Insights into the relationship between eye movements and personality traits in restricted visual fields

Xu, Kuangzhe

doi:10.1038/s41598-024-60992-w

Download PDF

Article
Open access
Published: 04 May 2024

Insights into the relationship between eye movements and personality traits in restricted visual fields

Kuangzhe Xu¹

Scientific Reports volume 14, Article number: 10261 (2024) Cite this article

435 Accesses
Metrics details

Subjects

Abstract

Previous studies have suggested behavioral patterns, such as visual attention and eye movements, relate to individual personality traits. However, these studies mainly focused on free visual tasks, and the impact of visual field restriction remains inadequately understood. The primary objective of this study is to elucidate the patterns of conscious eye movements induced by visual field restriction and to examine how these patterns relate to individual personality traits. Building on previous research, we aim to gain new insights through two behavioral experiments, unraveling the intricate relationship between visual behaviors and individual personality traits. As a result, both Experiment 1 and Experiment 2 revealed differences in eye movements during free observation and visual field restriction. Particularly, simulation results based on the analyzed data showed clear distinctions in eye movements between free observation and visual field restriction conditions. This suggests that eye movements during free observation involve a mixture of conscious and unconscious eye movements. Furthermore, we observed significant correlations between conscious eye movements and personality traits, with more pronounced effects in the visual field restriction condition used in Experiment 2 compared to Experiment 1. These analytical findings provide a novel perspective on human cognitive processes through visual perception.

Two common and distinct forms of variation in human functional brain networks

Article 30 April 2024

Memorability shapes perceived time (and vice versa)

Article 22 April 2024

Electrophysiological signatures of veridical head direction in humans

Article 06 May 2024

Introduction

In contemporary society, elucidating and understanding the factors that influence human behavior and decision-making is a crucial challenge. Among these factors, visual information processing plays a central role when interacting with the environment and engaging with others. Specifically, visual attention and eye movements play significant roles in individual cognitive processes, and the progress in comprehending how these factors impact an individual’s personality and decision-making continues.

Personality traits and eye movements have been extensively studied, particularly in the field of personality psychology. Research in this area aims to elucidate the impact of individuals’ inherent personality traits on their behavior and decision-making. Studies based on the Big Five model have defined fundamental personality traits, such as extraversion, agreeableness, conscientiousness, neuroticism, and openness^1,2,3. It has been revealed that these traits play crucial roles in individuals’ daily lives. For instance, individuals with different personality traits exhibited variations in performance in different occupations⁴. Moreover, personality traits were found to influence executive functions in information retrieval, extending beyond the realm of work⁵. Furthermore, in social settings, individuals were observed to make personality judgments based on others’ facial traits or expressions^6,7. In addition to facial features, a variety of social factors, such as age, gender, and the relationship with the observer, have been verified to attract the observer’s attention and draw their gaze⁸. Furthermore, it has been confirmed that a person’s personality traits are actually related to the formation of attention towards social situations, such as landscapes or objects⁹. Thus, personality traits not only serve as the starting point influencing people’s behavior but also act as the endpoint influencing impressions and evaluations of others. Understanding how individual differences in personality traits impact human behavior and decision-making provides essential insights into comprehending personality traits.

Research on eye movements is crucial for understanding the mechanisms of visual information processing and deepening our knowledge about behavior and cognitive processes. Previous studies have revealed associations between eye movement patterns and cognitive processes in various visual tasks^10,11. In particular, eye movements when observing human faces have been shown to closely relate to the physiological response of brain waves (N170), indicating a strong connection^12,13. Additionally, cultural influences have been suggested, as individuals from Eastern and Western cultures exhibit different eye movement patterns when observing faces, implying the impact of external factors, such as culture, on eye movements^14,15. Numerous studies have demonstrated the sensitivity of eye movements to stimuli that are highly specific to humans, such as faces^{16,17,18,19,20} and bodies^21,22. Studies related to saccadic trajectories, which are closely linked to eye movements, have revealed a diverse relationship with facial features in a social context²³. It was shown that eye movements demonstrate a significant correlation when observing faces. Particularly, it was confirmed that eye movements are effective in the learning and processing of new faces²⁴. Furthermore, leveraging insights gained from eye movement research has proven valuable in exploring therapeutic approaches for conditions like Autism Spectrum Disorder (ASD) and prosopagnosia (face blindness)^25,26,27,28. These findings provide a foundational understanding of how eye movements influence human behavior and cognition, contributing to ongoing efforts to develop interventions for various disorders.

In recent years, a growing body of research has focused on the relationship between eye movements and personality traits. Starting from early studies investigating the potential reflection of individual personality traits in eye movements²⁹, more recent research has suggested that eye movement data are also associated with the formation of individuals’ cognitive processes and emotional cognition^30,31,32. For instance, in various impression-rating tasks that involve facial images, strong correlations were observed between personality traits and eye movements³³. Additionally, tasks related to facial expression recognition revealed a robust correlation between personality traits and eye movements^34,35. These research findings have advanced our understanding of how objective physiological indicators, such as eye movements, are related to subjective evaluation indicators like personality traits.

Nevertheless, an experiment involving impression-rating tasks, where observers received explicit observation instructions, reported that despite engaging in unfamiliar observation behaviors due to these instructions, certain relationships between eye movements and personality traits were almost identical to those observed in free observation scenarios^36,37. This indicates that behavior was observed where eyes were unconsciously directed towards areas deemed “not to be looked at” or considered “information-free.” This result suggests the possibility that when humans observe their surroundings, conscious (when they are aware) and unconscious (when they are not aware) eye movements may blend together. Studies have also specifically focused on conscious eye movements. In one such study³⁸, researchers explored the effects of conscious observation behaviors on judging facial expressions in videos depicting faces represented by landmarks. To simulate conscious observation behavior, an experimental setup was created where only a small area around the mouse cursor was visible. Participants observed the facial stimuli by moving the cursor and made judgments about the expressions. The results revealed significant correlations between personality traits and conscious observation behaviors. Understanding how specific visual observation behaviors under certain conditions relate to individuals’ conscious processes offers new insights into visual behavior research. However, in this study, the movement of the cursor was considered a conscious observation behavior, hence no eye movement recordings were taken. Furthermore, while the effects of conscious observation behaviors were examined, they were not compared with free observation scenarios, indicating a need for more in-depth exploration in this field. Although concerns exist about the impact of methods that restrict the observation range on face recognition³⁹, reports suggest that in scenarios where the observation range can be freely moved, facial features are captured in a manner similar to free observation³⁸. It is believed that an experimental design that minimally interferes with participants’ observational intent can significantly reduce the impact on face recognition.

In this study, we aim to delve deeper into understanding how conscious eye movements induced by visual field restrictions relate to an individual’s personality traits. Specifically, drawing inspiration from prior research³⁸, we will observe faces under conditions of limited visual fields and conduct impression-rating tasks. We will analyze the relationships between personality traits and eye movements (a restricted visual field group and a free observation group) in different evaluation tasks using a hierarchical Bayesian model. By elucidating the relationships between eye movements under visual field restrictions and behavior and personality traits, we seek to enhance our understanding of visual behavior based on individual characteristics. We anticipate that this exploration will not only contribute to future clinical applications but also to cross-cultural studies and a more profound understanding of human behavior.

Experiment 1

In Experiment 1, two conditions were established: one with a restricted observation and one without restrictions. We conducted a behavioral experiment for impression-rating tasks. The size of the restricted observation area was determined with reference to prior research⁴⁰ and set as a circular region within 2 degrees above and below the center of gaze. All participants in this study were provided with a comprehensive explanation of the objectives, procedures, potential risks, and benefits of the research, and written informed consent was obtained. All methods in this research were carried out with approval from the Ethics Committee of Hirosaki University (Approval Number: 0002(2023)), in accordance with relevant guidelines and regulations.

Equipment

The experiment utilized the Tobii Pro Fusion eye tracker (120Hz) from Tobii Inc. to record participants’ eye movements. The experiment design was created using Psychopy. The experiment used a MacBook Pro (OS: Big Sur) and a 24-inch LG monitor (1080p). To stabilize the participants’ head positions, we employed a chinrest (TKD-UK1) from NAMOTO Inc.

Participant

The sample size for this study was set at a minimum of 30 participants, based on prior research^14,38,41. We recruited the participants in the experiment from Hirosaki University, with a total of 42 university students (29 males). All participants had normal vision (including corrected vision), and the average age was 19.8 (SD = 2.32). The details and purpose of the experiment were explained to participants in advance, and all participants underwent the experiment after demonstrating a clear understanding of the procedures. As compensation for their participation, each participant received a gift card of 1000 yen.

Stimuli

In this study, we used 50 facial images as stimuli, following the approach of previous studies^33,36,42. All of these facial images feature frontal views with neutral expressions. The individuals who provided these images are university students aged 18-22. To minimize extraneous factors, the images were converted to black and white using Photoshop, ensuring uniform brightness and contrast. Additionally, non-facial components (such as clothing below the neck, excluding hair) were removed, and the final images were standardized to a size of 412 $\times$ 558. Consistent with the experimental methodology of prior research, we fixed 10 facial images, randomly selected in advance, for each impression-rating item.

Procedure

All participants received an explanation and practice session before the start of the experiment. After gaining sufficient understanding of the content and procedures of the study, they proceeded to participate in the main experiment. The main experiment included two conditions: one with a restricted field of view and another without such restriction (details of the field restriction method are explained in the “Visual field restriction” Section). The procedures for the experiments under these two conditions were identical.

The experiment begins with the question presentation stage, in which an impression-rating question is presented. Participants confirm the question and, at their own timing, click the mouse to proceed to the stimulus observation stage. In the stimulus observation stage, for the restricted field-of-view condition, the cursor’s surroundings are masked, allowing visibility only in a specific area. Participants can move the cursor to observe the stimulus image. After 3 seconds of observation time, the session automatically transitions to the evaluation stage. In the unrestricted condition, there is no interference mask, and participants can directly observe the stimulus image for 3 seconds. In the evaluation stage, participants provide a 7-point rating for the impression question asked in the question presentation stage. The experiment comprises a total of 50 sessions, with stimulus images presented randomly.

The impression evaluation items were selected with reference to previous studies^1,2,3,4, and we adopted the Big Five items (extraversion, conscientiousness, agreeableness, neuroticism, and openness) as the evaluation items. Participants first underwent a behavioral experiment under the condition of restricted visual fields, followed by a short break, and then participated in a behavioral experiment under the condition of unrestricted visual fields. Participants first took part in the experiment under conditions with restricted visual fields to prevent a potential issue. If the experiment with restricted vision were conducted after participants had initially memorized the features of stimulus images under free observation conditions, it might lead to observation behavior focused more on identifying memorized facial features than on evaluating impressions. To circumvent this problem, participants were first introduced to the experiment under conditions with restricted visual fields. After completing all experiments, participants responded to the TIPI-J questionnaire⁴³ for a self-assessment of their personality traits.

Visual field restriction

The distance between participants and the monitor was fixed at 60 cm using a chin rest. Regarding the visual field restriction, based on a previous study⁴⁰, it was set within a range of 2 degrees above and below the central line of sight (the discrimination visual field). Considering the distance from the participant’s head (chin rest position) to the monitor, the radius of the visible area reflected on the monitor was approximately 80 pixels. The method for the experiment involving the visual field restriction followed a previous study⁴³, in which circular areas with a 40-pixel radius were created around the cursor’s center. Furthermore, a 40-pixel Gaussian filter was applied to smooth the peripheral areas, ensuring a gentle edge.

Data processing

Based on previous studies^36,44, eye movement data recorded by the eye tracker were formatted into an analyzable data structure. Specifically, participants’ eye movement coordinate data (1920 $\times$ 1080) for a single stimulus image were adjusted to match the size of the stimulus image (412 $\times$ 558). The adjusted data was then subjected to a Gaussian filter (SD = 10) and converted into weighted data with a unified scale. Subsequently, using pre-created masks for facial features, eye movement data corresponding to each facial part was extracted (the face mask includes the positions of the eyes, nose, mouth, eyebrows, forehead, and glabella for each stimulus image). Finally, the data for each extracted facial part was divided by the area (in pixels), resulting in the calculation of the weight of eye movement per pixel for each facial part. Additionally, as this study focused on eye movements during the observation of stimulus images, the analysis employed data solely from the eye movement during the stimulus observation stage.

Analysis

Based on the previous study³⁸, we analyzed the relationship between personality traits and eye movements using the ZIB model constructed according to Bayesian rules. The detailed model is presented in Eqs. (1)–(6). Here, G represents eye movements, and n denotes the index of each region (comprising a total of 6 locations: the eyes, nose, mouth, eyebrows, glabella, and forehead). P represents the matrix set of five personality traits for each participant; q corresponds to the parameter estimating the coefficients of the Bernoulli distribution model, representing the probability of focusing on a specific region. The parameters a, b, and $\mu _k$ are associated with estimating coefficients of the Beta distribution model, directly related to assessing the degree of focus on a specific region (representing dwell time), and $r^{subj}$ and $r^{pic}$ indicate random effects for participants and stimulus images, respectively.

$$\begin{aligned}{} & {} G_{n}\sim ZIB (q_{n},a_{n},b_{n}) \end{aligned}$$

(1)

$$\begin{aligned}{} & {} ZIB (G_{n}|q_{n},a_{n},b_{n}) = {\left\{ \begin{array}{ll} Bern (0|q_{n}) &{} ( G_{n} = 0 )\\ Bern (1|q_{n}) \times Beta (G_{n}|a_{n},b_{n}) &{} ( G_{n} > 0 ) \end{array}\right. } \end{aligned}$$

(2)

$$\begin{aligned}{} & {} q_{n} =\frac{1}{1+exp (-(\alpha ^{bern}_{k}+\sum ^{5}_{k=1}\beta ^{bern}_{k}P_{k(n)}+r^{subj(Bern)}_{i}+ r^{pic(Bern)}_{j}))} \end{aligned}$$

(3)

$$\begin{aligned}{} & {} a_{n}=\phi \cdot \mu _{n} \end{aligned}$$

(4)

$$\begin{aligned}{} & {} b_{n}=\phi (1-\mu _{n}) \end{aligned}$$

(5)

$$\begin{aligned}{} & {} \mu _{n} =\frac{1}{1+exp (-(\alpha ^{Beta}_{k}+\sum ^{5}_{k=1}\beta ^{Beta}_{k}P_{k(n)}+r^{subj(Beta)}_{i}+r^{pic(Beta)}_{j}))} \end{aligned}$$

(6)

We employed the Rstan package to construct our Bayesian model and estimate its parameters^45,46,47. The prior distribution for fixed effects adhered to a normal distribution with a mean of 0 and a standard deviation of 10. Meanwhile, the prior distribution for random effects followed a gamma distribution ($\alpha$ = 10, $\beta$ = 10). We ran each model with default Stan hyperparameter values: 4 chains, 1 thin, 2000 iteration steps, and 1000 warm-up steps. Consequently, we obtained 4000 MCMC samples.

To ensure the convergence of MCMC estimations, we calculated Rhat (${\hat{R}}$) for each parameter, a widely accepted criterion for convergence. In line with typical MCMC estimation practices, we deemed estimations as converged when the number of chains was three or more and the Rhat was below 1.1 for all parameters. Based on these criteria, we confirmed the convergence of all parameter estimations.

We utilized the highest density interval (HDI) to assess the statistical significance of the model estimation results^45,48. The HDI, a variant of a confidence interval, identifies the region with the highest density in the Bayesian posterior distribution, typically using a 95% HDI. This methodology allows us to consider the uncertainty associated with the model parameters and determine the range of parameter values considered most credible within the posterior distribution. Employing this approach facilitated the evaluation of the significance of the estimation results, providing a comprehensive perspective for interpreting the model outcomes.

Result

Free observation condition

Table 1 summarizes the significant results of eye movement and personality traits under the free observation condition. The Bernoulli model indicates whether specific facial features were observed, whereas the Beta model illustrates how much attention was given to particular facial features. When evaluating agreeableness, individuals with high conscientiousness tended not to look at the nose but exhibited a tendency to focus on the forehead. When evaluating conscientiousness, individuals with high neuroticism tended to look at the mouth but not at the eyebrows. Additionally, individuals with high conscientiousness did not focus on the eyes and nose. In evaluating extraversion, individuals with high conscientiousness tended to look at the eyebrows but not at the nose. When evaluating neuroticism, individuals with high conscientiousness tended to look at the eyebrows but not at the eyes and nose. Finally, when evaluating openness, individuals with high openness tended to focus on the eyes.

Table 1 Significant correlations between eye movements and personality traits (Free observation condition).

Full size table

Discrimination visual field condition

Table 2 summarizes the significant results of eye movements and personality traits in situations with restricted visual fields. The Bernoulli model and the Beta model respectively indicate whether specific facial areas were observed and to what extent. When evaluating agreeableness, individuals with high conscientiousness tended not to look at the nose but showed a tendency to look longer at the forehead. Additionally, individuals with high extraversion tended to look at the forehead, and those with high neuroticism tended to look at the forehead when evaluating openness. Moreover, individuals with high neuroticism tended to look at the nose when evaluating openness. Compared to the results of the free observation condition, the number of significant correlations notably decreased.

Table 2 Significant correlations between eye movements and personality traits (Discrimination visual field condition).

Full size table

Verify changes in eye movements through simulation

Using the estimated parameters from the model, we conducted simulations to verify changes in each eye movement for different impression evaluations. In this study, prioritizing the readability of results, confidence intervals and error bars were not included in the graphs, but the published result data does contain confidence intervals. Additionally, as an example explanation, we selected changes in conscientiousness with many correlations by referring to the results in Table 1. Henceforth, for consistency, all example explanations will focus on changes in conscientiousness. Further detailed results can be checked at the https://osf.io/8zfc5/. Figure 1 illustrates the trend of changes in whether each facial area was observed (Bernoulli distribution) during free observation with variations in conscientiousness. Figure 2 illustrates the trend of changes in how much each facial area was observed (Beta distribution) with variations in conscientiousness. Conscientiousness was scored on a 7-point scale from 1 to 7, whereas other personality traits were fixed with a score of 3. As a result, except for extraversion ratings, an increasing conscientiousness tended to be associated with a decreased tendency to look at the eyes. Moreover, all impression evaluations generally confirmed that individuals with higher conscientiousness tended not to look at the mouth. Conversely, excluding conscientiousness ratings, an increasing conscientiousness was associated with a tendency to look longer at the area between the eyebrows.

Figure 3 illustrates the changes in the perception of individual facial features (Bernoulli distribution) with variations in conscientiousness during restricted visual field conditions. Figure 4 demonstrates the trends in changes in how much each facial feature was observed (Beta distribution) with variations in conscientiousness. As a result, regardless of changes in conscientiousness, participants tended to look at all facial features, except for the mouth, with a high probability across all impression evaluations. Conversely, when evaluating conscientiousness, an increase in conscientiousness was associated with a tendency to look at the forehead region for most impression evaluations.

To clarify the comparison between changes observed during free observation and restricted visual field conditions, the results from the simulations are displayed after subtraction (results of restricted condition–results of free condition). Figure 5’s panel (A) illustrates the difference between Figs. 1 and 3, while panel (B) displays the difference between Figs. 2 and 4. A value greater than 0 indicates dominance of the restricted visual field condition; a value less than 0 indicates dominance of the free observation condition; and the closer the value is to 0, the lesser the change observed. For additional results, please check the provided https://osf.io/8zfc5/. According to Fig. 5’s panel (A), individuals with higher levels of conscientiousness are significantly more likely to focus on the eyes in the restricted visual field condition when evaluating neuroticism. Conversely, in the context of assessing extraversion, these individuals are more inclined to focus on the mouth under the restricted visual field condition. As per Fig. 5’s panel (B), individuals with higher levels of conscientiousness are observed to spend significantly more time looking at the space between the eyebrows in the restricted visual field condition when assessing conscientiousness.

The estimation results of the impact of variations in other personality traits on eye movements can be found in the supplementary data (https://osf.io/8zfc5/). The results of how much was observed (Beta distribution) showed different patterns depending on the specific personality trait. For example, unlike the changing trend in conscientiousness during the visual field restriction, neuroticism and openness exhibited a tendency to not look at the forehead as they increased. However, the results of whether each region was observed or not (Bernoulli distribution) showed a consistent trend across all personality traits. The probability of looking at each region was very high in all cases.

Discussion

The analysis results demonstrated a significant relationship between personality traits and eye movements during free observation, consistent with many aspects of prior studies. However, when the field of observation was restricted, there was a dramatic reduction in the number of significant correlations between personality traits and eye movements. In order to explore the cause, an examination of eye movement changes through model simulations was performed, which showed that attention was directed towards almost all facial regions. These findings suggest that the reduction in significant correlations could be attributed to the observation field being overly narrow and restrictive, potentially forcing participants to look at each facial area at least once. To more accurately capture conscious eye movements, the restricted field was modified, leading to the initiation of Experiment 2.

Experiment 2

In Experiment 2, considering the findings from Experiment 1 and to tackle the problem of a too-narrow observation range, we adjusted the size of the restricted observation area and conducted Experiment 2 under conditions similar to Experiment 1. The observation area was set to extend 3 degrees above and below the gaze center. We chose not to expand the effective visual field beyond 3 degrees because we anticipated that further enlargement could reduce the effectiveness of the visual field restriction. The experimental setup and the methods for organizing and analyzing data remained unchanged. Similar to Experiment 1, all participants in this study were provided with a comprehensive explanation of the objectives, procedures, potential risks, and benefits of the research, and written informed consent was obtained. All methods in this research were carried out with approval from the Ethics Committee of Hirosaki University (Approval Number: 0002(2023)), in accordance with relevant guidelines and regulations.