Investigating joint attention in a guided interaction between a child with ASD and therapists: A pilot eye-tracking study

Abstract

Introduction

Deficits in joint attention are commonly seen in children with autism spectrum disorder. Research examining joint attention in autism spectrum disorder commonly uses two broad strategies to cue and measure joint attention: behavioral observation and eye tracking. These strategies trade off prioritizing ecological validity vs. gaze measurement accuracy, with a focus on one sacrificing the other. The purpose of this case study was to develop a method to accurately measure gaze position while maintaining an ecologically valid dyadic interaction.

Methods

A child with autism spectrum disorder completed two developmentally appropriate tabletop activities. Each activity was guided by a different occupational therapist who purposefully used a different interaction style with the child. Mobile eye trackers worn by both the child and the therapist recorded the dyadic interactions. Data collection included audio and video recording of interactive behaviors, eye movements and visual fixations in regions of shared interest.

Results

Differences were detected in gaze use and interactive joint attention behaviors between the therapists working with the child and within the child’s respective dyadic interactions.

Conclusions

The proof of concept method maintained both ecological validity and measurement accuracy of therapist–child joint attention. This method has promise to be scaled for larger studies.

Keywords

Autism joint attention occupational therapist eye movements areas of interest

Introduction

Autism spectrum disorder (ASD) is a neurodevelopmental disorder with a prevalence of about 1 in 59 individuals (Baio et al., 2018). ASD is characterized by deficits in social communication and restricted or repetitive behaviors, leading to persistent impairment (American Psychiatric Association, 2013). Difficulties with joint attention (JA) are persistently seen in children with ASD (Mundy, 2018). JA is the coordination of attention within a dyadic interaction with an element external to the dyad (Mundy, 2018). JA is a pivotal behavior, supporting the development of social cognition (Mundy, 2018) and the acquisition of other developmental skills (see for example Casenhiser et al., 2011). There are two types of JA: initiated joint attention (IJA): the attempt by one partner to focus the other partner’s attention on an external element; and responded joint attention (RJA): the placing of attention on an external element in response to a partner’s attentional bid.

Both IJA and RJA are expressed through gaze shift, with an individual moving their eyes to look at the external element (Mundy, 2018). For example, when playing together, one child may attempt to initiate JA to an interesting toy by first looking to the other child, then looking toward the toy, and then back again to the other child. In addition to gaze shift IJA might also be expressed through contextually appropriate vocalization, pointing or showing (Gulsrud et al., 2014), and RJA might also be expressed through contextually appropriate vocalization, or by taking the action implied by the IJA attempt (Bean and Eigsti, 2012). Looking at a communication partner’s face is a precursor to IJA (Mundy, 2018); however, children with ASD look at a partner’s face less than typically developing children (Falck-Ytter et al., 2013; Hanley et al., 2014).

JA is a pivotal behavior, supporting the development of human social cognition (Mundy, 2018) and the acquisition of other developmental skills, like language (see for example Casenhiser et al., 2011). It is a goal-directed behavior with the purpose of sharing subjective experience with another individual (Mundy, 2018). Through repeated IJA and RJA interactions the developing child gains experience in adopting a common frame of reference with the social partner, with sharing information within that common frame of reference, and with appraising similarity between their own and the other’s perspective. The incremental development of these skills, though JA experiences, supports the development of intersubjectivity and social langue use (Mundy, 2018).

Recent systematic reviews have situated JA as an important component skill supporting occupational performance (Althoff et al., 2019; Eschenfelder and Gavalas, 2017; Tanner et al., 2015). During development, referencing a caregiver’s perspective helps children navigate moments of uncertainty when they are unsure how to appraise an event or object. This referencing helps children increase the range of experiences they comfortably encounter. Additionally, JA supports engagement in co-occupation, a feature of many childhood occupations occurring when two or more people share emotional and physical occupational engagement (Eschenfelder and Gavalas, 2017). Many childhood occupations, such as playing with peers and learning daily living skills with parents, are performed as co-occupations (Eschenfelder and Gavalas, 2017). Interventions targeting JA have been shown to improve social participation (Tanner et al., 2015), play, behavior and adaptive functioning (Althoff et al., 2019).

Children with ASD commonly show reduced quality and quantity of JA, using it less often and less skillfully compared to typically developing peers (Landa et al., 2013). Features of the cueing, the task and the environment all impact JA as demonstrated by children with ASD (Ambrose et al., 2020). Because of its key role in facilitating childhood development, JA is a frequent target for intervention among practitioners providing services to children with ASD (Eschenfelder and Gavalas, 2017). Practitioners can select from many intervention methods to help children with ASD build JA skills (Murza et al., 2016). One group of interventions that may be adopted to help children with ASD build JA are commonly called developmental social pragmatic interventions (Ingersoll, 2010). There are several developmental social pragmatic interventions, such as More than Words, DIR/Floortime, JASPER, and Relationship Development Intervention. Occupational therapists are among the professionals who use developmental social pragmatic interventions (Eschenfelder and Gavalas, 2017) and have published studies investigating their effectiveness (see for example Liao et al., 2014). Other disciplines interested in aiding children with ASD also use these interventions (Althoff et al., 2019; Oono et al., 2013; Tanner et al., 2015). Occupational therapists have used research from across disciplines to build a more comprehensive understanding of JA and its contribution to occupational performance in children with ASD (see for example Tanner et al., 2015). Developmental social pragmatic interventions are sometimes performed as therapist-mediated, and sometimes as parent-mediated (Oono et al., 2013; Tanner et al., 2015). In the latter, trained professionals coach parents in the use of these interventions, and the parents then apply them in daily living situations.

Arranging the environment to support development, rather than directly cueing for specific behavioral displays of developmental skills, is a core feature of these interventions (Ingersoll, 2010), in the belief that the skills that emerge will be more thoroughly integrated and generalized than those from direct cueing. The interaction style used by therapists is a component of the environment that developmental social pragmatic interventions seek to adapt (Kong and Carta, 2013), with the responsive interaction style (Mahoney and Perales, 2005) commonly employed within these interventions.

The responsive interaction style is a complex set of skills (Ruble et al., 2008), summarized as adults consistently using contingency, reciprocity, positive affect and non-directiveness when engaging with children (Mahoney and Solomon, 2016). When adults use a responsive interaction style, they respond sensitively and contingently to the child at a level appropriate for the child’s level of development (Ingersoll, 2010), minimizing communications that command or direct the child (Aldred et al., 2004) or the child’s attention (Shire et al., 2016). Adults prioritize following the child’s focus of attention more often during a task (Girolametto et al., 2007). Adults will allow longer times for a child to respond, and use more gestures, facial expressions and other non-verbal communications in place of verbal communication (Kossyvaki et al., 2014). Adults will also structure the physical and social environment to promote child initiations (Gulsrud et al., 2016), then elongate resulting synchronous interactions (Kong and Carta, 2013) to emphasize co-regulation, emotional expression and affect sharing (Casenhiser et al., 2011).

Consistent use of the responsive interaction style has shown a positive impact on the development of joint attention (Karaaslan et al., 2011; Shire et al., 2016). All the evidence supporting a responsive interaction style to impact JA skills in children with ASD has used observation of video recorded interactions (see for example Casenhiser et al., 2011; Green et al., 2010; Karaaslan et al., 2011). However, these observational methods have been critiqued as lacking precision when determining the target of gaze because some children with ASD demonstrate atypical gaze patterns (Guillon et al., 2014).

Fidelity testing of adults using the responsive interaction style has been similarly measured using observation of the video recorded interactions. While these observational ratings of responsive behaviors provide global ratings of a caregiver’s interaction style, they lack granular detail related to gaze use. Previous eye-tracking studies (see for example MacKenzie and Westwood, 2013) found therapists employed different gaze patterns when examining a clinical scene than did adults without healthcare training. One of the factors suggested for this difference is the result of top-down influences that training exerts on gaze use. Exposure to and practice with a framework for categorizing observations guides gaze use when collecting observations (MacKenzie and Westwood, 2013). It might therefore be expected that a therapist familiar with the responsive interaction style and one naïve to it would demonstrate different gaze use patterns when attempting to elicit JA in a child with ASD.

Both IJA and RJA are in part expressed through gaze (Mundy, 2018). As gaze is a signal of overt visual attention in dyads (Ho et al., 2015), its accurate measurement is important when investigating JA. Computerized eye tracking using corneal reflection is a widely used eye-tracking technology in which near-source infrared light is aimed at the eyes with video that captures the eyes and of the reflections off the corneas and pupils (Duchowski, 2007). Based on the images of the pupils and the reflections, computer algorithms calculate gaze position (Falck-Ytter et al., 2013). Eye tracking offers a measurement tool that more accurately tracks the target of an individual’s gaze than behavioral observation alone (Falck-Ytter et al., 2013; Guillon et al., 2014).

When used to examine JA, tabletop screen-based eye trackers precisely measure gaze position while participants interact with an image or video on a computer screen (see for example Swanson and Siller, 2013). However, the ecological validity of these investigations is questionable as they do not use real-world social interactions, only screen-based simulations (Falck-Ytter et al., 2013; Guillon et al., 2014). By contrast, using mobile eye tracking during dyadic interaction permits accurate measurement of gaze within real-world environments, maintaining ecological validity of live, face-to-face interactions. Some studies have used mobile eye tracking to investigate gaze in children with ASD during a face-to-face interaction (Hanley et al., 2014; Noris et al., 2012). These studies showed that children with ASD look at their partner’s face less than typically developing children. However, these studies used eye trackers only with children and not with the examiners guiding them. Additionally, these studies did not specifically investigate JA.

There is a need to understand how visual attention and interaction style impact JA in a real-world interaction. Mobile eye tracking offers a feasible tool for gaze measurement accuracy in real time while simultaneously permitting live, face-to-face interaction. The purpose of this case study is to explore gaze use and JA between a child and two different occupational therapists’ interaction styles during a real-world interaction using mobile eye-tracking devices.

Methods

This case study received Health Sciences Research Ethics Board approval. Written informed consent was provided by the occupational therapists and the child’s parent, and the child provided verbal assent prior to participation in this study.

Participants

Recruitment for child participants with ASD through a community partner yielded one participant who met the inclusion criteria. The child participant was a 9-year-old boy with a parent report of high functioning ASD diagnosis and no other comorbid developmental, psychiatric or medical diagnoses. The child spoke English as his primary communication method and attended school in a mainstream classroom.

Two occupational therapists provided signed informed consent prior to participation. One occupational therapist (OT1) had 14.5 years of experience working with children with ASD, and 9 years using a responsive interaction style with children with ASD. OT1 was purposefully recruited for their experience with the responsive interaction style. The second occupational therapist (OT2) had 12 years of experience working with children with ASD but no specific knowledge or training using a responsive interaction style. OT2 was purposefully recruited through local professional networking for their combination of clinical experience with children with ASD and lack of training and experience with the responsive interaction style.

Equipment and task

The study was conducted in the School of Occupational Therapy’s Home Simulation Suite, Dalhousie University. The suite is equipped with pan-tilt-zoom (PTZ) cameras, which provided an external scene video that captured the interactions between the therapist–child dyads. SensoMotoric Instruments (SMI) ETG 2w mobile eye trackers were worn by both the therapist and the child during the interactive task. The SMI ETG 2w glasses sample binocular eye position at a rate of 120 Hz and use corneal reflection to estimate gaze location (SMI, 2016a). An integrated scene camera also recorded the user’s audio and scene field of view. The eye tracker’s algorithms calculated gaze position and overlayed a cursor on the gaze target within the scene video in real time. The scene and gaze cursors were displayed on a dedicated laptop connected to the eye tracker via a closed, secure wireless network for monitoring during the experiment.

The task, developed specifically for this case study, was a tabletop block construction designed to resemble an animal. To minimize the learning effect between the child’s interactions with the two therapists, two different animal constructions with a similar level of difficulty were used (a dinosaur with 66 blocks and a spider with 63 blocks). The task was developmentally appropriate, engaging for children aged 8–12 years with ASD and challenging enough to afford opportunities for JA. The child had no previous experience with the experimental task.

A step-by-step pictorial instruction guide was developed for building the animal constructions. The pictorial instruction guide was available to the child and the therapist during their interaction. OT1 and OT2 practiced building the constructions one day prior to dyad interaction. During this practice period OT2 received no additional instruction on the responsive interaction style.

Data collection

After the informed consent process, the child–OT1 dyad each donned a pair of eye-tracking glasses and individually participated in single-point calibration. The SMI ETG 2w one-point calibration method was used, given its efficiency and equivalent measurement accuracy to a three-point calibration process (SMI, 2016a). The calibration process also served as a period permitting the child to become familiar with the glasses. Following calibration, the dyad was seated across from each other at the table. The unassembled blocks were arranged to the child’s right, and the pictorial instruction book was placed to the child’s left. First, OT1 instructed the child that they would work together to build the animal (dinosaur) that was shown in the instruction book. Following completion of the dinosaur build, the child–OT1 dyad eye trackers were removed. The child was then given a 10-minute break prior to the OT2 interaction. The same mobile eye-tracking donning and calibration procedure was then completed with the child–OT2 dyad. The child and OT2 then worked together to build the animal (spider) shown in the second instruction book. During the interactions, only the dyad members were in the area of the home simulation suite sectioned off for the study.

Data analysis

Each dyad’s task generated two mobile eye-tracker audio and video recordings: the child’s perspective and the respective OT1 or OT2 perspective. A third video recording of the dyad interaction was captured from the Home Simulation Suite’s PTZ cameras. Using video editing software, the three videos from each interaction were combined into a single video file – one that showed synchronized views of all three perspectives. These combined, multi-perspective video files were used to extract interactive behavior data.

BeGaze v 3.7 (SMI, 2016b) software was used in the analysis of eye-tracking data. BeGaze’s default settings were used for the low speed event detection dispersion-based algorithm to identify gaze fixations. This algorithm identifies fixation as having a minimum duration of 80 ms and a maximum dispersion of 100 pixels. BeGaze permits users to define specific areas of interest (AOI) on a static reference picture. BeGaze’s semantic gaze mapping (SMI, 2016b) permits algorithmically detected fixations within a perspective video to be mapped onto a reference picture to determine if the fixations fall within defined AOI.

While the mobile eye-tracking glasses were able to accurately identify the target of a user’s gaze during real-world interactions, operational definitions of interactive joint attention behaviors were required. All interactive behaviors considered for analysis are described in Table 1

Table 1.

Interactive behaviors identified for analysis.

Interactive behavior	Brief definition	Rationale
Verbal	First person perspective uses words or prosodic vocalizations.	Words and prosodic vocalizations are used to communicate about the task and about non-task topics.
Block building	First person perspective actively arranges blocks to help produce the emerging structure.	Block construction was the experimental task.
Initiated joint attention (IJA)	First person perspective is looking at partner’s face, then moves gaze to an external referent, and then back to partner’s face.	The face-referent-face pattern of gaze use is an accepted behavioral definition of IJA (see for example Mundy, 2018) that has been used in other eye-tracking studies of joint attention (JA) (see for example Johnson et al., 2012).
Block passing	When the occupational therapist passes blocks to the child.Not coded from child perspective, as the child did not pass a block to the occupational therapist during either variant.	Passing the block is part of the experimental task.
Mutual facial gaze (MFG)	When the occupational therapist and child gaze cursors simultaneously fall in each perspective’s head areas of interest.	Looking at a partner’s face is a common measure in eye-tracking studies of children with autism spectrum disorder (see for example Noris et al., 2012). It is a precursor to IJA (Mundy, 2018). As this study used two eye trackers, identifying MFG was an opportunity to examine interactive gaze at a partner’s face.
Partner response to IJA attempt	When the partner follows first person perspective’s IJA attempt, moving their gaze to referent indicated during IJA attempt.	This records the success of IJA attempts. IJA attempts are successful if the partner moves their gaze onto the referent. They are unsuccessful if the partner takes actions other than moving their gaze to the referent.

Data analysis proceeded in two steps. First, the BeGaze semantic gaze mapping process was used to map each algorithmically identified fixation onto a reference picture for each perspective. Figure 1

Figure 1.

Areas of interest reference.

illustrates the person, task and environment-based AOIs used to map the child’s gaze during the interaction with OT1. The AOIs defined in this image are OT1’s head, torso and hands, the build, the blocks in reserve and the instruction book. A similar reference picture was created for each participant in each interaction, in which the same AOIs were defined. An additional AOI, white space, was used in each reference picture. The white space AOI comprised all areas of the scene not attributed to any other AOI and represented the non-task and non-partner elements of the visual scene.

Once each fixation was mapped, the BeGaze software calculated the overall fixation count and duration in each AOI. This permitted comparison of the proportion of each engagement the user looked at each AOI; thus, the child’s time spent looking at each AOI could be compared between occupational therapists, and the occupational therapists’ time spent looking at each AOI could be compared.

Next, interactive behaviors occurring within each interaction were identified. Using the combined, multi-perspective video files, multi-pass frame-by-frame analysis was conducted. In digital playback, videos recorded by the eye trackers displayed the elapsed time to the millisecond and a cursor showing the user’s gaze location. Each pass through the video file identified exact start and stop times for one specific interactive behavior, from only one perspective. This resulted in frequency counts of each interactive behavior for each perspective, permitting comparison of the rate of use of each interactive behavior between interactions. Thus, the frequency of the child’s use of an interactive behavior could be compared between occupational therapists, and the occupational therapists’ frequency of interactive behaviors could be compared. Three of the interactive behaviors – mutual face gaze (MFG), IJA and RJA – were defined in part by the gaze cursor generated by the eye-tracking glasses. These definitions are the link between maintaining accuracy when measuring gaze location and ecological validity when examining JA.

IJA was operationally defined as a gaze sequence that started at the partner’s head AOI, moved to a referent, and then returned to the partner’s head AOI, consistent with the definition used by Mundy (2018). To be considered IJA, the initiator’s gaze was required to fixate on the referent before the responder’s gaze. Additionally, to be considered IJA, the gaze shift sequence had to intentionally attempt to direct the responder’s attention to the referent within the context of the interaction. For example, if one therapist moved their gaze from the child’s head AOI to their wristwatch and back to the child’s head AOI, it would not be considered IJA as the occupational therapist’s intent was not to direct the child’s gaze to the wristwatch but rather to check the time.

MFG is operationally defined as the times when the therapist and child gaze cursors are simultaneously in the other’s head AOI. In this case study’s definitions, MFG is not necessarily a prerequisite of IJA. Although IJA requires the initiator’s gaze cursor to start in the partner’s head AOI, the partner need not hold their gaze cursor in the initiator’s head AOI for the initiator to complete an IJA attempt. While other studies have used eye trackers with dyadic partners (see for example Ho et al., 2015; Schwarzkopf et al., 2016), to our knowledge none have identified MFG as a variable of interest. As children with ASD tend to look less at faces (Falck-Ytter et al., 2013; Guillon et al., 2014; Hanley et al., 2014; Noris et al., 2012), identifying MFGs provides an opportunity to examine joint, simultaneous face looking.

Results

Table 2

Table 2.

Gaze and interactive behavior.

	OT1 (Child)	OT2 (Child)
Area of interest	Fixation time as percent of total time
Head	8.5 (5.4)	4.7 (0.9)
Body	8.5 (0.7)	2.8 (0.4)
Hands	17.3 (4.5)	15.8 (2.5)
Build	23.3 (51.0)	35.8 (47.9)
Instructions	10.6 (18.5)	20.2 (16.0)
Blocks in reserve	6.2 (1.1)	1.0 (0.9)
White space	5.9 (6.3)	6.7 (17.6)
Interactive behavior	Interaction frequency per minute
Verbal	6.6 (6.8)	10.3 (6.8)
Pass blocks	1.4 (0.0)^a	0.7 (0.0)^a
Build blocks	0.6 (2.3)	2.4 (2.2)
Initiated joint attention (IJA)	1.7 (0.3)	0.8 (0.3)
Mutual facial gaze (MFG)	3.9^b	1.5^b
Partner response to IJA attempts	0.7 (0.8) c	1.0 (0.8) c

As the child did not pass blocks to the occupational therapist in either engagement, there is no child data for that interactive behavior.

MFG involves both the occupational therapist and the child, producing only one frequency measure for each engagement.

^cThe “partner response” to IJA attempts records the frequency of successful partner responses/IJA attempts.

presents the results for the gaze mapping behavior per AOI and interactive gaze behaviors.

Fixations by AOI

After semantic gaze mapping, BeGaze collated the total time each member of the dyad had fixated gaze within each AOI. Because the duration of the engagements differed between the dyads (11:32 with OT1 and 9:39 with OT2), the data in Table 2 is presented as percentages of the total fixation time.

The results show the occupational therapists used gaze differently when working with the child, and the child used gaze differently when working with each therapist. OT1, who had experience using the responsive interaction style, looked more at the child’s head and body AOIs, and at the blocks in reserve AOI. OT2 looked more at the instructions and build AOIs. When working with OT1, the child looked more at the occupational therapist head AOI. When working with OT2, the child looked more at the white space AOI.

Interactive gaze behaviors

This step of the analysis focused on understanding IJA, RJA and MFG use in each dyad. Interactive behaviors coded from video analysis are presented as frequency per minute. Table 2 presents the gaze and interactive behavior data collected across the two engagements. In addition to frequency measurements, contextual information can be extracted from the combined videos. Figure 2

Figure 2.

Comparison of child interactions with OT1 and OT2.

provides a visual representation of the respective therapist and child interactions during the experimental task completion.

OT1 attempted IJA 19 times, 12 of which were immediately preceded by MFG. The instruction book was the most common referent of OT1’s IJA (14 times). Other referents were the blocks in reserve (three times) and the build (two times). The child accurately responded to OT1’s IJA 14 times. During successful IJA, the child’s gaze entered an average of 3.3 AOIs between its location at the beginning of OT1’s IJA and its end point on the referent. These AOIs were uniformly along the scan path between the child’s gaze location at the start of OT1’s IJA and the referent, suggesting the child moved their gaze directly to the referent.

With OT1 the child used IJA three times, one of which was immediately preceded by MFG. All of the child’s IJA used the instruction book as the referent. OT1 accurately responded to two of the child’s IJA. When responding to the child’s IJA, OT1’s gaze moved through an average of 3.0 AOIs between its location at the beginning of the child’s IJA and its end point on the referent. These AOIs were uniformly along the scan path between OT1’s gaze location at the start of the child’s IJA and the referent, suggesting OT1 moved their gaze directly to the referent. The child and OT1 demonstrated MFG 45 times during the interaction. OT1 initiated MFG 17 times and the child initiated MFG 28 times. OT1 ended MFG 29 times, the child ended MFG 15 times, and OT1 and the child simultaneously ended MFG one time.

OT2 attempted IJA eight times, two of which were immediately preceded by MFG. All of OT2’s IJA used the instruction book as the referent. The child accurately responded to OT2’s IJA four times. During successful IJA, the child’s gaze entered an average of 2.8 AOIs between its location at the beginning of OT2’s IJA and its end point on the referent. These AOIs were uniformly along the scan path between the child’s gaze location at the start of OT2’s IJA and the referent, suggesting the child moved their gaze directly to the referent.

With OT2, the child used IJA three times. Two of these attempts were immediately preceded by MFG. All of the child’s IJA used the instruction book as the referent. OT2 accurately responded to all of the child’s IJA. When responding to the child’s IJA, OT2’s gaze moved through an average of 3.0 AOIs between its location at the beginning of the child’s IJA and its end point on the referent. These AOIs were uniformly along the scan path between the OT2’s gaze location at the start of the child’s IJA and the referent, suggesting OT2 moved their gaze directly to the referent.

The child and OT2 demonstrated MFG 15 times during the interaction. OT2 initiated MFG eight times and the child initiated MFG seven times. OT2 ended MFG three times, and the child ended MFG 12 times. When OT2 initiated MFG, five occurrences were contextually linked to the child’s actions: after the child viewed the pictorial instruction book (four times) and after completing a step in the construction (one time). Two were unrelated to the construction task. When OT2 ended MFG, two of the three occurrences immediately preceded IJA.

Discussion and implications

The purpose of this case study was to pilot a mobile eye-tracking method to explore gaze use and joint attention between a child and two different occupational therapists and interaction styles. This study introduced simultaneous recording from multiple mobile eye-tracking devices as a unique method that maintains both high measurement accuracy and ecological validity when examining JA in a single child with ASD. Measurement accuracy was maintained through the use of mobile eye-tracking glasses, worn by both members of each dyad (child–OT1/2) in the case study. The tabletop task created a natural physical boundary to the task, limiting the variability of the visual scene. This allowed a single and static reference picture to map all gaze fixations from a user’s perspective, which is dynamic and three dimensional. Semantic gaze mapping also provided aggregate gaze use data of fixation time per AOI for each member in each dyad, permitting comprehensive investigation of gaze allocation differences between dyads.

This case study used two occupational therapists and one child, with each therapist using a different interaction style. The study detected differences in gaze use and interactive behaviors between the occupational therapists. OT1 looked more at the child’s body and hands, and less at the build and instructions than did OT2. OT1 also spoke and built less than did OT2. Together, these differences are consistent with what would be expected if OT1 used a responsive interaction style and OT2 did not. The increased frequency of looking at the child displayed by OT1 is a prerequisite for responding sensitively and contingently to child communications, as the guide must attentively monitor the child in order to respond. Within the responsive interaction style, the reduced frequency of talking and building OT1 displayed could be interpreted as arranging the social environment to create opportunities for development, rather than specifically cueing for a display of developmental behavior (Ingersoll, 2010). When an adult guide talks less and moves slower, it creates more time and space for the child to respond.

Screen-based eye-tracking studies (Falck-Ytter et al., 2013; Guillon et al., 2014) and mobile eye-tracking studies (Hanley et al., 2014; Noris et al., 2012) have shown children with ASD spend less time looking at a partner’s face. In this case study, both partners wore eye trackers. This allowed for the operational definition of IJA and MFG based on eye-tracking data, permitting an analysis that joins the measurement accuracy of the eye tracker with the ecological validity of a face-to-face interaction. The identification of MFG, the times each partner looked at each other’s face, is a unique measure of interactive behavior. This is an important measure, as IJA is preceded by looking at a partner’s face (Mundy, 2018), and the responding partner must see the initiation in order to respond to it. OT1 and the child looked more at each other than did OT2 and the child, generating more frequent periods of MFG. The periods of MFG are more evenly distributed across the child’s interaction with OT1 than they are in the child’s interaction with OT2, where they cease approximately two-thirds of the way through the task.

The child also spent more time in the white space AOI when working with OT2 on the task than when working with OT1. Hanley et al. (2014) included “non-partner” AOIs in their face-to-face eye-tracking study of gaze use of children with ASD. This non-partner AOI seems similar to this case study’s white space AOI. In Hanley et al. (2014), children with ASD are found to fixate on the non-partner AOI more than all other AOIs in their engagement. The authors propose this finding could reflect children with ASD’s gaze aversion in social engagements, where children look away to manage physiological arousal associated with social gaze. Hanley et al. (2014) also propose distraction or lack of interest as possible explanations for the high dwell-time in their non-partner AOI. While this case study does not replicate Hanley et al.’s (2014) findings, the difference in the child’s proportional dwell-time in the white space AOI between occupational therapists is potentially consequential. It suggests that individuals with ASD can look at non-task, non-social targets at varying frequencies, dependent on the context. In this case study, an important part of the context is the interaction style employed by the occupational therapists, thus reiterating that the responsive interaction style can reliably be used to facilitate JA in children with ASD.

The methods developed and initial findings from this pilot study suggest the approach may be scaled to larger studies. Future investigations can employ this two-eye-tracker method to examine the impact of adults’ use of the responsive interaction style on JA in children with ASD. More broadly, this method can be adapted to investigate other gaze use, interaction style or task-related factors that may also influence JA in children with ASD.

Conclusion

This case study presents a novel method for studying JA in children with ASD using mobile eye trackers. By using the mobile eye tracker’s gaze cursor overlay in the video playback to operationally define JA, the method maintains both measurement accuracy and ecological validity. This method can be used by larger-scale studies to examine questions about JA in children with ASD. The results will inform discussion on the research and practical feasibility of conducting further investigations on a larger scale.

Key findings

• The mobile eye-tracking method piloted in this case study maintains both ecological validity and measurement accuracy when studying JA in children with ASD.

• The method detected differences in gaze and interactive behaviors between occupational therapists, and within a single child when working with different occupational therapists.

• A child with ASD can look at non-task, non-social elements of a scene at varying frequencies, depending on context.

What this study has added

This study introduces a method of examining JA in children with ASD using mobile eye trackers that maintains the ecological validity of real-time face-to-face interaction while accurately measuring gaze position. This method can be used in larger-scale studies to examine the influence of therapist interaction style or other factors on the JA demonstrated by children with ASD.

Footnotes

Acknowledgments

Shelley McKibbon, Information Services Librarian at the W.K. Kellogg Health Sciences Library at Dalhousie University, assisted with refining the literature search strategy used in this manuscript.

Research Ethics

This study received approval from the Health Sciences Research Ethics Board effective January 5, 2018. HREB file number 2017-4343.

Consent

Written informed consent was obtained from participants and guardians. Child verbal assent was obtained prior to participating in experimental tasks.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

This research was supported by Dalhousie University’s School of Occupational Therapy Student Research Award.

The equipment for this project was provided by Diane E MacKenzie’s grants from CFI: John R. Evans Leader’s Fund and the Nova Scotia Research and Innovation Trust.

Contributorship

David T Ambrose and Diane E MacKenzie contributed to the study design and methodology of the case study. David T Ambrose conducted the literature search. All authors contributed to the analysis, synthesis and interpretation of the findings. David T Ambrose wrote the first draft of the manuscript. All authors reviewed and edited the manuscript and approved the final version of the manuscript.

ORCID iDs

David Ambrose

Diane E MacKenzie

Parisa Ghanouni

References

Aldred

Green

Adams

(2004) A new social communication intervention for children with autism: Pilot randomised controlled treatment study suggesting effectiveness: Randomised social communication treatment for autism. Journal of Child Psychology and Psychiatry 45(8): 1420–1430.

Althoff

Dammann

Hope

, et al. (2019) Parent-mediated interventions for children with autism spectrum disorder: A systematic review. American Journal of Occupational Therapy 73(3): 7303205010p1.

Ambrose

MacKenzie

Ghanouni

(2020) The impact of person–environment–occupation transactions on joint attention in children with autism spectrum disorder: A scoping review. British Journal of Occupational Therapy 83(6): 350–362.

American Psychiatric Association (2013) Diagnostic and Statistical Manual of Mental Disorders, 5th ed. Washington, D.C.: American Psychiatric Association.

Baio

Wiggins

Christensen

, et al. (2018) Prevalence of autism spectrum disorder among children aged 8 years – Autism and developmental disabilities monitoring network, 11 sites, United States, 2014. Morbidity and Mortality Weekly Report. Surveillance Summaries (Washington, D.C.: 2002) 67(6): 1–23.

Bean

Eigsti

(2012) Assessment of joint attention in school-age children and adolescents. Research in Autism Spectrum Disorders 6(4): 1304–1310.

Casenhiser DM, Shanker SG and Stieben J (2011) Learning through interaction in children with autism: Preliminary data from asocial-communication-based intervention. Autism: The International Journal of Research & Practice 17(2): 220–241. DOI: 10.1177/1362361311422052.

Duchowski

(2007) Eye Tracking Methodology: Theory and Practice. Berlin: Springer-Verlag.

Eschenfelder

Gavalas

(2017) Joint attention and occupations for children and families living with autism spectrum disorder: A scoping review. The Open Journal of Occupational Therapy 5(4): 1–18.

10.

Falck-Ytter

Bolte

Gredeback

(2013) Eye tracking in early autism research; PMC3849191. Journal of Neurodevelopmental Disorders 5(1): 1–13.

11.

Girolametto

Sussman

Weitzman

(2007) Using case study methods to investigate the effects of interactive intervention for children with autism spectrum disorders. Journal of Communication Disorders 40(6): 470–492.

12.

Green

Charman

McConachie

, et al. (2010) Parent-mediated communication-focused treatment in children with autism (PACT): A randomised controlled trial. Lancet 375(9732): 2152–2160.

13.

Guillon

Hadjikhani

Baduel

, et al. (2014) Visual social attention in autism spectrum disorder: Insights from eye tracking studies. Neuroscience and Biobehavioral Reviews 42: 279–297.

14.

Gulsrud

Hellemann

Freeman

SFN

, et al. (2014) Two to ten years: Developmental trajectories of joint attention in children with ASD who received targeted social communication interventions. Autism Research 7(2): 207–215.

15.

Gulsrud

Hellemann

Shire

, et al. (2016) Isolating active ingredients in a parent‐mediated social communication intervention for toddlers with autism spectrum disorder. Journal of Child Psychology and Psychiatry 57(5): 606–613.

16.

Hanley

Riby

McCormack

, et al. (2014) Attention during social interaction in children with autism: Comparison to specific language impairment, typical development, and links to social cognition. Research in Autism Spectrum Disorders 8(7): 908–924.

17.

Foulsham

Kingstone

(2015) Speaking and listening with the eyes: gaze signaling during dyadic interactions; PMC4550266. PLoS One 10(8): e0136905.

18.

Ingersoll

(2010) Teaching social communication: A comparison of naturalistic behavioral and developmental, social pragmatic approaches for children with autism spectrum disorder. Journal of Positive Behavior Interventions 12(1): 33–43.

19.

Johnson AL, Gillis JM and Romanczyk RG (2012) A brief report: Quantifying and correlating social behaviors in children with autism spectrum disorders. Research in Autism Spectrum Disorders 6(3): 1053–1060. DOI: 10.1016/j.rasd.2012.01.004.

20.

Karaaslan

Diken

Mahoney

(2011) a randomized control study of responsive teaching with young Turkish children and their mothers. Topics Early Childhood Special Education 33(1): 18–27.

21.

Kong NY and Carta JJ (2013) Responsive Interaction Interventions for Children With or at Risk for Developmental Delays: A Research Synthesis. Topics Early Child.Spec.Educ. 33(1): 4–17. DOI: 10.1177/0271121411426486.

22.

Kossyvaki

Jones

Guldberg

(2014) Training teaching staff to facilitate spontaneous communication in children with autism: Adult interactive style intervention (AISI). Journal of Research in Special Education Needs 16(3): 156–168.

23.

Landa

Gross

Stuart

(2013) Developmental trajectories in children with and without autism spectrum disorders: The first 3 years. Child Development 84(2): 429–442.

24.

Liao

Hwang

Chen

, et al. (2014) Home-based DIR/Floortime intervention program for preschool children with autism spectrum disorders: Preliminary findings. Physical & Occupational Therapy in Pediatrics 34(4): 356–367.

25.

MacKenzie

Westwood

(2013) Occupational therapists and observation: What are you looking at? OTJR: Occupation, Participation and Health 33(1): 4–11.

26.

Mahoney

Perales

(2005) Relationship-focused early intervention with children with pervasive developmental disorders and other disabilities. Journal of Developmental and Behavioral Pediatrics 26(2): 77–85.

27.

Mahoney

Solomon

(2016) Mechanism of developmental change in the PLAY project home consultation program: Evidence from a randomized control trial. Journal of Autism & Developmental Disorders 46(5): 1860–1871.

28.

Mundy

(2018) A review of joint attention and social‐cognitive brain systems in typical development and autism spectrum disorder. European Journal of Neuroscience 47(6): 497–514.

29.

Murza

Schwartz

Hahs-Vaughn

, et al. (2016) Joint attention interventions for children with autism spectrum disorder: A systematic review and meta-analysis. International Journal of Language and Communication Disorders 51(3): 236–251.

30.

Noris

Nadel

Barker

, et al. (2012) Investigating gaze of children with ASD in naturalistic settings. PLoS One 7(9): e44144.

31.

Oono

Honey

McConachie

(2013) Parent-mediated early intervention for young children with autism spectrum disorders (ASD). Cochrane Database of Systematic Reviews 4: CD009774.

32.

Ruble

McDuffie

King

, et al. (2008) Caregiver responsiveness and social interaction behaviors of young children with autism. Topics in Early Childhood Special Education 28(3): 158–170.

33.

Schwarzkopf

Buchner

Holscher

, et al. (2016) Perspective tracking in the real world: Gaze angle analysis in a collaborative wayfinding task. Spatial Cognition and Computation 17(1–2): 143–162.

34.

SensoMotoric Instruments (SMI) (2016a) IViewETG User Guide Version 2.7. Boston, MA, USA.

35.

SensoMotoric Instruments (SMI) (2016b). BeGaze Manual Version 3.7. Boston, MA, USA.

36.

Shire

Gulsrud

Kasari

(2016) Increasing responsive parent-child interactions and joint engagement: Comparing the influence of parent-mediated intervention and parent psychoeducation; PMC4826805. Journal of Autism and Developmental Disorders 46(5): 1737–1747.

37.

Swanson

Siller

(2013) Patterns of gaze behavior during an eye-tracking measure of joint attention in typically developing children and children with autism spectrum disorder. Research in Autism Spectrum Disorders 7(9): 1087–1096.

38.

Tanner

Hand

O’Toole

, et al. (2015) Effectiveness of interventions to improve social participation, play, leisure, and restricted and repetitive behaviors in people with autism spectrum disorder: A systematic review. American Journal of Occupational Therapy 69(5): 6905180010p1.