parietal dysfunction could affect performance in a way that biases those with LPD process space and length differently.

In addition, subcortical brain pathology may play a role in altered perception in PD. Where humans perceive themselves to be looking is determined by a combination of factors, including a series of presumed corollary discharges offered by the brain during each eye movement. These discharges have been proposed to be altered in PD38. If the movements that are executed do not match the ones that are shared in a corollary discharge (due to motor pathology in PD), it could alter individuals’ perceptual map of the world dramatically. Two brain systems that may be involved with the corollary discharge, the superior colliculus and thalamus, are affected in PD. The superior colliculus is crucial for determining location of eye movement and maintaining a map of salient locations in space39. This area is directly connected to substantia nigra40, and has been proposed to be dysfunctional in PD41. In addition, PD is associated with alterations in the functional connectivity and morphology of the thalamus42,43,44, affecting basal ganglia-thalamocortical circuits. It is possible that alterations in thalamic connectivity and function may contribute to the manifestation of perceptual bias in PD patients, a point that necessitates further research following the findings identified here.

Limitations

The present work was subject to some limitations. First, it would have been useful to see if gaze bias in Experiment 1 (using eye tracking but not requiring central fixation) predicted line bisection performance on the task, but eye tracking was not used in that experiment. A second limitation relates to the line lengths used in our stimuli. Neglect in LPD previously had been shown primarily using longer line lengths than used in the present study. We used lines of approximately 16 degrees of visual angle due to constraints of a standard computer monitor. Future studies could attempt replication of the present study by using a projector or other methods to increase the visual angle of lines. Another limitation of the present study is that we collected the data without a clear sense of what our final sample size would be, opening ourselves to potential bias if we had been running analyses after every few subjects and waiting till they turned out to be significant. Fortunately, we were able to resist that temptation. Finally, replicating the present results in comparison with a task that allowed or required exploratory eye movements, such as a traditional pen and paper line bisection test or other perceptual neuropsychological tests such as judgement of line orientation, might reveal interesting truths about the role that eye movements play in hemineglect-like performance in LPD, especially since a large recent study did find evidence for neglect in LPD8.

Conclusions

The results of the present study indicate that the perception of length is altered in LPD, particularly for vertical lines, even for briefly presented stimuli that preclude examination with exploratory eye movements. No significant hemineglect effect for LPD was found in a horizontal version of the task, despite a trend in this direction. Understanding the mechanisms underlying spatial biases in PD, as well as their impact on daily life, may offer insight into possible routes toward remediation.

Table 1 Subject demographic and clinical information.
Full size table