Introduction

As society ages rapidly, older adults face health concerns that extend beyond physical well-being. Research has shown that a considerable proportion of older adults exhibit symptoms of depression, referred to as late-life depression (LLD), with an estimated point prevalence ranging from 0.9 to 42% 1,2,3,4. Despite the significant detrimental consequences associated with LLD5,6,7, depressive symptoms in older adults are often overlooked due to more pronounced physical health issues. Therefore, it is imperative to develop an objective and cost-effective means to assist in the diagnosis of depression in older individuals.

The eye-tracking method, which monitors eye movements, provides an effective means for developing behavioral biomarkers for clinical investigations8,9,10,11. Microsaccades, a type of fixational eye movement, are small saccadic eye movements (< 2 degrees) that occur approximately 1–3 times per second during the period of visual fixation12,13,14. These microsaccades are not merely random noise of the oculomotor system; instead, they are modulated by various sensory, affective, and cognitive processes [e.g.,15,16,17,18,19,20,21,22,23,24]. For example, research has shown that microsaccade rates and metrics are influenced by arousal levels21,25,26,27. More specifically, higher arousal levels, associated with fatigue-related or task demands factors, often result in lower microsaccade rates, higher peak velocities and amplitudes, and a steeper main sequence slope between peak velocity and amplitude, underscoring the arousal modulation on microsaccade behavior. Moreover, microsaccades respond transiently to visual stimuli, exhibiting an immediate decrease in occurrence (referred to as microsaccadic inhibition) followed by an increase (referred to as microsaccadic rebound) [e.g.,18,19,26,27]. Research in behaving animals has further suggested that microsaccadic inhibition is primarily mediated by the midbrain superior colliculus (SC) and brainstem, and the frontal eye field is more involved in microsaccadic rebound30,31,32,33,34,35,36,37,38. Although microsaccades have been utilized in clinical studies39,40,41,42,43,44,45, they have yet to be examined in individuals with depression.

Depression is associated with autonomic dysregulation46, often indicative of heightened sympathetic activity, such as reduced heart rate variability (HRV), which is a well-established marker of increased arousal. For example, during heightened sympathetic activation, such as in a fight-or-flight response, HRV decreases as the body shifts to a heightened arousal state. More specifically, studies have demonstrated reduced HRV in depressed individuals47,48, and HRV has been shown to correlate with increased sympathetic tone and reduced parasympathetic control, both of which are indicative of heightened arousal [e.g.,49,50,51].

The goal of this study is to investigate microsaccade behavior in patients with LLD using a high-resolution video-based eye tracker to detect microsaccades during a visual fixation task with varying background luminance to induce microsaccadic inhibition and rebound. We hypothesize that compared to age-matched healthy controls, LLD patients will exhibit heighten arousal levels due to depressive symptoms47,48. Consequently, LLD patients will produce lower microsaccade rates, higher peak velocities and amplitudes, and steeper main sequence slopes. As for microsaccadic inhibition and rebound, these aspects will be explored without specific hypotheses.

Methods and materials

Experimental setup

Data collected from previously published research were used to analyze microsaccade behavior in the current study, where the method regarding task, procedure as well as pupillary results are described in detail52. Briefly, experimental procedures were approved by the Institutional Review Board of the Taipei Medical University, Taiwan, in accordance with the Declaration of Helsinki53. All participants were naïve regarding the purpose of the experiment and provided informed consent with compensation for their participation. Twenty-five LLD patients from Shuang Ho Hospital, and twenty-nine age-matched healthy older adults (referred to hereafter as CTRL) participated in the study. Four LLD and six CTRL participants were excluded from data analysis due to an insufficient number of microsaccades (see Data Analysis below), resulting in a final sample of twenty-one LLD patients (mean age = 72 years, range: 65–81) and twenty-three CTRL participants (mean age = 73 years, range: 65–85) for analysis. LLD inclusion criteria focused on adults aged 60 years or older with a current DSM-5 diagnosis of nonpsychotic unipolar major depressive episode and the first lifetime depressive episode at age 60 or older. Participants were required to be cognitively intact, without a clinical diagnosis of mild cognitive impairment or dementia. To exclude comorbid cognitive disorders, inclusion criteria included scores of the Mini-Mental State Examination (MMSE)54 of 24 or above (for years of education > 6), 21 or above (for between 1 and 6 years of education), and 17 or above (for no education). Other common exclusion criteria included: (1) Current or past diagnoses of other psychiatric disorders, except for depression. (2) History of cognitive disorders, major neurological illnesses, and brain injuries. (3) Physically unstable patients. LLD cases were confirmed with a comprehensive collection of correlated symptoms and signs, instead of structured interviews. Final DSM-5 diagnoses were determined through diagnostic interviews conducted by co-author Y.L., a geriatric psychiatrist, and co-author S.C., a neurologist. As mentioned, age-matched healthy older adults, with no history of major psychiatric disorders or neurological illnesses, were also recruited. These participants were spouses or friends of LLD participants or community members who responded to advertisements. All subjects were assessed using neuropsychological tests to evaluate somatic symptoms (Patient Health Questionnaire, PHQ-15), cognitive status (Montreal Cognitive Assessment, MoCA), and disease severity (Geriatric Depression Scale, GDS-15)55,56,57, with previously validated the Chinese versions of these tests58,59,60. There were no significant differences between the CTRL and LLD groups in terms of age, years of education, or MoCA scores. Clinical data and participant demographics are presented in Table 1. LLD patients did not discontinue their medications for the study, adhering to approved institutional review board ethical guidelines. Because there are no prior studies examining microsaccades in individuals with depression, sample sizes were determined based on previous microsaccade studies in healthy individuals with comparable microsaccade measurements and trial numbers per participant17,20,21,27,29,61,62.

Table 1 Demographics and clinical score of participants.
Full size table

Recording and apparatus

Participants were seated in a dark room, with their head stabilized in a chin and forehead rest. Eye position was measured with a video-based eye tracker (Eyelink-1000 plus binocular-arm, SR Research, Osgoode, ON, Canada) at a rate of 500 Hz with binocular recording, and stimulus presentation and data acquisition were controlled by the Eyelink Experiment Builder. Stimuli were presented on an LCD monitor at a screen resolution of 1920 × 1080 pixels with a 60 Hz refresh rate, subtending a viewing angle of 43° x 24°, with the distance from the eyes to the monitor set at 80 cm.

Interleaved light and darkness reflex task

We used an interleaved light and darkness reflex task27 to compare microsaccade responses between LLD and CTRL. Each trial began with the appearance of a central fixation point (FP) (0.5° diameter, 25 cd/m2) on a gray background (10 cd/m2). After 900–1100 ms of central fixation, background luminance either increased to 15–20 cd/m2 (referred to as Light and White, respectively), decreased to 5–0.1 cd/m2 (referred to as Dark and Black, respectively) (both with 50 and 100% contrast relative to the gray background), or stayed the same (10 cd/m2, referred to as NoChange). Participants were required to maintain steady fixation for an additional 2–2.5 s. Background luminance conditions were randomly interleaved, and each condition had 35 trials.

Data analysis

Following our previous studies27,63,64,65,66, we detected microsaccades using a well-established velocity-based algorithm19,67. Briefly, horizontal and vertical eye position were used, and a velocity threshold was defined flexibly depending on the noise level of each trial (900 ms before to 2000 ms after background luminance change onset) (threshold: 6 median SDs). To reduce the noise in the analyses, we only included microsaccades that also occurred simultaneously in both eyes during at least one data sample (2 ms), and the amplitude was also used as a threshold to exclude macrosaccades, such that we only included microsaccades that were between 0.1 and 2 degrees. Each subject retained at least 10 microsaccades for analysis, 4 LLD and 6 CTRL subjects were thus excluded from the analysis. As shown in Fig. 1B, the identified microsaccades showed the characteristic relation between peak velocity and amplitude (i.e. the main sequence). The amplitude (degree), peak velocity (deg/s), and main sequence slope (peak velocity/magnitude) and intercept of microsaccades were analyzed. Microsaccade rate was first calculated on an individual participant basis (averaged across all trials in each condition), and then rates for the corresponding conditions were averaged across all participants. Similar to previous research [e.g.,19,26,62], the histogram of microsaccades was scaled to a rate-per-second measure, which was computed within a moving window of 100 ms. Similarly, we also calculated temporal dynamics of microsaccade metrics within a moving window of 100 ms. We further analyzed microsaccade orientation (or direction) using a polar histogram to calculate the relative frequency of microsaccades in a given direction. Direction was defined by the angle of the polar coordinates of the microsaccade vector, and we used twenty-four equally spaced directional bins to calculate probability density. Microsaccadic inhibition and rebound are well-documented12,13,14. We therefore analyzed two time windows to characterize these microsaccade responses between LLD and CTRL. The epochs for inhibition and rebound were 60–330 ms and 350–550 ms after background appearance, respectively, to approximately capture the modulation of microsaccadic inhibition and rebound. These epochs were quantitatively similar to our previous research examining human microsaccades following changes in background luminance in healthy young adults64. Furthermore, research has shown differences in microsaccadic rebound, but not inhibition, between positive and negative visual contrast stimuli69. We thus examined microsaccadic inhibition and rebound in the positive and negative background luminance change conditions separately.

Fig. 1
figure 1

Microsaccade dynamics between LLD and CTRL. (A) Dynamics of microsaccade rates following background luminance changes between LLD and CTRL, and mean rates (− 700 to 1700 ms) for each group. (B) Microsaccade main sequence, intercept, and slope between LLD and CTRL. (C) Dynamics of microsaccade peak velocity following background luminance changes between LLD and CTRL, and mean peak velocities (− 700 to 1700 ms) for each group. (D) Dynamics of microsaccade amplitude following background luminance changes between LLD and CTRL, and mean amplitudes (− 700 to 1700 ms) for each group. (E) Microsaccade direction density between LLD and CTRL. In (A), (C) and (D), the shaded colored regions surrounding microsaccadic dynamics curves represent the ± standard error range (across participants) for different groups. In (A–D), the color-filled squares and error-bars represent mean value ± standard error (across participants) for each group, and the small circles represent mean value for each subject. Circle color dots represent each subject data point. In (E), the colored-bars represent ± standard error (across participants) for each angle condition. CTRL: healthy age-matched older adults, LLD: late-life depression patients, Bkgd: background. *Indicates differences are statistically significant.

Full size image

A two-tailed student t-test was performed to compare the general microsaccade differences between the two groups. A mixed ANOVA (2 × 2 ANOVA: between-subjects factor: LDD/CTRL × within-subjects factor: background luminance level 50%/100% contrast) was also performed for statistical analysis. The simple main effect was further used to specifically test our hypothesis that the modulation of microsaccadic inhibition and rebound was different between two groups. Effect sizes (partial eta squared or Cohen’s d), where appropriate, are also reported. Statistical tests were performed using R70,71, JASP72 and MATLAB (The MathWorks Inc., Natrick, MA, USA).

Results

Microsaccade rates and metrics

To first examine microsaccade behavior between LLD (N = 21) and CTRL (N = 23), all conditions were collapsed and we examined microsaccade responses during the period of central fixation in the epoch between 700 ms before and 1700 ms after background luminance change onset. As mentioned, we hypothesized that arousal level would be higher in LLD compared to CTRL, resulting in lower microsaccade rates, larger peak velocities and amplitudes, and steeper main sequence slope. As illustrated in Fig. 1A, while microsaccade rates were lower in LLD compared to CTRL, these differences were not statistically significant (t(42) = 1.506, p = 0.139, d = 0.455). The relationship between microsaccade peak velocity and amplitude was clearly seen in both LLD and CTRL (Fig. 1B), however, differences in intercept and slope of the main sequence were similar between LLD and CTRL (intercept: t(42) = 0.705, p = 0.484, d = 0.213; slope: t(42) = 0.586, p = 0.561, d = 0.177). In contrast, higher peak velocities (Fig. 1C) as well as larger amplitudes (Fig. 1D) were observed in LLD compared to CTRL (Fig. 1C: t(42) = 2.472, p = 0.018, d = 0.746; Fig. 1D: t(42) = 2.577, p = 0.014, d = 0.778). Note that similar effects were observed in microsaccades prior to the background luminance change (500 ms before to the onset of the background luminance change), with lower microsaccade rates, higher peak velocities, and larger amplitudes in LLD compared to CTRL, although only differences in peak velocity were significant (Supplementary Fig. 1). As shown in Fig. 1E, both groups exhibited a typical horizontal bias seen in the literature13. While microsaccade direction density results were not identical between LLD and CTRL, differences were not significant, as indicated by a similar continuous distribution in the Kolmogorov–Smirnov test (K = 0.167, p = 0.861).

Microsaccadic inhibition and rebound after background luminance decreases

To examine whether microsaccade responses evoked by the global luminance changes64,69 are different between LLD and CTRL, we first focused on trials with decreases in background luminance, and the Dark (-50% contrast) and Black (-100% contrast) conditions were collapsed. Consistent with the literature17,20,27,29,64,65,69, microsaccade rates were quickly suppressed after background luminance changes, followed by increased rates of microsaccade generation both in CTRL and LLD (Fig. 2A, B). In the inhibition epoch (Fig. 2C), microsaccade rates were significantly lower when background luminance level was changed compared to it was unchanged (background change: F(1,42) = 10.19, p = 0.003, ηp2 = 0.195). The simple main effects further demonstrated that these differences were only significant in CTRL (p = 0.002), but not in LLD (p = 0.293). All other effects were not significant (p > 0.2). Similarly, in the rebound epoch (Fig. 2D), microsaccade rates were significantly higher when background luminance level was changed compared to when it was unchanged (background change: F(1,42) = 12.369, p = 0.001, ηp2 = 0.227). The simple main effects further demonstrated that these differences were only significant in CTRL (p = 0.04), but not in LLD (p = 0.099). All other effects were not significant (p > 0.2). Following our previous research29, we normalized microsaccade rates (Change conditions subtracted from the NoChange condition). Figure 2E illustrates normalized microsaccade rate, highlighting microsaccadic inhibition and enhancement. Larger suppression in microsaccade occurrence was observed in a higher intensity of background luminance change (Fig. 2F: background luminance intensity: F(1,42) = 6.184, p = 0.017, ηp2 = 0.128). All other effects were not significant (p > 0.19). Background luminance intensity did not modulate microsaccade occurrence in the rebound epoch (Fig. 2G: p > 0.13).

Fig. 2
figure 2

Microsaccadic inhibition and rebound after background luminance decreases between LLD and CTRL. (A) Dynamics of microsaccade rates following background luminance changes in different conditions in CTRL. (B) Dynamics of microsaccade rates following background luminance changes in different conditions in LLD. (C) Mean microsaccade rates in the inhibition epoch (60–330 ms) in the NoChange (no background luminance change) and Change (background luminance changes: Dark and Black) condition between LLD and CTRL. (D) Mean microsaccade rates in the rebound epoch (350 to 550 ms) in the NoChange and Change condition between LLD and CTRL. (E) Normalized microsaccade rate (Change minus NoChange condition) between LLD and CTRL. (F) Mean normalized microsaccade rates in the inhibition epoch in different conditions between LLD and CTRL. (G) Mean normalized microsaccade rates in the rebound epoch in different conditions between LLD and CTRL. In (A), (B) and (E), the shaded colored regions surrounding microsaccade dynamics curves represent the ± standard error range (across participants) for different groups. The gray area represents the epoch selected for analyses. In (C), (D), (F) and (G), the color-filled squares and error-bars represent mean value ± standard error (across participants) for each group, and the small circles represent mean value for each subject. Circle color dots represent each subject data point. CTRL: healthy age-matched older adults. LLD: late-life depression patients. Dark: 50% decrease in contrast relative to the gray background. Black: 100% decrease in contrast relative to the gray background. Bkgd: background. *Indicates differences are statistically significant.

Full size image

Microsaccadic inhibition and rebound after background luminance increases

Similarly, the Light (+ 50% contrast) and White (+ 100% contrast) conditions were first collapsed to examine whether microsaccade responses modulated by increases in background luminance were different between LLD and CTRL. Microsaccade rates were suppressed transiently after background luminance changes, followed by increased rates of microsaccade generation both in CTRL and LLD (Fig. 3A,B). In the inhibition epoch (Fig. 3C), microsaccade rates were significantly lower when background luminance level was changed compared to it was unchanged (background change: F(1,42) = 8.101, p = 0.007, ηp2 = 0.162). The simple main effects further demonstrated that these differences were only significant in CTRL (p = 0.023), but not in LLD (p = 0.138). All other effects were not significant (p > 0.1). While higher microsaccade rates were seen in the rebound epoch (Fig. 3D) in the Change condition compared to the NoChange condition, these effects were not significant (background change: F(1,42) = 0.685, p = 0.422, ηp2 = 0.015), and all other effects were not significant (p > 0.2). Figure 3E illustrates normalized microsaccade rate (Change conditions subtracted from the NoChange condition), showing larger microsaccadic inhibition in the White condition compared to the Light condition (Fig. 3F: background luminance intensity: F(1,42) = 13.93, p < 0.001, ηp2 = 0.249), and all other effects were not significant (p > 0.3). Background luminance intensity did not modulate microsaccade occurrence in the rebound epoch (Fig. 3G: p > 0.2).

Fig. 3
figure 3

Microsaccadic inhibition and rebound after background luminance increases between LLD and CTRL. (A) Dynamics of microsaccade rates following background luminance changes in different conditions in CTRL. (B) Dynamics of microsaccade rates following background luminance changes in different conditions in LLD. (C) Mean microsaccade rates in the inhibition epoch (60–330 ms) in the NoChange (no background luminance change) and Change (background luminance changes: Light and White) condition between LLD and CTRL. (D) Mean microsaccade rates in the rebound epoch (350–550 ms) in the NoChange and Change condition between LLD and CTRL. (E) Normalized microsaccade rate (Change minus NoChange condition) between LLD and CTRL. (F) Mean normalized microsaccade rates in the inhibition epoch in different conditions between LLD and CTRL. (G) Mean normalized microsaccade rates in the rebound epoch in different conditions between LLD and CTRL. In (A), (B) and (E), the shaded colored regions surrounding microsaccadic dynamics curves represent the ± standard error range (across participants) for different groups. The gray area represents the epoch selected for analyses. In (C), (D), (F) and (G), the color-filled squares and error-bars represent mean value ± standard error (across participants) for each group, and the small circles represent mean value for each subject. Circle color dots represent each subject data point. CTRL: healthy age-matched older adults, LLD: late-life depression patients. Light: 50% increase in contrast relative to the gray background. White: 100% increase in contrast relative to the gray background. Bkgd: background. *Indicates differences are statistically significant.

Full size image

Background luminance polarity modulation on microsaccadic inhibition and rebound

Research in behaving monkeys has shown that negative visual contrast stimuli compared to positive visual contrast stimuli induce larger microsaccade responses, particularly during an epoch of microsaccadic rebound69. To examine this polarity effect between LLD and CTRL, we collapsed conditions with different levels of background luminance contrast. As illustrated in Fig. 4A, decreases in microsaccade occurrence (normalized microsaccade rates) were clearly seen in both the negative and positive contrast conditions in both groups. In the inhibition epoch (Fig. 4B), while suppression in microsaccade rates seemed larger in CTRL, these differences were not significant (group: F(1,42) = 1.368, p = 0.249, ηp2 = 0.032). Furthermore, no differences between the negative and positive contrast conditions were noted (visual contrast polarity: F(1,42) = 0.028, p = 0.867, ηp2 = 0.001), and the interaction was not significant (p > 0.3). In the rebound epoch (Fig. 4C), while, as expected, microsaccadic enhancement was larger in the negative condition compared to the positive condition, these effects were not significant (visual contrast polarity: F(1,42) = 2.831, p = 0.100, ηp2 = 0.063). Similarly, though larger polarity effects seemed observed in CTRL compared to LLD, differences were not significant (group: F(1,42) = 2.325, p = 0.135, ηp2 = 0.052), and the interaction was not significant (p > 0.8).

Fig. 4
figure 4

Microsaccadic inhibition and rebound in different polarity conditions between LLD and CTRL. (A) Normalized dynamics of microsaccade rates following background luminance decreases and increases between LLD and CTRL. (B) Mean normalized microsaccade rates in the inhibition epoch in different polarity conditions between LLD and CTRL. (C) Mean normalized microsaccade rates in the rebound epoch in different polarity conditions between LLD and CTRL. In A, the shaded colored regions surrounding microsaccadic dynamics curves represent the ± standard error range (across participants) for different groups. The gray area represents the epoch selected for analyses. In (B) and (C), the color-filled squares and error-bars represent mean value ± standard error (across participants) for each group, and the small circles represent mean value for each subject. Circle color dots represent each subject data point. CTRL: healthy age-matched older adults, LLD: late-life depression patients, Neg: background luminance decrease conditions, Pos: background luminance increase conditions, Bkgd: background.

Full size image

Discussion

To investigate microsaccade behavior in late-life depression (LLD) patients, we used a high-resolution eye tracker to detect microsaccades during a visual fixation task with varying background luminance. Lower microsaccade rates were observed in LLD compared to CTRL, though differences were not significant. No differences in main sequence intercept and slope were obtained between LLD and CTRL. In contrast, higher peak velocities and larger amplitudes were observed in LLD compared to CTRL. Microsaccade rates transiently varied following background luminance changes in both LLD and CTRL, exhibiting signatures of microsaccadic inhibition and rebound. However, rate differences between the change and no-change conditions in microsaccadic inhibition and rebound in the negative background luminance conditions were only significant in CTRL, not in LLD. Background luminance intensity affected microsaccadic inhibition, with larger microsaccade suppression in the high-intensity condition (-100% contrast) compared to the low-intensity condition (-50% contrast). Similar findings were seen in the positive background luminance conditions, though these effects were less pronounced. Furthermore, larger microsaccadic rebound was observed in the negative condition compared to the positive condition, though these effects were not significant. Together, our results demonstrate that microsaccade behavior is altered in LLD compared to CTRL, highlighting the potential of using microsaccades as a behavioral biomarker for LLD.

Altered microsaccade behavior in LLD

While the relationship between depression and arousal state is indeed complex and multifaceted, individuals with depression often exhibit heightened and disrupted arousal levels (e.g., reduced HRV)48. Reduced HRV reflects a fight-or-flight mode in the autonomic nervous system, characterized by increased sympathetic tone and reduced parasympathetic control, correspondingly increasing arousal level51. A reduction in HRV correlates with symptom severity73, and treatment with selective serotonin reuptake inhibitors may potentially normalize some aspects of cardiac dysregulation in patients with depression48. This reduced HRV is also seen in depressed older adults47. Higher arousal levels could correspondingly affect microsaccade behavior. As mentioned, lower microsaccade rates, as well as higher peak velocities and larger amplitudes, are seen in the high arousal condition in healthy adults21,25,26,27. Consistently, we observed lower microsaccade rates in LLD compared to CTRL, though differences were not significant. Moreover, as expected, higher peak velocities and larger amplitudes were obtained in LLD compared to CTRL. Furthermore, microsaccadic inhibition and rebound after background luminance changes were observed, but these effects were only significant in CTRL, not in LLD, suggesting a disruption of microsaccadic inhibition and rebound in LLD. Interestingly, similar to findings in behaving monkeys69, larger microsaccadic rebound was seen in the negative background luminance condition compared to the positive background luminance condition, though differences were not significant. Overall, altered microsaccade behavior was indeed observed in LLD compared to CTRL. What are the neural mechanisms underlying the alteration in microsaccade behavior in LLD?

Neural mechanisms underlying microsaccade behavior in LLD

Research primarily from behaving monkeys has demonstrated a network of brain structures centrally involved in the control of microsaccade generation30,31,32,33,34,35,36,37,38,64,65, including the superior colliculus (SC), brainstem, frontal eye fields, and cerebellum. Among these regions, the SC is causally involved in microsaccade generation30,31. The SC receives important signals from various brain areas74,75,76,77,78,79,80,81,82,83, including the locus coeruleus (LC)82,83. The LC known to modulate global arousal levels and brain states releases noradrenaline to a wide range of brain structures84,85,86,87,88,89,90,91,92. The connection between the LC and SC likely plays a significant role in modulating the microsaccade effects observed in the current study. Research has shown that the LC is affected in LLD93,94,95, and this disruption could, in turn, affect the projections from the LC to the SC, resulting in the altered microsaccade behavior observed in LLD patients.

Limitations and future directions

The current study is limited by the characterization of disrupted microsaccade behavior in a small study cohort. Future work with larger cohorts is certainty needed to confirm these microsaccade effects in LLD. Additionally, we relied on a comprehensive collection of correlated symptoms and signs, rather than structured interviews, to confirm LLD cases, and the absence of a structured clinical interview is a notable limitation. Moreover, the time windows for analysis were selected based on previous work. It is evident that the time course of microsaccade response dynamics between LLD and CTRL were not identical, and the response dynamics differed across the two intensity conditions. Future work using a different approach to determine the time window of microsaccadic inhibition and rebound across various conditions and groups is needed to examine these effects in LLD. As noted, LLD patients recruited for the study were not discontinued from their medications. While the effects of pharmacological treatment on microsaccade behavior have yet to be systematically examined, it is possible that medication could influence microsaccade responses. To address this, we performed an ANCOVA, using medication type (serotonin/noradrenaline reuptake inhibitors, selective serotonin reuptake inhibitors, norepinephrine-dopamine reuptake inhibitors, tricyclic antidepressants, and monoamine oxidase inhibitors) as a covariate in the analyses of microsaccade peak velocity and amplitude. The results remained consistent, with significant differences observed in peak velocity (F(1,37) = 6.106, p = 0.018) and amplitude (F(1,37) = 6.493, p = 0.015), even after accounting for medication type. Future studies should systematically investigate the effects of pharmacological treatments on microsaccade behavior to better understand their potential influence.

Conclusions

Eye-tracking is a powerful and cost-effective means for understanding brain functions8,9,10,11. While microsaccades have been extensively studied behaviorally and neurophysiologically over the past few decades, their clinical investigation remains very limited. This limitation may be due to the demand for high-resolution eye trackers to measure eye position both temporally and spatially to accurately detect microsaccades40. Nevertheless, compared with high-cost neuroimaging methods, microsaccades remain an easy-to-measure technique that can assist in the objective diagnosis of LLD and other clinical disorders, as well as in evaluating the effectiveness of treatment outcomes.