A Scoping Review of Behavioral Interventions for Promoting Social Gaze in Individuals with Autism Spectrum Disorder and Other Developmental Disabilities

Abstract

Background

Individuals diagnosed with autism spectrum disorder (ASD) commonly experience difficulties maintaining social gaze with others during interactions. Although behavioral interventions targeted to promote social gaze in ASD are evident in the literature, to our knowledge, no review of the literature has been conducted to summarize and evaluate the evidence for these interventions.

Methods

We reviewed and summarized behavioral intervention studies designed to promote social gaze in individuals diagnosed with ASD and other developmental disabilities published in English between 1977 and January 2022 using PsychINFO and PubMed databases.

Results

41 studies met the inclusion criteria describing interventions conducted on 608 individuals. A variety of intervention strategies were employed to promote social gaze in these individuals including discrete trial instruction, prompting, modeling, and imitation. Most studies employed single-case research designs and reported successful outcomes, but limited data were available concerning the generalization, maintenance and social validity of these interventions. An increasing number of studies utilized technology-based procedures including computer application gameplay, gaze-contingent eye tracking devices and humanoid robots.

Conclusions

The present review indicates that behavioral interventions can be successfully employed to promote social gaze in individuals with ASD and other developmental disabilities. However, future research is needed to establish the generalization, maintenance and social validity of these interventions. There are also important ethical issues to be addressed given the increasing divide between treatment advocates and proponents of the neurodiversity movement.

1. Introduction

1.1. Social gaze behavior

The ability to maintain eye contact with others (i.e., eye-to-eye gaze) during social interactions is considered a key prerequisite skill for the development of more complex social skills. In social contexts, one can infer an individual’s identity, familiarity, emotional expression and potentially their intentions and mental states by making eye contact with another person (Emery, 2000). In learning contexts, the probability of attending to relevant instructional stimuli may be improved by sustained social gaze, increasing the probability of correctly responding to instructions and potentially accelerating the rate of skill acquisition (Carbone et al., 2013; Cook et al., 2017; Fonger & Malott, 2019; Kraus et al., 2012; Tiegerman & Primavera, 1984). For example, Hamlet and colleagues (1984) demonstrated that establishing social gaze prior to delivering instructions to classroom students increased their compliance by 2-3 times that of baseline observations (Hamlet et al., 1984).

Similarly, difficulties with maintaining social gaze can impede the acquisition of more complex skills, such as joint attention, an important precursor to communication. These findings have led researchers to suggest that social gaze may need to be established in the behavioral repertoire prior to teaching these more complex social communication skills (Koegel et al., 2009; Taylor & Hoch, 2008). For these reasons, the acquisition of social gaze has been proposed as an important behavioral cusp (Cook et al., 2017), a behavior change that facilitates contact with broader learning opportunities (Rosales-Ruiz & Baer, 1997). The ability to maintain social gaze with others can therefore facilitate engagement with a wide range of learning and reinforcement opportunities in the natural world, potentially increasing quality of life over time.

1.2. Autism spectrum disorder

Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by two broad sets of criteria: 1) impairments in social-communication and social interaction and 2) restricted, repetitive behaviors (American Psychiatric Association, 2013). Engaging in atypical social gaze is one of the earliest indicators of an ASD diagnosis in children (Baron-Cohen et al., 1992; Chawarska et al., 2014; Zwaigenbaum et al., 2005) and has become synonymous with the disorder itself (Ninci et al., 2013). In fact, research has indicated that experienced clinicians and parents alike can detect signs of ASD, such as atypical social gaze, as early as 12 months in infants later diagnosed with ASD (Luyster et al., 2009; Ozonoff et al., 2010; Szatmari et al., 2016; Zwaigenbaum et al., 2005). Over the past few decades, the prevalance of ASD has increased dramtically with current estimates indicating that as many as 1 in 44 children now receive this diagnsosis (Maenner et al., 2021).

1.3. Mechanisms underlying social gaze in ASD

Given the importance of social gaze to social and learning contexts, and the difficulties it can pose when not established as part of an individual’s repertoire, a growing body of research has attempted to explore the underlying mechanisms responsible for atypical social gaze in ASD. Studies have suggested that social gaze in individuals with ASD may result from deficits in attentional or associative mechanisms that ultimately manifests as a form of social gaze indifference (Moriuchi et al., 2017). That is, social interactions may not function as a reinforcer for individuals with ASD and this could lead to a lack of social gaze given the role that social gaze plays in social interactions. Other studies have shown that social gaze in ASD may be associated with high levels of anxiety that ultimately manifests as a form of social gaze aversion (Trevisan et al., 2017). A recent synthesis of neural evidence for atypical social gaze in ASD indicates that this behavior may also be used to reduce amygdala-related hyperarousal in ASD (see Stuart et al., 2022 for a review). Gaze cuing research suggests that directing gaze to the eye region may modulate activity in key areas of the social brain and that individuals with ASD may avoid social gaze to prevent negative affective arousal (Perlman et al., 2011; Zürcher, Donnelly, et al., 2013; Zürcher, Rogier, et al., 2013). As noted by Moriuchi et al. (2017), understanding the mechanisms underlying social gaze avoidance may be important for designing effective interventions. For example, behavioral interventions that increase exposure to social gaze may be indicated when the underlying mechanism may be related to hyperarousal/gaze aversion. In contrast, behavioral interventions that enhance the reinforcing value or motivation to engage in social gaze may be indicated in cases when the underlying mechanism may be related to hypoarousal/gaze indifference. In other words, operant processes likely play a role in the mechanisms in the lack of development in social gaze, which may be evidenced by the effectiveness of behavioral interventions in increasing social gaze.

Research indicates that interventions that directly target mutual gaze may adjust the pattern of brain activity toward a more typical trajectory (Dawson et al., 2012; Johnson et al., 2015; Jones & Klin, 2013). This line of research suggests that interventions to improve social gaze may have ongoing benefits for social functioning in ASD (Lassalle et al., 2017; Stuart et al., 2022). Increasing social gaze may therefore help reduce missed social and emotional learning opportunities, mitigating adverse impacts on social-cognitive development and real-world social functioning later in life.

1.4. Aims and outline

Taken together, there appears to be a strong clinical rationale for promoting social gaze in early interventions for individuals with ASD. Indeed, social gaze is often one of the first behavioral targets selected for assessment and treatment of ASD based on the principles of Applied Behavior Analysis (ABA) which is widely considered the gold standard treatment for children with ASD (Vismara & Rogers, 2010). However, several recent papers have called into question whether ABA approaches to facilitate acquisition of social skills in this population are necessary (Sandoval-Norton & Shkedy, 2019) and have suggested that interventions designed to promote social gaze behavior in particular may not be appropriate (Trevisan et al., 2017). It is therefore important to examine the available evidence for behavioral interventions designed to promote social gaze in these conditions and the extent to which these interventions are acceptable to patient stakeholders.

In this paper, we review the current body of literature on behavioral interventions to promote social gaze in individuals with ASD, as well as individuals with other forms of developmental disability, and provide a summary of approaches and outcomes. We also include reports of maintenance, generalization, and social validity data to advance our understanding of the overall effectiveness and acceptability of the procedures. Finally, we discuss important gaps in the literature on behavioral interventions for promoting social gaze in individuals with ASD, and future directions for this line of research including the implications for neurodiversity. Given that the primary purpose of the current investigation was to summarize the body of literature on behavioral interventions to promote social gaze, clarify key concepts, and identify knowledge gaps, a scoping review was warranted (Munn et al., 2018). To guide the review, we included the most recent reporting guidelines for scoping reviews, specifically the Preferred Reporting Items for Systematic Reviews and Meta-Analyses – Scoping Reviews (PRISMA-ScR) checklist. This checklist is provided in Supplementary Material File 1. The specific questions addressed by the review are as follows:

1. To what extent are behavioral interventions conducted to promote social gaze in individuals with ASD and other developmental disabilities?

2. What are the characteristics of individuals who receive these interventions?

3. What types of behavioral interventions have been employed to promote social gaze in these conditions?

4. How is social gaze defined and measured in these studies?

5. Where are these interventions typically being conducted (e.g., home, school, clinic)?

6. What is the typical duration of these interventions and what are the outcomes?

7. What differences in outcomes (if any) have been found between individuals with ASD and individuals with other forms of developmental disability?

8. Are methods typically included to measure maintenance, generalization, and social validity in these studies?

2. Methods

2.1. Search Strategy

Studies employing procedures designed to promote social gaze in individuals with ASD and other developmental disabilities (e.g., fragile X syndrome) were identified from two databases: PsychINFO and PubMed. Search terms included the following descriptives: “autism”, “autism spectrum disorder”, “fragile X syndrome”, “development* delay”, “eye contact training”, “social gaze training”, alongside the Boolean operators AND and OR. A preliminary search was conducted in December 2021 and the main search was conducted in January 2022. The databases were last searched on or before January 17th, 2022, yielding a total of 328 papers. After checking for duplicates, the title and abstracts of papers were screened for inclusion by the second author. The first author independently screened the titles and abstracts for 141 articles (47%) for inclusion and interobserver agreement was acceptable at 91%. There was a total of 13 rater discrepancies. In all discrepancies, the title and abstract did not clearly indicate the use of a behavioral procedure for promoting social gaze. Discrepancies were resolved to consensus through discussion and consultation with the third author resulting in one of the 13 articles being included in the final review. The remaining articles (N = 52) were advanced to full text screening. Studies that did not attempt to promote social gaze or focused on another aspect of social skills training such as joint attention were excluded. We also excluded studies that had not been subject to peer-review or had not been published in English. Additional articles were identified through ancestral search and expert nomination by the third author, wherein records were identified from reference lists of relevant articles or from literature review and were included if they met the inclusion criteria above. Figure 1 presents the PRISMA flow diagram.

Figure 1.

PRISMA Flow Diagram

2.2. Coding

A data extraction form was developed and refined during the review stage of the analysis. The second author charted the data and the first author independently confirmed the data. Articles were coded for the following variables: number (N) of participants who received the intervention (excluding controls), sex of the participants, participant age range (in years), diagnosis (ASD or other developmental disability), response definition (how social gaze was defined), response measurement (how social gaze was measured), design (single-subject research design or group design), intervention (description of the main components of the intervention), setting (where the intervention took place), duration of the intervention, outcome (outcome of intervention, coded as positive, mixed, or negative), generalization and/or maintenance (whether the intervention generalized to other settings or people and/or effects were maintained over time), and social validity (acceptability of the intervention). Articles coded as a single-subject research design (SSRD) were subsequently coded as either an AB, ABA, ABAB reversal design, multiple baseline design (MBL) or an alternating treatments design (ATD). Articles coded as a group design (GD) were coded as a randomized controlled trial (RCT) if participant assignment to groups was randomized or non-RCT. For studies that employed SSRD, positive results were those in which the authors reported that all the participants demonstrated improvements in social gaze. Mixed results indicated that the authors reported that some of the participants demonstrated improvements. Negative results indicated that no participants made improvements in social gaze. For studies that employed group designs (treatment versus control), positive results indicated a statistically significant difference in favor of the treatment over control. Outcomes were coded as no difference if no differences were found between treatments, or between treatment and control. Finally, we separated articles into those that employed human-to-human social interaction versus those that employed technology-based procedures (i.e., high technology such as robotics or eye tracking technology). The coding for each study included in the final analysis was completed independently by the first and second author. Any discrepancies in coding were resolved by consensus through discussion and consultation with the third author.

3. Results

A total of 41 articles met the criteria for inclusion in the review. Table 1 shows the studies that employed human-to-human social interactions and Table 2 shows the studies that employed technology-based procedures. Narrative summaries of the information extracted, trends and patterns, are identified in the following sections.

Table 1.

Characteristics and main results of studies that involved human-to-human interaction.

Authors and year of publication	Participants (Sex)	Age (in years)	Diagnosis	Response definition	Response measurement	Design	Intervention	Setting	Duration	Outcome	Generalization/ Maintenance	Social validity
Alexander et al. (2011)	4 (4 M, 0 F)	14-24	ASD, Down Syndrome, ID and ADHD	Eye contact rated as acceptable, unacceptable, or not attempted	Direct: Observation using the Observation Skills Rating Form	SSRD (ABCA)	Social skills and sports program that combined classroom instruction with soccer activities	Community	Twice weekly 90 min sessions across 14 weeks	Mixed	NR	Parent Skill Rating Form indicated decreases or no change in eye contact
Bagaiolo et al. (2017)	34 (25 M, 9 F)	3-6	ASD	Looking at parent	Indirect: Parent records	GD (RCT)	Use of video modeling to teach parents to implement discret-trial instruction and prompt-fading	Home	22 weeks of parental training intervention	Not reported	NR	Compliance to the intervention was regarded as favorable
Bordini et al. (2020)	34 (29 M, 5 F)	3-6	ASD	Not defined	Indirect: Vineland	GD (RCT)	Parent training via video modeling on discrete-trial instruction with errorless learning	Clinical/Home	Weekly 90 min sessions across 22 weeks	Positive	NR	NR
Carbone et al. (2013)	1 (1 M, 0 F)	3	ASD	Movement by participant's head and eyes to make direct contact with eyes of person they were interacting with	Direct: Observation	SSRD (AB)	Extinction and differential reinforcement in the context of mand training	Clinical	3 weekly 3-hour sessions across 3 months	Positive	NR	NR
Chien et al. (2021)	41 (35 M, 6 F)	18-45	ASD	Eye contact rated on a 9-point scale based on frequency	Indirect: Communicative Behavior Observatory Scale (CBOS)	GD (RCT)	PEERS intervention program administered across groups	Clinical	Weekly 90 min in sessions across 16 weeks	Positive	Improvements maintained 3- and 6-months post intervention	Four participants dropped out due to dissatisfaction with some aspect of treatment
Chou et al. (2016)	2 (2 M, 0 F)	6	ASD and ADHD	Looking at another person's face for at least 3 seconds while speaking to them	Direct: Observation	SSRD (MBL)	Gestural prompting and differential reinforcement in the form of praise and token reinforcement	Clinical	Seventeen 60-minute sessions	Positive	Improvements maintained at follow-up	Vineland scores indicated that parents and teachers perceived improvem ent in social skills after the art program
Cook et al. (2017)	21 (Not reported)	3-12	ASD	Eye contact defined as 1 s or more of gaze at therapist's eyes or face within 3 s of a name call	Direct: Observation	SSRD (ABAB)	Contingent praise, contingent edible plus praise, edible prompt, contingent video, varied consequences, and alternative stimulation removal	Clinical	Duration varied from 50-1,720 trials across participants	Positive	Improvements maintained for some participants at follow-up	NR
Fonger & Malott (2019)	3 (1 M, 2 F)	2.5	ASD	Eye contact with instructor for a duration of 3 seconds	Direct: Observation	SSRD (MBL)	Preferred item removed and returned for 15 s with edible item contingent on eye contact	School	60 or more 5 min sessions across 6 months	Positive	Improvements maintained at follow-up	NR
Foxx (1977)	3 (2 M, 1 F)	6-8	ASD	Eye contact within 5 sec of the therapist's verbal prompt: "Look at me."	Direct: Observation	SSRD (ATD)	Overcorrection (functional movement training avoidance) with reinforcement in the form of social and edible items, or reinforcement alone	Clinical	Five daily sessions totaling 100 trials across several weeks	Positive	No generalization to positive consequences only	NR
Gannon et al. (2018)	20 (20 M, 0 F)	8-18	FXS	Orienting head toward the therapist and looking directly at the therapist’s face within 5 s of prompt	Direct: Tobii X120 eye tracker	GD (RCT)	Discrete-trial instruction with least-to-most prompting and percentile schedule reinforcement in the form of praise tangibles, and tokens	Clinical	Four 60 min sessions per day across 2 days	Mixed	Greater generalization of improvements with high-dose group	NR
Hall et al. (2009)	6 (6 M, 0 F)	8-17	FXS	Orienting head toward the experimenter so that the eyes looked directly at the experimenter’s eyes	Direct: Observation	SSRD (ABAB)	Overcorrection (functional movement training avoidance) with percentile schedule reinforcement in the form of edibles or tangibles	Clinical	Two daily 60 min sessions across 2 days	Mixed	NR	NR
Helgeson et al. (1989)	3 (2 M, 1 F)	13-16	ID	Orientation of the head and eyes towards the eyes of another person for at least 2 seconds	Direct: Observation	SSRD (MBL)	Positive reinforcement in the form of praise, tokens, and edible items contingent upon each incident of eye contact during or after questions	Community	25 days	Positive	NR	Staff member ratings indicated a decrease in gaze aversion and more engagement in social activity following the intervention, and that the procedures were reasonable.
Hupp & Reitman (2000)	1 (1 M, 0 F)	8	PDD-NOS	Looking at conversation partner's face without looking away for more than 3 seconds	Direct: Observation	SSRD (ABA)	Modeling, rehearsal, reinforcement in the form of praise and money, and parent training	Clinical/Home	Twice weekly sessions across 16 weeks	Positive	Improvements maintained 4- and 12-months post intervention	Parents reported conducting over 30 training sessions at home
Hwang & Hughes (1995)	1 (0 M, 1 F)	2	Developmental Disability	Looking at experimenter's eyes or face for at least 2 continuous seconds without a prompt	Direct: Observation	SSRD (ABAB)	Contingent imitation and social reinforcement	Clinical	Daily 15-min sessions across several weeks	Positive	NR	Social validity ratings indicated that procedures were considered acceptable
Hwang & Hughes (2000)	3 (3 M, 0 F)	2-4	ASD	Looking at the interactive partner's eyes or face for at least 2 continuous seconds without a prompt	Direct: Observation	SSRD (MBL)	Contingent imitation and social reinforcement	Clinical	Twice weekly 10-min sessions across 30 weeks	Positive	Improvements in eye contact generalized to new partners and settings	Social validity ratings indicated that procedures were considered acceptable
Killmeyer et al. (2019)	3 (3 M, 0 F)	<3	ASD or at risk	Spontaneous visualgaze during play	Direct: Observation	SSRD (MBL)	Parent training on contingent imitation	Clinical	3 weekly 10-30 min sessions across approximately 3 months	Positive	Partial maintenance of improvements at follow-up	All parents reported that CI was useful, enjoyable for their child, and said that they would continue to implement it in the future
Koegel & Frea (1993)	1 (1 M, 0 F)	16	ASD	Gaze directed toward partner or relevant referent in conversation	Direct: Observation	SSRD (MBL)	Modeling, rehearsal, discrimination training, self-monitoring, and reinforcement in the form of points that could be exchanged for gameplay	Community	4-7 sessions per day, 1 day per week across 14 weeks	Positive	NR	Appropriateness of procedures was rated higher after treatment
Kornacki et al. (2013)	3 (Not reported)	21-23	ASD, Down Syndrome, and ID	Making eye contact during each step of a conversation task analysis	Direct: Observation	SSRD (MBL)	Behavioral Skills Training (BST) including instruction, modeling, rehearsal, and feedback administered in training phases	Clinical	Daily 5-20 min sessions across 3 months	Positive	Partial maintenance of improvements at follow-up	NR
Matson et al. (1988)	3 (2 M, 1 F)	12-14	ASD and hearing impaired	Looking at the face of the adult communi cating with him/her throughout an entire observation period	Direct: Observation	SSRD (ATD)	Reinforcement in the form of social and edible items	School	4 weekly 10-min sessions across several weeks	Positive	Partial generalization to the classroom setting	NR
McParland et al. (2021)	10 (7 M, 3 F)	5-11	ASD	Gaze directed at face for 1 second following social stimuli	Direct: SMI remote eye tracking glasses	SSRD (AB)	Verbal praise and points delivered as reinforcement for meeting response criterion	School	One 30 min session	Positive	NR	NR
Ninci et al. (2013)	1 (1 M, 0 F)	4	PDD-NOS	Unprompted gaze at therapist's eyes	Direct: Observation	SSRD (MBL)	Differential reinforcement and most-to-least prompting in the context of mand training	Clinical	17 sessions of baseline/intervention and 3 follow up sessions	Positive	Partial maintenance of improvements at 1-, 2-, and 3-months post intervention	NR
O'Handley et al. (2015)	6 (6 M, 0 F)	16-19	ASD and ID	Total duration of eye contact during 3-minute conversation probe	Direct: Observation	SSRD (MBL)	Social stories, video modeling, and a combination of social stories and video modeling with non-contingent reinforcement	School	up to 24 sessions	Positive: Video modeling and video modeling + social stories	Improvements generalized to a familiar teacher	NR
Rapp et al. (2019)	15 (Not reported)	3-7	ASD	One second or more of visual orientation to therapist within 3 seconds of a name call	Direct: Observation	SSRD (MBL)	Contingent praise, contingent edibles, prompt plus contingent edibles, contingent video, and varied consequences	Clinical	Multiple sessions per week across several months	Mixed	Improvements maintained at follow-up	NR
Rollins et al. (2021)	78 (54 M, 24 F)	1-3	ASD	Looking in the parent’s eyes for 2 seconds, coupled with a smile	Direct: Hidden-camera eyeglasses worn by the parent	GD (RCT)	Pathways (parent training on strategies to establish eye gaze without verbal, visual, or physical prompts); Communication (parent training on strategies to facilitate communication)	Home	Weekly 90 min sessions across 12 weeks	Positive: Pathways	NR	NR
Tang et al. (2021)	26 (24 M, 2 F)	Mean of 7	ASD	Area of interests (AOI) were set on the human faces, including eyes, mouths, facial expression in photos and videos	Direct: Tobii X2-60 Eye Tracker	GD (RCT)	Storytelling with and without social contextual information	Clinical/School	8 sessions across 4 weeks	Positive: Storytelling with social contextual information	NR	NR
Taras et al. (1988)	2 (1 M, 1 F)	9-10	ASD	Looking at the therapist while speaking	Direct: Observation	SSRD (MBL)	Modeling, instructions, performance feedback, role-playing, rehearsal, and tangible and social reinforcement	Clinical	15 min sessions across several weeks	Positive	Partial maintenance of improvements at 4- and 12-month follow-up	NR
Tetreault & Lerman (2010)	3 (2 M, 1 F)	4-8	ASD	Eye contact before, during, or within 2 seconds of target vocal behavior	Direct: Observation	SSRD (MBL)	Point-of-view video modeling, reinforcement in the form of food items, a combination of video modeling and reinforcement, and least-to-most prompting	Clinical	Daily sessions across several months	Positive	Improvements in eye contact maintained at follow-up and generalized to new stimuli	NR
Vernon et al. (2012)	3 (3 M, 0 F)	2-4	ASD	Looking at facial region of parent's face	Direct: Observation	SSRD (MBL)	Parent training on Pivotal Response Treatment (PRT) with embedded social interaction	Home	3-5 weekly 60 min sessions for 16 total sessions	Positive	Partial generalization and partial maintenance of Improvements at follow-up	Increases in parent positive affect indicate that they enjoyed implementing the intervention
Wong & Kwan (2010)	9 (8 M, 1 F)	1-3	ASD	Scores on B1 Unusual Eye Contact, B4 Integration of gaze and other behaviors during social overtures, and B7 Requesting from the ADOS	Indirect: ADOS	GD (RCT)	Modeling, prompt-fading, and reinforcement, and parent training	Clinical	Ten 30-min sessions	Positive	NR	Parenting Stress Index showed statistically significant decrease in parental stress post-intervention

ASD: Autism spectrum disorder; FXS: fragile X syndrome; ID: Intellectual disability; SSRD: single-subject research design; ATD: alternating treatments design; MBL: multiple baseline; GD: group design; RCT: randomized controlled trial; NR: not reported

Table 2.

Characteristics and main results of studies that employed technology-based procedures.

Authors and year of publication	Participants (Sex)	Age (in years)	Diagnosis	Response definition	Response measurement	Design	Intervention	Setting	Duration	Outcome	Generalization/ Maintenance	Social validity
Chung (2019)	14 (14 M, C F)	9-11	ASD	Duration and frequency of eye gaze during communication	Direct: Observation	GD	Structured interactive social games structured story-based activities, and singing and movement to music facilitated by a humanoid robot	Clinical	Weekly 30-minute sessions across 12 weeks	Positive	Partial maintenance of Improvements at follow-up	NR
Chung (2021)	15 (13 M, 2 F)	5-11	ASD	Duration and frequency of eye gaze while communicating	Direct: Observation	SSRD (ABA)	Structured interactive social games structured story-based activities, and singing and movement to music facilitated by a humanoid robot	Clinical	Weekly 30 min sessions across 12 weeks	Positive	Partial maintenance of Improvements at follow-up	NR
Hopkins et al. (2011)	49 (44 M, 5 F)	6-15	ASD	Touching the object at which the avatar is gazing	Not measured	GD (RCT)	FaceSay computer application with gameplay and reinforcement in the form of praise contingent on eye contact	School	Twice weekly 10-25 min sessions across 6 weeks	Unknown	NR	NR
Jeffries et al. (2016)	3 (3 M, 0 F)	3-5	ASD	Making eye contact with therapist of any duration while requesting an item	Direct: Observation	SSRD (MBL)	Look in My Eyes Steam Train application delivered reinforcement for matching the number found in a photo of a person’s eyes to a grid; Differential reinforcement in the form of praise and requested item when making eye contact when manding	Clinical	Six 10-trial sessions during tablet application, five 10 min sessions during differential reinforcement	Positive: Differential reinforcement	Improvements maintained and generalized to a new setting after a 1-hour delay	Parents reported an increase in eye contact for all the children that was maintained after the completion of the study
Liu et al. (2017)	2 (2 M, 0 F)	8-9	ASD	Gaze directed at a partner’s face	Indirect: Parent report	SSRD (AB)	Brain Power Systems smart glasses with the Face Game application uses real-time camera feed to detect human faces in the user’s field of view, overlays and gradually fades an augmented cartoon face to gain attention and delivers reinforcement in the form of points for looking at their partner’s face	Clinical	1 session per participant (length unspecified)	Positive	NR	Caregivers reported high tolerability and engagement and reported the system to be enjoyable
Rakhymbayeva et al. (2021)	11 (10 M, 1 F)	4-11	ASD and ADHD	Looking at robot	Direct: Observation	GD	Robot mediated activities designed to improve social skills, tailored individually to participants by therapists	Clinical	Ten 15 min sessions across 21 days	No difference	NR	Interviews suggested that parents were satisfied but suggested Improvements
Rice et al. (2015)	16 (16 M, 0 F)	5-11	ASD	Looks into the eyes of another child	Direct: Observation	GD (RCT)	FaceSay computer application with gameplay and reinforcement in the form of praise contingent on eye contact	Clinical	Weekly 25 min session across 10 weeks	No difference	NR	NR
Scassellati et al. (2018)	12 (7 M, 5 F)	6-12	ASD	Eye gaze directed at eyes of social robot or of the child’s caregiver	Direct for robot: eye tracker Indirect for caregiver : Likert-scale rating	SSRD (ABA)	In-home use of robot system intended to teach social skills to participants via modeling and interactive gameplay	Home	30-minute sessions every day for 30 days	Positive	Generalization of eye contact to caregivers following training with the robot	Caregivers reported ease of implementing the robot therapy with the participant
Sosnowski et al. (2021)	54 (47 M, 7 F)	4-14	ASD	Gaze on the eye and mouth regions of game characters	Direct: Tobii 4C eye tracker	GD (RCT)	Lookware gameplay using discrete-trial instruction with prompting and token reinforcement	Clinical/School/Home	3-5 weekly 15 min sessions across 6 weeks	Unknown	NR	Survey data indicated that participants found the intervention enjoyable
Tanaka et al. (2010)	42 (34 M, 8 F)	Mean of 10.5	ASD	Not defined	Indirect: Skills Battery subtests	GD (RCT)	Let’s Face It! Computer application gameplay and token reinforcement	Home	100 min per week for a total of 20 hours	Unknown	NR	NR
Wang et al. (2020)	35 (31 M, 4 F)	Mean of 3	ASD	Proportion of time looking at actress face	Direct: SR Eyelink 1000 Plus 500 Hz eye-tracking system	GD (RCT)	Gaze-contingent trials with adaptive cueing videos with darkening and blurring effects	Clinical	One 15 min session	Positive	NR	NR
Yun et al. (2017)	15 (15 M, 0 F)	4-7	ASD	Percentage of intervals during which eye contact occurred at least once	Direct: Observation	GD (RCT)	Discrete-trial instruction with differential reinforcement and prompting delivered by either a therapeutic robot or human therapist	Clinical	Eight 30-40 min sessions	Positive: Robot-facilitated and human-facilitated treatment	Improvements did not maintain at follow-up for either group	NR

ASD: Autism spectrum disorder; SSRD: single-subject research design; MBL: multiple baseline; GD: group design; RCT: randomized controlled trial; NR: not reported

3.1. Participants

A total of 608 participants received methods to promote social gaze across the included studies. Of these, 479 were reported to be males and 90 were females. Three studies (39 total participants) did not report the sex of the participants. Of the included articles, thirty-five (85%) included participants with ASD (n = 567; 93%). Other diagnoses of participants included FXS (n = 26; 4%) (Gannon et al., 2018; Hall, Maynes, et al., 2009), ADHD (n = 2; <1%) (Alexander et al., 2011; Chou et al., 2016), intellectual disability (n = 7; 1%) (Kornacki et al., 2013; O’Handley et al., 2015), Down Syndrome (n = 2; <1%) (Alexander et al., 2011; Kornacki et al., 2013), and other developmental delays (n = 4; <1%) (Hupp & Reitman, 2000; Hwang & Hughes, 1995; Matson et al., 1988; Taras et al., 1988). Six studies (15%) included participants with multiple diagnoses. Participants in the reviewed articles ranged from 17 months to 45 years, with the majority of articles reviewed (30; 73%) including participants 10 years or younger.

3.2. Response definition and measurement

Considerable variability was observed across studies in terms of how social gaze was defined. For most studies (21; 51%), appropriate social gaze was defined, but temporal details about the response were not included (i.e., when social gaze should occur following a prompt or discriminative stimulus or the minimum duration of social gaze required to be counted) and simply defined social gaze as any form of eye gaze (or “looking”) during an interaction with another person (Bagaiolo et al., 2017; Carbone et al., 2013; Chung, 2019, 2021; Hall, Maynes, et al., 2009; Jeffries et al., 2016; Killmeyer et al., 2019; Koegel & Frea, 1993; Kornacki et al., 2013; Liu et al., 2017; Matson et al., 1988; O’Handley et al., 2015; Rakhymbayeva et al., 2021; Rice et al., 2015; Scassellati et al., 2018; Sosnowski et al., 2022; Tang et al., 2022; Taras et al., 1988; Vernon et al., 2012; Wang et al., 2021; Yun et al., 2017). Five studies (12%) specified when social gaze should occur following a stimulus in the definition, recording responses if they occurred within 2 seconds (Tetreault & Lerman, 2010), within 3 seconds (Cook et al., 2017; Rapp et al., 2019), or within 5 seconds (Foxx, 1977; Gannon et al., 2018) of the prompt. Nine studies (22%) specified the minimum duration of social gaze in the definition, recording responses if they occurred for at least 1 second (Cook et al., 2017; McParland et al., 2021; Rapp et al., 2019), at least 2 seconds (Helgeson et al., 1989; Hwang & Hughes, 2000; Rollins et al., 2021) or at least 3 seconds (Chou et al., 2016; Fonger & Malott, 2019; Hupp & Reitman, 2000). Two studies (5%) focused specifically on unprompted or spontaneous social gaze and included this detail in the response definition (Killmeyer et al., 2019; Ninci et al., 2013). Finally, there were four studies (10%) that did not explicitly define social gaze (Bordini et al., 2020; Tanaka et al., 2010) or used an indirect measure to rate the quality of social gaze (Alexander et al., 2011; Wong & Kwan, 2010).

Most studies (26; 63%) employed direct measurement systems involving direct live observation or recording from video. In addition, seven of the studies (17%) reviewed utilized technology-based eye tracking devices to measure responding (Gannon et al., 2018; McParland et al., 2021; Rollins et al., 2021; Scassellati et al., 2018; Sosnowski et al., 2022; Tang et al., 2022; Wang et al., 2020). Seven studies (17%) employed indirect measures of social gaze such as norm-referenced assessment and rating scales (Bagaiolo et al., 2017; Bordini et al., 2020; Chien et al., 2021; Liu et al., 2017; Scassellati et al., 2018; Tanaka et al., 2010; Wong & Kwan, 2010) and one study did not provide a measure of social gaze (Hopkins et al., 2011).

3.3. Design

Most studies (26; 63%) employed a single-subject research design. Three studies (7%) utilized a basic AB design including a baseline phase with repeated measurements and an intervention phase continuing the same measures (Carbone et al., 2013; Liu et al., 2017; McParland et al., 2021). Three studies (7%) utilized an ABA design integrating a posttreatment follow-up (Chung, 2021; Hupp & Reitman, 2000; Scassellati et al., 2018). Three studies utilized an ABAB withdrawal or reversal design by integrating a second intervention phase (Cook et al., 2017; Hall, Maynes, et al., 2009; Hwang & Hughes, 1995). One study utilized an ABCA multiple treatments design (Alexander et al., 2011). Two studies (5%) utilized an alternating treatments design (ATD) to compare two different treatments simultaneously with participants (Foxx, 1977; Matson et al., 1988). The most commonly utilized SSRD was the multiple-baseline design (MBL), which was employed in 13 studies (32%; (Chou et al., 2016; Fonger & Malott, 2019; Helgeson et al., 1989; Hwang & Hughes, 2000; Jeffries et al., 2016; Killmeyer et al., 2019; Koegel & Frea, 1993; Kornacki et al., 2013; Ninci et al., 2013; O’Handley et al., 2015; Rapp et al., 2019; Taras et al., 1988; Tetreault & Lerman, 2010; Vernon et al., 2012). Group designs were utilized to demonstrate experimental control in 15 studies (37%). Of these, 13 (32%) were randomized controlled trials in which participants were randomly assigned to either the treatment group, or the control or comparison group (Bagaiolo et al., 2017; Bordini et al., 2020; Chien et al., 2021; Gannon et al., 2018; Hopkins et al., 2011; Rice et al., 2015; Rollins et al., 2021; Sosnowski et al., 2022; Tanaka et al., 2010; Tang et al., 2022; Wang et al., 2020; Wong & Kwan, 2010; Yun et al., 2017).

3.4. Intervention strategies and settings

Intervention strategies to promote social gaze varied widely across studies. Twenty-three studies (56%) employed discrete trial instruction (DTI; Bagaiolo et al., 2017; Bordini et al., 2020; Carbone et al., 2013; Chou et al., 2016; Cook et al., 2017; Fonger & Malott, 2019; Foxx, 1977; Gannon et al., 2018; S. S. Hall, Maynes, et al., 2009; Helgeson et al., 1989; Jeffries et al., 2016; Liu et al., 2017; Matson et al., 1988; McParland et al., 2021; Ninci et al., 2013; Rapp et al., 2019; Sosnowski et al., 2022; Taras et al., 1988; Tetreault & Lerman, 2010; Vernon et al., 2012; Wang et al., 2020; Wong & Kwan, 2010; Yun et al., 2017). DTI is a standardized teaching procedure utilizing instruction, prompting, and reinforcement, with a brief pause between consecutive trials, such that a high rate of learning opportunities can be administered during each session. Prompting and prompt-fading strategies were also utilized in eight studies (20%; (Bagaiolo et al., 2017; Bordini et al., 2020; Chou et al., 2016; Gannon et al., 2018; Ninci et al., 2013; Sosnowski et al., 2022; Tetreault & Lerman, 2010; Wong & Kwan, 2010). Modeling with rehearsal of appropriate social gaze was included in four studies (10%; Hupp & Reitman, 2000; Koegel & Frea, 1993; Kornacki et al., 2013; Taras et al., 1988). Three studies (7%) employed contingent imitation of play, in which experimenters or therapists imitated the participants’ interactions with toys within the participants’ field of vision immediately following participants’ interactions to gain attending and eye contact (Hwang & Hughes, 1995, 2000; Killmeyer et al., 2019).

Parents or caregivers were trained to implement intervention procedures in seven of the studies (17%) reviewed (Bagaiolo et al., 2017; Bordini et al., 2020; Hupp & Reitman, 2000; Killmeyer et al., 2019; Rollins et al., 2021; Vernon et al., 2012; Wong & Kwan, 2010). Most of the studies (35; 85%) reviewed included some form of reinforcement for emitting social gaze or other appropriate social skills. Six of the studies reviewed (15%) did not include delivery of reinforcers in the intervention procedures (Chien et al., 2021; Chung, 2019, 2021; Rakhymbayeva et al., 2021; Scassellati et al., 2018; Wang et al., 2020). Two studies also included a punishment procedure in the form of overcorrection to decrease failures to respond to bids for social gaze (Foxx, 1977; Hall, Maynes, et al., 2009). Twelve studies (29%) utilized technology-based procedures ranging from computer application gameplay (Hopkins et al., 2011; Jeffries et al., 2016; Rice et al., 2015; Sosnowski et al., 2022; Tanaka et al., 2010) to gaze-contingent eye tracking devices (Liu et al., 2017; Wang et al., 2020), to the use of humanoid robots to facilitate training (Chung, 2019, 2021; Rakhymbayeva et al., 2021; Scassellati et al., 2018; Yun et al., 2017).

Figure 2 shows the cumulative number of studies published according to whether they employed human-to-human social interaction versus technology-based procedures.

Figure 2.

Cumulative Record of Included Publications Across Years

In all studies utilizing technology-based procedures, training involved directing social gaze toward two-dimensional actors, avatars, or three-dimensional robots, rather than real people, with one exception. The behavioral intervention described by Scassellati et al. (2018) involved participants interacting with a social robot alongside a caregiver, providing opportunities for the participants to interact and share experiences with the caregiver during intervention.

Interventions to promote social gaze occurred in a variety of settings. Five studies (12%) implemented procedures in the school setting (Fonger & Malott, 2019; Hopkins et al., 2011; Matson et al., 1988; McParland et al., 2021; O’Handley et al., 2015). Five studies (12%) implemented procedures in the participants’ homes (Bagaiolo et al., 2017; Rollins et al., 2021; Scassellati et al., 2018; Tanaka et al., 2010; Vernon et al., 2012). Two studies (5%) implemented procedures in a community setting (Alexander et al., 2011; Helgeson et al., 1989). The most common setting for intervention was a clinical setting, typically a treatment program or university-based classroom or clinic, which was utilized in 24 studies (59%) (Carbone et al., 2013; Chien et al., 2021; Chou et al., 2016; Chung, 2019, 2021; Cook et al., 2017; Foxx, 1977; Gannon et al., 2018; Hall, Maynes, et al., 2009; Hwang & Hughes, 1995, 2000; Jeffries et al., 2016; Killmeyer et al., 2019; Kornacki et al., 2013; Liu et al., 2017; Ninci et al., 2013; Rakhymbayeva et al., 2021; Rapp et al., 2019; Rice et al., 2015; Taras et al., 1988; Tetreault & Lerman, 2010; Wang et al., 2020; Wong & Kwan, 2010; Yun et al., 2017). Four studies (10%) implemented procedures across multiple settings, including clinical, home, and school settings (Bordini et al., 2020; Hupp & Reitman, 2000; Sosnowski et al., 2022; Tang et al., 2022).

3.5. Outcome

The outcome of the interventions was positive for most studies (31; 76%) indicating that a variety of intervention strategies may be effective for promoting social gaze. Four studies (10%) reported mixed results indicating that at least one participant did not respond to treatment (Alexander et al., 2011; Gannon et al., 2018; Hall, Maynes, et al., 2009; Rapp et al., 2019), including both reviewed studies including participants with FXS. Results with respect to improvements in social gaze specifically were not reported in four studies (10%), therefore the results were coded as Unknown (Bagaiolo et al., 2017; Hopkins et al., 2011; Sosnowski et al., 2022; Tanaka et al., 2010). Two studies (5%) reported no significant difference in social gaze between the intervention and control groups following treatment (Rakhymbayeva et al., 2021; Rice et al., 2015). Of the 12 studies that employed technology-based procedures, seven (58%) reported positive outcomes with the technology-based procedure utilized, one reported no difference (Rakhymbayeva et al., 2021), and for three studies, the outcome was unknown (Hopkins et al., 2011; Sosnowski et al., 2022; Tanaka et al., 2010). Jeffries et al. (2016) utilized both a technology-based procedure (involving a tablet application) and a non-technological procedure (differential reinforcement) separately, reporting no difference when the technology-based procedure was applied and a positive outcome when the non-technological procedure was applied. Of the remaining 29 studies that involved human-to-human social interactions, 24 studies (83%) reported positive outcomes, four reported mixed outcomes (Alexander et al., 2011; Gannon et al., 2018; Hall, Maynes, et al., 2009; Rapp et al., 2019), and for one study changes in social gaze specifically was not reported (Bagaiolo et al., 2017).

3.6. Maintenance and generalization

Generalization and/or maintenance of improvements was reported in 22 of the reviewed studies (54%). In seven studies (17%), the investigators made attempts to test for generality with new conversational partners, to new settings, or new stimuli and were at least partially successful generalizing the skill in these new contexts (Hwang & Hughes, 2000; Jeffries et al., 2016; Matson et al., 1988; O’Handley et al., 2015; Scassellati et al., 2018; Tetreault & Lerman, 2010; Vernon et al., 2012). Foxx et al. (1977) was the only study to report a generalization failure. In 16 studies (39%), investigators made attempts to test for maintenance of improvements in social gaze (Chien et al., 2021; Chou et al., 2016; Chung, 2019, 2021; Cook et al., 2017; Fonger & Malott, 2019; Hupp & Reitman, 2000; Jeffries et al., 2016; Kornacki et al., 2013; Ninci et al., 2013; Rapp et al., 2019; Taras et al., 1988; Tetreault & Lerman, 2010; Vernon et al., 2012; Yun et al., 2017). Of these, all but one study employing technology-based procedures (Yun et al., 2017) reported at least partial maintenance of improvements with follow-up phases ranging from 1-hour post-intervention to 12 months post-intervention.

3.7. Social validity

Social validity data were reported in 17 of the reviewed studies (41%). Five of these studies (12%) provided information indicating that caregivers found the procedures appropriate or acceptable (Hwang & Hughes, 1995, 2000; Koegel & Frea, 1993; Rakhymbayeva et al., 2021; Scassellati et al., 2018). One study reported statistically significant decreases in parental stress post intervention (Wong & Kwan, 2010). Two studies (5%) reported improvements following the completion of participation (Chou et al., 2016; Jeffries et al., 2016). Three studies (7%) indicated a high degree of compliance with the procedures and a willingness to continue implementing the procedures following the completion of participation (Bagaiolo et al., 2017; Jeffries et al., 2016; Killmeyer et al., 2019). Four studies (10%) indicated participant enjoyment with the procedures, including caregiver enjoyment (Vernon et al., 2012), child enjoyment as reported by the caregiver (Killmeyer et al., 2019; Liu et al., 2017) or child enjoyment by self-report (Sosonowski et al., 2021). Two studies (5%) reported unfavorable social validity data, both of which utilized human-to-human social interaction. Chien et al. (2021) reported the loss of four participants due to dissatisfaction with some aspect of the intervention, and Alexander et al., (2011) reported decreases or no change in social gaze based on parent ratings.

4. Discussion

The purpose of this study was to review and summarize published interventions for promoting social gaze in individuals diagnosed with ASD and other developmental disabilities. A total of 41 studies published between 1977 and 2022 were identified as meeting inclusion criteria and were subsequently analyzed. We included all studies from this period given that the purpose of the review was to identify (scope) all types of available evidence on behavioral interventions to promote social gaze in ASD. We acknowledge however, that this long (45 year) time span included the development of diagnostic criteria for ASD (American Psychiatric Association, 1994, 2003). Also, older studies may have included only one or two participants which may not be representative. Most studies (63%) employed single-case research designs with only 37% of studies employing group designs. The results therefore need to be set in this context. Overall, 76% of studies reviewed reported positive outcomes with 10% of studies reporting mixed results indicating that at least one participant in those studies did not respond to treatment. The results of this analysis indicates that there are a variety of treatment approaches that show favorable outcomes for these populations, including treatment approaches that utilized technology-based procedures. In our analysis, 58% of studies utilizing technology-based procedures reported positive outcomes. Technology-based procedures to improve social gaze have gained momentum over the last decade. Robotics in particular could be considered a potentially more accessible and personalized intervention that may be more enjoyable and effective than interacting live with therapists or peers (Scassellati et al., 2018). A recent meta-analysis of face-to-face and technology-based interventions to teach social skills to individuals with ASD found no significant differences between these methodologies, with both approaches producing improved social skills from pre- to -post-treatment (Soares et al., 2021). Taken together with the results of the current analysis, there appears to be at least initial support for some technology-based interventions for promoting social gaze and other important social skills. Though promising, the use of technology-based interventions is a relatively new field of study in the ASD population. Many studies employing this approach measure social gaze toward faces in video, on computer screens, with robots or avatars, which may be distinctly different than social gaze in real world settings. While it has been hypothesized that technology-based interventions may provide an opportunity to practice social gaze in a more preferred setting (with minimal anxiety), the intervention setting may not generalize to human communication partners. Therefore, additional literature is necessary to determine the extent to which improvements in social skills translate or generalize to real live interactions with other people.

The low number of studies including maintenance, generalization, and in particular, social validity data reveals important gaps in our understanding of the efficacy of procedures and important areas for future research. Much of the research in this area rests on the rationale that social gaze is a critical prerequisite skill to developing more complex and sophisticated social skills, which in turn leads to increased opportunities to experience social interactions with others. Therefore, generalization of improvements to other social skills, people, and contexts, as well as long-term maintenance of improvements is necessary for determining whether the ends justify the means. Given the growing body of research supporting the hyperarousal/gaze aversion model to explain atypical social gaze in ASD, methods to continuously monitor participant tolerability and social validity of procedures are needed in future research. Treatment acceptability is an important form of social validity that refers to the judgements about the treatment procedures by stakeholders, caregivers, and participants in the treatment on whether the treatment is fair, reasonable, appropriate and unintrusive, and whether they would recommend the treatment for others (Kazdin, 1980). Researchers should include these measures in future research, prioritizing participant perspective on treatment acceptability whenever possible to monitor assent and willingness to participate in such procedures.

The majority if interventions were conducted on individuals diagnosed with ASD. However, a few studies included individuals diagnosed with genetic disorders commonly associated with ASD (i.e., fragile X syndrome). Notably, mixed results were reported for both studies that included participants with FXS (Gannon et al., 2018; and Hall et al., 2009). It has been suggested that while individuals with FXS show similar gaze patterns as those with ASD, there may be distinct differences in gaze path and brain activity between the groups which could explain why social gaze may be more difficult to teach in FXS (Dalton et al., 2008). However, due to the small number of investigations aimed at promoting social gaze in individuals with FXS, firm conclusions about their relative responsiveness to these interventions (as compared to individuals with ASD) is unknown. Certainly, the research reviewed lends support for the acquisition of social gaze in at least some individuals with FXS, as 14 out of the total 26 participants with FXS across studies (54%) demonstrated significant improvements in social gaze. Additional research exploring procedures to promote social gaze in FXS are needed. Similarly, the studies included a disproportionate number of males compared to females, with female participants making up just 15% of the total participants across studies. Given that the male:female ratio in ASD has been projected to be 3:1 (Loomes et al., 2017), it appears that female participants were under-represented. Further, none of the studies reviewed included females with FXS, representing an important evidence gap in this line of research. Therefore, the generality of the findings to these subgroups is limited.

There were similarities in the way social gaze was defined across studies. For example, most studies defined the target response as looking at a face or eyes for some duration of time (1-5 seconds). These similarities in response definition are interesting given the wide age-range and different developmental stages, including toddlers, adolescents and adults included in the analysis because the characteristics of social communication and gaze interaction are very different across these stages. However, there was also considerable variability among the included studies in the way that authors defined appropriate social gaze, though these differences did not appear to be related to participants’ age. This variability in response definition could be problematic for at least two reasons. First, depending on the temporal specificity of the definition, when and for how long social gaze had to occur to meet the response criteria, the results of the intervention could have been inflated or deflated in comparison to other studies that employed less or more stringent definitions, respectively. Therefore, without a standard, established definition of social gaze as a dependent variable, the generality of findings is somewhat limited in this line of research.

Secondly, this finding reveals that there is no consensus on how to define appropriate social gaze. Walker and Hops (1976) described the use of normative peer data as a standard for evaluating treatment effects, which may allow clinicians a better source of information for evaluating the adequacy of post-treatment improvements. Normative data provides a measure of variability and trends over time and a basis for evaluating the practical significance and durability of treatment effects (Walker & Hops, 1976). Wang et al. (2019) presented an innovative method for utilizing normative data. These authors collected eye-tracking data from a group of typically developing children and used their gaze patterns to develop a model of prototypical looking. They then created adapted stimuli designed to direct attention toward locations of prototypical gaze. Using normative data to create a “prototypical map” of social scene looking, the authors combined looking positions across the sample to generate “normative heatmaps” which were used to construct corresponding filters that were applied to videos. Regions with high values in the normative heatmap were left bright and sharp, whereas the regions with low or no values in the heatmap were darkened or blurred to help direct attention accordingly. Approaches to measuring and training to a normative sample may be useful in producing results that are most appropriate to the participants’ specific social context, age group, and culture.

Future investigations could also employ multiple outcome measures of both objective and subjective information to improve differentiation of treatment approaches, and which alternatives are likely to produce superior results with similar populations. Hall and Venema (2017) evaluated the reliability, validity, and factor structure of a parent-report screening tool designed to measure eye contact avoidance in individuals with FXS and reported good psychometric properties. This measure specifically quantifies eye contact avoidance in five domains of functioning during social interactions with informants, friends/family and unfamiliar people (Hall & Venema, 2017). A tool such as this one could be particularly helpful as a concurrent measure of treatment effectiveness, as it would lend evidence for the amelioration of hyperarousal if implementation of the training procedures assuages discomfort from increased social gaze over time.

4.1. Implications

An important consideration concerns how individuals with ASD experience social gaze themselves and whether interventions to promote social gaze are appropriate. In a qualitative analysis of first-hand accounts of adults and teens with self-declared ASD, Trevisan and colleagues (2017) analyzed videos and texts posted on publicly accessible internet sites to explore the question, “How do people with self-declared ASD experience eye contact?” The most common emotions reported in their data included anxiety, panic, and fear. These qualitative data would seem to corroborate data that points to the hyperarousal/gaze aversion model as an explanation for atypical social gaze in ASD (as opposed to other theories that suggest an indifference to social gaze such that faces and eyes are not preferentially attended to in ASD; Moriuchi et al., 2017). Trevisan et al. (2017) also reported that, “Some advocated for increased awareness of “neurodiversity,” and the belief that people with ASD should not have to engage in eye contact just to conform with societal expectations.” (p. 13). Given the growing voice of the neurodiversity movement and, in particular, the account of procedures to promote social gaze as traumatic experiences, practitioners of ABA are likely at a crossroads when it comes to intervening on this pivotal behavioral target. Interestingly, Trevisan et al. (2017) also explored the strategies individuals with self-declared ASD used to improve their own social gaze or compensate for difficulties while making social gaze. Several informants recognized that difficulties with social gaze had created personal and professional barriers for them, which led them to employ various strategies to improve their own social gaze. Many described how they began practicing social gaze, gradually becoming increasingly comfortable over time (Trevisan et al., 2017).

While there appears to be a growing divide between treatment advocates and proponents of the neurodiversity movement, others have questioned the assumption that contemporary behavioral interventions necessarily run counter to the goals of the neurodiversity movement, arguing that collaborative partnerships which center the Autistic perspective and focus on co-constructed goals can improve their overall quality of life (Schuck et al., 2021). Research on the long-term effects of interventions to promote social gaze and whether procedures implemented in early childhood mitigate the need for supports later in life, lead to improved social functioning throughout the lifespan, and are correlated with improved quality of life overall, are clearly needed and would elucidate the necessity and appropriateness of these teaching strategies as part of early intervention for individuals with neurodevelopmental disorders. To this end, future research may be improved by including measures of social and ecological validity, including patient stakeholders, disclosing conflicts of interest, reporting adverse events including psychological distress, and allowing for flexibility in research such that interventions can be individualized (Schuck et al., 2021).

To our knowledge, this is the first scoping review of behavioral interventions to promote social gaze in ASD, providing a synthesis of key information on a complex and heterogeneous body of literature. The scoping review approach is broad in nature, limiting the degree to which firm conclusions can be made about intervention effectiveness with various participant profiles, notwithstanding a variety of strategies appear to be effective in promoting social gaze in individuals with ASD across age groups and in various settings. This topic maintains the interest of researchers, with the number of publications rising across the last 45 years and investigations of technology-based interventions gaining momentum over the last decade. Data are still limited in terms of the maintenance and generalization of results, and social validity of procedures. To better assess and improve upon the acceptability of or preference among alternative procedures, researchers should employ multiple measures to assess these aspects of treatment, particularly because a growing body of research indicates that individuals with ASD may experience increased physiological arousal or aversive private events when making eye contact with other people. Further research on the utility of interventions to promote social gaze in early childhood and the long-term impact it has on social functioning throughout the lifespan in ASD is needed.

Highlights.

• Behavioral interventions to promote social gaze may be beneficial for autistic individuals

• We summarized and evaluated the evidence for these interventions in ASD

• The review indicated that successful outcomes were reported in the majority of studies

• Further studies are needed to examine treatment maintenance, generalization, and social validity