The effectiveness of social training in individuals with autism spectrum disorder (ASD): a systematic review and transfer analysis

Abstract

Autism spectrum disorders (ASD) are characterized by impaired social communication and interactions, as well as constrained and repetitive manifestations of interests and behaviors. Various interventions at cognitive and behavioral levels aim to address impaired social communication and interaction in individuals with ASD. This study systematically explores the transferability of social training in individuals with ASD, guided by the conceptual model known as the FIELD framework (Function, Implement, Ecology, Level, and Durability). Employing the PRISMA methodology, 52 original experiments were included in the study. The transfer analysis, formulated as a conceptual meta-analysis based on the effect sizes, underscores the significant impact of variables including age, severity of autism, intervention tools, intervention intensity, intervention context, and intervention duration on the transferability of social training in individuals with ASD. The transfer of skills was particularly conspicuous among younger individuals, especially in face-to-face interventions, in contrast to digital alternatives. Moreover, cognitive interventions exhibited superior transferability compared to behavioral interventions, especially when administered with a higher intervention dose.

Introduction

Autism spectrum disorders (ASD), classified as a neurodevelopmental condition, exhibit two primary behavioral features: compromised social communication and interactions, along with constrained and repetitive manifestations of interests and behaviors ¹. These symptoms could be followed at behavioral, cognitive, and neural levels. Apart from the core behavioral symptoms of ASD, there are the potential to experience a range of maladaptive behaviors and challenges that extend beyond the defining features of the disorder. A variety of maladaptive social behaviors has been described in individuals with ASD, such as self-injurious acts², social withdrawal³, oppositional conduct⁴, irritability⁵, and aggression⁶. At the cognitive level, individuals with ASD have abnormal social information processing and struggle with impaired spatial abilities⁷, joint attention⁸, imitation⁹, emotion recognition¹⁰, self-referential processing¹¹, and theory of mind¹². At the neural level, Several meta-analyses of functional magnetic resonance imaging studies have described disruptions in the functioning of social brain regions^13–16

Regarding these atypical behavioral, cognitive, and neural characteristics, interventions aim to address and enhance these abnormal functions at these three levels. In this context, it’s important to consider, on the one hand, the fundamental impairments should be targeted for intervention with respect to a cause-and-effect relationships for improvement. On the other hand, recognizing the interconnectedness between these levels, any adjustments made at one level should spread to the others levels. This concept known as “transferability” in the field of intervention literature is widely recognized as a hallmark of effectiveness. An intervention lacking a transfer effect may be likened to a practice effect rather than constituting genuine improvement. Transferability explores the extent to which the effects of training can be applied beyond the specific domain in which the training occurred¹⁷. In the lens of the FIELD model, transfer encompasses five key dimensions: Function, Implement, Ecology, Level, and Durability¹⁸. “Function” refers to the ability to transfer of performance from a trained function, such as imitation, to an untrained function, like theory of mind. The “Implement” dimension deals with the various assessment and intervention materials and methods used in the assessment and training process. Compared to intervention materials and methods, showcasing improvement through diverse assessment tools signifies the occurrence of implement transfer. “Ecological transfer” involves the transference of intervention effects from one specific setting, for example, a clinical environment, to an entirely different and new setting, such as home. “Level transfer” focuses on the transfer of training effects across different levels, which can include neural, cognitive, and behavioral levels. Lastly, “Durability” is concerned with how long the effects of training persist after discharge, Table 1.

Table 1.

Description of different transfer domains based on FIELD model.

Dimension	Description
Function	The transfer of training effect from trained function to an untrained function(s)
Implement	Improvement demonstrated through diverse assessment tools during the assessment and training process
Ecology	The transference of intervention effects from one specific setting (e.g., clinical environment) to an entirely different and new setting (e.g. home)
Level	The transfer of training effects across different levels, including neural, cognitive, and behavioral levels
Duration	Concerned with how long the effects of training persist after discharge

Numerous interventions have been developed to enhance social functioning in individuals with autism, addressing both behavioral and cognitive aspects. On the behavioral front, for instance, intervention to improve social interaction skills^19–28, language skills^20,29,30, communication skills^19,31,32, symbolic play^33,34, cooperative play^20,35, social stories³⁶, social motivation, problem-solving abilities, and self-confidence skills¹⁹, and academic skills²⁰, or intervention to modulate abnormal behaviors such as obsessions, rituals, phobias, temper tantrums, and over-activity, as well as the teaching of constructional, play, and social skills³⁷.

Likewise, at the cognitive level, several cognitive training studies aimed remediation of abnormal information processing in children and adolescents with ASD with targeting emotion recognition^19,38–46, joint attention^{33,34,45,47–51}, imitation^{45,50,52–55}, eye contact^44,56, perspective taking^19,45, mentalizing^45,57, and social norm perception^19,45.

Viewing individuals with autism through a neurodiversity lens emphasizes the need to avoid pathologizing their distinctive behaviors and information processing styles^58–61. Consequently, interventions should refrain from targeting these inherent traits. In the realm of therapeutic interventions, the primary objective is to address abnormal functions that directly influence symptoms and, in turn, enhance the patient’s overall well-being by addressing their chief complaint. With this ultimate objective in mind, the transferability of interventions could prove pivotal in effectively supporting individuals with autism. Therefore, understanding how interventions can effectively adapt across various contexts becomes essential in ensuring that interventions honor the principles of neurodiversity and ultimately contribute to the well-being of individuals with autism.

The current study endeavors to conduct a comprehensive review of interventions targeting social cognition in individuals with autism. In addition to this, we aim to delve into the concept of transfer effect within the framework of the FIELD model. By analyzing the transferability of these interventions, we seek to shed light on the broader impact of such programs, their adaptability across various life domains, and their potential to foster lasting improvements in social cognitive skills among individuals on the autism spectrum.

Materials and methods

Search strategy

Initially, keywords associated with the theory of mind were identified through a thorough examination of review articles. Subsequently, systematic searches were performed on ScienceDirect, Google Scholar, and PubMed, without imposing any limitations on publication dates. The keywords employed in these searches encompassed terms related to autism spectrum disorder (ASD) and various types of training, such as "theory of mind training," "joint attention training," "spatial ability training," "emotion recognition training," "inhibition training," "imitation," "language training," "social interaction training," "flexible behavior training," and "communication skills training." In a comprehensive framework for social training terms, various domains of social cognition have been taken into account. These include higher social skills such as Theory of Mind, as well as basic social functions encompassing emotional components like emotional recognition, causation, learning, desire, belief, hiding, regulation, mixing, and morality, along with physical components such as imitation and self-other distinction. Additionally, perspective taking, communication/language, and joint attention are considered essential components within this framework. Each keyword was also explored individually. Publications were exclusively considered if they were in the English language.

Selection criteria

Initially, one reviewer (AP) screened articles by evaluating their titles and abstracts. Subsequently, duplications were removed, and search results were consolidated. To ensure rigorous eligibility assessment, two independent reviewers (FA and AP) thoroughly examined full-text articles. The selection process is visually represented in Fig. 1 through the PRISMA diagram. Studies were considered eligible if they satisfied the following inclusion criteria: (1) they offered training based on the social function, (2) they involved participants diagnosed with ASD, (3) they have a control group, and (4) they were published in the English language.

Fig. 1.

Data extraction diagram for review. 52 studies were entered into our study out of 580 initial candidate studies. 1 study was excluded in the screening phase, and 485 studies in the eligibility phase.

Data extraction and transfer analysis

The outcome measures in each study included behavioral rating scales and cognitive tests, which were used to assess the results. The specific scales and tests utilized are presented in Tables 1 and 2. Two independent reviewers (NN and FA) extracted the relevant data from each study, utilizing an Excel spreadsheet for organization. Subsequently, AP integrated the extracted data and resolved any discrepancies that arose between the initial reviewers. Summary statistics reported in most of the studies encompassed the number of participants in the control and test groups, mean age, and standard deviations (SDs).

Table 2.

The risk of bias in the included studies. RSG: random sequence generation, AC: allocation concealment, SR: selective reporting, PB: participants blinding, OB: outcome blinding, IO: incomplete outcome data, O: other biases, color coding; green/ + : low risk, yellow/?: unclear, red/-: high risk.

When assessing transferability across the five FIELD domains, we organized test results into individual domains. Utilizing a conceptual meta-analysis methodology, we sorted tests into their respective FIELD categories, and the cumulative effect sizes represented the overall effect size for each category. More precisely, we compared intervention and training domains within each FIELD category. If there was a complete overlap, it indicated no transfer, whereas differences suggested the presence of transfer. Subsequently, effect sizes for tests were computed within their respective FIELD domains. To calculate the effect size, we applied the following formula for Cohen’s d in each instance:

$$d=\frac{\left(M_2,-,M_1\right)}{\sqrt{\frac{{(n}_2-1)\times {\mathit{SD}}_2^2+{(n}_1-1)\times {\mathit{SD}}_2^2}{n_1+n_2-2}}}$$

Based on effect size magnitude, we categorize transfer effects as follows: an effect size below 0.2 is considered a small transfer effect, an effect size between 0.21 and 0.51 is classified as a medium transfer effect, and an effect size exceeding 0.51 is labeled as a large transfer effect.

Risk of bias

To evaluate the risk of bias in each study, the Cochrane Collaboration tool was employed. This tool facilitated the assessment of various factors including random sequence generation, allocation concealment, selective reporting, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, and other potential biases. The final judgments regarding the risk of bias for each study are displayed in Table 1. The risk of bias was categorized as low, high, or uncertain based on the assessment.

Results

A total of 52 studies were included in this review, encompassing a participant pool of 2350 individuals diagnosed with ASD. Among the participants, there were 1653 children, 415 adolescents, 239 adults, and 43 individuals whose age was not specified. The gender distribution comprised 411 females, 1885 males, and 54 cases where gender was not specified. The diagnosis of ASD was based on various tools, including DSM-4 in 8 studies, DSM-5 in 9 studies, DSM-4-TR in 5 studies, ICD-10 in 2 studies, DSM-4 & ICD-10 in 5 studies, DSM-4-TR & ICD-10 in 1 study, DSM-4 or DSM-5 in 1 study and not specified in 21 studies. In terms of study design, 37 studies utilized a randomized clinical trial (RCT) design and 13 studies employed a non-randomized controlled (NRC) design, and not specified in 2 studies. control groups in the studies consisted of wait-list controls (20 studies), active control groups (32 studies) (see Table 2 for details).

Regarding the assessment procedures employed in the included studies, 31 of them utilized two assessment sessions, consisting of a pre-test and post-test evaluation. On the other hand, 21 studies incorporated three assessment sessions, encompassing a pre-test, post-test, and follow-up evaluation. The duration of the follow-up period varied across studies, ranging from 1 week to 3 years. The assessments conducted in the studies can be categorized into three levels: neural, cognitive, and behavioral (as presented in Tables 3 and 4). Cognitive assessments, such as the facial emotion recognition test, involved tasks aimed at measuring the accuracy and/or speed of cognitive performance. Behavioral assessments, such as Autism spectrum Quotient, employed objective measures in the form of questionnaire-based evaluations to assess behavioral performance. The assessments in the included studies were conducted in various settings, including home (7 studies), clinical settings (18 studies), and school setting (1 study). Some studies utilized assessments in multiple settings, such as home and clinic (16 studies) and home and school (10 studies). As for the interventions, they were carried out in different settings as well, including home (3 studies), clinical settings (30 studies), and school (8 studies). Similar to assessments, some studies implemented interventions in multiple settings, such as home and clinic (12 studies), home and school (1 study), school and clinic (1 study), and home, clinic and school (2 studies). In Table 3 and 4, the behavioral and cognitive assessment tools are classified based on the agent/responder (self-, parent-, teacher-, clinician-rating), objectivity, material (computerized, motion, observational, paper and pencil, and robot tasks), and the setting in which they were performed (clinic, home, or school).

Table 3.

Demographic and diagnostic characteristics of studies.

Authors, Year	Study Design, Control1	N (I:C), Age Mean (SD), Age Range, Gender (F/M), Education (Yr)	Diagnosis
Chien et al., 2023	RCT, TAU	82 (41:41), 26.45 (5.25), 18–45, 13/69, ns	DSM-4
Nejati, 2023	RCT, ACG	30 (15:15), 8.76 (1.61), ns, 0/30, 2.21	DSM-5
Afsharnejad et al., 2022	RCT, SC	90 (46:44), 13.77 (1.60), 12–17, 25/65, ns	DSM-4 or DSM-5
Elhaddadi et al., 2022	ns, VM	18 (9:9), 6.45 (0.8), 5–8, 4/14, ns	ns
Mirzaei et al., 2022	RCT, RC	50 (25:25), 10.51 (2.23), 6–12, 9/41, ns	DSM-4
Mohammadi et al., 2022	RCT, RT	52 (25:27), 11.38 (2.51), 6–12, 19/33, ns	DSM-4
Sosnowski et al., 2022	RCT, CVG	54 (25:29), 8.56 (2.84), 4–14, 7/47, ns	DSM-5
Bashirian et al., 2021	RCT, TEFT	54 (27:27), 11.51 (1.29), 6–12, 21/33, ns	DSM-5
Beaumont et al., 2021	RCT, CIA	70 (35:35), 9.89 (1.37), 7–12, 10/60, ns	DSM-5
Cheung et al., 2021	NRCT, WC	74 (45:29), 9.91 (24.7), 6–14, 7/67, ns	DSM-4
Yamada et al., 2020	NRCT, WC	28 (14:14), 13.08 (13.66), 11–15, 9/19, ns	ns
Chester et al., 2019	NRCT, WC	45 (15:15), 10.16 (1.26), 8–12, 9/36, 4.66 (1.2)	ns
Corbett et al., 2019	RCT, WC	77 (44:33), 10.89 (2.44), 8–16, 18/59, ns	DSM-5
Dekker et al., 2019	RCT, TAU	122 (98:24), 10.96 (0.78), 9.6–13.0, 19/103, ns	DSM-4-TR
Holopainen et al., 2019	RCT, WC	135 (72:63), 9.5 (1.67), 8–13, 16/119, ns	DSM-4-TR
Carlson et al., 2018	ns, ACG	20 (10:10), 5.3 (0.7), 4–6, 0/20, 0	ns
Einfeld et al., 2018	NRCT, TAU	84 (26:58), 10.71 (1.46), 8.2–14.6, 9/75, ns	DSM-4-TR & ICD-10
Garcia-Villamisar, 2018	RCT, WC	43 (22:21), 33.71 (5.98), 22–44, 16/27, ns	ns
Kumazaki et al., 2018	RCT, JA-HA	28 (16:12), 5.82 (0.44), 5–6, 9/19, ns	DSM-5
Becker et al., 2017	NRCT, TSST	31 (17:14), 10.97 (1.84), 8–14, 3/28, ns	ns
Yun et al., 2017	RCT, FER-HF	15 (8:7), 6.01 (1.04), 4–7, 0/15, ns	DSM-5
Al-Dawaideh, 2016	RCT, WC	20 (10:10), 6.45 (ns), 6–12, 0/20, ns	ns
Koehne et al., 2016	RCT, CMI	51 (27:24), 32.7 (9.1), 18–55, 19/ 32, ns	DSM-5 & ICD-10
Russo-Ponsaran et al., 2016	RCT, WC	25 (12:13), 11.53 (1.80), 8–15, 5/20, ns	ns
Srinivasan et al., 2016	RCT, TTA	36 (24:12), 7.63 (2.24), 5–12, 4/32, ns	ns
Adibsereshki et al., 2015	RCT, WC	24 (12:12), 9.74 (1.59), 7–12, 12/12, ns	ns
Beaumont et al., 2015	NRCT, SASP	69 (35:34), 9.54 (1.55), 7–12, 5/64, ns	ns
Golzari et al., 2015	RCT, WC	30 (15:15), ns (ns), 6–12, 0/30, ns	ICD-10
Karst et al., 2015	RCT, WC	64 (32:32), 13.75 (1.4), 11–16, 11/53, ns	DSM-5
Laugeson et al., 2015	RCT, WC	22 (12:10), 20.4 (1.85), 18–24, 4/18, ns	ns
Laugeson et al., 2014	NRCT, VGMS	73 (40:33),13 (0.7), 12–14, 9/64, 7.58 (0.5)	DSM-4
Lerna et al., 2014	NRCT, CLT	14 (7:7), 5.72 (0.98), 1.5–5, ns, 0	DSM-4
Schohl et al., 2014	RCT, WC	58 (29:29), 13.65 (1.50), 11–16, 11/47, ns	ns
Warreyn & Roeyers, 2014	NRCT, ACG	36 (18:18), 5.73 (0.65), 4.38–6.86, 9/27, 0	DSM-4-TR
Yoo et al., 2014	RCT, WC	47(23:24), 13.78 (1.57), 12–18, 3/44, ns	DSM-4-TR
Ingersoll, 2012	RCT, ACG	27 (14:13), 3.16 (0.63), 2.25–3.91, 3/24, 0	DSM-4-TR
Kasari et al., 2012	RCT, WC	60 (30:30), 8.14 (1.56), 6–11, 6/54, 2.65	ns
Lawton & Kasari, 2012	RCT, ABA-EIP	52 (36:16), 3.5 (0.5), 3–4, 40/12, 0	DSM-4 & ICD-10
Lerna et al., 2012	NRCT, CLT	18 (9:9), 3.23 (0.67), 1.5–5, 1/17, 0	DSM-4
Lerner & Mikami, 2012	RCT, SSI	13 (7:6), 11.07 (1.65), ns, 0/13, 5.30 (1.76)	ns
Derosier et al., 2011	RCT, TSST	55 (27:28), 10.0 (1.2), 8–12, 1/54, ns	ns
Laugeson et al., 2009	RCT, WC	33 (17:16), 14.6 (1.4), 13–17, 5/28, ns	ns
Kasari & Paparella, 2008	RCT, TAU	58 (41:17), 3.55 (0.53), 3–4, 12/46, 0	DSM-4 & ICD-10
Owens et al., 2008	NRCT, WC	47 (31:16), 8.3 (1.61), 6–11, 1/46, ns	ns
Gulsrud et al., 2007	RCT, SPT	35 (17:18), 3.57(0.58), 2.75–4.5, 7/28, ns	DSM-4 & ICD-10
Kroeger et al., 2007	NRCT, PAGF	25 (13:12), 5.2 (0.8), 4–6, 5/20, 0	ns
Golan & Baron-Cohen, 2006	RCT, WC	41 (19:22), 30.71 (10.78), 21–43, 10/31, ns	ns
Heimann et al., 2006	RCT, NSFP-CI	20 (10:10), 6.5 (2.2), 4.4–12.9, 1/19, ns	ICD-10
Sallows & Graupner, 2005	RCT, PDG	23 (13:10), 2.80 (0.36), 2–3.5, 4/19, 0	DSM-4
Solomon et al., 2004	RCT, WC	18 (9:9), 112 (ns), 8–12, ns, ns	DSM-4
Silver & Oakes, 2001	RCT, TAU	22 (11:11), ns, 10–18, ns, ns	DSM-4 & ICD-10
HOWLIN, 1981	NRCT, WC	32 (16:16), 6.28 (2.39), 3–11, 0/32, ns	ns

1. ABA-EIP: Applied Behavior Analysis-Early Intervention Program, ACG: Active Control Group, CIA: Central Intelligence Agency, CLT: Conventional Language Therapy, CMI: Control Movement Intervention, CVG: Control Video Game, FER-HF: Facial Emotion Recognition-Human Facilitated, JA-HA: Joint Attention-Human Agent, NRCT: Non-Randomized Control Trial, NSFP-CI: Nadels ‘Still-Face Paradigm-Contingent Interaction, PAGF: Play Activities Group Format, PDG: Parent Direct Group, RC: Routine Care, RCT: Randomized Control Trial, RT: Routine Training, SASP: Secret Agent Society Program, SC: Super Chef (Interactive Group Cooking Program), SPT: Symbolic Play Therapy, SSI: Social Skills Intervention, TAU: Treatment As Usual, TEFT: Traditional Emotional Facial Training , TSST: Traditional Social Skill Training, TTA: TableTop activities, VGMS: Village Glen Middle School, VM: Video Modeling, WC: Waitlist Control

Table 4.

Description of interventions and properties.

Intervention	Description (Respective included studies)
Animal-assisted Social Skills Training (ASST)62	A therapist-guided and curriculum-based intervention for children with ASD included getting acquainted, making friends and conversation, play skills, empathy, self-regulation, and conflict management 63
Attention Remediation of Theory of Mind (ARTOM) 45	A therapist-guided computerized intervention to train imitation, face recognition, emotion recognition, Joint attention, perspective taking, mentalizing, and social norm perception 45
Coach Assistant MiX (CA-MiX)64	A therapist-guided robotic emotion recognition training through videos and imitation exercise 38
CommU Robotic Intervention (CRI)47	Joint attention training through alternating gaze training 47
Computer-facilitated Emotion Recognition Training (CERT)65	A self-administered computerized emotion perception training using dynamic scenes, schematic drawings, situations, desires, and beliefs and informational states training through conceptual and visual perspective taking, knowledge tracking, and false beliefs 19
CuDDler 66	A therapist-guided robotic joint attention training through an imbedded robot 48
Emotion Trainer (ET)46	A self-administered computerized emotion recognition program using faces, scenes or objects 46
Direct Teaching Group Format (DTGF)35	A therapist-guided mental and motor group-based training for socialization, simple motor behaviors, joint play, pretend play, and interactive play 35
Facial Emotion Recognition (FER)67	A parent- and therapist-guided paper–pencil facial emotion recognition training 40
Facial Emotional Recognition and Eye Contact Training (FER-ECT)65	A robotic computerized educational program for eye contact and facial emotion recognition 44
Home-based Language Training Program (HLTP)68	A parental paper–pencil training to modify a child’s behavior—such as obsessions, rituals, phobias, temper tantrums, and over-activity, as well as the teaching of constructional, play, and social skills 37
Imitative Interaction Training (IIT)52	A therapist-guided mental training to target imitation of the child movements and sounds, including stereotypes, involves immediately reflecting the child’s behavior by mimicking the actions demonstrated by the child 52
Intensive Behavioral Treatment (IBT)69	A parental- and therapist-guided, paper–pencil, and behavioral intervention to target receptive and expressive language, social interaction, cooperative play, and academic skills were targeted. 20
Joint Attention and Imitation Training (JAIT)50	A therapist-based, paper–pencil, and behavioral intervention to target gestural, vocal, object, and symbolic imitation and joint attention 50
Joint Attention Training (JAT)70	A teacher- and therapist-guided motion and paper–pencil behavioral intervention, with emphasis on embedding strategies, targeting joint attention into teachers’ everyday classroom routines and activities 34,49,71
KONTAKT® 72	KONTAKT® is a therapist-based paper–pencil behavioral intervention to improve communication and social interaction skills, social motivation, problem-solving abilities, and self-confidence 19
Language Training Program (LTP)73	A therapist-based, paper–pencil, and question guided language training with providing praise feedback 29
LEGO Therapy (LT)51	A self-administered and paper–pencil training of joint attention, turn taking, sharing, joint problem solving, listening and general social communication skills 51
Let’s Face It (LFI)74	A parental- and therapist-guided computerized training, which consists of eight gamified task to target facial emotion recognition. (Bashirian et al., 2021)
LookWare (LW)75	A self-administered computerized gamified emotion recognition training using auditory and visual prompts to encourage the player to look at specific emotionally expressive areas of the face 42
Mind Reading (MR)76	A self-administered intervention consists of a software program that teaches participants to systematically analyze emotions by comparing facial expressions and vocal tones 43
Picture Exchange Communication Scale (PECS)77	A therapist-based, paper–pencil, and behavioral intervention based on reinforcement techniques for functional communication in a social context 31,32
Program for the Education and Enrichment of Relational Skills (PEERS)78	A parental- and therapist-guided paper–pencil manualized caregiver-assisted social skills program to establish and maintain friendships, and managing peer conflict and rejection 28,79,79–84
Reciprocal Imitation Training (RIT)85	A parental- and therapist-guided, paper–pencil, and behavioral intervention to teach imitation within a social-interactive context 53,54
Rhythm Training (RhT)56	A parental- and therapist-guided, motor and paper–pencil, and behavioral intervention for eye contact, turn-taking, greeting/farewell, responding to questions, commenting, asking for help, and the use of gestures, as well as targeting gross motor skills 56
Robot Training (RT)56	A robot-assistant motor intervention to train social communication skills such as eye contact, turn-taking, greetings/farewells, responding to questions, commenting, asking for help, and using gestures, while also targeting gross motor skills 56
Secret Agent Society (SAS)21	A teacher- parent- and therapist-guided set of computerized games to target social and emotional skills 21,22
Sense Theatre (ST)24	A self-administered, mental, and theater-based intervention using learning theory behavioral strategies, and peer mediation, to explore and practice social interaction skills 24
Social Adjustment Enhancement Intervention (SAEI)23	A self-administered and mental intervention to target emotional understanding, receptive and expressive body language, problem solving, friendship, and conversational skills 23
Social Cognitive Intervention Program (SCIP)25	A therapist-guided computerized social skill training through visually scaffolded materials (such as comic-style short stories) 25
Social Skills Program (SSP)26	A therapist-guided paper–pencil pretend play for social skill training (e.g., eye contact, tone of voice, body language, etc.) 86
Social Skills Training (SST)87	A parent- teacher- therapist-guided mental social skill training based on behavioral therapeutic principles and the social learning theory 88,87
Social Stories Intervention (SSI)36	A therapist-guided paper–pencil social skill training based on social stories 89
Social Use of Language Program (SULP)30	A therapist-guided mental educational program based on stories, group activities, and games to target social and communication skills, including eye contact, listening, turn taking, proxemics and prosody
Sociodramatic Affective Relational Intervention (SDARI)27	A therapist-guided motion and mental game-based social skill training, with instrumental reinforcement 90
Symbolic Play (SP)34	A therapist-guided, paper–pencil, and behavioral intervention aimed at incorporating symbolic play into teachers’ daily classroom routines and activities 33,70
Synchronize Dance/Movement Intervention (SDMI)55	A therapist-guided dance training based on imitation 91
Theory of Mind Training (ToMT)57	A therapist-guided paper–pencil intervention to target emotion, situational emotions, desire, and desire- beliefs 57,92

Table 4 provides an overview of the characteristics of intervention programs, which are classified according to how they are administered (by clinicians, parents, teachers, or self-administered), their objectivity (whether they are based on objective tasks or subjectivity), and the materials used (such as computerized, paper and pencil, educational, or motoric materials with a physical activity demand). These attributes of assessment tools and intervention programs were incorporated into our framework to investigate various aspects of knowledge transfer across different domains.

In the context of assessing and intervening in various domains, we have evaluated the transfer of training effects, Table 5. Broadly, any distinctions observed between assessment and intervention domains have been interpreted as indicative of transfer within the specific domain. To elaborate further, for the cognitive intervention, concerning the concept of functional transfer in the FIELD model, we have classified any transfer from a trained function to an untrained function according to specific conceptual models. In our model, we have delineated cognitive functions into distinct blocks based on their structural foundations and functional similarities. Transfer within the same block has been termed "near transfer," while transfer between different blocks has been denoted as “far transfer”, Fig. 2.

Table 5.

Properties of assessments and intervention in the included studies.

Author, Year	Intervention			Assessment
	Name1(Agent2)	Setting3	Dose	Characteristic4: Name5	Setting3	Time6
Chien et al., 2023	PEERS(C)	C	24	P/SSP: AQ; C/SSP: SRS; SSP: EQ, TYASSK & SIAS; PSP: ESQ	CH	PPF
Nejati, 2023	ARTOM(C)	C	10	PSP: SASQ & GARS; SOC: TMT, EEFT, NSFERT & CFERT	CH	PPF
Afsharnejad et al., 2022	KONTAKT(C)	C	24	PSP: SRS; SSP: SIAS, CSIE, PALS, PQLI, ESM & ERSSQ; SOC: MB	CH	PPF
Elhaddadi et al., 2022	RIT(CP)	CH	18.5	CSO: PSS	C	PPF
Mirzaei et al., 2022	FER(CP)	C	7.5	PSP: MCRQ	H	PP
Mohammadi et al., 2022	FER(CP)	C	3.33	SOC: BFRS	C	PPF
Sosnowski et al., 2022	LW(S)	CSH	6.5	SOC: EEFT	C	PP
Bashirian et al., 2021	LFI(CP)	C	5	SOC: BFRS	C	PPF
Beaumont et al., 2021	SAS(P)	H	ns	P/TSP: SSQ & ERSSQ; PSP: SCAS & ECBI	SH	PPF
Cheung et al., 2021	SCIP(C)	S	20	PSP: TMI, SSRS & GAS; SOP: SST	CH	PP
Yamada et al., 2020	PEERS(CP)	CH	21	PSP: SRS, VABS, CBCL & SCQ; SSP: TASSK & DSS; P/SSP: QSQ	H	PPF
Chester et al., 2019	SSP-S(C), SSP-U(S)	C	16	P/TSP: SSRS; P/T/SSP: SSQ	SH	PPF
Corbett et al., 2019	ST(S)	S	40	SOC: IFM; CSO: PIP	C	PP
Dekker et al., 2019	SST(C), SSTPTI(CPT)	C, CHS	27	P/TSP: SSRS; PSP: VABS	SH	PP
Holopainen et al., 2019	ToMT(C)	C	8	CSO: SOER	C	PP
Carlson et al., 2018	CuDDler(R)	C	1.33	CSO: ESCS	C	PP
Einfeld et al., 2018	SAS(PT)	SH	21.5	P/TSP: SSQ & ERSSQ; SOP: JMT & DIBT	SH	PP
Garcia-Villamisar, 2018	CERT(S)	C	180	SSP: SSS; SOC: EMBA-A	CH	PP
Kumazaki et al., 2018	CRI(R)	C	0.25	COR: AJA	C	PP
Becker et al., 2017	ASST(C)	C	12	SSP: CDI; SOC: RMET; SOP: SLDT	CH	PP
Yun et al., 2017	FER-ECT(R)	C	4.75	CSO: ADOS, FEC & FER; PSP: SRS, SCQ & CBCL	CH	PP
Al-Dawaideh, 2016	LTP(C)	C	24	SOP: ELT	C	PPF
Koehne et al., 2016	SBDMI(C)	C	15	SSP: IRI; SOC: IS & MET; CSO: AI & ASIM	CH	PP
Russo-Ponsaran et al., 2016	CA-MiX(C)	CS	16	SOC: CATS, DANVA, CASPS & NEPSY; CSO: DASE; SOP: EF & ES; SSP: BEQI	CH	PPF
Srinivasan et al., 2016	RT(RP), RhT(CP)	CH, CH	24	COM: BOT; CSO: TTIP	C	PP
Adibsereshki et al., 2015	ToMT(C)	C	ns	P/TSP: SSRS	SH	PP
Beaumont et al., 2015	SAS(T)	S	15	P/TSP: ERSSQ, SSQ & CAPES; PSP: SCAS; SOP: JMT & DIBT	SH	PPF
Golzari et al., 2015	SSI(C)	S	5	PSP: TSSA	H	PP
Karst et al., 2015	PEERS(CP)	CH	21	PSP: CHOS, PSOC & SIPA	H	PP
Laugeson et al., 2015	PEERS(SC)	CH	24	PSP: SRS, EQ & SSRS; P/SSP: QSQ; SSP: TASSK	H	PP
Laugeson et al., 2014	PEERS(T)	S	35	P/TSP: SRS & SSRS; P/SSP: QSQ & SAS; SSP: FQS, TASSK & PHSS	SH	PP
Lerna et al., 2014	PECS(C)	C	288	PSP: VABS; CSO: ADOS; SOP: GMDS; SSO: UFPE	CH	PPF
Schohl et al., 2014	PEERS(CP)	CH	21	SSP: TASSK, SIAS & FQS; P/SSP: QSQ; P/TSP: SRS & SSRS	SH	PP
Warreyn & Roeyers, 2014	JAIT(C)	C	12	CSO: OIJA	C	PP
Yoo et al., 2014	PEERS(CP)	CH	21	CSO: ADOS; PSP: VABS, SRS, CBCL & SCQ; SSP: TASSK, CDI, STAIC & SSRS; P/SSP: QSQ	CH	PPF
Ingersoll, 2012	RIT(C)	C	30	CSO: ESCS; PSP: BSID	CH	PPF
Kasari et al., 2012	PEERS(C)	S	20	TSP: TPSS; CSO: SNS	S	PPF
Lawton & Kasari, 2012	JAT(C), SP(C)	C, C	13.5	CSO: ESCS	C	PPF
Lerna et al., 2012	PECS(C)	C	288	PSP: VABS; CSO: ADOS; SOP: GMDS; SSO: UFPE	CH	PP
Lerner & Mikami, 2012	SDARI(C)	S	6	CSO: SIOC; P/TSP: SSRS; PSP: SRS & SCQ; SSP: SN	SH	PP
Derosier et al., 2011	SST(CP)	CH	15	PSP: SRS & ALQ; SSP: SDQ	H	PP
Laugeson et al., 2009	PEERS(CP)	CH	18	P/TSP: SSRS; P/SSP: QSQ; SSP: TASSK & FQS	SH	PPF
Kasari & Paparella, 2008	JAT(C), SP(C)	C, C	13.5	CSO: SPA, RDLS & ESCS; PSO: MCI	C	PPF
Owens et al., 2008	LT(S), SULP(C)	C, C	18	PSP: VABS & GARS; CSO: OIJA	CH	PPF
Gulsrud et al., 2007	JAT(C)	C	16	COO: JAP	C	PP
Kroeger et al., 2007	DTGF(C)	C	15	CSO: SIOC & ABLLS	C	PP
Golan & Baron-Cohen, 2006	MR(S)	H	20	SOC: CMFVB, RMET & RMVT	C	PP
Heimann et al., 2006	IIT(C)	C	1.75	CSO: PEP	C	PP
Sallows & Graupner, 2005	IBT(CP)	CH	7680	CSO: RDLS; PSP: VABS; CSP: ADI	C	PP
Solomon et al., 2004	SAEI(S)	CH	75	SOC: DANVA & FPS; SOP: SST & TPS; SSP: CDI & BDI	CH	PP
Silver & Oakes, 2001	ET(S)	S	5	SOP: SST, ERC & FEP	C	PP
Howlin, 1981	HLTP(P)	H	Ns	SSO: ALLLU	H	PPF

Fig. 2.

A conceptual framework for functional transfer at the cognitive level.

In the context of behavioral intervention, we assessed the transfer of training effects with regard to ASD symptoms and categorized it as "near transfer." Conversely, when the transfer extended to more comprehensive dimensions of performance outside the immediate scope of the intervention, such as improvements in overall quality of life, it was regarded as "far transfer." In the context of implemental transfer, any differences between assessment and intervention tools, concerning the materials and methods used, have been designated as implemental transfer. In the realm of ecological transfer, distinctions in the settings between intervention and assessment have been referred to as ecological transfer. Regarding the concept of transfer at the cognitive and behavioral levels, given that we employ both cognitive and behavioral tests and interventions, the results of cognitive assessments for behavioral interventions and the results of behavioral assessments for cognitive interventions have been categorized as "transfer of level." Finally, concerning durability, if an intervention includes follow-up assessments, the outcomes of these tests have been described as "durability transfer." Overall, for all FIELD’s domains, we find that 14 studies did not yield any transfer effects ^{20,28,40,45,53,63,74,79,82,87,90,93–95}, while 17 studies exhibited minor transfer effects ^{19,24,25,37,43,44,48–51,54,56,71,81,88,92,96}. Moreover, 10 studies unveiled moderate transfer effects ^{21–23,35,42,46,47,52,89,91}, and 11 studies demonstrated significant transfer effects ^{31–33,38,39,57,70,73,80,83,86}, Table 6.

Table 6.

The effect sizes of included studies in the FIELD’s domains.

Author (Year)	Cohen’s D (95% Confidence Interval)
	F	I	E	L	D	S
Chien et al., 2023	-0.00 (-0.23, 0.23)	-0.28 (-0.57, 0.00)	-0.06 (-0.29, 0.17)	-0.28 (-0.57, 0.00)	-0.28 (-0.57, 0.00)	-0.18 (-0.3, -0.06)
Nejati, 2023	-0.94 (-2.61, 0.71)	-1.55 (-2.93, -0.17)	-1.55 (-2.93, -0.17)	-0.94 (-2.61, 0.71)	-0.94 (-2.61, 0.71)	-1.19 (-1.86, -0.51)
Afsharnejad et al., 2022	0.11 (0.05, 0.17)	0.11 (0.05, 0.17)	0.11 (0.05, 0.17)	0.11 (0.05, 0.17)	0.11 (0.05, 0.17)	0.11 (0.08, 0.14)
Elhaddadi et al., 2022	-0.43 (-1.87, 0.99)	-0.43 (-1.87, 0.99)	-0.43 (-1.87, 0.99)	0 (0, 0)	0 (0, 0)	-0.26 (-0.72, 0.2)
Mirzaei et al., 2022	-1.45 (-9.12, 6.22)	-1.45 (-9.12, 6.22)	-1.45 (-9.12, 6.22)	0 (0, 0)	0 (0, 0)	-0.87 (-3.31, 1.57)
Mohammadi et al., 2022	9.4 (5.21, 13.6)	9.4 (5.21, 13.6)	0 (0, 0)	9.4 (5.21, 13.6)	9.4 (5.21, 13.6)	7.52 (5.37, 9.67)
Sosnowski et al., 2022	0.55 (-)	0 (-)	0.55 (-)	0 (-)	0 (-)	0.22 (-0.04, 0.48)
Bashirian et al., 2021	-3.75 (-8.09, 0.59)	-7.54 (-11.07, -4.02)	0 (0, 0)	-7.54 (-11.07, -4.02)	-7.54 (-11.07, -4.02)	-5.28 (-7.1, -3.45)
Beaumont et al., 2021	-0.02 (-0.47, 0.43)	0.14 (-0.37, 0.66)	0.05 (-0.02, 0.13)	0.14 (-0.37, 0.66)	0.14 (-0.37, 0.66)	0.09 (-0.09, 0.28)
Cheung et al., 2021	0.08 (-0.09, 0.26)	0.08 (-0.09, 0.26)	0.08 (-0.09, 0.26)	0 (0, 0)	0 (0, 0)	0.05 (-0.01, 0.11)
Yamada et al., 2020	-0.02 (-0.21, 0.15)	-0.02 (-0.21, 0.15)	-0.02 (-0.21, 0.15)	-0.02 (-0.21, 0.15)	-0.02 (-0.21, 0.15)	-0.02 (-0.1, 0.05)
Chester et al., 2019	0 (0, 0)	1.14 (0.92, 1.37)	1.14 (0.92, 1.37)	1.14 (0.92, 1.37)	1.14 (0.92, 1.37)	0.92 (0.76, 1.07)
Corbett et al., 2019	0 (0, 0)	0.22 (0.08, 0.35)	0 (0, 0)	0 (0, 0)	0 (0, 0)	0.04 (0.00, 0.08)
Dekker et al., 2019	0 (0, 0)	0.43 (0.37, 0.5)	0.43 (0.37, 0.5)	0 (0, 0)	0 (0, 0)	0.17 (0.1, 0.25)
Holopainen et al., 2019	0.28 (0.05, 0.51)	0.28 (0.05, 0.51)	0 (0, 0)	0 (0, 0)	0 (0, 0)	0.11 (0.02, 0.2)
Carlson et al., 2018	0.75 (0.48, 1.01)	0.75 (0.48, 1.01)	0 (0, 0)	0 (0, 0)	0 (0, 0)	0.3 (0.05, 0.54)
Einfeld et al., 2018	0.38 (0.11, 0.64)	0.38 (0.11, 0.64)	0.38 (0.11, 0.64)	0 (0, 0)	0 (0, 0)	0.22 (0.12, 0.33)
Garcia-Villamisar, 2018	0.41 (-0.61, 1.44)	-0.28 (-0.85, 0.27)	-0.28 (-0.85, 0.27)	0 (0, 0)	0 (0, 0)	-0.03 (-0.29, 0.22)
Kumazaki et al., 2018	0.84 (-)	0.84 (-)	0 (-)	0 (-)	0 (-)	0.33 (-0.06, 0.74)
Becker et al., 2017	-0.3 (-0.49, -0.1)	-0.26 (-0.5, -0.02)	-0.26 (-0.5, -0.02)	0 (0, 0)	0 (0, 0)	-0.16 (-0.25, -0.08)
Yun et al., 2017	0.28 (0.17, 0.4)	0.28 (0.17, 0.4)	0.21 (0.1, 0.33)	0 (0, 0)	0 (0, 0)	0.15 (0.11, 0.2)
Al-Dawaideh, 2016	21.55 (14.84, 28.26)	21.55 (14.84, 28.26)	0 (0, 0)	21.55 (14.84, 28.26)	21.55 (14.84, 28.26)	17.24 (11.26, 23.22)
Koehne et al., 2016	0.58 (0.28, 0.88)	0.58 (0.28, 0.88)	0 (0, 0)	0 (0, 0)	0 (0, 0)	0.23 (0.12, 0.34)
Russo-Ponsaran et al., 2016	0.66 (0.23, 1.1)	0.66 (0.23, 1.1)	0.66 (0.23, 1.1)	0.66 (0.23, 1.1)	0.66 (0.23, 1.1)	0.66 (0.48, 0.85)
Srinivasan et al., 2016	0.13 (-0.1, 0.38)	0.13 (-0.1, 0.38)	0.13 (-0.1, 0.38)	0 (0, 0)	0 (0, 0)	0.08 (0.00, 0.16)
Adibsereshki et al., 2015	1.05 (0.72, 1.38)	1.05 (0.72, 1.38)	1.05 (0.72, 1.38)	0 (0, 0)	0 (0, 0)	0.63 (0.28, 0.97)
Beaumont et al., 2015	0.35 (-0.02, 0.73)	0.39 (0.01, 0.76)	0.39 (0.01, 0.76)	0.39 (0.01, 0.76)	0.39 (0.01, 0.76)	0.38 (0.22, 0.54)
Golzari et al., 2015	0.36 (0.1, 0.61)	0.46 (0.28, 0.65)	0.46 (0.28, 0.65)	0 (0, 0)	0 (0, 0)	0.25 (0.15, 0.36)
Karst et al., 2015	-0.13 (-0.58, 0.32)	-0.13 (-0.58, 0.32)	-0.13 (-0.58, 0.32)	-0.13 (-0.58, 0.32)	-0.13 (-0.58, 0.32)	-0.13 (-0.58, 0.32)
Laugeson et al., 2015	-0.48 (-1.03, 0.06)	-0.48 (-1.03, 0.06)	-0.51 (-1.04, 0.02)	-0.48 (-1.03, 0.06)	-0.48 (-1.03, 0.06)	-0.49 (-0.72, -0.25)
Laugeson et al., 2014	0.88 (0.39, 1.38)	0.88 (0.39, 1.38)	0.88 (0.39, 1.38)	0 (0, 0)	0 (0, 0)	0.53 (0.29, 0.76)
Lerna et al., 2014	1.52 (-0.07, 3.12)	1.52 (-0.07, 3.12)	0.09 (-0.04, 0.23)	1.52 (-0.07, 3.12)	1.52 (-0.07, 3.12)	1.24 (0.6, 1.87)
Schohl et al., 2014	0.33 (-0.08, 0.75)	0.33 (-0.08, 0.75)	0.33 (-0.08, 0.75)	0 (0, 0)	0 (0, 0)	0.20 (0.05, 0.34)
Warreyn & Roeyers, 2014	0.51 (0.26, 0.75)	0.51 (0.26, 0.75)	0 (0, 0)	0 (0, 0)	0 (0, 0)	0.2 (0.11, 0.29)
Yoo et al., 2014	-0.02 (-0.5, 0.46)	-0.02 (-0.5, 0.46)	-0.02 (-0.5, 0.46)	-0.02 (-0.5, 0.46)	-0.02 (-0.5, 0.46)	-0.02 (-0.23, 0.18)
Ingersoll, 2012	0.9 (0.66, 1.13)	0.9 (0.66, 1.13)	0.9 (0.66, 1.13)	0.9 (0.66, 1.13)	0.9 (0.66, 1.13)	0.9 (0.82, 0.97)
Kasari et al., 2012	0.25 (0.04, 0.47)	0.25 (0.04, 0.47)	0 (0, 0)	0.25 (0.04, 0.47)	0.25 (0.04, 0.47)	0.2 (0.12, 0.29)
Lawton & Kasari, 2012	1.15 (0.91, 1.39)	1.15 (0.91, 1.39)	0 (0, 0)	1.15 (0.91, 1.39)	1.15 (0.91, 1.39)	0.92 (0.69, 1.14)
Lerna et al., 2012	1.18 (0.28, 2.08)	1.18 (0.28, 2.08)	0.12 (-0.07, 0.32)	1.18 (0.28, 2.08)	1.18 (0.28, 2.08)	0.97 (0.6, 1.33)
Lerner & Mikami, 2012	-0.51 (-1.21, 0.19)	-0.51 (-1.21, 0.19)	-0.57 (-1.24, 0.1)	0 (0, 0)	0 (0, 0)	-0.31 (-0.56, -0.07)
Derosier et al., 2011	-0.59 (-0.9, -0.28)	-0.44 (-1.04, 0.14)	-0.44 (-1.04, 0.14)	0 (0, 0)	0 (0, 0)	-0.29 (-0.49, -0.1)
Laugeson et al., 2009	0.12 (-0.12, 0.37)	1.08 (0.5, 1.65)	1.08 (0.5, 1.65)	1.08 (0.5, 1.65)	1.08 (0.5, 1.65)	0.89 (0.62, 1.61)
Kasari & Paparella, 2008	0.87 (0.54, 1.2)	0.87(0.54, 1.2)	0 (0, 0)	0 (0, 0)	0.87 (0.54, 1.2)	0.52 (0.38, 0.67)
Owens et al., 2008	0.06 (-0.25, 0.39)	0.06 (-0.25, 0.39)	0.06 (-0.25, 0.39)	0.06 (-0.25, 0.39)	0.06 (-0.25, 0.39)	0.06 (-0.07, 0.2)
Gulsrud et al., 2007	0.33 (-0.99, 1.65)	0.33 (-0.99, 1.65)	0 (0, 0)	0 (0, 0)	0 (0, 0)	0.13 (-0.19, 0.46)
Kroeger et al., 2007	1.24 (1.03, 1.44)	1.24 (1.03, 1.44)	0 (0, 0)	0 (0, 0)	0 (0, 0)	0.49 (0.17, 0.81)
Golan & Baron-Cohen, 2006	0.31 (0.03, 0.58)	0 (0, 0)	0.31 (0.03, 0.58)	0 (0, 0)	0 (0, 0)	0.12 (0.03, 0.21)
Heimann et al., 2006	0.52 (-)	0.52 (-)	0 (-)	0 (-)	0 (-)	0.21 (-0.04, 0.46)
Sallows & Graupner, 2005	-0.03 (-0.56, 0.49)	-0.03 (-0.56, 0.49)	-0.03 (-0.56, 0.49)	0 (0, 0)	0 (0, 0)	-0.01 (-0.19, 0.15)
Solomon et al., 2004	0.62 (0.19, 1.04)	0.63 (0.22, 1.04)	0 (0,0)	0 (0, 0)	0 (0, 0)	0.25 (0.1, 0.39)
Silver & Oakes, 2001	0.74 (0.01, 1.47)	1.06 (0.95, 1.18)	0.39 (-0.37, 1.15)	0 (0, 0)	0 (0, 0)	0.44 (0.15, 0.72)
Howlin, 1981	0.22 (-0.15, 0.61)	0.22 (-0.15, 0.61)	0 (0, 0)	0 (0, 0)	0 (0, 0)	0.09 (-0.01, 0.2)

F: function, I: implements, E: ecology, L: level, D: durability, S: sum of all domains

In detail, regarding functional transfer, 16 studies demonstrated no transfer effect ^{20,24,28,40,45,53,63,74,79,82,86,87,90,93,94,96}, 7 studies exhibited small transfer effects ^{19,25,51,54,56,83,88}, 10 studies unveiled medium transfer effects^{21,22,37,43,44,49,71,81,89,92}, and 19 studies showcased large transfer effects^{23,31–33,35,38,39,42,46–48,50,52,57,70,73,80,91,95}. For implemental transfer, 16 studies demonstrated no transfer effect ^{20,28,40,42,43,45,53,63,74,79,82,87,90,93–95}, 6 studies exhibited small transfer effects ^{19,25,51,54,56,96}, 10 studies unveiled medium transfer effects^{21,22,24,37,44,49,71,81,89,92}, and 20 studies showcased large transfer effects^{23,31–33,35,38,39,46–48,50,52,57,70,73,80,83,86,88,91}.

Ecological transfer reveals that 31 studies presented no transfer effect^{20,23,24,28,33,35,37,39,40,45,47–50,52,53,56,63,70,71,73,74,79,82,87,90–95}, while 7 studies displayed minor transfer effects^{19,25,31,32,51,54,96}. Furthermore, 8 studies uncovered moderate transfer effects^{21,22,43,44,46,81,89}, and 6 studies illustrated substantial transfer effects^{38,42,57,80,83,86,88}. For the level domain, we observe that 38 studies failed to demonstrate any transfer effect^{20,22–25,28,35,37,40,42–50,52,53,56,57,63,70,74,79–82,87–95}, whereas 4 studies exhibited minor transfer effects ^19,51,54,96. In addition, 1 study revealed moderate transfer effects⁷¹, and 9 studies showcased significant transfer effects^{21,31–33,38,39,73,83,86}.

For the durability, it is evident that 37 studies did not show any transfer effects^{20,22–25,28,35,37,40,42–50,52,53,56,57,63,74,79–82,87–95}, while 4 studies displayed small transfer effects ^19,51,54,96. In addition, 2 studies revealed moderate transfer effects^21,71. Furthermore, 9 studies presented substantial transfer effects^{31–33,38,39,70,73,83,86}.

Table 7 provides a subgroup classification of transfer effects based on several potential factors, including participant age, intervention implements, intervention level, and intervention setting. For participant age, studies were categorized into four groups: early childhood, childhood, adolescents, and adults. Autism severity was considered in the grouping of transferability, categorized based on IQ levels (above or below 70). Although six studies did not report IQ, the severity of symptoms was assessed using rating scales for classification. Notably, a trend emerged where younger participants demonstrated higher transfer effects in total, 0.45, 0.4, 0.14, -0.04 for early childhood, childhood, adolescents, and adults in order. This pattern generally holding true across all FIELD domains. With respect to autism severity, the transferability was higher in mild autism compared to moderate autism, 0.25 versus -0.17. Regarding intervention implements, face-to-face interventions exhibited greater transferability compared to digital intervention, 0.39 versus -0.13. In terms of level transfer, behavioral interventions displayed moderate transfer effects (0.4), while cognitive interventions revealed low transfer effect (0.17). The greater transfer effect for behavioral intervention was applicable for all transfer domains. Intervention settings exhibited a moderate transfer effect for clinical interventions (0.35), whereas minimal transfer effects were noted for other settings, lower than 0.2, and this pattern extended across all transfer domains. Lastly, higher dose of intervention leads to greater transfer in all domains. For the total transfer effect, it was strong (0.64), moderate (0.25) and null (-0.18) for above 51 h, between 16 and 50, and below 15 in order.

Table 7.

Subgroup classification of transfer effect size.

Potential Factors	Groups (n)	Cohen’s D (95% Confidence Interval)
		F	I	E	L	D	S
Age	<6 (12)	.75 (.48, 1.03)	.75 (.48, 1.03)	.09 (.02, .15)	.1 (.03, .18)	.56 (.28, .84)	.45 (.35, .55)
	6-12 (25)	.55 (.00, 1.1)	.55 (-.02, 1.12)	.16 (-.08, .39)	.25 (-.18, .68)	.5 (-.02, 1.03)	.4 (.19, .61)
	13-18 (8)	.17 (.03, .31)	.23 (.08, .38)	.2 (.06, .35)	.03 (-.00, .07)	.08 (-.02, .19)	.14 (.09, .2)
	>18 (5)	.12 (-.08, .32)	-.07 (-.28, .13)	-.11 (-.27, .04)	.04 (-.01, .1)	-.19 (-.35, -.04)	-.04 (-.12, .03)
Autism Severity	Mild (19)	.42 (.04, .81)	.32 (-.1, .76)	.18 (.08, .29)	.17 (-.26, .6)	.17 (-.26, .6)	.25 (.08, .42)
	Moderate (3)	-.29 (-2.7, 2.11)	-.29 (-2.7, 2.11)	-.29 (-2.7, 2.11)	0 (0, 0)	0 (0, 0)	-.17 (-.99, .63)
Intervention Implements	Digital (12)	.01 (-.33, .36)	-.28 (-.71, .14)	.07 (-.14, .28)	-.1 (-.29, .08)	-.35 (-.77, .07)	-.13 (-.28, .02)
	Face-to-face (38)	.56 (.26, .87)	.63 (.32, .93)	.56 (-.08, .19)	.19 (-.05, .43)	.53 (.26, .81)	.39 (.28, .51)
Intervention Level	Cognitive (26)	.3 (-.06, .67)	.21 (-.18, .61)	.05 (-.18, .28)	.15 (-.19, .49)	.15 (-.19, .49)	.17 (.02, .32)
	Behavioral (25)	.53 (.2, .85)	.55 (.23, .88)	.06 (-.01, .14)	.42 (.1, .74)	.45 (.13, .77)	.4 (.27, .53)
Intervention Setting	Clinic (29)	.63 (.21, 1.05)	.55 (.11, .99)	-.03 (-.22, .15)	.15 (-.17, .48)	.45 (.04, .86)	.35 (.18, .51)
	Home (3)	.16 (-.06, .39)	.14 (-.08, .37)	.09 (.01, .18)	.05 (-.06, .24)	.52 (-.12, .22)	.11 (.02, .19)
	School (7)	.12 (-.05, .29)	.21 (.02, .41)	.17 (-.02, .37)	0 (0, 0)	.12 (.01, .22)	.12 (.05, .19)
	Home & Clinic (7)	.06 (-.13, .26)	.06 (-.13, .26)	.06 (-.13, .26)	.01 (-.01, .04)	.01 (-.09, .12)	.04 (-.02, .11)
	All three (4)	.05 (-.18, .28)	.2 (-.04, .45)	.23 (-.01, .47)	.02 (-.02, .07)	.17 (-.00, .34)	.13 (.04, .22)
Intervention Dose (Hour)	15> (21)	.41 (-.02, .84)	.25 (-.2, .72)	-.07 (-.33, .19)	.11 (-.28, .51)	.2 (-.2, .6)	-.18 (0, .36)
	16-50 (23)	.34 (-.02, .72)	.38 (.01, .76)	.13 (.03, .23)	.29 (-.07, .67)	.1 (.02, .18)	.25 (.12, .38)
	>51 (5)	.86 (.31, 1.42)	.82 (.26, 1.37)	.05 (-.08, 1.18)	.73 (.18, 1.27)	.73 (.18, 1.27)	.64 (.41, .86)

F: function, I: implements, E: ecology, L: level, D: durability, S: sum of all domains, N: number of studies, 0 indicates no transfer, an effect size below 0.2 is considered a small transfer effect, an effect size between 0.21 and 0.51 is classified as a medium transfer effect, and an effect size exceeding 0.51 is labeled as a large transfer effect.

Although we classify the studies based on potential factors, these factors can co-occur. Table 8 shows the transfer effect size in a two-factorial classification. In this table, we consider the total transfer effect and reclassify the factors based on their significance in the initial classification to simplify interpretation. The results revealed a higher transfer effect for adolescents and adults compared to children for digital intervention, while face-to-face intervention showed a higher transfer effect in children compared to adolescents and adults (0.86 versus 0.04). Cognitive intervention exhibited a weak transfer effect in both age groups (0.18 and 0.14), whereas behavioral intervention was more transferable for children compared to adults and adolescents (0.69 versus 0.06). Furthermore, the results showed a higher transfer effect for adolescents and adults compared to children in a non-clinical setting (0.33 compared to 0.11), while clinical intervention had a higher transfer effect in children compared to older age groups (0.49 versus 0.03). Additionally, higher doses led to more significant transfer effects in children (0.27, 0.68, 0.7 for low, medium, and high doses, respectively), while the reverse order was found in adolescents and adults (0.28, 0.06, -0.32 for low, medium, and high doses, respectively). Regarding the implementation, similar transfer effects were found for face-to-face and digital interventions at each behavioral level (0.38 and 0.42, respectively) and cognitive level (-0.17 and 0.05, respectively). In terms of setting, face-to-face intervention was more effective in clinical settings compared to digital intervention (0.51 versus -0.27), whereas digital intervention was more transferable in clinical settings compared to non-clinical settings (0.28 versus -0.27). Concerning the dose, the transfer effect of face-to-face intervention increased with dose (0.43, 0.33, 0.56 for low, medium, and high doses, respectively), while medium doses had a higher effect in digital intervention (-0.46, 0.26, 0.07 for low, medium, and high doses, respectively). Regarding the level of intervention, medium and weak transfer effect were found for clinical and non-clinical behavioral interventions (0.42 versus 0.11), while there was a low and negative transfer effect for cognitive interventions (0.19 versus 0.06). Regarding the dose, behavioral intervention showed similar transfer effects for low and medium doses (0.33), which were lower than high-dose interventions (0.63). Meanwhile, at the cognitive level, the medium dose transfer was greater than low and high doses (0, 0.23, 0.21 for low, medium, and high doses in order). Finally, concerning clinical settings, higher doses revealed a greater transfer effect (0.11, 0.31, 0.51 for low, medium, and high doses, respectively), whereas in non-clinical settings, the medium dose had a greater transfer effect (0.19) compared to low (-0.38) and high (0) doses, Fig. 3.

Table 8.

Two-factorial classification of potential factors and the total transfer effect size.

Potential Factors		n, Cohen’s D (95% Confidence Interval)
		Age		Implement		Level		Setting
		≤12	>12	Digital	Face-to-face	Cognitive	Behavioral	Clinical	Nonclinical
Implement	Digital	8, -.01, (-.08, .05)	3, .41, (.28, .55)	-	-	-	-	-	-
	Face-to-face	18, .86, (.59, 1.13)	9, .04, (-.01, .09)	-	-	-	-	-	-
Level	Cognitive	22, .18, (.00, .36)	4, .13, (.07, .18)	11, -.17, (-.34, .01)	1, .05, (-.01, .11)	-	-	-	-
	Behavioral	16, .69, (.46, .91)	9, .06, (0.00, .12)	13, .41, (.16, .66)	26, .38, (.25, .50)	-	-	-	-
Setting	Clinical	30, .49, (.30, .67)	9, .03, (-.01, .08)	8, -.27, (-.46, -.07)	27, .51, (.35, .66)	23, .11, (-.06, .28)	18, .42, (.26, .59)	-	-
	Non-clinical	5, .11, (.06, .16)	4, .33, (.21-.49)	4, .28, (.20, .36)	4, .04, (-.04, .13)	3, -.06, (-.18, .06)	7, .19, (.14, .25)	-	-
Dose	15>	17, .27, (.03, .52)	3, .28, (.17, .39)	7, -.46, (-.73, -.19)	13, .43, (.18, .67)	16, .00, (-.26, .26)	5, .33, (.25, .41)	19, .11, (-.09, .32)	2, -.08, (-.24, .09)
	16-50	13, .68, (.4, .95)	9, .06, (.01, .12)	4, .26, (.18, .34)	17, .33, (.16, .5)	5, .23, (.17, .28)	5, .33, (.25, .41)	15, .33, (.14, .51)	6, .19, (.13, .25)
	>51	4, .7, (.46, .94)	1, -.32, (-.16, .09)	1, .07, (-.27, .36)	4, .56, (.34, .77)	2, .21, (.07, .34)	3, .63, (.36, .9)	5, .51, (.32, .71)	0, 0, (0,0)

Fig. 3.

Impact of potential factors on the total transfer effect size.

Discussion

In this study, our goal was to perform a thorough review and analysis of the transferability of social cognition training in individuals with ASD, with the FIELD model serving as the framework for assessment. The findings indicate that age, intervention level, settings, and implements have a significant impact on the outcomes of transferability.

The relevance of age

The age is an influential factor in the transferability of interventions. The analysis demonstrates a more prominent transfer effect in younger participants when compared to their older counterparts. In summary, the transfer effect was moderate during early childhood and childhood but declined to a lower level during adolescence and adulthood. Prior research has consistently emphasized the adaptability and plasticity of the developing brain ^97,98, a factor that likely contributes to the observed findings. Earlier studies in children with autism revealed younger child’s age at start of intervention predicted greater cognitive gains with intervention ^99,100. Similarly, a meta-analysis of early interventions in children with ASD revealed that participants of a younger age exhibited a greater treatment effect size on social communication outcomes ¹⁰¹. Moreover, from a broader perspective, social competency requires the utilization of resources both externally (from the environment and all individuals within it) and internally (from individuals’ existing skills, capabilities, and potential for growth) ¹⁰². Indeed, various environmental factors influence adaptation in these age groups. For example, adolescence entails exposure to a variety of new social interaction contexts and roles ¹⁰², while social support tends to decrease during middle adolescence ¹⁰³. Therefore, the lower transferability observed in adolescents and adults could be interpreted as a result of the heightened environmental and social demands faced by these age groups. The interaction of potential factors reveals that digital interventions seem to yield a higher transfer effect for adolescents and adults compared to children, while face-to-face interventions demonstrate greater efficacy in children than in older age groups. Lower digital literacy among youth and reduced social interaction tendencies in adults due to increased demands of social interactions could be defined as two influential factors contributing to this pattern. An earlier meta-analysis of digital health interventions for weight management supports this notion, indicating that the intervention is more suitable for adults compared to children and adolescents ¹⁰⁴. Aligned with this assertion, the findings demonstrated a greater transfer effect for adolescents and adults in non-clinical settings compared to children, whereas clinical interventions exhibited a higher transfer effect in children compared to older age groups. Moreover, higher doses were associated with more pronounced transfer effects in children, contrasting with the inverse trend observed in adolescents and adults.

The relevance of autism severity

The transferability was higher in mild autism compared to moderate autism. A previous study discovered an inverse relationship between the gains achieved by four distinct behavioral interventions for children with ASD and the severity of their autism, while also noting a direct correlation with the amount of time input ¹⁰⁵. The initial capability and baseline performance, whether for a targeted function or other general cognitive functions, should be considered in assessing the propensity for transferability. A cognitive training study revealed that baseline working memory performance is a crucial factor for the transferability of working memory training ¹⁰⁶. An earlier study in autistic children with higher initial cognitive levels and children with better early social interaction deficits showed better acquisition of skills after intensive behavioral intervention ¹⁰⁷.

The relevance of intervention implements

The results indicate that face-to-face interventions demonstrate greater transferability in comparison to digital ones. This disparity may be attributed to the reduced therapist-patient interaction in computerized interventions, suggesting that paper-based interventions possess an additional social component that enhances the effectiveness and transferability of the training effect. Furthermore, the educational materials employed in paper-based interventions often incorporate social scenarios, facilitating ecological transfer. A prior systematic review of computer-based interventions aimed at improving social and emotional skills in individuals with ASD yielded unsatisfactory outcomes following the intervention¹⁰⁸. It is worth noting that this study primarily assessed the efficacy of training, whereas our study primarily evaluated transferability as a secondary outcome. Another systematic review assessed the effects of cognitive training programs on executive function in children and adolescents with ASD and found both computerized and non-computerized executive function training to be effective, particularly computerized programs. However, this study reported limited evidence regarding the generalization to untrained skills, such as social abilities, as well as the long-term effects. It is essential to recognize that while our study indicated lower transferability at the level of function and duration, executive functions can be trained effectively through computer-based methods, unlike social cognition, which is fundamentally communication-based¹⁰⁹. Another meta-analysis revealed a medium effect size for digital interventions in autistic children but identified a negative correlation between the duration of the interventions and the effect sizes reported in the studies¹¹⁰, indicating a low durability transfer in digital interventions. However, a meta-analysis found comparable effects of digital and face-to-face interventions in social skills training for children with ASD¹¹¹. With respect to the interaction of implement with other potential factors, when considering settings, face-to-face interventions demonstrated greater effectiveness in clinical settings compared to digital interventions, whereas digital interventions exhibited higher transferability in clinical settings compared to non-clinical settings. Regarding dosage, the transfer effect of face-to-face interventions increased with higher doses, whereas medium doses showed a stronger effect for digital interventions. In line with this study, a previous systematic review and meta-analysis focusing on face-to-face communication found that face-to-face settings showed greater effectiveness, particularly in clinical contexts, compared to digital interventions¹¹².

The relevance of intervention setting

The clinical intervention exhibited a moderate transfer effect, while minimal transfer effects were observed in other settings, and this pattern was consistent across all transfer domains. An earlier study compared the outcomes of individualized home-based programs and small group center-based programs with parent training and support¹¹³. The study found that the center-based program yielded better outcomes for some children and families. However, it also noted that not all children and families were suitable for this type of intervention. Parents who receive adequate educational, financial, and social support have reported feeling empowered when home educating their autistic child¹¹⁴. However, a study revealed that combining a home-based program with a center-based program resulted in greater cognitive development and behavioral improvement in children, particularly those from more stressed families¹¹⁵. A review of 23 studies examining social communication outcomes in children with ASD revealed that studies employing context-bound measures, where settings, materials, communication partners, or interaction styles similar to the treatment context, had an 82% likelihood of producing a positive treatment effect. In contrast, studies using generalized measures had a 33% probability of positive outcomes¹¹⁶. A systematic review and meta-analysis of spoken language interventions in children with ASD found that early intervention had positive effects on spoken language. The largest effects were observed for interventions delivered collaboratively by both parents and clinicians (g = 0.42), followed by parents only (g = 0.11) and clinicians only (g = 0.08). It is important to note that this review focused solely on spoken language outcomes and did not include assessments of social communication outcomes¹¹⁷. Moreover, the clinical interventions outlined in this review entail structured programs with active parental involvement, distinguishing them from interventions administered solely by parents. Additionally, it is crucial to note that in the current study, the concept of “setting” extends beyond the agent of intervention; rather, it encompasses a broader context that includes elements such as location, time, physical features, and social participants. In terms of the level of intervention and setting, both interventions exhibit lower transferability in non-clinical settings. This indicates that regardless of the type of intervention or setting, transferability tends to be lower in non-clinical environments.

The relevance of intervention level

The result of the transfer analysis revealed greater transfer effect in all FIELD’s domains in behavioral intervention compared to cognitive interventions. In summary, the transfer effect was moderate for behavioral interventions and mild for cognitive interventions. While cognitive interventions aim at enhancing specific cognitive functions, behavioral interventions exhibit a tendency for greater transferability. Several factors should be considered in this context. Most cognitive interventions focus on a narrow function, like facial emotion recognition, making them incomparable to behavioral training that encompasses multiple cognitive domains. Additionally, in the studies included, the duration of behavioral interventions was significantly longer, 18 times more than cognitive interventions (362.42 vs. 20.11), potentially influencing transferability. An earlier RCT compared Eclectic-Developmental (ED) and ABA in very young children with ASD, matched for severity, cognitive abilities, and socio-economic background. The results indicated that the ABA group exhibited significantly greater improvements than the ED group at the post-intervention time¹¹⁸.

The relevance of dose

The results of the transfer analysis indicate that long interventions, exceeding 50 h, exhibit a high transfer effect, whereas medium-duration interventions (16–49 h) show a moderate transfer effect, and small-duration interventions (below 15 h) demonstrate a low transfer effect. this pattern was applicable for all transfer domains except functional transfer. It can be concluded that transfer at the functional level occurs more rapidly than in other domains. In a meta-analysis examining outcomes of interventions employing ABA, children who underwent longer-term interventions (lasting at least 45 weeks with 10 h of therapy per week) exhibited more favorable results ¹¹⁹. Regarding the interaction of dose with other potential factors, the high dose of behavioral intervention and medium dose of cognitive intervention exhibit optimal transfer effect. This could be attributed to the more specific cognitive interventions compared to behavioral interventions. Similarly, the same dose pattern was observed for settings: higher doses revealed a greater transfer effect in clinical settings, whereas in non-clinical settings, the medium dose had a greater transfer effect. This may be due to the more personalized intervention offered in clinical settings.

Rating of studies

While transfer is a crucial aspect of effectiveness, it is important to note that not all included studies were specifically designed to investigate transfer effects. To address this limitation, we have introduced an index to assess the study’s power in uncovering transferability, Table 9.

Table 9.

The study power for discover the transfer effect. F: Function, I: Investigation tools, E: Environment, L: Level, D: Durability, S: Sum of transferability index, color coding; for F, I, E, L, and D: green (1): strong, yellow (0.33 or 0.67): moderate, red (0): weak power of transfer; for S; green (3.34–5): strong, yellow (1.66–3.33): moderate, red (0–1.65): weak power of transfer.

According to this index, approximately 34.61% of the studies demonstrated strong power, 63.46% exhibited moderate power, and the remaining 1.92% showed weak power in evaluating transferability. In our results table, we distinguish the assessment of transferability by considering the availability of assessment tests. Specifically, we employ a scoring system using 0 or 1 to signify the presence or absence of a transfer based on test results when assessment tools are accessible for both near and far transfers. However, in instances where these tests are unavailable, a missing value is used to signify a lack of data concerning transferability. This approach enables us to provide a precise representation of transferability in each study while also recognizing the constraints imposed by the availability of assessment tools.

Conclusion, limitation and future directions

This study identified age, autism severity, intervention implements, intervention level, intervention setting, and intervention dose as influential factors in the transferability of social skill training in individuals with ASD. The transfer of skills was notably more pronounced among younger individuals, particularly in face-to-face interventions as opposed to digital alternatives. Moreover, behavioral interventions outperform cognitive interventions in terms of transferability. Additionally, cognitive interventions exhibited greater transfer compared to behavioral interventions, especially when coupled with a higher dose of intervention. The effective transference of skills is a pivotal element in the success of interventions for individuals with autism. Earlier studies have revealed that individuals with ASD face challenges in transferring communicative behaviors to untrained contexts^120,121. This difficulty can be attributed to theories such as weak central coherence theory, social cognition theory, or executive functioning theory. According to weak central coherence theory, the generalization of intervention outcomes necessitates global information processing across domains. In line with social impairment theory, generalization involves transferring skills from one’s own world to others, and limited social capabilities can hinder this ability. From the perspective of executive function theory, generalization is defined as an abstraction that requires disengagement from current, local pieces of information and engagement with new, global information. Incorporating a transferability framework within the context of the neurodiversity lens underscores the importance of recognizing and respecting the unique traits and processing styles of individuals with autism. By acknowledging the diverse ways in which individuals with autism experience the world, interventions can be designed to complement rather than suppress their inherent characteristics. This aligns with the neurodiversity perspective, which advocates for embracing the differences among individuals with autism and refraining from pathologizing their behaviors.

Future studies are proposed to incorporate a variety of assessment tools to explore the transferability of social training interventions. Additionally, factors such as the age and severity of children with ASD, as well as the type of intervention, level of intervention, setting, and dosage, should be considered for the development of effective interventions for social training in autism.

Certain limitations must be considered in this study. Firstly, the heterogeneity across the included studies, involving variations in intervention protocols, assessment tools, and participant characteristics, could introduce a level of bias. We utilize a conceptual meta-analysis approach to compute similar features of different measurements and develop a general conceptualization tailored to our objectives. However, it is important to recognize that the variability in measurement tools utilized across studies presents a notable limitation to our analysis. Secondly, in this study, we analyze our results by examining potential influential factors separately. However, it is important to recognize that these factors are intertwined, and caution should be taken into account when considering them individually. Regarding future avenues of research, there is a need for further exploration of the mechanisms that underlie the observed transfer patterns. This could potentially be achieved through methods such as neuroimaging or neurophysiological assessments. Additionally, future studies should account for the variety of assessments to account for the potential variability in the evaluation of transfer.

Appendix A. FIELD’s properties of assessments in the included studies

Measurement (abbreviation; developer)	Measure(s)	Properties
		Agent 1	Objectivity 2	Materail 3	Setting 4
Achieved Learning Questionnaire (ALQ)122	Social skill attainment	P	S	P	H
Assessment of Basic Language and Learning Skills (ABLLS)123	Life skills	C	S	O	P
Achievement of Joint Attention (AJA)47	Joint attention	C	O	R	C
Assessment of Language Level end Language Usage (ALLLU)68	Language	S	S	O	H
Assessment of Spontaneous Interaction in Movement (ASIM)91	Interactional movements	C	S	O	C
Autism Diagnostic Interview (ADI)20	Social functioning	C	S	P	C
Autism Diagnostic Observation Schedule (ADOS)124	Communication skills	C	S	O	C
Autism spectrum Quotient (AQ)125	Autistic sym	PS	S	P	H
Automatic Imitation (AI)126	Imitation	C	S	C	C
Bar-on Emotional Quotient Inventory (BEQI)127	Emotional intelligence	S	S	P	H
Bayley Scales of Infant Development (BSID)128	Social development	P	S	P	H
Beck Depression Inventory (BDI)129	Depression	S	S	P	H
Benton Face Recognition Scale (BFRS)[130]	Face recognition	S	O	C	C
Bruininks-Oseretsky Test of motor proficiency (BOT)131	Gross and fine motor	C	O	M	C
Cambridge Mindreading Face-Voice Battery (CMFVB)76	Mind reading	S	O	C	C
Child Adjustment and Parent Efficacy Scale (CAPES)132	Behavioral problems	P	S	P	H
Child and Adolescent Social Perception Scale (CASPS)133	Social emotion awareness	S	O	C	C
Child Behavior Checklist (CBCL)134	Behavioral problems	P	S	P	H
Children’s Depression Inventory (CDI)135	Depression	S	S	P	H
Children Facial Emotion Recognition Test (CFERT)136	Facial emotion recognition	S	O	C	C
Circumplex Scale of Interpersonal Efficacy (CSIE)137	Social interaction	S	S	P	H
Comprehensive Affective Testing System (CATS)138	Emotion perception	S	O	C	C
Confusion, Hubbub, and Order Scale (CHOS)139	Environmental confusion	P	S	P	H
Depression Self-rating Scale (DSS)140	Depression	S	S	P	H
Diagnostic Analysis of NonVerbal Accuracy (DANVA)141	Facial emotion recognition	S	O	C	C
Direct Assessment of Self-Expression (DASE)38	Emotional expressions	C	S	O	C
Dylan Is Being Teased (DIBT)142	Anxiety and anger manage	S	O	P	C
Early Social-Communicative Scale (ESCS)143	Nonverbal communication	C	S	O	C
Electronic Ekman 60 Faces Test (EEFT)138	Emotion regulation	S	O	C	C
Emotion Fluency (EF)38	Emotional vocabulary	S	O	P	C
Emotion Multimedia Battery Assessment for adults with Autism (EMBA)144	Emotional capabilities	S	O	C	C
Emotion Recognition Cartoons (ERC) 145	Emotion recognition	S	O	P	C
Emotion Regulation and Social Skills Questionnaire (ERSSQ)146	Social engagement	SPT	S	P	H
Emotion Storybook (ES)38	Emotion vocabulary	S	O	P	C
Empathizing/Systemizing Quotient (ESQ)147	Empathy	P	S	P	H
Empathy Quotient (EQ)148	Empathy	PS	S	P	H
Experience Sampling Method (ESM)149	Daily emotional state	S	S	P	H
Expressive Language Task (ELT)73	Expressive language	S	O	P	C
Eyberg Child Behavior Inventory (ECBI)150	Child’s behavior	P	S	P
Facial Emotion Recognition (FER)44	Facial emotion	C	S	O	C
Facial Expression Photographs (FEP)151	Facial recognition	S	O	P	C
Faux Pas Stories (FPS)152	Theory of mind	S	O	C	C
Frequency of Contact (FEC)44	Eye contact	C	S	O	C
Friendship Qualities Scale (FQS)153	Friendships perceptions	S	S	P	H
Gilliam Autism Rating Scale (GARS)154	Autistic behavior	P	S	P	H
Griffiths’ Mental Developmental Scales (GMDS)155	Language	S	O	P	C
Goal attainment scaling (GAS)156	Social skills	P	S	P	H
Incidental Face Memory (IFM)157	Face Memory	S	O	C	C
Interpersonal Reactivity Level (IRL)158	Empathic concern	S	S	P	H
Interpersonal Synchronization (IS)159	Interaction dynamic	S	O	C	C
James and the Maths Test (JMT)142	Anxiety and anger manage	S	O	P	C
Joint Attention Probe (JAP)49	Joint attention	C	O	O	C
Mindreading Battery (MB)160	Emotion recognition	S	O	C	C
Mother–Child Interaction (MCI)161	Joint attention	P	S	O	C
Mother–Child Relationship Questionnaire (MCRQ)162	Mother–child interaction	P	S	P	H
Mullen Scales of Early Learning (MSEL)163	Learning	C	S	O	C
Multifaceted Empathy Test (MET)164	Empathic feelings	S	O	C	C
NeuroPsychology emotion recognition (NEPSY)165	Emotion recognition	S	O	C	C
Nim Stim Facial Emotion Recognition Test (NSFERT)166	Facial emotion recognition	S	O	C	C
Observation of Imitation and Joint Attention (OIJA)50	Joint attention & imitation	C	S	O	C
Parenting Sense of Competence Scale (PSOC)167	Parenting self-efficiency	P	S	P	H
Pediatric Quality of Life Inventory (PQLI)168	Quality of life	S	S	P	H
Peer Interactive Paradigm (PIP)169	Cooperative play	C	S	O	C
Perth A-Loneliness Scale (PALS)170	Isolation and friendship	S	S	P	H
Piers-Harris Self-concept Scale (PHSS)171	Self-concept	S	S	P	H
Play Skill Scale (PSS)172	Play type	C	S	O	C
Psycho Educational Profile (PEP) 173	Learning problems	C	S	O	C
Quality of Socialization Questionnaire (QSQ)174	Socialization	PS	S	P	H
Reading the Mind in the Eyes Task (RMET) 43	Mind reading	S	O	C	C
Reading the Mind in the Voice Task (RMVT)76	Mind reading	S	O	C	C
Reynell Developmental Language Scale (RDLS)175	Language	C	S	O	C
Screening Autism Spectrum Questionnaire (SASQ)176	Autistic behavior	P	S	P	H
Social Anxiety Scale (SAS)177	Anxiety	PS	S	P	H
Social Communication Questionnaire (SCQ)178	Communication skills	P	S	P	H
Social Dissatisfaction Questionnaire (SDQ)179	Loneliness	S	S	P	H
Social Interaction Anxiety Scale (SIAS)180	Anxiety	S	S	P	H
Social Interaction Observation Code (SIOC)181	Social interaction	C	S	O	C
Social Language Development Test (SLDT)182	Language skills	S	O	P	H
Social Network Survey (SNS)71	Friendship	C	S	O	S
Social Responsiveness Scale (SRS)183	Communication skills	PT	S	P	H
Social Skills Rating System (SSRS)184	Social skills	PT	S	P	SH
Social Skills Questionnaire (SSQ)151	Social skills	PT	S	P	SH
Sociometric Nominations (SN)185	Friendship patterns	S	S	P	H
Spence Children’s Anxiety Scale (SCAS)186	Anxiety sym	P	S	P	H
State and Trait Anxiety Inventory for Children (STAIC)187	Anxiety	S	S	P	H
Strange Stories Test (SST)188	Theory of mind	S	O	P	C
Stress Index for Parents of Adolescents (SIPA)189	Stress	P	S	P	H
Stress Survey Schedule (SSS)190	Stress	S	S	P	H
Structured Observations of Empathic Responsiveness (SOER)92	Empathy	C	S	O	C
Structured Play Assessment (SPA)191	Play type	C	S	O	C
Teacher Perception of Social Skills (TPSS)192	Social skills	T	S	P	S
Test of Adolescent Social Skills Knowledge (TASSK)193	Social skills	S	S	P	H
Test of Problem Solving (TPS)194	Critical thinking	S	O	P	C
Test of Young Adult Social Skills Knowledge (TYASSK)78	Social skills	S	S	P	C
Theory of Mind Test (TMT)195	Theory of mind	S	O	C	C
Theory of Mind Inventory (TMI)196	Theory of mind	P	S	P	H
Training-specific Test of Imitation/Praxis (TTIP)56	Imitation	C	S	O	C
Triad Social Skills Assessment (TSSA)197	Social skills	P	S	P	H
Unstructured Free-Play with Examiner (UFPE)198	Social skills	S	S	O	C
Vineland Adaptive Behavior Scale (VABS)199	Communication skills	P	S	P	H

1. C: clinician, P: parent, S: self-administered; 2. O: objective, S: subjective; 3. C: computerized, M: motion, O: observational, P: paper and pencil, R: robot; 4. C: clinic, H: home, S: school

Appendix B. FIELD’s properties of interventions in the included studies

Intervention	Properties
	Agent 1	Objectivity 2	Materail 3	Setting 4
Animal-assisted Social Skills Training (ASST)62	C	O	M	C
Attention Remediation of Theory of Mind (ARTOM) 45	S	O	C	C
Coach Assistant MiX (CA-MiX)64	C	O	C	CS
CommU Robotic Intervention (CRI)47	R	O	CM	C
Computer-facilitated Emotion Recognition Training (CERT)65	S	O	C	C
CuDDler 66	R	O	C	C
Emotion Trainer (ET)46	S	O	C	C
Direct Teaching Group Format (DTGF)35	C	O	MMo	C
Facial Emotion Recognition (FER)67	PC	O	P	C
Facial Emotional Recognition and Eye Contact Training (FER-ECT)65	R	O	C	C
Home-based Language Training Program (HLTP) 68	P	O	P	H
Imitative Interaction Training (IIT)52	C	O	M	C
Intensive Behavioral Treatment (IBT)69	CP	O	P	CH
Joint Attention and Imitation Training (JAIT)50	C	O	P	C
Joint Attention Trining (JAT)70	CT	O	MoP	CS
KONTAKT® 72	C	O	P	C
Language Training Program (LTP)73	C	O	P	C
LEGO Therapy (LT)51	S	O	P	C
Let’s Face It (LFI)74	PC	O	C	C
LookWare (LW)75	S	O	C	CHS
Mind Reading (MR)76	S	O	C	H
Picture Exchange Communication Scale (PECS)77	C	O	P	C
Program for the Education and Enrichment of Relational Skills (PEERS)78	CP	O	P	CHS
Reciprocal Imitation Training (RIT)85	CP	O	P	CH
Rhythm Training (RhT)56	CP	O	MoP	CH
Robot Training (RT)56	PR	O	MoR	CH
Secret Agent Society (SAS)21	CTP	O	C	CSH
Sense Theatre (ST)24	S	O	M	C
Social Adjustment Enhancement Intervention (SAEI)23	S	O	M	C
Social Cognitive Intervention Program (SCIP)25	C	O	C	S
Social Skills Program (SSP)26	C	O	P	C
Social Skills Training (SST)87	CPT	O	M	CHS
Social Stories Intervention (SSI)36	C	O	P	S
Social Use of Language Program (SULP)30	C	S	M	C
Sociodramatic Affective Relational Intervention (SDARI)27	C	O	MMo	S
Symbolic Play (SP)34	C	O	P	C
Synchronize Dance/Movement Intervention (SDMI)55	C	O	Mo	C
Theory of Mind Training (ToMT)57	C	O	P	C

1. C: clinician, P: parent, R: robot, S: self-administered, T: teacher; 2. O: objective, S: subjective; 3. C: computerized, P: paper and pencil, M: mental, Mo: motion; 4. C: clinic, H: home, S: school

Author contributions

V.N.: Conceptualization, writing, methodology, and supervision. A.P., N.N., F.A.: Data gathering and data analysis.

Data availability

The datasets generated during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

V.N. is the corresponding author of one of the studies included in the review. The authors declare that there are no other financial or non-financial conflicts of interest.