Gaze perception is a foundational social skill. Here, we investigated how head and eye cue integration in gaze perception changes across development. Across five experiments involving 195 Japanese children (ages 4–16) and 126 adults (ages 18–58), we tested eye gaze perception using both Wollaston illusion images, where eye regions remain identical across head orientations, and Normal images with naturally varying eye regions. We found that the attractive influence of head orientation, whereby perceived gaze is biased toward the head direction, decreased from early childhood to adolescence. Notably, children aged 10–16 years did not show the attractive effect of head orientation characteristic of the Wollaston illusion. Adults consistently showed the illusion as expected. These findings highlight adolescence as a critical transitional period. A follow-up experiment with 7–9-year-old children showed that perceived gaze direction was more strongly biased toward head orientation in smaller images where iris and pupil positional details are less clearly visible, suggesting that the influence of head orientation is flexibly modulated by the clarity of eye region as well as age. The findings provide new insights into the dynamic development of social cue integration and perceptual decision-making across childhood, adolescence, and adulthood.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-34625-9.

A well-known finding from research on gaze perception in triadic gaze tasks is the overestimation of horizontal gaze directions. In general, a looker model's gaze appears to deviate more from the straight line of sight than is objectively the case. Although there is, up to now, a substantial amount of evidence for what Anstis et al. (1969) termed the overestimation effect, results vary regarding the absolute overestimation factor. Starting from the occlusion hypothesis by Anstis et al. (1969), the present study examines the influence of horizontal iris movement range, operationalized as the sclera size index on overestimation factors acquired for a sample of 40 looker models. The study rendered two main findings. First, horizontal iris movement range (sclera size index: M = 2.02, SD = 0.11, min = 1.79, max = 2.25) proved not useful for the explanation of variance in the overestimation factors (M = 1.79, SD = 0.16, min = 1.49, max = 2.24) obtained separately for each of the looker models. Second, intraclass correlations revealed that variance in perceived gaze directions between observers was roughly 10 times larger (ICC = 0.189) than variance between looker models (ICC = 0.019). The results strongly emphasize the need for larger and more diverse observer samples and may serve as a post hoc justification for using only a few or no different looker models in triadic gaze judgment tasks.

The gaze of other people is of interest to human observers, particularly in cases of direct gaze, that is, when it targets the observer. Gaze direction research has successfully clarified some of the mechanisms underlying gaze perception, but little is known about the active perception of direct gaze. Three eye-tracking experiments were conducted in which fixations and scan paths were recorded during the task to judge direct gaze. Somewhat surprisingly, judgments were issued after a single eye fixation only in a minority of trials. In most cases, observers fixated both eyes of a looker model, sometimes even scanning them repeatedly. Fixation duration showed a consistent pattern, where first fixations were longer when the task response followed immediately, and second fixations were shorter just before the response. A direct-gaze bias was tested but was not found: visiting the second eye was even more likely when the first fixation was on a straight-gazing rather than an averted eye. There was no systematic pattern in the final fixation, contradicting the expectation that it would fall on the abducting (leading) eye. It is argued that overt looking behavior during direct gaze judgments reflects a cumulative decision process that spans over consecutive fixations. Several factors may contribute to the high incidence of multiple-eye scans, including vergence and angle kappa. Vergence, in particular, is considered an important candidate, because the depth of fixation is ambiguous when only one eye is visible, but can be limited by probing the gaze direction of both eyes.

Social cognitive impairments were well-documented in schizophrenia and individuals with clinical high risk for psychosis (CHR). While eye movement abnormalities during gaze perception in schizophrenia have been suggested, such understanding in individuals with CHR is limited. This study recruited 36 CHR individuals and 50 healthy controls (HC) to perform a gaze perception task with eye-tracking techniques analyzed with hidden Markov models for predictability and pattern recognition. Eye movement variables and behavioral responses, self-referential gaze perception (SRGP) rates, were compared between groups and examined with discriminant analyses. Associations between eye movements and SRGP rates with multiple symptom dimensions and social functioning were examined. Results indicated that CHR group displayed nose-focused (vs eye-focused) and erratic eye movement patterns compared to HC. The combination of behavioral and eye movement variables exhibited a discriminatory ability of 0.893 area under the curve (AUC) in classifying CHR and HC, significantly outperforming models using either SRGP rates (AUC = 0.824) or eye movement variables (AUC = 0.781) alone. The combined model achieved high sensitivity (0.861) and specificity (0.820), with eyes-nose scale, fixation duration, and ambiguous SRGP rate emerging as the most discriminative features. Self-referential gaze perception rates were primarily associated with schizotypy and social anxiety, while eye movement patterns were mainly associated with delusional ideations and social functioning. This study comprehensively examined eye movement patterns during gaze perception in CHR and their associations with clinical manifestations, suggesting the potential of visual attention patterns as both diagnostic biomarkers and therapeutic targets in the CHR population.

Perceiving social signals: The similarity of direct gazing and pointing.

The role of task-irrelevant configural and featural face processing in the selective attention to gaze.

This study examined the perceptual dynamics that influence the evaluation of eye contact as a communicative display. Participants (n = 137) completed a task where they decided if agents were inspecting or requesting one of three objects. Each agent shifted its gaze three times per trial, with the presence, frequency and sequence of eye contact displays manipulated across six conditions. We found significant differences between all gaze conditions. Participants were most likely, and fastest, to perceive a request when eye contact occurred between two averted gaze shifts towards the same object. Findings suggest that the relative temporal context of eye contact and averted gaze, rather than eye contact frequency or recency, shapes its communicative potency. Commensurate effects were observed when participants completed the task with agents that appeared as humans or a humanoid robot, indicating that gaze evaluations are broadly tuned across a range of social stimuli. Our findings advance the field of gaze perception research beyond paradigms that examine singular, salient and static gaze cues and inform how signals of communicative intent can be optimally engineered in the gaze behaviours of artificial agents (e.g. robots) to promote natural and intuitive social interactions.

Human gaze-based dual teacher guidance learning for semi-supervised medical image segmentation.

Heartfelt gaze: Cardiac afferent signals and vagal tone affect gaze perception.

Examining the joint and unique contributions of contour integration and visuospatial memory to altered eye gaze perception in schizophrenia.

A widely known result from gaze perception research is the overestimation effect where gaze direction-or more precisely gaze endpoints-is seen farther to the side than they actually are. A common gain factor reported in the literature is 1.5, that is, an overestimation of gaze endpoint by 50%. Gaze endpoint, however, must be a joint function of gaze angle and distance. Results from data collected between 2022 and 2024 show that a strong overestimation for photographed models at short distances turns into almost perfect perception at larger distances. This was equally true when gazing was done with the eyes only (head straight relative to observer) and with the head only (eyes straight relative to head). A new method measures gaze angle by triangulation from fixation points at varying distances and separates two components: (a) a slope and (b) an intercept. This triangulation indicates that the overestimation of gaze angle (slope) is very moderate and that the strong effects in gaze endpoints are mainly due to the intercept. Further experiments indicate that the intercept effects are confined to two-dimensional pictures of lookers and are not observed in physical three-dimensional lookers. The results are interpreted with reference to the distinction between picture space and physical space. Moreover, the present results do not fully comply with the classic partial-occlusion explanation for the overestimation effect. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Individuals with schizophrenia (SZ) and bipolar disorder (BD) show disruptions in self-referential gaze perception-a social perceptual process related to symptoms and functioning. However, our current mechanistic understanding of these dysfunctions and relationships is imprecise. The present study used mathematical modeling to uncover cognitive processes driving gaze perception abnormalities in SZ and BD, and how they relate to cognition, symptoms, and social functioning. We modeled the behavior of 28 SZ, 38 BD, and 34 controls (HC) in a self-referential gaze perception task using drift-diffusion models parameterized to index key cognitive components: drift rate (evidence accumulation efficiency), drift bias (perceptual bias), start point (expectation bias), threshold separation (response caution), and nondecision time (encoding/motor processes). Results revealed that aberrant gaze perception in SZ and BD was driven by less efficient evidence accumulation, perceptual biases predisposing self-referential responses, and greater caution (SZ only). Across SZ and HC, poorer social functioning was related to greater expectation biases. Within SZ, perceptual and expectancy biases were associated with hallucination and delusion severity, respectively. These findings indicate that diminished evidence accumulation and perceptual biases may underlie altered gaze perception in patients and that SZ may engage in compensatory cautiousness, sacrificing response speed to preserve accuracy. Moreover, biases at the belief and perceptual levels may relate to symptoms and functioning. Computational modeling can, therefore, be used to achieve a more nuanced, cognitive process-level understanding of the mechanisms of social cognitive difficulties, including gaze perception, in individuals with SZ and BD.

Non-verbal communication, especially eye contact, plays a crucial role in human interaction. Integrating eye contact capabilities into robotic behavior presents challenges that require comprehensive interdisciplinary collaboration from the fields of vision science and robotics. This study introduces an innovative methodology to improve the user's perception of the tabletop social robot Haru's direct gaze by explaining how eye and head movements can be managed through simulating visual perception processing. By considering the mechanisms of eye and head movements during visual perception processing, the robot's eye and head gaze behaviors are intended to replicate human gaze movements with Vestibulo-Ocular Movement (VOM) during face-to-face interaction, ultimately enhancing the user's perception of the robot's direct gaze.

Individuals with schizophrenia (SZ) and bipolar disorder (BD) show disruptions in self-referential gaze perception-a social perceptual process related to symptoms and functioning. However, our current mechanistic understanding of these dysfunctions and relationships is imprecise. The present study used mathematical modeling to uncover cognitive processes driving gaze perception abnormalities in SZ and BD, and how they relate to cognition, symptoms, and social functioning. We modeled the behavior of 28 SZ, 38 BD, and 34 controls (HC) in a self-referential gaze perception task using drift-diffusion models (DDM) parameterized to index key cognitive components: drift rate (evidence accumulation efficiency), drift bias (perceptual bias), start point (expectation bias), threshold separation (response caution), and non- decision time (encoding/motor processes). Results revealed that aberrant gaze perception in SZ and BD was driven by less efficient evidence accumulation, perceptual biases predisposing self-referential responses, and greater caution (SZ only). Across SZ and HC, poorer social functioning was related to greater expectation biases. Within SZ, perceptual and expectancy biases were associated with hallucination and delusion severity, respectively. These findings indicate that diminished evidence accumulation and perceptual biases may underlie altered gaze perception in patients and that SZ may engage in compensatory cautiousness, sacrificing response speed to preserve accuracy. Moreover, biases at the belief and perceptual levels may relate to symptoms and functioning. Computational modeling can, therefore, be used to achieve a more nuanced, cognitive process-level understanding of the mechanisms of social cognitive difficulties, including gaze perception, in individuals with SZ and BD.

Mentalizing, or theory of mind (ToM), impairments and self-referential hypermentalizing bias are well-evident in schizophrenia. However, findings compared to individuals with at-risk mental states (ARMS) are inconsistent, and investigations into the relationship between social cognitive impairments and social anxiety in the two populations are scarce. This study aimed to examine and compare these deficits in first-episode schizophrenia-spectrum disorder (FES) and ARMS, and to explore potential specific associations with neurocognition and symptomatology. Forty patients with FES, 40 individuals with ARMS, and 40 healthy controls (HC) completed clinical assessments, a battery of neurocognitive tasks, and three social cognitive tasks. The comic strip and hinting tasks were used to measure non-verbal and verbal mentalizing abilities, and the gaze perception task was employed to assess self-referential hypermentalizing bias. FES and ARMS showed comparable mentalizing impairments and self-referential hypermentalizing bias compared to HC. However, only ambiguous self-referential gaze perception (SRGP) bias remained significantly different between three groups after controlling for covariates. Findings suggested that self-referential hypermentalizing bias could be a specific deficit and may be considered a potential behavioral indicator in early-stage and prodromal psychosis. Moreover, working memory and social anxiety were related to the social cognitive impairments in ARMS, whereas higher-order executive functions and positive symptoms were associated with the impairments in FES. The current study indicates the presence of stage-specific mechanisms of mentalizing impairments and self-referential hypermentalizing bias, providing insights into the importance of personalized interventions to improve specific neurocognitive domains, social cognition, and clinical outcomes for FES and ARMS.

Videoconferencing has become a ubiquitous medium for collaborative work. It does suffer however from various drawbacks such as zoom fatigue. This paper addresses the quality of user experience by exploring an enhanced system concept with the capability of conveying gaze and attention. Gazing Heads is a round-table virtual meeting concept that uses only a single screen per participant. It enables direct eye contact, and signals gaze via controlled head rotation. The technology to realise this novel concept is not quite mature though, so we built a camera-based simulation for four simultaneous videoconference users. We conducted a user study comparing Gazing Heads with a conventional “Tiled View” video conferencing system, for 20 groups of 4 people, on each of two tasks. The study found that head rotation clearly conveys gaze and strongly enhances the perception of attention. Measurements of turn-taking behaviour did not differ decisively between the two systems (though there were significant differences between the two tasks). A novel insight in comparison to prior studies is that there was a significant increase in mutual eye contact with Gazing Heads, and that users clearly felt more engaged, encouraged to participate and more socially present. Overall, participants expressed a clear preference for Gazing Heads. These results suggest that fully implementing the Gazing Heads concept, using modern computer vision technology as it matures, could significantly enhance the experience of videoconferencing.

Impairment in mentalization is implicated in the development and maintenance of depression. Major depressive disorders showed significant impairment in social cognition and such impairment appears to be positively associated with the severity of depression. Self-referential gaze perception (SRGP), a measurement of mentalization, was predominantly measured in patients with psychosis but rarely examined in late-life depression (LLD). To assess the effect of cognition on the interpretation bias of mentalization, 29 LLD patients and 29 healthy controls were asked to judge if various gaze directions were directed to self in SRGP. Patients with better cognition showed less unambiguous-SRGP bias than those with worse cognitive scores; this difference was not found in healthy controls. Global cognition and executive function contributed to the SRGP rate in patients. The current study is the first study to explore the relationship between cognition and SRGP in the LLD population. Our study findings suggested that the cognitive function of LLD patients may contribute to the modulation of interpretation bias, which in turn underlie the role of SRGP bias. Greater SRGP bias in patients may reflect social cognition deterioration, impairing the social interaction and functioning of LLD patients. This highlights the need for early intervention and cognitive decline identification to facilitate better prognosis and treatment effectiveness; thus, further studies could navigate the potential of SRGP task as a screening tool for high-risk group of LLD likely to develop dementia.

IntroductionImpairment in mentalization, interpreting and perceiving social relevant information has been found to play a part in the development and maintenance of depression. Major depressive disorders showed significant impairment in social cognition and such impairment appears to be positively associated with the severity of depression. Self-referential gaze perception, as a measurement of mentalization, was predominantly measured in patients with psychosis but rarely examined in late-life depression (LLD).ObjectivesTo assess the effect of cognition on the interpretation bias of mentalizationMethodsThis will be a cross-sectional case-controlled study on Chinese older adults with major depressive disorder recruited from outpatient departments of the public mental health service in Hong Kong. The same inclusion and exclusion criteria, with the exception of the history of major depressive disorder, will be used to recruit the control group. Assessments included sociodemographics, cognitive assessments and depressive symptoms. The primary experimental task was Gaze Perception Task using E- prime Professional 2.0. The stimuli of task are photographs of six Chinese models (3 men and 3 women) facing straight to camera with 13 different gaze directions (0°, 5°, 10°, 15°, 20°, 25° and 30° to the left and to the right, respectively). Participants shall be instructed to respond with a “yes” or a “no” to the question (for self-referential gaze): ‘Do you feel as if the person in the picture is looking at you?’.Results41 patients and 41 healthy controls have been recruited. The group comparison in SRGP revealed that there was only significant difference in the unambiguous-SRGP (U=561.000, Z=-2.62, N=82, p=0.009). Patients had higher unambiguous self-referential gaze accuracy (Mean=0.16) than controls (Mean= 0.075). With a cut-off score of 22, patients with better HK-MoCA scores had better unambiguous SRGP scores than those with lower HK-MoCA scores (p=0.024). This difference was not observed in healthy controls. HK-MoCA could predict ambiguous SRGP rate F(1,80)=14.85, p<.001, R square=15.7%. and redict unambiguous SRGP rate F(1,80)=14.85, p<.001, R square=15.7%.ConclusionsLLD subjects had a significant interpretation bias in the unambitious averted gaze (20°, 25° and 30°) interpretation compared with healthy controls. LLD subjects tend to have more self-referential perception of the clear averted gaze. This misinterpretation of the eye gaze is probably due to the interpretation bias in processing external information, which is commonly reported as mentalization impairment in depression (Weightman et al., 2014).Disclosure of InterestNone Declared

The demand for kiosk systems with embodied conversational agents has increased with the development of artificial intelligence. There have been attempts to utilize non-verbal cues, particularly virtual human (VH) eye contact, to enable human-like interaction. Eye contact with VHs can affect satisfaction with the system and the perception of VHs. However, when rendered in 2D kiosks, the gaze direction of a VH can be incorrectly perceived, due to a lack of stereo cues. A user study was conducted to examine the effects of the gaze behavior of VHs in multi-party conversations in a 2D display setting. The results showed that looking at actual speakers affects the perceived interpersonal skills, social presence, attention, co-presence, and competence in conversations with VHs. In a second study, the gaze perception was further examined with consideration of the Mona Lisa effect, which can lead users to believe that a VH rendered on a 2D display is gazing at them, regardless of the actual direction, within a narrow range. We also proposed the camera rotation angle fine tuning (CRAFT) method to enhance the users’ perceptual accuracy regarding the direction of the VH’s gaze.The results showed that the perceptual accuracy for the VH gaze decreased in a narrow range and that CRAFT could increase the perceptual accuracy.

A core feature of autism involves difficulty perceiving and interpreting eye gaze shifts as nonverbal communicative signals. A hypothesis about the origins of this phenotype is that it emerges from developmentally different social visual attention (SVA). We developed Social Games for Autistic Adolescents (SAGA; Scherf et al. BMJ Open 8(9):e023682, 2018) as a serious game intervention for autistic individuals to discover the significance of eye gaze cues. Previously, we demonstrated the effectiveness of SAGA to improve the perception and understanding of eye gaze cues and social skills for autistic adolescents (Griffin et al. JCPP Adv 1(3):e12041, 2021). Here, we determine whether increases in social visual attention to faces and/or target gazed-at objects, as measured via eye tracking during the same Gaze Perception task in the same study sample, moderated this improvement. In contrast to predictions, SVA to faces did not differentially increase for the treatment group. Instead, both groups evinced a small increase in SVA to faces over time. Second, Prior to the SAGA intervention, attention to faces failed to predict performance in the Gaze Perception task for both the treatment and standard care control groups. However, at post-test, autistic adolescents in the treatment group were more likely to identify the object of directed gaze when they attended longer to faces and longer to target objects. Importantly, this is the first study to measure social visual attention via eye tracking as a treatment response in an RCT for autism. NCT02968225