1 Background

Autism spectrum disorder (ASD) is characterized by difficulties in social communication and restricted, as well as repetitive, patterns in behaviors, interests, and activities [1]. In recent studies. Individuals with ASD have been reported to have sensory processing problems, motor dysfunction, impaired executive functioning, and deficits in risk perception [2,3,4,5,6,7]. Their susceptibility to injury is one difficulty caused by these dysfunctions in the context of daily living. A heightened risk of injury in individuals with ASD has been identified in previous studies [8, 9]. These injuries are frequently the result of falls or collisions with objects [10]. The susceptibility to injury in individuals with ASD has primarily been examined during their early childhood [11, 12]. The motivation of the present study was to address whether this susceptibility factor would also be the case in adulthood.

Impaired perceptual–motor coordination may underlie the high susceptibility to injury observed in individuals with ASD. Perceptual–motor coordination refers to the utilization of perceptually derived information (e.g., from vision or touch) in the control of ongoing movements [13]. With regard to perceptual and motor integration, individuals with ASD are known to have an inaccurate perception of the spatial relationship between the body and the environment. Linkenauger et al. [14] found that adolescents and adults with ASD showed a greater degree of dissociation between estimated movement and actual movement than controls matched for age and IQ. Participants performed three tasks: a reachability task, in which they judged the maximum distance they could reach with their arm; a graspability task, in which they judged the maximum object size they could grasp with their fingers; and an aperture passability task, in which they judged the minimum aperture size their hand could pass through. The results showed that the errors between the estimated movement and the actual movement in the ASD group were much greater than in the controls for all tasks, particularly for the aperture-passability task. This suggested that individuals with ASD had an inaccurate perception of the spatial relationship between the body and the environment (the aperture size). Given that the most common types of injuries for children with ASD could be related to being struck by or against objects [10], contexts such as the aperture-passability task implied that they have difficulties perceiving an appropriate distance from their surrounding objects.

The perception of individuals with ASD is also classically described as a detail-focused processing style, which is characterized by attention to detail [15]. This perceptual style is characterized by a tendency to focus attention on specific details rather than on the overall context. This becomes particularly apparent during tasks that require divided attention, in which individuals need to pay attention to multiple locations at the same time [16,17,18]. In general, when attention is focused on a single stimulus, such as a spotlight, it is processed preferentially at the expense of other stimuli [19, 20]. If the non-prioritized information is important for body-related spatial perception, individuals with ASD might not be good at appropriately organizing and dealing with all of the perceptual information. Such difficulties in processing or integrating perceptual information may affect motor planning during voluntary movement [21, 22].

Gowen and Hamilton [23] proposed that, in individuals with ASD, their detail-focused processing style might contribute to their anticipatory motor planning strategies. Several studies have shown that individuals with ASD have difficulty planning a sequence of actions to achieve a goal [24, 25]. The abilities relative to motor planning skills were often examined in tasks known as chained tasks, which were characterized by a sequence of actions: an initial movement, followed by a subsequent purposeful movement [26, 27]. Fabbri-Destro et al. [24] used a task in which participants reached out to grasp an object (the initial movement) and then placed it in a larger or smaller container (the subsequent movement). Their results showed that, for typically developing children, their initial reach to grasp the object was slower when the target container was smaller, indicating that they adjusted their initial movement depending on the difficulty of the final goal. However, for children with ASD, there was no difference in the reach-to-grasp time when the container size was made smaller, suggesting that such children prioritize the processing of positional information about the initial movement goal and, thus, have no time to predictably plan the subsequent movement after the initial movement. Cattaneo et al. [25] used electromyography (EMG) to record muscle activity associated with mouth opening during a sequence of eating actions in children with ASD. Participants were instructed to lift food and either carry it to their mouths or to a container on their shoulders. The results showed that, for typically developing children, the EMG activity for the mouth movement began before their hands grasped the object. In contrast, for children with ASD, the EMG activity began after their hands brought the food to their mouths. These previous studies have shown that difficulties in motor planning in individuals with ASD can be revealed from a chained task that requires the integration of the entire action rather than planning and executing each component of the action separately, which is reminiscent of the detail-focused processing style.

The present study was designed to examine the relationship between impaired perceptual–motor coordination and autistic traits in healthy adults. A previous study examining autistic traits in a large adult population sample indicated that different profiles of autistic traits, including detail-focused processing styles, tend to occur in adult nonclinical populations [28]. Consistent with this idea, significant correlations between ASD traits and perceptual function in typically developing individuals have been reported recently [29,30,31]. For the examination, an original action-selection task was developed based on a chained task (Fig. 1). In this task, the movement of reach-to-grasp for the marble was regarded as the initial movement (Fig. 1a, b). The subsequent movement was transporting the marble from a right-side container to a left-side container, along with the action selection for avoiding obstacles (Fig. 1a). The task was designed to assess the perceptual accuracy and predictive attentional properties for action selection of subsequent movements in a sequence of actions. In line with the study of Linkenauger et al. [14] the task involved crossing an aperture with the dominant right hand. However, unlike the previous study in which participants were only asked to estimate their ability to pass through the aperture, participants were asked to actually pass through the aperture. Participants were instructed to select a passage through or above the aperture to transport a handheld marble to a final goal while avoiding collisions with an obstacle (Fig. 1c). The shape of the obstacle was manipulated to vary the width of two apertures created between two objects (an aperture at the entrance and one at the exit) (Fig. 1d). The aperture sizes were manipulated with the aim of inducing detail-focused processing in participants with higher autistic traits. To successfully complete the task, participants needed to plan their subsequent movement (i.e., whether to pass through the aperture at the exit). In some trials, the aperture at the entrance became very wide so that participants would be likely to try to pass through it, even though the size of the aperture at the exit was too narrow to pass through. We hypothesized that adults with higher autistic traits would misjudge apertures that were actually impassable to be passable, resulting in frequent collisions with the aperture at the exit. Furthermore, we hypothesized that, because of their detail-focused processing style, participants with higher autistic traits would fixate on the aperture at the entrance more frequently than participants with lower autistic traits. We believed that this attentional style would cause participants with higher autistic traits not only to misjudge the aperture size but also to have insufficient motor planning for choosing the way to avoid an obstacle, leading to hesitation behavior (i.e., an attempt to pass through an impassable aperture before deciding to pass above it). If these hypotheses were supported, then the findings would suggest that impaired perceptual–motor coordination, which could occur due to difficulties in perceiving spatial relationships and anticipatory motor planning, might be an important factor leading to the susceptibility to injury in individuals with ASD.

Fig. 1
figure 1

Experimental setup and condition. Regarding the experimental setup, panel a is seen from above, panel b is from the side, and panel c and panel d are seen from the front. In Fig. 1c, the areas of interest (AOI) for eye-tracking data analysis are defined as the aperture at the entrance (the orange rectangle), the aperture at the exit (the green rectangle) and the container (the blue circle)

2 Methods

2.1 Participants

Sixteen young adults (six females, 25.31 ± 4.35 years of age) participated in this study. Inclusion criteria for participants were being between the ages of 18 and 30, being right-handed, and obtaining a score of 24 or higher on Raven’s Colored Progressive Matrices (RCPM) test, which indicates no intellectual impairment [32, 33]. To measure participants’ autistic traits, the Japanese version of the Autism Spectrum Quotient (AQ) was administered [34, 35]. The AQ consists of 50 questions divided into five subsections: social skills, communication, imagination, attention switching, and attention to detail. This study was approved by the Ethics Review Committee of Tokyo Metropolitan University (approval number: H3-66). Informed consent was obtained from all participants prior to the experiments, and this study was performed in line with the principles of the Declaration of Helsinki.

2.2 Apparatus

The experimental setup is shown in Fig. 1. The apparatus consisted of two obstacles, a marble placed in the container, and a target container. The aperture was created between the upper and lower obstacle. The bottom-line height of the aperture (the lower obstacle height) was fixed at 15 cm on the right side and 9 cm on the left side from the desk (Fig. 1d). The lower obstacle was a 19.5 cm square (Fig. 1a). The upper obstacle (height 6 cm, depth 4 cm and width 19 cm) had a variable height structure that allowed the width of the aperture on the left side (i.e., the exit) to be adjusted to the first decimal place (Fig. 1b, d). The width of the aperture on the right side (i.e. the entrance) was 2, 4 or 6 cm wider than the exit width (Fig. 1d). A 14.5 mm diameter marble placed on a plastic container was positioned on the right side of the apparatus. A target container was positioned on the left side of the apparatus. In addition, two tapes were prepared to make participants constantly control their hand movement. A white tape was placed 7 cm away from the front side of the lower object (as a line parallel to the placed marble). The tape was used as the guide on which participants aligned a hand-held marble while moving their arm in the space between the two obstacles (Fig. 1b). A piece of black tape, on which participants put their index finger before starting the trial, was placed 20 cm from the marble (Fig. 1a). Participants wore Tobii Pro Glasses 2 (Tobii AB, Sweden) and sat on a stool placed in front of the desk. Eye-tracking data measured by Tobii Pro Glasses 2 were processed with Tobii Pro Lab software (version 1.181, Tobii AB, Sweden) at a frequency of 100 Hz.

2.3 Task and procedure

Participants were tasked with transporting a marble placed on the right side of the apparatus to a plastic container positioned on the left side of the apparatus, either by passing through a space between the two obstacles or by passing over the upper obstacle. Participants were instructed to move the marble as fast as possible, choosing to pass through the aperture if they deemed it possible to do so without colliding with the obstacle. Before starting a trial, participants sat on a stool in front of the desk and positioned their right index finger on the black tape affixed to the desk. The height of the desk was adjusted to align with the participant’s xiphoid process. They kept their eyes closed prior to initiating the task and opened their eyes as soon as they heard the auditory cue to start the task.

Based on the configuration of the two obstacles, the participant selected whether to pass through the aperture situated between the obstacles or to pass above the upper obstacle (i.e., take a detour). Throughout the trials, to maintain the location of the hand and the marble in the sagittal dimension when passing through the aperture between the two obstacles, participants were instructed to ensure that they were traversing the handheld marble over the line of the trajectory guide placed on the lower obstacle (white tape; see Fig. 1b). The arm movement of each participant was recorded with a video camera so that the number of times they chose to pass through the aperture could be counted. For the trials in which participants chose to pass through an aperture, we counted the number of collisions with the obstacle and obvious deviations from the trajectory guide tape. For trials in which they chose to pass above the upper obstacle, we counted the number of trials in which they tried to pass through an aperture before they finally chose to pass above the upper obstacle (i.e., hesitation).

The aperture width at the exit (i.e., the exit width) was set relative to the hand width as defined by the distance between the second metacarpal head and the first metacarpal head when participants pinched the marble with the thumb and index finger. There were four widths for the aperture at the exit: 0.95, 1.10, 1.25, and 1.40 times the width of the hand. There were three widths for the aperture at the entrance (i.e., the entrance width): 2, 4, and 6 cm wider than the aperture at the exit. There were 48 main trials: four repeat trials for four widths for the aperture at the exit and three widths for the aperture at the entrance. The trials were blocked for each width of the aperture at the exit. The block order was randomized.

In this study, participants were presented with the same aperture width at the exit for 12 consecutive trials. One may assume that this could affect the results, such as consistently selecting the same behavior once participants chose the behavior of either passing through it or passing above. We believe that the impact of the procedure would not be critical for several reasons. First, participants were not informed that the presentation of the aperture width at the exit was blocked. In fact, since the aperture width at the entrance was randomly changed, at least some participants were not aware that the aperture at the exit was constant. Second, if participants repeated the same behavior, then switching the behavior would not occur as the trials progressed. However, comparisons of the number of switching the behavior between the first half of trials (6 trials) and the second half of trials for each aperture width at the exit showed no significant difference for all widths (mean switching times in the first and second halves were0.50 trials ± 0.87 and 0.88 trials ± 1.41 for the 0.95 times exit width condition, 1.56 trials ± 1.94 and 1.81 trials ± 1.29 for the 1.10 times exit width condition, 1.06 trials ± 1.43 and 1.19 trials ± 1.33 for the 1.25 times exit width condition, and 0.38 times ± 0.78 and 0.38 times ± 0.78 for the 1.40 times exit width condition). Taken collectively, although we could not fully deny the impact on the present findings, the impact itself would be small, if any.

2.4 Dependent variables

There were four types of dependent variables: the number of collisions, the number of hesitation behaviors, the body-related spatial perception, and the eye-tracking data.

The number of collisions with obstacles was counted. Increasing collisions in conditions where the entrance width was very wide indicated that participants were affected by the entrance width. A trial in which the participant tried to pass through an aperture before finally choosing to pass above the upper obstacle is described as hesitation behavior. This behavior was observed during the time between grasping the marble and just before passing through the entrance of the aperture. The number of hesitation behavior was also measured to examine how participants accurately planned the subsequent action-selection movement from the initial movement phase. In other words, hesitation behavior was considered to represent a lack of consideration of obstacle avoidance pathways when reaching for the marble (i.e., anticipatory motor planning).

Following previous relevant studies [14, 36], we evaluated the ability to perceive the spatial relationship between the body and the environment which was calculated as the sensitivity, consistency, and accuracy of the body-related spatial perception. These dependent variables were calculated as follows. First, the performance in each trial was evaluated and assigned to one of three categories: success (S), collision (C), or passing above (A). Success (S) meant that participants successfully passed through the exit aperture without striking the obstacle. Collision (C) indicated that they chose to pass through the aperture but collided with the obstacle or deviated from the trajectory guide tape. Passing above (A) entailed participants taking a detour (i.e., going over the upper obstacle) instead of passing through the aperture. Using the percentage of performances in each category among all trials, we calculated the attempt rate and the success rate. For each participant’s hand-width ratio multiplied by the aperture at the exit, the attempt rate was calculated as the percentage of trials among all trials in which the participant chose to pass through the aperture. The success rate was calculated as the percentage of trials in which the participant successfully passed through the aperture without striking any obstacles.

$$The\, attempt\, rate\, = \frac{S+C}{S+C+A}$$
$$The\, success\, rate\, = \frac{S}{S+C}$$

Attempt rate and success rate trends were approximated using a logistic curve. Figure 2 showed the example of the logistic curve of participants with lower autistic traits who had 7-point AQ scores, and participants with higher autistic traits who had 34-point AQ scores. For each participant, the attempt rate and success rate were plotted on each ratio of the exit width for hand size. The decision threshold (DT) as the ratio of the exit width at which the attempt rate reached 50% and the affordance threshold (AT) as the ratio of the exit width at which the success rate reached 50% were estimated using the following equations[37].

Fig. 2
figure 2

The approximation of the logistic curves for the 6 cm entrance-width condition. Panel a shows that the slopes of the decision threshold (DT) and affordance threshold (AT) curves were steeper and the difference between DT and AT was smaller for participants with lower autistic traits. Conversely, panel b illustrates that the DT and AT curves are shallower and the difference between DT and AT is larger for participants with higher autistic traits

$$P\left(r\right)=\frac{1}{1+{\text{exp}}\left(-a\left(r-DT\right)\right)}$$
$$P\left(r\right)=\frac{1}{1+{\text{exp}}\left(-a\left(r-AT\right)\right)}$$

In the above formulas, r denoted each ratio of each exit width, P(r) was the estimated probability for each ratio of the exit width, a was the slope of the approximated curve, and DT or AT was the 50% probability. The a of the slopes of the two curves for the attempt and success rates was used as the index for a(DT): sensitivity of the body-related spatial perception and a(AT): consistency of the body-related spatial perception. The difference between DT and AT was calculated as an index of |AT-DT|: accuracy of the body-related spatial perception. Although AT was usually greater than DT, AT could be less than DT when a participant chose taking a detour very frequently and succeeded all trials without collision when passing through the aperture. Therefore, this difference value was absolutized and used as an accuracy (estimation error) value. The estimation of a, DT, and AT was performed by approximating the logistic curve using nonlinear least squares in the curve-fitting toolbox in MATLAB R2020b (MathWorks Inc., Natick, MA, USA).

Gaze behavior was measured using the eye tracker, and the fixation at an area of interest (AOI) was quantified during the initial movement phase (time of interest: TOI). Fixations were defined as a gaze that remained on a location according to the Tobii I-VT (attention) filter, which incorporated a velocity threshold of 100 visual degrees per second (̊/s), a criterion for merging the adjacent fixation (maximum angle between fixations: 0.5 degrees; maximum time between fixations: 75 ms) and a window length of 20 ms for the I-VT fixation classifier, as set by default in Tobii Pro Lab [38]. The AOI was designated at the aperture at the exit, the aperture at the entrance, and the marble container (Fig. 1d). To quantify the amount of fixation, the total duration of fixation (TDF), which is the accumulated duration of fixations on each AOI, was converted to a percentage as the proportion of TDF (PTDF).

$$PTDF\, in\, the\, aperture\, at\, the\, exit\, or\, entrance\, or\, marble\, container\, \left[\%\right]= \frac{TDF in\, the\, aperture\, at\, the\, exit\, or\, entracne\, or\, marble\, container\, \left[msec\right]}{TOI \left[msec\right]}\times 100$$

To quantify the priority of each AOI, the time to first fixation (TFF) was defined as the time elapsed before each AOI was initially viewed from the moment participants looked at the experimental object.

2.5 Statistical analysis

Poisson regression was used to model count data, such as the number of collisions and hesitation behavior, in this experiment. Poisson regression analysis was conducted to examine whether collisions and hesitation behavior are more frequently observed in participants with higher autistic traits, using the AQ score as the independent variable and the number of collisions and hesitation behavior as the dependent variable for each entrance-width condition. In addition to the regression coefficient, the incident rate ratio (IRR) and its 95% confidence interval were calculated.

To investigate whether it would be difficult for adults with higher autistic traits to perceive the spatial relationship between the body and the environment, we performed a Pearson’s correlation analysis between the AQ score and the indexes of the body-related spatial perception (sensitivity, consistency, and accuracy) to the exit width for each entrance-width condition. The eye-tracking data was also analyzed using Pearson’s correlation to investigate the properties of attention to the aperture for each entrance-width condition relative to the AQ score. In correlation analysis, the eye-tracking data was averaged for each participant in each condition, making it difficult to understand the relationship between gaze behavior during the initial movement and obstacle collision in the subsequent movement for each trial. Therefore, a generalized linear mixed model (GLMM), which allows us to examine this issue based on the variability of behavior for each trial, was used to examine the relationship. Although the sample size for this study was small (N = 16), 16 repeated trials were included for each subject in each exit-width condition. In an experiment with such a small sample size and many repeated trials, the GLMM is capable of extracting the characteristics of each trial [39,40,41]. Specifically, the presence or absence of collisions with obstacles (binomial responses) in the conditions, in which significant correlations were found between the AQ score and the body-related spatial perception indices, was predicted by autistic traits, the exit width condition, and the eye-tracking data. The data was analyzed with the lme4 package [42] of the R system [43] using the GLMM. Helmert coding was one kind of contrast coding that compared each level of a categorical variable to the mean of the subsequent levels. The exit width condition was used to adapt the Helmert coding: 0.95 times as − 1, 1.1 times as − 0.5, 1.25 times as 0.5, and 1.40 times as 1. Our analysis included the factor of the AQ score (autistic traits), the exit width-condition factor, the PTDF (at the exit, entrance, and marble container), the TFF (at the exit, entrance, and marble container), and their two-way interactions (AQ score x each PTDF and AQ score x each TFF). If a significant interaction was found, a simple slope test was performed. For the random effects, a random intercept for participant and a random by-participant slope for the AQ score were included. In terms of the most parsimonious model to explain the collision rate variances, Akaike’s information criterion (AIC) was used to compare the estimated model and the null model. All statistical analyses were performed using R-4.2.0 [43].

3 Results

Poisson regression analysis results revealed that the AQ score was a significant predictor of the number of collisions under the 6 cm entrance-width condition (β = 0.075, p = 0.002, IRR = 1.078 [1.034–1.127]), but there was no significance in the 2 cm (β = 0.037, p = 0.098, IRR = 1.037 [0.985–1.097]) and 4 cm (β = 0.030, p = 0.264, IRR = 1.030 [0.990–1.075]) entrance-width condition. Unlike the number of collisions, the AQ score was a significant predictor of the number of hesitation behaviors under the 2 cm (β = 0.150, p = 0.005, IRR = 1.162 [1.056–1.308]) and 4 cm (β = 0.242, p = 0.011, IRR = 1.274 [1.093–1.627]) entrance-width condition, but there was no significance in the 6 cm entrance-width condition (β = 0.069, p = 0.377, IRR = 1.071 [0.932–1.280]) (Fig. 3).

Fig. 3
figure 3

Poisson regression analysis results for predicting the hesitation behaviors from AQ scores. Each line of the figure is the regression line (red: the 2 cm, green: the 4 cm, blue: the 6 cm entrance-width condition), and each line shading (red: the 2 cm, green: the 4 cm, blue: the 6 cm entrance-width condition) indicates the prediction interval. Each dots (red: the 2 cm, green: the 4 cm, blue: the 6 cm entrance-width condition) represent individual data

Correlation coefficients between the body-related spatial perception and AQ score for each entrance-width condition are shown in Table 1. There was a negative correlation with sensitivity of body-related spatial perception (r = − 0.612, p < 0.05) only in the 6 cm entrance-width condition. A negative correlation was also observed for consistency of body-related spatial perception under the 4 cm and 6 cm entrance-width conditions (r = − 0.505, p < 0.05; r = − 0.676, p < 0.01). A positive correlation (r = 0.502, p < 0.05) with the absolute value of the difference between the DT and AT for accuracy of body-related spatial perception was also identified in the 6 cm entrance-width condition. Thus, as the AQ score was higher, sensitivity and consistency were lower, and accuracy was higher in the 6 cm entrance-width condition.

Table 1 Correlations between the AQ score and the dependent variables in each entrance width condition

Table 1 also showed the results of the correlation analysis between gaze behavior and the AQ score for each entrance-width condition. A significant positive relationship was observed between the AQ score and the PTDF in the aperture at the entrance in the 2 cm entrance-width condition (r = 0.596, p < 0.05). Figure 4 shows a heat map of participants’ gaze fixations, with the data from the bottom three AQ scorers used as the figure for participants with lower autistic traits and the data from the top three AQ scorers used as the figure for participants with higher autistic traits. Comparing the figures, participants with higher autistic traits seemed to fixate their gaze near the entrance width in both conditions.

Fig. 4
figure 4

Points of gaze fixation on the experimental apparatus shown on the heat map. The heat map for participants with lower autistic traits in the upper panel is drawn from the data of three low AQ scores, and the heat map for participants with higher autistic traits, in the lower panel, is drawn from the data of three high AQ scores. The green dots indicate the locations where participants held their gaze, and the red dots indicate fixation on the location for a longer period

In GLMM analysis, the data set was 256 trials of the 6 cm entrance-width condition for all participants. The GLMM results of the estimated model (AIC = 135.1), which converged and were significantly better than the null model (AIC = 215.5, p < 0.01), were shown in Table 2. No significant main effects were found in this model. We found a significant interaction between the AQ score and the PTDF at the exit (z = − 2.571, p = 0.01) and between the AQ score and the TFF at the exit (z = − 3.087, p < 0.01). A simple slope test, which was shown in Fig. 5, revealed that the collision rate with the low PTDF at the exit was higher in participants with higher autistic traits than that in participants with lower autistic traits (p < 0.01). In addition, the collision rate with the high TFF at the exit was higher in participants with higher autistic traits than that in participants with lower autistic traits (p < 0.01). The collision rate in participants with higher autistic traits varied depending on the value of the PTDF and TFF at the exit (PTDF: p = 0.02, TFF: p < 0.01), unlike in participants with lower autistic traits.

Table 2 Results of mixed-effects logistic regression in the estimated model
Fig. 5
figure 5

Interaction effect of the AQ score and eye-tracking data on the collision rate. Both graph shows the mean of the collision rate between the AQ score (+ 1SD and − 1SD) and the eye-tracking data (+ 1SD and − 1SD). Graph 5a shows the interactions between the AQ score and the PTDF, and graph 5b shows the interactions between the AQ score and the TFF. Light blue bars represent average values in participants with lower autistic traits, and orange bars represent average values in participants with higher autistic traits. Error bars represent standard errors of the mean. An asterisk (*) indicates a significant difference in group means (**: p < 0.01, *: p < 0.05)

4 Discussion

Participants with higher autistic traits had frequent collisions with the obstacles in the 6 cm entrance-width condition. Correlations between autistic traits and the body-related spatial perception were significant only for the 6 cm entrance-width condition. This suggests that, when the aperture at the entrance is wide enough to explore the possibility to pass through it, then the quality of the judgment and performance to pass through it would be dependent on the autistic traits. In contrast, no significant correlations were observed for relatively narrow apertures (i.e., the 2 and 4 cm entrance-width condition). A possible explanation for the result was that, when the aperture was so narrow that it would not afford successful passage without collisions, then participants would consistently choose passing above, regardless of the autistic traits, resulting in no significant correlations. The results of correlation analyses between gaze behavior and autistic traits showed that participants with higher autistic traits are prone to directing their gaze toward the entrance width even in the 2 cm entrance-width condition. Although the correlation between the gaze analysis and AQ scores in the other conditions did not reach statistical significance, a visual inspection of Fig. 4 appears to show that participants with higher autistic traits shifted their gaze more toward the aperture at the entrance in the 6 cm entrance-width condition. Therefore, we focused on the 6 cm entrance-width condition to investigate the predictive attention that causes collisions with obstacles. The results showed that participants with higher autistic traits were more likely to have collisions with the obstacles when the total time of the fixations on the exit width was short and the time to first fixate on the exit width was late. This result suggests that they were unable to focus their predictive attention on the exit width due to their detail-focused processing style. Furthermore, Poisson regression results of hesitation behavior showed that AQ score was a significant predictor of increased behavioral adjustments just before obstacle avoidance, which implies that the subsequent action was not planned until just before approaching the obstacle. These results suggest that participants with higher autistic traits are likely to exhibit impaired perceptual–motor coordination due to being influenced by their detail-focused processing style.

As hypothesized, the correlation results between body-related spatial perception and AQ score indicated that the sensitivity, consistency, and accuracy of body-related spatial perception became more difficult for participants with higher autistic traits as the entrance width was wider. Participants with higher autistic traits did not seem to understand the clumsiness of their own hand movements and seemed to make incorrect action selections, resulting in collisions even when the aperture width was wide enough to pass through. This may be due to problems with movement execution and control in individuals with ASD [3, 44,45,46]. In addition, this finding may reflect the fact that participants with higher autistic traits tend to focus on immediate distracters and show inaccuracy in body-related spatial perception. A previous review study suggested that individuals with ASD exhibit decreased neural activity and behavioral habituation to stimulus repetition as compared to neurotypical individuals [47, 48]. This can be interpreted as a reduction in the utilization of stimulus repetition expectations to enhance stimulus discrimination. Additionally, another review study demonstrated that reduced neural activity and behavioral habituation to repeated stimulation may be contingent upon attentional processes [49]. A previous study revealed that scattered fixations during the biological motion task correlated with low perceptual adaptation performance [50]. This finding indicates that perceptual adaptation is less likely to occur due to increased scatter in gaze direction, which suggests that attentional function plays a vital role in perceptual adaptation.

Participants with higher autistic traits may have difficulty with anticipatory gaze behavior. Difficulties with predictive feedforward control in children with ASD have previously been shown by kinematic measures [51,52,53]. On the other hand, the lack of a significant correlation in the 6 cm entrance-width condition may provide important suggestion that participants with low autistic traits may have focused on the entrance width to compare it with the exit width. Participants with lower autistic traits may have performed exploratory gaze behavior for perceiving the whole of the aperture information, such as estimating the exit width by referring to the entrance width. Such behavior has also been observed in a visually guided reaching task with self-centered (egocentric condition) or other-centered (allocentric condition) coordinates. According to Umesawa et al. [54] typically developing individuals tend to prioritize other-centered coordinates concerning the surrounding spatial coordinates in their reaching behavior, whereas individuals with ASD tend to prioritize self-centered coordinates. Individuals with higher autistic traits are also more likely to prioritize self-centered coordinates, which may be associated with inflexibility in their eye-gaze behavior [55]. In other words, individuals with higher autistic traits may determine the fixation point according to their own rules, regardless of the surrounding situation. These previous findings support the possibility that individuals with ASD may have determined their action selection by fixating near the entrance width in the narrower entrance-width conditions (i.e., 2 and 4 cm entrance-width conditions) through detail-focused processing, which also resulted in no significant correlation with the body-related spatial perception.

Hesitation behaviors were considered to represent failure to consider obstacle-avoidance pathways while reaching for the marble until just before approaching the obstacle. It was more frequently identified in the narrower entrance-width condition for participants with higher autistic traits. This finding may support previous studies that have suggested that individuals with ASD have difficulty anticipating and planning for a subsequent movement while planning and executing an initial movement [51, 56]. Another possibility is conceivable. The aperture width in subsequent movements was uncertain because it varied with each trial. The human brain inherently generates motor planning that optimizes task performance in the face of uncertainty [57]. Previous studies have shown the differences in exploratory gaze behavior during anticipatory sensorimotor control and uncertainty-related behavioral adjustments in the ASD group [58, 59]. It has also been noted that individuals with ASD exhibit visuomotor patterns that differ from those of typically developing individuals due to their inability to effectively use prior contextual cues and prior verbal instructions [60, 61]. Given that participants with higher autistic traits tended to fixate on the aperture at the entrance, it is possible that they exhibited hesitation behavior to avoid collision by judging that they could not pass through according to the size of the entrance width. This may mean that the action was selected just before the subsequent aperture task, which could contribute to their impaired perceptual–motor coordination.

This study has several limitations. First, we sampled neurotypical individuals in whom AQ scores could be high even without a diagnosis of ASD [35]. Therefore, it is uncertain whether a similar trend is observed in individuals with a diagnosis of ASD. It is necessary to include the ASD group in future studies. Second, the sample size was relatively small. Given that the results of this study were obtained from a sample of 16 young adult participants, it is quite possible, considering the first limitation, that individuals with ASD may not exhibit behavioral patterns similar to the present findings. A more diverse, heterogeneous, and sufficient sample may provide more robust results in future studies. Third, we have only prepared two methods of obstacle avoidance (via passage through or above an aperture). As such, if participants solely judge the passage through the aperture or pass above the obstacle, despite the exit aperture ratio being 1.40 times their hand width, there is a likelihood that logistic curves will not be suitable for estimating indices of body-related spatial perception. In future studies, it will be necessary to determine the minimum exit-width ratio at which participants will select passage through the aperture. Last, there were no interviews about daily susceptibility to injury. Therefore, whether the performances of this experimental task directly affect susceptibility to injury should be examined in future studies. In addition, future studies require adapting this experimental system to children with ASD and examining whether the results are consistent with those of the present study.

5 Conclusions

Our findings indicate that individuals with higher autistic traits exhibit impaired perceptual–motor coordination, which could stem from difficulties in perceiving spatial relationships and anticipatory motor planning. Additionally, analyses of gaze behavior suggested that they tended to prioritize immediate and local information over global information for motor planning. Based on our results, it appeared that individuals with higher autistic traits may be particularly susceptible to problems related to impaired perceptual–motor coordination, which in turn may contribute to a heightened risk of injury in individuals with ASD.