We investigated the association between autistic traits in neurotypical adults and the visual processing of an approaching life-size avatar’s gait. Autism spectrum disorder (ASD) is a neurodevelopmental disorder and is reported to be related to difficulties in visual processing of human motion, such as biological motion and gestures. It has been reported that neurotypical adults with higher autistic traits were clumsier than those with lower autistic traits when passing by others. To clarify such atypical visual motion processing along with higher autistic traits in daily life, we analyzed the Subthreshold Autistism Trait Questionanaire (SATQ) score, a 24-item self-reported scale of ASD and event-related potentials (ERPs), for 26 neurotypical adults, in response to walking motion while passing the avatar. Videos of a walking life-sized virtual avatar approaching and retreating were presented as visual stimuli. The association between participants' scores of autistic traits and the latencies and amplitudes of ERPs was examined. ERP components (N170 and P200) were identified in the occipito-temporal region. As a result, adults with higher autistic traits had longer latencies and smaller amplitudes of P200 for the approaching avatar in the occipito-temporal region than those with lower autistic traits. These findings indicate that adults with higher autistic traits have delayed and less sensitive visual processing of the approaching avatar. It suggests that while passing by another person, these individulas have atypical visual processing of another person’s approach. This study may contribute to elucidating autistic traits from the perspective of visual processing in an environment mimicking daily life.
Research Article
Neurotypicals with Higher Autistic Traits have Delayed Visual Processing of an Approaching Life-size Avatar’s gait: An Event-Related Potentials Study
https://doi.org/10.21203/rs.3.rs-2037171/v3
This work is licensed under a CC BY 4.0 License
Version 3
posted
You are reading this latest preprint version
Autism spectrum disorder
higher autistic traits
Event-Related Potentials
Visual Motion Processing
Approaching Life-size Avatar
Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by two different features: (a) persistent deficits in social communication and interaction across multiple contexts and (b) restricted, repetitive patterns of behavior, interests, or activities (Frith and Happé, 2005; American Psychiatric Association, 2013; Lord et al., 2018). Social interaction disorders also include difficulty in maintaining interpersonal relationships and understanding the expressions and gestures of others (Vukusic et al., 2017; Lord et al., 2018). In the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), which is generally used as classification and diagnostic criteria for autism, the diagnosis is based on impaired social interaction, impaired communication, and clinical observations of restricted, repetitive, and stereotyped interests and behaviors (Walter et al., 2009; American Psychiatric Association, 2013). In addition, recent studies measured the percentage of correct or reaction time necessary for observers to discriminate the biological motions and reported that people with ASD have atypical visual processing of the biological motions (Freitag et al., 2008; Nackaerts et al., 2012). Appropriate visual motion processing is one of the requirements in social interaction, and improper visual motion processing negatively impacts daily non-verbal interactions, such as understanding the gestures of others and avoiding conflicts while passing each other. Although deficits in social communication and other areas described in the DSM-5 are also important autistic traits, deficits of perceptual processing, such as the ability to discriminate biological motion not reflected in the DSM-5 checklist, have also been noted in previous ASD study (Blake et al., 2003). It is necessary to focus on a quantitative understanding of visual motion processing during interactions with others to better elucidate the atypicality of social interactions in individuals with higher autistic traits.
Previous studies have shown that individuals with higher autistic traits have specific physical motion and visual processing in situations where another person is approaching. Shigeta et al. (2018) focused on gait characteristics in individuals with higher autistic traits and conducted a walking experiment using a motion capture system. They reported that waist movements while passing another person were correlated with higher autistic traits. Participants with higher autistic traits had larger standard deviations from the norm of angular velocity in the waist than those with lower autistic traits (Shigeta et al., 2018). The findings suggest that individuals with higher autistic traits have difficulty perceiving others approaching them and have delayed physical responses to avoid a collision; thus, it is suggested that individuals with higher autistic traits have difficulty in visual processing of another person’s approach.
The event-related potentials (ERPs) have been used to evaluate brain function and clarify the mechanisms of various neurological disorders because ERPs appear in response to perceptual processing and reflect visual information processing (Tobimatsu and Celesia, 2006; Horvath et al., 2018). In previous studies, ERPs have been used to understand visual motion processing in ASD (Jeste and Nelson, 2009; Yamasaki et al., 2011). For example, Yamasaki et al. measured ERPs for visual stimuli, including radial and horizontal motion, in adults with ASD and neurotypical adults (Yamasaki et al., 2011). Optic flow containing dots moving radially outward were used as the radial motion, and dots moving to the right were used as the horizontal motion. The results showed that the latencies of both ERP components (N170 and P200) to radial optic flow were significantly prolonged in adults with ASD compared with neurotypical adults. However, the latencies of ERPs to horizontal motion were not prolonged in either group. In addition, other previous studies have measured ERPs for a point-light walker (PLW) that simulated the person approaching. PLW is a visual stimulus consisting of moving dots that record joint positions during human body movement. The results showed that the latency of N1 in the occipito-temporal region was longer in individuals with higher autistic traits than in those with lower autistic traits (Ichikawa et al., 2019; Inokuchi et al., 2022). These ASD studies indicate that individuals with higher autistic traits have atypical visual processing of expansion movements, such as radial optic flow with an outward pattern and approaching PLW.
Most studies of visual motion processing in ASD used moving dot stimuli, such as radial optic flow, which represents nonbiological motion (Yamasaki et al., 2011) or approaching PLW (Ichikawa et al., 2019; Inokuchi et al., 2022). However, the visual processing of natural and smooth gait motion by a more realistic avatar remains unclear in ASD. One study compared the visual processing of biological and non-biological motions in neurotypical participants (Pelphrey et al., 2003). Functional magnetic resonance imaging (fMRI) was used to evaluate whether biological motion activates the superior temporal sulcus (STS) more than non-biological motion. The results confirmed that the STS was more active for biological motion (for example, the walking robot) than for nonbiological motion (for example, a disjointed mechanical figure, the movements of a grandfather clock). The findings revealed that STS is sensitive to biological movement itself, not merely to the surface features of the stimulus (Pelphrey et al., 2003). On the other hand, even though individuals with high autistic traits have been shown to have low STS function (Zilbovicius et al., 2000; Boddaert et al., 2004), It is not clear whether they have atypical visual processing for biological movements with realistic avatars. To further elucidate autistic traits for biological motion from the perspective of visual motion processing, we postulated that it would be useful to measure the neural responses to gait motion displayed by a realistic avatar rather than the conventional point stimuli.
Herein, we investigated the association between higher autistic traits and visual processing of another person’s approach by measuring the ERPs for videos using a life-size virtual avatar. Three different patterns of avatar movements were used: approaching walking movement (AW), receding walking movement (RW), and standing (ST). By using the visual stimuli with the virtual avatar—instead of using videos of an actual person's gait or employing a real-life situation of passing others—some task-irrelevant factors that affect visual processing were minimized, such as the avatar’s facial information (i.e., gaze shift, gaze direction, facial expression, face attractiveness, and asymmetric gait traits of an actual person). Next, the latency and amplitude of N170 and P200 in the occipito-temporal region were examined, and their association with higher autistic traits and ERPs for more realistic biological motion were examined. The spatial structure and information in actual walking movements are more complicated and ecologically valid than those in radial optic flow or PLW. Therefore, measuring ERPs for another person’s approach in a more real-life setting provides important clues for investigating behavioral characteristics of ASD in daily life, such as atypical gait traits while passing each other.
2.1 Participants
Thirty neurotypical male university students, with a mean (± standard deviation) age of 22.17 ± 1.19 years participated in this study. Consistent with a previous study (Shigeta et al., 2018) that suggested an association between atypical visual processing of other's gait motion and autistic severity, the participants enrolled for this study were neurotypical men, and they had never been diagnosed with ASD. Thus, the confounding factors, such as comorbidities and autism-related symptoms, were not considered. As the gender ratio in the university where this study was conducted is quite unbalanced (most students were male), we employed only male participants in this study.
The participants completed the Subthreshold Autism Trait Questionnaire (SATQ), a 24-item self-reported scale to quantitatively measure autistic traits, with each question rated on a 4-point Likert scale that ranges from ‘‘false, not at all’’ to ‘‘very true” (Nishiyama et al., 2014). The SATQ is especially sensitive to individuals that are not clinically diagnosed but show subliminal autistic traits. The SATQ score was calculated by summing the scores for the 24 items. A higher SATQ score indicates higher autistic traits. The mean and standard deviation of the SATQ score in the participants was 27.77±8.02 out of 72 points.
Informed consent was obtained from each participant prior to the experiment. The experimental procedures were approved by the Ethics Committee of the Tokyo University of Science (approval number 16027).
2.2 Visual Stimuli and Procedure
We created visual stimuli using a life-sized virtual avatar to imitate a more realistic setting of “passing others” compared to other studies (Figure 1). Unity 3D game engine (version 2020.2.3f1, Unity Software Inc., San Francisco, USA) was used to create visual stimuli. The visual stimuli consisted of a virtual avatar and a background that imitated a typical street scene. Three different patterns of avatar movement were created (approaching walking movement, AW; receding walking movement, RW; just standing and not walking, ST). AW refers to the video of the virtual avatar approaching; the visual angle of the height of the person changed from 62.4° to 95.6°. RW refers to the video of a person moving away from the participant; the visual angle of the height of the person changed from 62.4° to 45.2°. In the experiment, these two types of stimuli were used as the target stimuli. In addition to the two types of visual stimuli, ST refers to a video of a person swaying from side to side and standing in place. The ST was presented as the non-target stimulus to prevent participants from predicting the avatars’ movements and to prevent them from preparing to press the button in the attention task. Figure 1 shows the three types of visual motion stimuli and the experimental procedure of one trial. First, a still image of the virtual avatar standing was presented for 1200 ms for all visual stimuli. Then, AW, RW, and ST visual motion stimuli were initited, and these lasted for 1220 ms each. After the video for each condition ended, a white fixation point was displayed for 1800–2200 ms. Each trial was presented for 4220–4620 ms. Five sets of visual stimuli were performed, each set consisting of 60 trials. Therefore, the participants randomly observed the AW and RW videos for 120 trials and the ST video for 60 trials. Participants were given a break between sets to prevent fatigue from affecting the EEG (visual stimulus video, Supplementary Material).
The visual stimuli used in the experiment were projected onto a screen using a projector. The screen was 2.7 m wide and 1.9 m high, and the refresh rate of the projector was 60 Hz. Participants stood 0.6 m away from the screen and were instructed to focus on the fixation point in the center of the stimulus to adjust the height of the screen.
2.3 Attention Task
During the EEG recording, participants performed the button-pressing task. The purpose of this task was to get the participants to concentrate on the visual stimuli. In this task, participants were instructed to predict the direction in which the virtual avatar on the screen moved and to answer the direction using a numeric keypad. The participants were instructed to press "left (4 on the numeric keypad)" if they judged that the virtual avatar went to the left from their point of view, and to press "right (6 on the numeric keypad)" if they judged that the virtual avatar went to the right, as quickly as possible. The participants were made to wear the numeric keypad by using a fixing belt around their thighs on the dominant side. The numeric keypad was positioned so that they could press the buttons while standing. Participants were allowed to answer intuitively to have them respond to as many trials as possible. Because the task was performed such that the participants' concentration was maintained during the experiment, analysis of participants’ performance in the additional task is beyond the scope of this study and has not been mentioned here.
2.4 Electroencephalography Recording
An 8-channel simple electroencephalograph (Ultracortex "Mark IV" EEG Headset; OPENBCI) was used for EEG measurements. The active dry electrodes were placed at six locations (C3, C4, T5, T6, O1, and O2) according to the 10-20 system. Two other electrodes were placed above the right eyebrow and below the right eye to measure the electrooculogram (EOG) signal. The EOG data were used to detect artifacts from eye movements and blinking. The reference marker was placed on the tip of the nose far from the temporal area. A ground electrode was placed on the right earlobe. Both the EEG and EOG were measured at a sampling frequency of 250 Hz.
2.5. ERP Analysis
The bandpass filter was 1–20 Hz. After filtering, EEG data were divided into epochs from 200 ms before to 600 ms after the onset of each visual stimulus (AW, RW, and ST). Non-neural artifacts, such as blinking and eye movements, were rejected by visual inspection and independent component analysis. Following this, the ERP waveforms were calculated by averaging all epochs, excluding artifacts. EEG data from 200 ms before each visual stimulus (AW, RW, and ST) were used as baseline. The ERP components in each channel for each participant were identified from the grand-average waveform by visual inspection. In this study, we identified the ERP component N170, which appears as a negative peak around 170 ms from onset, and the ERP component P200, which appears as a positive peak around 200 ms from onset. Participants and electrodes with strong noise were excluded from analysis. The ERP components at the T5 and T6 electrodes were used as the evaluation indices, because a previous study reported a longer ERP latency in the occipito-temporal region for the PLW [13], and the N170 and P200 components were intensely observed at the T5 and T6 electrodes in this study. At each T5 and T6 electrode, the data of participants whose ERPs were not identified were excluded from the analysis. Thus, ERPs data from four participants were excluded at both electrodes, and data from 26 participants were used for the analyses at the T5 and T6 electrodes. Figure 2 shows histograms of the SATQ scores of the 26 participants used in the ERP analysis at each of the T5 and T6 electrodes. The acquired data were pre-processed using EEGLAB (MATLAB, MathWorks, MA, USA; https://sccn.ucsd.edu/eeglab/) (Delorme and Makeig, 2004).
2.6 Statistical Analysis
We conducted a multiple regression analysis with the exhaustive search (ES) method to select effective variables to explain the SATQ score. ES is a method of variable selection in which the coefficient of each variable is 0 or other, by exhaustively searching N N 1 N 2 N K N N models when the number of variables is N (Ichikawa et al., 2014; Nagata et al., 2015; Igarashi et al., 2018). The ES method is also used as an evaluation method for studies with multichannel functional continuous-wave near-infrared spectroscopy (fNIRS) measurements and is suitable for evaluating brain activity (Ichikawa et al., 2014). In this study, multiple regression analysis using the exhaustive search (ES) method was performed at each of the T5 and T6 electrodes.We used the SATQ scores of each participant as the objective variables. Eight ERP indicators at electrodes T5 were used as explanatory variables, when analyzing at the T5 electrode. Therefore, 255 different models were exhaustively searched for at the T5 electrode; the data obtained at T6 electrode were analyzed in the same manner.
Cross-validation was then used to examine the predictive accuracy of each multiple regression model. In the cross-validation, the dataset was split into two parts: one for training and one for testing. We used leave-one-out cross-validation (LOOCV), which is effective when the number of samples is small. LOOCV is a method in which only one of the entire datasets is the test data, and then the test data and training data are swapped and repeated for validation until all cases have been examined as the test data. The following is a procedure for LOOCV for the data of the 26 participants with identified ERPs. First, we defined one data (eg., Participant 1’s data) as test data, and the remaining 25 (eg., Data from participants 2 to 26) as training data. Multiple regression analysis was conducted using the training data and a regression model was obtained. The regression model was then applied to the test data, and the cross-validation error (CVE) was calculated. Next, we defined the other data (eg., Participant 2’s data) as the test data, and the remaining 25 (eg., Data of Participant 1 and Participant 3 to Participant 26) as training data, conducted regression analysis and cross-validation, and calculated the CVE. The procedure was repeated 26 times, and the CVE for the regression model was obtained by averaging the 26 CVEs. In addition, to compare the impact of each explanatory variable on the SATQ score, standardized partial regression coefficients were calculated for each explanatory variable for each model and are shown in the color map within the weight diagram. The standardized partial regression coefficients are those when the objective variable and each explanatory variable are standardized to mean = 0 and variance = 1. In the weight diagram, a color map shows the value of the standardized partial regression coefficient for each explanatory function included in the model. In accordance with the color bar, red and yellow indicate positive values, and blue indicates negative values.
Herein, CVE was defined as the mean absolute error (MAE) between the SATQ score predicted by the model obtained from multiple regression analysis and the actual SATQ score. The formula for calculating the MAE is shown in Equation 1. MAE has the advantage of being less susceptible to outliers than the root mean square error or the mean squared error. To evaluate the effective variables that explain SATQ scores, we created a weight diagram plot using the CVE and standardized partial regression coefficient in each model. All statistical analyses were conducted using MATLAB version 2021a (MathWorks, Inc.).
3.1 The grand-average waveform
Figure 3 shows the grand-average waveforms at the T5 and T6 electrodes. The negative ERP component N170 around 170 ms and the positive ERP component P200 around 200 ms appeared in both the AW and RW conditions. However, neither the N170 nor the P200 component appeared in the ST condition, in which the avatar was standing upright on the screen.
3.2 Multiple regression analysis
Multiple regression analysis with the ES method was performed using the amplitude and latency of each ERP at the T5 and T6 electrodes. LOOCV was performed to evaluate the accuracy of each model.
First, a multiple regression analysis was conducted with eight ERP indicators (N170 and P200 latencies and amplitudes under AW and RW conditions) at the T5 electrode (Figure 4). The upper panel in Figure 4 shows the CVE results for each model as an evaluation indicator of prediction accuracy. On the horizontal axis, 50 of the 255 models were arranged based on the model with the smallest CVE. The lower panel in Figure 4 shows a weight diagram plot with ES. Each of the 255 models with multiple regression analyses was evaluated based on CVE. White areas in the color map indicate that the explanatory variable is not included in the model. Out of the 50 multiple regression models with small CVE, “the P200 amplitude in the AW condition” was most often selected of the eight explanatory variables; the variable was selected for 40 of the 50 models. In addition, the color map in the Weight diagram shows the value of the standardized partial regression coefficient for each explanatory function included in the model. The magnitude of the standardized partial regression coefficients in each explanatory variable means that when the coefficient has a positive value, the higher the value of the explanatory variable, the higher the SATQ score for the objective function. When the coefficient has a negative value, it implies that the smaller the value of the explanatory variable, the lower the SATQ score. The value of the standardized partial regression coefficients for “P200 amplitude in the AW condition” were negative for all 40 models selected. Additionally, the second most frequently selected variable was “the N170 amplitude in the AW condition,” which was selected 21 times of the 50 models. Table 1 shows the multiple regression model with the smallest CVE at the T5 electrode. Three variables were entered as predictors: “P200 latency in RW condition,” “N170 amplitude in AW condition,” and “P200 amplitude in AW condition.” The multiple regression model was statistically significant (F(3,22) = 3.094, p = 0.048, adjusted R2 = 0.200).
Second, multiple regression analysis was performed using eight ERP indicators at the T6 electrode. The weight diagram plot with ES is shown in Figure 5, where “the P200 latency in the AW condition” was the most frequently selected explanatory variable among the eight explanatory variables: 36 out of the 50 multiple regression models. The value of the standardized partial regression coefficients for “the P200 latency in the AW condition” were positive for all 36 models selected. The second most frequently selected variable was “the P200 amplitude in the AW condition,” which was selected 28 times among the 50 models. Table 2 shows the multiple regression model with the smallest CVE at the T6 electrode. Three variables were entered as predictors: “N170 latency in AW condition,” “P200 latency in AW condition,” and “P200 amplitude in AW condition.” The multiple regression model was statistically significant (F(3,22) = 3.135, p = 0.046, adjusted R2 = 0.204).
We investigated the association between higher autistic traits and visual processing of another person’s approach by measuring the ERPs for walking videos of the life-size virtual avatar. Two ERP components (N170 and P200) were identified for each participant in the AW and RW conditions, respectively. Multiple regression analysis was performed using the latency and amplitude of ERP components at the T5 and T6 electrodes. A weight diagram was calculated using ES. The results showed that “P200 amplitude in the AW condition” was the most frequently selected variable among the 50 models at the T5 electrode, and “P200 latency in the AW condition” was the most frequently selected variable at the T6 electrode. Furthermore, the standard partial regression coefficient for “P200 amplitude in the AW condition at the T5 electrode” showed the tendency to adopt negative values for the selected models in the weight diagram. The standard regression coefficients for “P200 latency in the AW conditions at the T6 electrode” showed the tendency to adopt positive values. These results indicated that the longer the P200 latency for the AW condition at the T5 electrode and the smaller the P200 amplitude for the AW condition at the T6 electrode the participant has, the higher the SATQ score. These findings suggest that P200 latency and amplitude in the video of the approaching person in the occipito-temporal region could explain the magnitude of the autistic trait. They indicated that individuals with higher autistic traits have more delayed and less sensitive processing of another person’s approach. This finding is supported by previous studies that showed that patients with ASD have longer latency of ERP to radial optical flow (Yamasaki et al., 2011), and that individuals with higher autistic traits have longer latency of ERP to the PLW (Ichikawa et al., 2019; Inokuchi et al., 2022).
The association between ASD tendencies and visual processing of approaching avatar shown in this study may be related to the atypical gait traits suggested in the previous ASD study (Shigeta et al., 2018). A previous study suggested that individuals with higher autistic traits have an atypical gait trait, in which they approach their counterparts more closely and twist their waist greatly to avoid a collision when walking past each other (Shigeta et al., 2018). Appropriate visuomotor control is required to assume appropriate motion patterns in response to obstacles in situations such as walking past each other. First, in the process of walking past each other, it is necessary to understand the intention of the other person and predict what they may do next (Honma et al., 2015). Furthermore, when an opponent is present at a distance, humans are required to avoid the opponent by modifying their own movement pattern and making anticipatory gait adjustments based on accurately perceived visual information (Higuchi, 2013). Individuals must use visual information and predict when and where a collision will occur (Patla, 1997; Gray et al., 2000). Thus, in cooperative behavior with a partner and obstacle avoidance behavior, such as walking past someone, predicting and accurately perceiving changes in the external world underlies adaptive locomotor adjustments. The visual processing of the walking behavior of others is closely related to the appropriate motor coordination in walking past another person. Delayed and less sensitive visual processing of another person’s approach may be related to the atypical gait characteristics of ASD while passing by, as described by Shigeta et al. (2018).
For the ERPs in this study, N170 and P200 were identified for each participant in the AW and RW conditions. These two ERPs (N170 and P200) appeared similarly in previous studies using optic flow as visual stimuli (Yamasaki et al., 2011), therefore, our results support that these ERPs are appropriate as indicators to evaluate visual processing of the visual stimuli. In addition, These ERPs results support a previous study that suggested selective impairment in the dorsal pathway in visual information processing of individuals with ASD (Spencer et al., 2000; Yamasaki et al., 2011). The human visual motion processing involves two main pathways: the ventral and dorsal pathways. The ventral visual pathway is involved in object and facial recognition. The dorsal visual pathway is primarily concerned with the spatial location of objects and motion stimuli (Vaina et al., 2001; Rizzolatti and Matelli, 2003). The visual processing of the approaching virtual avatar in this experiment suggests that the dorsal path is primarily used because it is necessary to understand the spatial position of the avatar. Thus, the delayed visual processing in response to an approaching human, as observed in this study, may be due to the reduced function of the dorsal visual pathway associated with individuals with autistic traits. Furthermore, this consideration is supported by a study that showed that N170 and P200 latency to radial optic flow (v-d lateral pathway stimulation) was delayed (Yamasaki et al., 2011).
Additionally, the previous studies have shown that individuals with autistic traits may have functional anomalies in the social brain, including the STS (Zilbovicius et al., 2000; Boddaert et al., 2004). The social brain is the region related to the processing of biological motion, such as perceiving movements and predicting the direction of the actions and intentions of others (Pelphrey et al., 2011). Previous studies have shown that individuals with autistic traits have localized bilateral hypoperfusion (Zilbovicius et al., 2000) and bilateral decreases in gray matter thickness in the STS, which is part of the social brain (Boddaert et al., 2004). In fact, the T5 and T6 electrodes that showed low sensitivity and delayed ERP component in this study were located near the STS. Therefore, these results suggest that individuals with autistic traits take longer to perceive and process biological motion in the social brain, resulting in delayed visual processing in response to someone approaching them.
Furthermore, we found a longer latency and smaller amplitude of P200 in the AW condition, but not in the RW condition. This may be because the visual stimulus in the AW condition was a radial motion similar to radial optic flow showing expansion. In other words, the AW condition was perceived as an expanding motion when the virtual avatar was approaching. Individuals with higher autistic traits may have a more specific visual processing for expanding movements (someone approaching them). On the other hand, the present results did not show a longer latency or reduced amplitude of the N170 component. This result differs from those of a study using radial optic flow (Yamasaki et al., 2011) and PLW (Ichikawa et al., 2019; Inokuchi et al., 2022) as the visual stimulus. This difference may be because the visual stimulus is more natural and smoother in a virtual avatar video than in the dot stimuli. The V5/MT in the brain region where the N170 component is evoked has been suggested to play a role in integrating the local motion signal from V1 and converting it to global motion (Snowden et al., 1991). However, because visual stimuli with virtual avatar have already been converted to global biological motion, it is possible that longer latency and smaller amplitude were observed in P200 than in N170.
The methodological limitation of this study was that the participants in this experiment were neurotypical male university students, and not patients with ASD. This is because we considered that ASD has a wide spectrum, and there is a lot of heterogeneity between those that are diagnosed with it. Additionally, as the variation in SATQ scores in this study was smaller than that in other ASD studies, the visual motion processing of individuals diagnosed with ASD could not be adequately verified. Therefore, future experiments with patients having ASD and using study samples with equal sex ratios may contribute to further clarification of autistic traits. Furthermore, the spatial resolution of ERPs is far inferior to that of fMRI, and ERP measurements alone are limited in their spatial considerations of the brain. Therefore, a more detailed spatial consideration of visual motion processing using fMRI is required.
In this study, we investigated the association between autistic traits in neurotypical male university students and visual processing of another person’s approach, by measuring the ERPs for videos of a life-size virtual avatar; approaching and retreating walking motions of the avatar were used as visual stimuli. As a result, the N170 and P200 for each participant in the occipito-temporal region were identified. Multiple regression analysis for each ERP indicator and cross-validation for each model were performed. The results showed that the P200 amplitude in the AW condition was the most selected variable for predicting the SATQ score at the T5 electrode, while the P200 latency in the AW condition was the most selected variable at the T6 electrode. Furthermore, the weight diagram results showed that the longer the P200 latency at the T5 electrode and the smaller the P200 amplitude at the T6 electrode in the AW condition, the higher was the SATQ score. These results suggest that the P200 latency and amplitude in the occipito-temporal region in the AW condition is the variable that explains the severity of autistic traits, and individuals with higher autistic traits have delayed and less sensitive processing regarding the approach of others than those with lower autistic traits. Previous studies on autistic traits have focused on simplified visual stimuli, such as radial optic flow or PLW. Therefore, investigating the association between autistic traits and visual processing of another person’s approach by using the more realistic life-size virtual avatar may be useful to clarify the atypical gait traits of individuals with ASD in daily life. In future, studies should verify the association between autistic traits and brain activity in more detail by measuring ERPs during actual interactions involving people walking past one another.
6 Data availability
The dataset analysed during the current study are available from the corresponding author on reasonable request.
7 Conflict of Interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
8 Author Contributions
R.I. designed the study. R.I., H.I., M.Y., and H.T. contributed to programming, testing, and data collection. R.I. performed data analysis. R.I., H.I., M.Y., and H.T. interpreted the data. All authors contributed to writing and editing the manuscript. All authors have approved the final version of the manuscript for submission.
9 Funding
This research was supported by JSPS KAKENHI (grant number JP20K03504).
10 Acknowledgments
We would like to appreciate English language editing of Editage (www.editage.com). The pre-print of this article is released on Research Square (Inokuchi et al., 2022).
American Psychiatric Association (2013). Diagnostic and Statistical Manual of Mental Disorders. 5th ed. (American Psychiatric Association).
Blake, R., Turner, L. M., Smoski, M. J., Pozdol, S. L., Stone, W. L. (2003). Visual recognition of biological motion is impaired in children with autism. Psychol. Sci. 14, 151–157. doi: 10.1111/1467-9280.01434.
Boddaert, N., Chabane, N., Gervais, H., Good, C. D., Bourgeois, M., Plumet, M. H. et al. (2004). Superior temporal sulcus anatomical abnormalities in childhood autism: a voxel-based morphometry MRI study. Neuroimage. 23, 364–369. doi: 10.1016/j.neuroimage.2004.06.016.Delorme, A., Makeig, S. (2004). EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J. Neurosci. Methods. 134, 9–21. doi: 10.1016/j.jneumeth.2003.10.009.
Freitag, C. M., Konrad, C., Häberlen, M., Kleser, C., von Gontard, A., Reith, W. et al. (2008). Perception of biological motion in autism spectrum disorders. Neuropsychologia. 46, 1480–1494. doi: 10.1016/j.neuropsychologia.2007.12.025.
Frith, U., Happé, F. (2005). Autism spectrum disorder. Curr. Biol. 15, R786–R790. doi: 10.1016/j.cub.2005.09.033.
Gray, R., Regan, D. (2000). Estimating the time to collision with a rotating nonspherical object. Vision Research, 40(1), 49-63. doi: 10.1016/S0042-6989(99)00157-1.
Higuchi, T. (2013). Visuomotor control of human adaptive locomotion: understanding the anticipatory nature. Front. Psychol. 4, 277. doi: 10.3389/fpsyg.2013.00277.
Honma, M., Koyama, S., Kawamura, M. (2015). Hesitant avoidance while walking: an error of social behavior generated by mutual interaction. Front. Psychol. 6, 1013. doi: 10.3389/fpsyg.2015.01013.
Horvath, A., Szucs, A., Csukly, G., Sakovics, A., Stefanics, G., Kamondi, A. (2018). EEG and ERP biomarkers of Alzheimer’s disease: a critical review. Front. Biosci. (Landmark Ed.). 23, 183–220. doi: 10.2741/4587.
Ichikawa, H., Kitazono, J., Nagata, K., Manda, A., Shimamura, K., Sakuta, R. et al. (2014). Novel method to classify hemodynamic response obtained using multi-channel fNIRS measurements into two groups: exploring the combinations of channels. Front. Hum. Neurosci. 8, 480. doi: 10.3389/fnhum.2014.00480.
Ichikawa, T., Shigeta, M., Urakawa, T., Sawatome, A., Tanaka, M., Kurita, Y. et al. (2019). Correlation between gait perception and autistic traits in the general population: a study on event-related evoked potentials. Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. 41st Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). 2019, 3135–3138. doi: 10.1109/EMBC.2019.8857577.
Igarashi, Y., Ichikawa, H., Nakanishi-Ohno, Y., Takenaka, H., Kawabata, D., Eifuku, S. et al. (2018). ES-DoS: exhaustive search and density-of-states estimation as a general framework for sparse variable selection. J. Phys.: Conf. Ser. 1036, 01200. doi: 10.1088/1742-6596/1036/1/012001.
Inokuchi, R., Ichikawa, T., Ichikawa, H., Yamamoto, M., Takemura, H. (2022). Gait perception of life-size point-light walker is associated with autistic traits: evidence from event-related evoked potentials 44st Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). 2022. https://pubmed.ncbi.nlm.nih.gov/36086078/. doi: 10.1109/EMBC48229.2022.9871294.
Inokuchi, R., Ichikawa, H., Yamamoto, M., Takemura, H. Neurotypicals with Higher Autistic Traits have Delayed Visual Processing of an Approaching Life-size Avatar’s gait: An Event-Related Potentials Study [Preprint] (2022). Available at:
Jeste, S. S., Nelson, C. A. (2009). Event related potentials in the understanding of autism spectrum disorders: an analytical review. J. Autism Dev. Disord.. 39, 495–510. doi: 10.1007/s10803-008-0652-9.
Lord, C., Elsabbagh, M., Baird, G., Veenstra-Vanderweele, J. (2018) Autism spectrum disorder. Lancet. 392, 508–520. doi: 10.1016/S0140-6736(18)31129-2.
Nackaerts, E., Wagemans, J., Helsen, W., Swinnen, S. P., Wenderoth, N., Alaerts, K. (2012). Recognizing biological motion and emotions from point-light displays in autism spectrum disorders. PLOS ONE. 7, e44473. doi: 10.1371/journal.pone.0044473.
Nagata, K., Kitazono, J., Nakajima, S., Eifuku, S., Tamura, R., Okada, M. (2015). An exhaustive search and stability of sparse estimation for feature selection problem. IPSJ Online Trans. 8, 25–32. doi: 10.2197/ipsjtrans.8.25.
Nishiyama, T., Suzuki, M., Adachi, K., Sumi, S., Okada, K., Kishino, H. et al. (2014). Comprehensive comparison of self-administered questionnaires for measuring quantitative autistic traits in adults. J. Autism Dev. Disord. 44, 993–1007. doi: 10.1007/s10803-013-2020-7.
Patla, A. E. (1997). Understanding the roles of vision in the control of human locomotion. Gait & posture, 5(1), 54-69. doi: 10.1016/S0966-6362(96)01109-5.
Pelphrey, K. A., Mitchell, T. V., McKeown, M. J., Goldstein, J., Allison, T., McCarthy, G. (2003). Brain activity evoked by the perception of human walking: controlling for meaningful coherent motion. J. Neurosci. 23, 6819–6825. doi: 10.1523/JNEUROSCI.23-17-06819.2003.
Pelphrey, K. A., Shultz, S., Hudac, C. M., Vander Wyk, B. C. (2011). Research review: constraining heterogeneity: the social brain and its development in autism spectrum disorder. J. Child Psychol. Psychiatry. 52, 631–644. doi: 10.1111/j.1469-7610.2010.02349.x.
Rizzolatti, G., Matelli, M. (2003). Two different streams form the dorsal visual system: anatomy and functions. Exp. Brain Res. 153, 146–157. doi: 10.1007/s00221-003-1588-0.
Shigeta, M., Sawatome, A., Ichikawa, H., Takemura, H. (2018). Correlation between autistic traits and gait characteristics while two persons walk toward each other. Adv. Biomed. Eng. 7, 55–62. doi: 10.14326/abe.7.55.
Snowden, R. J., Treue, S., Erickson, R. G., Andersen, R. A. (1991). The response of area MT and V1 neurons to transparent motion. J. Neurosci. 11, 2768–2785. doi: 10.1523/JNEUROSCI.11-09-02768.1991.
Spencer, J., O’Brien, J., Riggs, K., Braddick, O., Atkinson, J., Wattam-Bell, J. (2000). Motion processing in autism: evidence for a dorsal stream deficiency. NeuroReport. 11, 2765–2767. doi: 10.1097/00001756-200008210-00031.
Tobimatsu, S., Celesia, G. G. (2006). Studies of human visual pathophysiology with visual evoked potentials. Clin. Neurophysiol. 117, 1414–1433. doi: 10.1016/j.clinph.2006.01.004.
Vaina, L. M., Solomon, J., Chowdhury, S., Sinha, P., Belliveau, J. W. (2001). Functional neuroanatomy of biological motion perception in humans. Proc. Natl Acad. Sci. U. S. A. 98, 11656–11661. doi: 10.1073/pnas.191374198.
Vukusic, S., Ciorciari, J., Crewther, D. P. (2017). Electrophysiological correlates of subliminal perception of facial expressions in individuals with autistic traits: A backward masking study. Front. Hum. Neurosci. 11, 256. doi: 10.3389/fnhum.2017.00256.
Walter, E., Dassonville, P., Bochsler, T. M. (2009). A specific autistic trait that modulates visuospatial illusion susceptibility. J. Autism Dev. Disord. 39, 339–349. doi: 10.1007/s10803-008-0630-2.
Yamasaki, T., Fujita, T., Ogata, K., Goto, Y., Munetsuna, S., Kamio, Y. et al. (2011). Electrophysiological evidence for selective impairment of optic flow perception in autism spectrum disorder. Research in Autism Spectrum Disorders. 5, 400–407. doi: 10.1016/j.rasd.2010.06.002.
Zilbovicius, M., Boddaert, N., Belin, P., Poline, J. B., Remy, P., Mangin, J. F. et al. (2000). Temporal lobe dysfunction in childhood autism: a PET study. Positron emission tomography. Am. J. Psychiatry. 157, 1988–1993. doi: 10.1176/appi.ajp.157.12.1988.
|
|
Unstandardized Coefficients |
standardized Coefficients |
|
|
|
Variable at T5 |
B |
Std.Error |
Beta |
P value |
|
P200 latency in RW condition |
0.04 |
0.04 |
0.18 |
0.35 |
|
N170 amplitude in AW condition |
0.59 |
0.48 |
0.22 |
0.23 |
|
P200 amplitude in AW condition |
-1.24 |
0.42 |
-0.53 |
0.01 * |
Table 1. Multiple regression analysis table in the model with the smallest CVE at the T5 electrode.
adjusted R2 = 0.20, * indicates p < 0. 05
CVE, cross-validation error
|
|
Unstandardized Coefficients |
standardized Coefficients |
|
|
|
Variable at T6 |
B |
Std.Error |
Beta |
P value |
|
N170 latency in AW condition |
-0.19 |
0.12 |
-0.36 |
0.12 |
|
P200 latency in AW condition |
0.16 |
0.07 |
0.50 |
0.04 * |
|
P200 amplitude in AW condition |
-1.06 |
0.41 |
-0.49 |
0.02 * |
Table 2. Multiple regression analysis table in the model with the smallest CVE at the T6 electrode.
adjusted R2 = 0.20, * indicates p < 0. 05
CVE, cross-validation error
No competing interests reported.
- supplementaryfilemovie.mp4
Supplementary movie (Visual Stimuli)