INTRODUCTION
Autism spectrum disorder (ASD) is a complex neurodevelopmental condition characterized by persistent deficits in social communication, restricted, repetitive behaviors, and atypical patterns of social attention and emotional face processing [1]. First defined as a distinct condition in the Diagnostic and Statistical Manual of Mental Disorders (DSM-III) in the 1980s, ASD has since become a focal point of cognitive, and behavioral research. The prevalence of ASD has risen dramatically over the past decades, currently affecting approximately 1% of the population globally [2]. ASD is characterized by challenges in social communication and shows strong familial aggregation [3-6]. This pattern suggests that certain cognitive and behavioral traits associated with ASD may be present, in milder or different forms, across the family spectrum. Investigating these traits in unaffected parents can help illuminate the nature of the familial susceptibility. In line with this goal, research has increasingly turned to first-degree relatives to identify intermediate or familial traits that may lie on the spectrum of ASD-related characteristics.
Aberrant eye gaze mechanisms have been consistently documented in individuals with ASD [7]. These difficulties are thought to stem from atypical patterns of visual attention to faces and a reduced reliance on information from the eye region when interpreting emotional expressions [8]. Research suggests that parents of individuals with ASD may exhibit similar patterns of atypical visual attention. Emotional stimuli naturally modulate visual attention processes [9]. However, the characteristics of visual attention to emotional faces in relatives of ASD individuals remain unclear, especially in culturally distinct populations such as Chinese families, where social norms and familial roles may influence social attention and perceptual processes differently [5,10].
Eye-tracking technology provides an objective and sensitive method for investigating visual attention to emotional faces. By analyzing gaze behavior, this method reveals patterns of attention allocation that reflect underlying cognitive mechanisms. Prior studies have shown that individuals with ASD often exhibit atypical gaze patterns, such as reduced fixation on the eyes and increased attention to the mouth, particularly when processing emotionally charged expressions like anger or fear [11-13]. While these findings provide valuable insights, research on the gaze behavior of parents of children with ASD remains limited, particularly in non-Western contexts.
This study seeks to address these gaps by investigating the gaze behavior of Chinese parents of children with ASD, a culturally distinct population that has been underrepresented in prior research. Using a validated eye-tracking paradigm, we focused on two key areas of interest (AOIs): the eyes and the mouth, which are critical for visual processing of emotional facial expressions and social communication. Emotional facial expressions are the cornerstone of non-verbal communication, and experiencing emotions involves different components, including physiological responses, mental states, and explicit behavior [14]. Specific emotional experiences (e.g., positive or negative states) can alter our cognitive function by changing decision-making, attention, learning, memory deletion, and perceptual processes [15-18]. In this study, facial emotional expressions (happiness, anger, sadness, and fear) are employed as stimuli to investigate patterns of gaze allocation, thereby elucidating the relationship between emotional expressions and visual attention within this population.
We hypothesize that parents of children with ASD will exhibit distinct gaze behaviors compared to parents of typically developing children (TDC). Specifically, we predict reduced fixation on the eyes and increased attention to the mouth, particularly when processing negative emotions such as anger and fear. And analyses demographic, psychometric and eye-tracking data respectively, this study aims to advance the understanding of cognitive and emotional correlates of parents of children with ASD. While preliminary and not directly testing an intervention, our findings offer a potential avenue for future applied work. Future intervention research could consider whether training more typical scanning strategies improves emotion perception and, subsequently, aspects of caregiver coping or family dynamics.
METHODS
Participants
We recruited 35 parents of children with ASD (11 males, 24 females) from autism rehabilitation centers as the experimental group. To be eligible, participants had to meet the following criteria: 1) their children had a confirmed ASD diagnosis based on the Diagnostic and Statistical Manual of Mental Disorders, 5th Edition (DSM-5) [1], validated by at least two experienced child psychiatrists, and 2) they reported no diagnosis of ASD, other neurodevelopmental disorders, or major psychiatric conditions. No structured clinical interview was administered to parents in this study. Subclinical autistic traits were assessed using the Broad Autism Phenotype Questionnaire, which was administered as a standardized self-report measure [19].
The control group consisted of 35 parents of TDC (12 males, 23 females), recruited from local primary schools. These participants were matched with the experimental group for age, gender, and educational level (Table 1). Two psychiatrists independently assessed the control group participants and their children, confirming that none met the diagnostic criteria for ASD or other psychiatric disorders.
All participants, regardless of group, were required to be right-handed, have no history of neuropsychiatric disorders, head injury, or substance abuse, and have no abnormalities in vision, hearing, or language abilities. The study protocol was approved by the Institutional Review Board (IRB) of Bengbu Medical University (IRB No. 2019-061). All participants voluntarily enrolled after providing written informed consent, adhering to the Declaration of Helsinki guidelines.
Materials
Neuropsychological assessments
To assess participants’ cognitive function, this study employed three standardized neuropsychological tests [20], including the Digit Span Test (Forward, Backward), Trail Making Test (TMT, Parts A and B), and Stroop Test (Color, Word, and Disturbance Tests). These tests assessed core domains of attention, processing speed, working memory, and executive function.
Digit Span Test (Forward and Backward)
Derived from the Wechsler Adult Intelligence Scale, this test includes two distinct tasks administered in a fixed order: Digit Span (DS) Forward (repetition of number sequences in the same order) and DS Backward (repetition in reverse order). The total score (sum of forward and backward spans) was used as a composite measure of auditory-verbal working memory. A higher score indicates better working memory.
Trail Making Test (Parts A and B)
TMT consists of two parts, A and B. In Part A, participants connect numbered circles in ascending order, whereas in Part B they alternate between numbers and letters in sequence. The primary outcome was completion time (seconds), with longer times indicating poorer performance. Part A primarily reflects processing speed and visuomotor tracking, while Part B additionally assesses cognitive flexibility and task switching.
Stroop Color and Word Interference Test
This test is an adaptation of the Victorian version and includes three components: Stroop Color, Stroop Word, and Stroop Interference. Participants completed three sequential conditions: 1) Color Naming (control), in which participants named the color of solid dots; 2) Word Reading (control), in which participants read color words printed in black ink; and 3) Interference (experimental), in which participants named the ink color of incongruent color words. This condition evaluates inhibitory control, with shorter completion times indicating better performance.
Eye-tracking apparatus
Eye-tracking data were collected using a remote infrared eye-tracking device (iView XTM RED; Senso Motoric Instruments), consisting of an operating computer and a monitoring computer. Stimuli were presented on a 22-inch Dell monitor with a resolution of 1,920×1,080 pixels. The eye-tracker recorded data at a sampling rate of 250 Hz, with gaze accuracy better than 0.4° of visual angle and pupil-tracking resolution of less than 0.03°. The high-resolution sensor allowed participants to move their heads freely within a wide operating range. The infrared detector was fixed below the display screen, eliminating the need for head fixation.
Eye-tracking data were processed using BeGaze software (Gaze Intelligence), which facilitated delineation of AOIs and extraction of key eye-movement metrics, including fixation duration, fixation count, entry time, and first fixation duration.
Design and procedure
The experiment utilized a free-viewing paradigm in which participants were randomly presented with 24 monochrome photographs of faces. The stimulus set comprised six distinct identities per emotion (happy, angry, sad, and fearful), balanced across two genders and three age groups. All stimuli were adapted from a validated database of emotional expressions. To minimize background distractions, images were cropped to include only the facial region above the neck (Figure 1).
Participants completed the task individually in a quiet, dimly lit room. They were seated approximately 60-80 cm from the display screen, with the physical position and angle verified using the RED Tracking Monitor calibration display to ensure steady binocular alignment. A five-point calibration followed by a four-point validation was conducted before the experiment. Calibration was deemed successful if the deviation error for all points was less than 1 on both the X and Y axes. Participants whose calibration failed were excluded from further analysis.
The experiment began with a fixation cross displayed at the center of the screen for 2,000 ms to standardize initial gaze position. Following the disappearance of the fixation cross, four emotional photographs were presented in random order, each for 4,000 ms. Participants were instructed to view the images freely. The entire task lasted approximately 20 seconds per participant, excluding calibration and instructions. The trial sequence ensured counterbalancing to minimize order effects.
Data analysis
To examine the allocation of visual attention to different facial regions, we manually delineated two AOIs: the eyes and the mouth (Figure 2). For each target face, AOIs were individually defined using rectangular bounding boxes. The eye AOI included both eyes and eyebrows, extending vertically from the upper boundary of the eyebrows to the lower eyelid margin and horizontally from the outer canthus of one eye to that of the other. The mouth AOI included the full lip contour, extending vertically from the upper vermilion border to the lower lip margin and horizontally to encompass the corners of the mouth [21]. Following data export, quantitative analyses were performed based on these two AOIs—the “eye AOI” and the “mouth AOI”—to compare attentional allocation to socially salient facial regions. Four widely adopted eye-tracking metrics were extracted to assess different aspects of visual attention: 1) entry time, defined as the latency from stimulus onset to the first fixation within an AOI; 2) first fixation duration, defined as the duration of the first fixation within an AOI; 3) fixation count, defined as the total number of fixations within an AOI; and 4) fixation time, defined as the sum of all fixation durations within an AOI. Demographic data were analyzed using independent sample t-tests and chi-square tests to assess group differences.
For eye-tracking data, a 2 (Group: ASD parents, TDC parents) ×2 (AOI: eyes, mouth)×4 (Emotion: happy, angry, sad, fearful) repeated measures analysis of variance (ANOVA) was conducted to evaluate differences across groups, AOIs, and emotions. For repeated-measures ANOVAs, main effects and interactions were first tested. If a significant two-way interaction (e.g., Emotion×AOI) was detected, post-hoc simple effect analyses were conducted to decompose the interaction-specifically, and Bonferroni correction was applied. All statistical tests were two-tailed, with p<0.05 considered significant after correction.
RESULTS
Neuropsychological background measures
There was no significant difference between parents of children with ASD and parents of TDC in age, gender, years of education (p>0.05). The results of the DS Forward test showed that parents of children with ASD scored lower than parents of TDC on the DS Forward test (p<0.05). The TMT B and STROOP 1 test results showed that parents of children with ASD scored higher than parents of TDC (p<0.05) (Table 1).
Entry time
To explore the differences in entry time between groups, AOIs, and emotional types, a 2 (Group: ASD parents, TDC parents)×2 (AOI: mouth, eyes)×4 (Emotion: happy, angry, sad, and fearful) repeated measures ANOVA was conducted. Descriptive statistical results for the comparison of entry time by different emotion types and AOIs are shown in Table 2. Figure 3 illustrates the entry time differences across conditions.
The analysis revealed no significant group main effect (F(1, 68)=0.169, p=0.681, ηp2=0.001, 95% confidence interval [CI]=[0.000, 0.032]). However, descriptively, the total entry time for the TDC parents (mean [M]=712.092 ms, standard deviation [SD]=35.914 ms) was slightly longer than for the ASD parents (M=692.112 ms, SD=32.718 ms). A significant main effect of AOI was observed (F(1, 68)=208.416, p<0.001, ηp2=0.384, 95% CI=[0.265, 0.491]): across all conditions, participants spent significantly longer entry times for the mouth region (M=1,052.787 ms, SD=41.114 ms) compared to the eye region (M=351.416 ms, SD=25.884 ms). Emotional type also significantly influenced entry time (F(3, 204)=4.174, p=0.006, ηp2=0.012, 95% CI=[0.002, 0.065]), with the longest entry time observed for anger (M=792.253 ms, SD=41.125 ms) and shorter durations for other emotions.
The interaction between group and emotional type was not significant (F(3, 204)=2.246, p=0.081, ηp2=0.007, 95% CI=[0.000, 0.053]). A significant interaction was found between group and AOI (F(1, 68)=4.478, p=0.035, ηp2=0.013, 95% CI=[0.001, 0.078]), and between emotion type and AOI (F(3, 204)=20.726, p<0.001, ηp2=0.058, 95% CI=[0.023, 0.118]). These interactions suggest that group-related differences, where present, were conditional on the specific AOI. Post hoc simple-effects analyses were conducted and are reported below.
Within the mouth AOI, ASD parents had significantly longer entry time than TDC parents (t(68)=2.31, p=0.023, ηp2=0.072, 95% CI=[0.005, 0.189]), with the significant differences observed for anger (ASD: M=1,124.56 ms, SD=48.32 ms; TDC: M=987.89 ms, SD=45.17 ms; t(68)=2.87, p=0.005) and fear (ASD: M=1,068.42 ms, SD=46.25 ms; TDC: M=992.17 ms, SD=43.81 ms; t(68)=2.15, p=0.035).
Within the eye AOI, the group difference in entry time was not significant (t(68)=1.24, p=0.218, ηp2=0.022).
Across all emotional types, entry time for the mouth region was significantly longer than for the eye region (all p<0.001).
First fixation duration
The first fixation duration was analyzed using the same repeated measures ANOVA design. Descriptive statistical results for the comparison of first fixation duration by different emotion types and AOIs are shown in Table 3 and Figure 4.
The results showed no significant main effect for group (F(1, 68)=1.454, p=0.228, ηp2=0.003, 95% CI=[0.000, 0.048]). Descriptively, TDC parents exhibited slightly longer first fixation durations (M=219.474 ms, SD=8.959 ms) compared to ASD parents (M=204.303 ms, SD=8.834 ms). A significant AOI main effect was observed (F(1, 68)=50.133, p<0.001, ηp2=0.082, 95% CI=[0.025, 0.183]), with first fixation durations were significantly longer in the eye region (M=256.430 ms, SD=8.896 ms) compared to the mouth region (M=167.347 ms, SD=8.896 ms). The main effect of emotion type was not significant (F(3, 204)=2.234, p=0.082, ηp2=0.004, 95% CI=[0.000, 0.052]). Although happiness had the numerically longest fixation duration (M=227.223 ms, SD=11.567 ms), this difference did not reach statistical significance.
A significant interaction was found between group and AOI (F(1, 68)=4.591, p=0.033, ηp2=0.008, 95% CI=[0.001, 0.069]). Post hoc simple effects analysis results are as follows.
Within the eye AOI, TDC parents had significantly longer first fixation durations than ASD parents (t(68)=2.42, p=0.018, ηp2=0.078, 95% CI=[0.007, 0.196]), with a prominent difference for happiness (TDC: M=279.68 ms, SD=10.24 ms; ASD: M=248.76 ms, SD=9.89 ms; t(68)=2.64, p=0.010).
Within the mouth AOI, no significant group difference was observed (t(68)=0.98, p=0.331, ηp2=0.014).
Across all emotions, first fixation durations were consistently longer in the eye region than in the mouth region (all p<0.001).
Fixation count
We conducted an additional repeated measures ANOVA to examine fixation counts across groups, AOIs, and emotional types. Descriptive statistical results for the comparison of fixation count by different emotion types and AOIs are shown in Table 4 and Figure 5.
The results indicated no significant group main effect (F(1, 68)=3.138, p=0.077, ηp2=0.006, 95% CI=[0.000, 0.061]). Descriptively, however, the ASD parents had slightly more fixations (M=4.701, SD=0.145) compared to the TDC parents (M=4.335, SD=0.147). A significant AOI main effect was observed (F(1, 68)=715.560, p<0.001, ηp2=0.559, 95% CI=[0.448, 0.645]): the number of fixations in the eye region (M=7.279, SD=0.146) was significantly higher than in the mouth region (M=1.756, SD=0.146). The main effect of emotion type was not significant (F(3, 204)=2.012, p=0.110, ηp2=0.004, 95% CI=[0.000, 0.050]). Although anger showed the numerically highest fixation count (M=4.703, SD=0.135), the difference was not significant.
A significant interaction was found between AOI and emotion type (F(3, 204)=25.676, p<0.001, ηp2=0.044, 95% CI=[0.016, 0.098]). Post hoc simple effects analysis results are as follows.
In the mouth region, fixation counts for anger (M=2.14, SD=0.18) and sadness (M=2.07, SD=0.17) were significantly higher than for happiness (M=1.52, SD=0.16) and fear (M=1.69, SD=0.15) (all p<0.001); no significant group difference was found (p=0.284).
In the eye region, anger elicited the highest number of fixations (M=7.86, SD=0.19) compared to other emotions (all p<0.001); no significant group difference was observed (p=0.215).
Fixation time
Fixation time was also analyzed using repeated measures ANOVA. Descriptive statistical results for the comparison of fixation time by different emotion types and AOIs are shown in Table 5 and Figure 6.
The results demonstrated no significant main effect for group (F(1, 68)=0.072, p=0.788, ηp2=0.000, 95% CI=[0.000, 0.029]). Descriptively, TDC parents showed shorter fixation times (M=1,109.200 ms, SD=37.339 ms) compared to ASD parents (M=1,123.309 ms, SD=36.816 ms). A significant AOI main effect was found (F(1, 68)=649.203, p<0.001, ηp2=0.535, 95% CI=[0.420, 0.621]), with fixation times significantly longer in the eye region (M=1,784.285 ms, SD=37.078 ms) than in the mouth region (M=448.223 ms, SD=37.078 ms). The main effect of emotion type was not significant (F(3, 204)=0.347, p=0.791, ηp2=0.001, 95% CI=[0.000, 0.024]). Although anger showed the numerically longest fixation times (M=1,135.947 ms, SD=32.058 ms), the main effect of emotion was not significant.
A significant interaction was found between AOI and emotion type (F(3, 204)=26.160, p<0.001, ηp2=0.044, 95% CI=[0.016, 0.098]). Post hoc simple effects analysis results are as follows.
Simple-effects analyses revealed that, across all emotional types, fixation times were consistently longer in the eye region than in the mouth region (all p<0.001).
In the eye region, fixation times for anger (M=1,856.32 ms, SD=39.45 ms) were significantly longer than those for the other emotions (all p<0.001). No significant group differences were found in either AOI (mouth: p=0.276; eye: p=0.342). In the eye region, fixation times for fear (M=1,812.47 ms, SD=38.76 ms) were significantly longer than those for happiness and sadness (all p<0.01). No significant group difference was observed (p=0.415).
DISCUSSION
This study utilized eye-tracking technology to explore differences in gaze behavior during emotional face processing between parents of children with ASD and parents of TDC. Specifically, we investigated visual attention patterns across key facial AOIs (eyes and mouth) during the observation of four basic emotional expressions (happy, angry, sad, and fearful). Our findings provide valuable insights into the gaze behavior among parents of children with ASD, highlighting its cognitive and emotional dimensions in non-Western cultural contexts.
Our results revealed nuanced patterns of visual attention allocation shaped by the interaction of group, AOI, and emotion type—rather than robust overall group differences. No significant group main effects were observed for any gaze metrics (entry time, first fixation duration, fixation count, and fixation time). Instead, significant Group×AOI interactions emerged for entry time and first fixation duration, indicating that group differences are conditional on specific facial regions.
Notably, within the mouth AOI, parents of children with ASD exhibited significantly longer entry time than TDC parents, with this difference being more pronounced for angry and fearful. In contrast, within the eye AOI, no significant group differences were found for entry time. This pattern is consistent with previous findings suggesting that individuals with ASD and their relatives may allocate attention differently across facial regions [22], including increased attention to the mouth during emotion processing [23-25]. The more pronounced differences for angry and fearful expressions suggest that emotion type may modulate region-specific attentional allocation in parents of children with ASD [26].
Conversely, for first fixation duration, TDC parents showed significantly longer durations in the eye region compared to ASD parents, with the most prominent difference observed for happy expressions [27,28]. This finding indicates differential initial fixation dynamics within the eye region as a function of group and emotion type, rather than generalized differences in overall gaze behavior [29,30].
Neuropsychological assessments further contextualized these findings. While the Stroop test results indicated comparable general cognitive functioning between groups, ASD parents exhibited slight deficits in short-term or working memory, as reflected by lower scores on the DS. Additionally, TMT test results suggested slower information processing in ASD parents, despite normal executive functioning. These findings are consistent with previous studies that first-degree relatives of individuals with ASD have subtle cognitive and emotional processing challenges [5,9].
The observed gaze patterns, derived from interaction effects, indicate region-specific differences in attentional allocation, with variations in the eye and mouth regions depending on the type of emotion displayed. While these findings are consistent with previous reports of region-specific gaze differences in ASD populations, particularly for negative emotions such as anger and fear [31-33], no overall group main effects were observed. Therefore, the present results should be interpreted as conditional distribution patterns rather than generalized differences in overall gaze behavior [34].
This study offers several important contributions to ASD research. The use of eye-tracking technology provided objective and robust measures of visual attention. Furthermore, the inclusion of neuropsychological assessments enriched the interpretation of the findings. By examining the gaze behavior of parents of children with ASD, this study also explored their related cognitive and emotional characteristics.
Several limitations warrant consideration. First, the relatively small sample size may have limited the statistical power to detect subtle group differences. Future studies with larger, more diverse samples are needed to validate these findings and explore potential moderating factors, such as gender, age, and socioeconomic status. Second, the use of static facial stimuli, while effective, but it has limited ecological validity and may not represent real-life social situations well. In particular, the findings may not fully generalize to the fluid, contingent, and emotionally rich context of naturalistic parent-child interactions. Dynamic stimuli, such as video clips, better mimic real-life emotional interactions and provide richer sensory information, enhancing the study of emotion recognition. Third, while ASD parents exhibited a divergent gaze pattern, this study was not designed to conclusively link this pattern to impaired behavioral recognition accuracy. Future studies should combine eye-tracking with explicit behavioral emotion-recognition tasks to more directly test the link between attentional patterns and recognition accuracy. Finally, the study examined only basic emotions. Expanding the range of emotional stimuli to include more nuanced or culturally specific expressions could provide a deeper understanding of emotion recognition in populations with ASD and their relatives.
Conclusion
This study provides important insights into gaze behavior during emotional face processing in parents of children with ASD. It enhances our understanding of their eye-tracking characteristics as well as the associated cognitive and emotional characteristics, as revealed through neuropsychological and eye-tracking assessments. Future studies should build on these findings, combining dynamic stimulation and neuroimaging techniques to further elucidate the characteristics of social communication and cognitive behavior in such populations and provides direction for targeted intervention research. These efforts will enhance our ability to support families of individuals with ASD, improving their social and emotional well-being.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Print
