Understanding Neurodiverse Social Play Between Autistic and Non-Autistic Children

Social play supports children to develop essential life skills and foster friendships. However, autistic and non-autistic children often do not have equal opportunities to engage in social play. Previous research to improve these opportunities tends to invoke social skill interventions solely for autistic children or is focused on designing for only one group, rather than considering the interactions or needs of all children in neurodiverse groups1. In order to understand the different experiences of children during social play, we conducted interviews with 6 professionals who support neurodiverse social play and undertook observation sessions of 36 autistic and non-autistic children during unstructured social play. Our findings move beyond the existing characterizations of autistic social play and build upon the double empathy problem to capture and consider the needs of all children in neurodiverse playgroups. We argue these findings could be used to inform future neurodiverse social play technology design in HCI.


INTRODUCTION
Social play represents a play activity driven by intrinsic motivation and voluntary participation, encompassing various degrees of social interaction [46].While it holds a pivotal role in children's growth, aiding them in acquiring essential social competencies crucial for success in the broader world [1,11], it is also an important and enjoyable experience for many children 2 2 .Unsuccessful or negative social interactions during play often lead to feelings of isolation and can have detrimental effects on mental well-being [32,51,57].Autistic children 3 equally benefit from engaging in social play with their peers and forming reciprocal friendships.However, they may encounter obstacles to social play due to certain differences from their non-autistic peers, such as different communication styles or play preferences, as well as other environmental, attitudinal, and institutional barriers [33].
Previous research has demonstrated that children who interact in neurodiverse groups consisting of both autistic and non-autistic individuals tend to derive greater benefits from these interactions, such as engaging in more complex coordinated play and experiencing higher levels of cohesion, compared to interactions within non-neurodiverse groups [5].However, these studies also highlight certain challenges that some children, particularly autistic children, may encounter, during social play [26].It is also important to note that social play has often been defined by non-autistic norms, with autistic play often being described as non-autistic play with certain deficits [29].Distinguished in those terms, we contend that autistic play tendencies have their own characteristics, just like non-autistic play, and should be treated as its own entity.Combining autistic and non-autistic play could present its own challenges, as often preferred play styles and needs during play may not align.
Current research concerning interactions in neurodiverse social play tends to attribute the barriers to successful social play to the characteristics of autism [22].These studies frequently adopt a medical model of disability, perceiving the differences associated with autism as deficiencies residing within the individuals that necessitate intervention to align with more typical development [2,18].
Many of these investigations place the responsibility for adaptation on the autistic child, often employing intervention approaches focused on teaching and adjusting social skills [6].There are comparatively fewer studies that emphasize the self-directed nature of social play or acknowledge the diversity within neurodiverse groups [21,53,55].Some researchers have made efforts to work with autistic children and create tools to facilitate social play by involving them in the design process through participatory design methods [21].However, these more recent approaches often exclusively include groups of autistic or neurodivergent children [21] (further discussed in 2.4).Furthermore, the findings of these studies seem to primarily revolve around designing for or addressing the challenges faced by autistic or neurodivergent children, without necessarily considering the needs and experiences of all children in a neurodiverse setting [21,53].
In this paper, we aim to understand and consider the different experiences of autistic and non-autistic children during neurodiverse social play.To this end, we present two studies: (1) An interview study with 6 professionals who support neurodiverse social play in educational and recreational settings to learn about their experiences of facilitating social play.(2) An observation study of unstructured social play involving a total of 36 children (9 autistic children) to gain insights into different experiences in neurodiverse social play.
By combining and comparing the findings from both studies: (1) We present and discuss characterizations of autistic and neurodiverse social play that exceed the existing definitions in the literature, such as solitary detours (2) We expand the notion of the double empathy problem to address and bridge the different experiences of autistic and non-autistic children in neurodiverse social play.
We believe our findings could serve as design insights for shaping the design of future neurodiverse social play technology and research in HCI.

BACKGROUND 2.1 Social Play
Social play is defined as a play activity in which two or more peers interact with one another [21].Social play can combine other play types, such as constructive, physical, or imaginative.Over time, researchers have developed various theories to categorize social play behaviors.Parten, for example, outlined distinct play categories, starting with unoccupied, solitary, and onlooker and progressing in social complexity to parallel play, associative play, and the most intricate form, cooperative play [46].Most of the above forms of social play entail a certain degree of communication among participants through play interactions or verbal and visual means.Previous research has shown that social play has numerous positive developmental effects that persist throughout a child's life [8].Successful social play experiences can foster friendships, reduce loneliness, and enhance self-confidence and well-being [8].

Autism and Play
The lived experience of autistic individuals is highly diverse and varies both between individuals and across a lifespan.It is important to recognize that autistic individuals all have unique experiences of sensory processing, interaction, and communication, which may differ from both their non-autistic and autistic peers.However, autistic play is often defined through deficit-focused descriptions based on a normalized understanding of play [29].Autistic play is often characterized by a lack of imaginative play, as well as increased patterns of repetitive play and a stronger preference for solitary play [28].These definitions of play align with the medical model of autism, as certain play behaviors are seen as needing to be corrected or changed to better align with more 'typical' play characteristics.Additionally, these definitions often do not consider the unique strengths or potential diversity of play experiences for autistic individuals.
Despite these deficit-focused characterizations of autistic play highlighting that autistic children may have a stronger preference for solitary play, research has shown that autistic children do possess a desire for socialization, sharing experiences, and forming friendships, particularly through play [45,60].

Neurodiverse Social Play
Previous research has demonstrated that engaging in neurodiverse social play, which is social play that includes both neurodivergent and neurotypical children, has a significant positive impact on the play patterns and social interactions of both groups.The benefits include increased responsiveness, enhanced social orientation and cohesion, and improved language skills compared to non-neurodiverse groups4 [5].
However, differences in play patterns and needs between autistic and non-autistic children often pose challenges for both groups when engaging in social play.For example, autistic children frequently exhibit differences in sensory processing, which may make certain social play environments that may be enjoyable for nonautistic children less enjoyable for autistic children.This may be due to overwhelming stimuli like loud sounds, particular textures, and strong smells present in the play environment.
In addition, differences in interaction and communication may cause more significant challenges when socializing or playing with non-autistic individuals [26].Variations in communication styles, including non-verbal cues, between autistic and non-autistic children often lead to misunderstandings during social play, further contributing to feelings of isolation [37].Heasman et al. report on the unique ways in which groups of solely autistic individuals establish mutual understanding with each other, which may not be compatible with non-autistic peers [25].
As a result, the difficulties mentioned above can reduce opportunities for positive social play experiences [31].Lack of successful social play experiences can lead to social withdrawal in autistic children, further reducing opportunities for the development of essential social skills and friendships within natural play environments [51,57].In addition, similarly to non-autistic children, social withdrawal can result in feelings of loneliness and isolation, which may have adverse effects on mental health and self-esteem [32].
To improve social play outcomes for autistic children, some institutions offer social skill support sessions that primarily aim to teach autistic children to better understand and mimic nonautistic social play behavior [9], and similarly, technology has been designed to achieve the same goals [7,13,38,48].However, this approach places the primary responsibility of deciphering nonautistic social situations on the autistic child rather than prioritizing the development of mutual understanding among children and professionals [9].
Instead, we argue that it is crucial to understand the diverse characteristics and needs of social play within neurodiverse groups to foster inclusive and beneficial social interactions for all children.In addition, understanding both autistic and non-autistic children's perceptions of and aspirations for friendship and supporting mutual understanding between peers is then vital in creating a mutually satisfying social play environment [12,15,16].

Neurodiversity
Within autism research, there has been a focus on moving away from deficit-focused language toward more neutral terminology that recognizes the diversity of cognitive styles within our society [40].The term 'neurodivergent,' which was coined by Kassiane Asasumasu [3], is used to describe cognitive styles that do not fall under the general societal standards of 'typical,' which includes those who may, for example, be autistic, dyslexic, dyspraxic, or have ADHD or learning differences [56].The term 'neurodiversity, ' which was coined by Judy Singer, can be used to represent the diversity in cognitive styles and includes both those who are neurotypical and those who are neurodivergent [52].In social play technology research, the term neurodiverse has been used both to describe solely neurodivergent populations [54] and populations consisting of both neurodivergent and neurotypical individuals [21], both of which are discussed further in section 3.1.We propose using the phrase 'neurodiverse social play' to represent play between individuals with different cognitive styles, both neurotypical and neurodivergent, in line with the definitions in [56].In this paper, although we focus on autistic and non-autistic children, we have chosen to use the term 'neurodiverse' to represent all cognitive styles present in our participant list, which is discussed in section 6.2.

RELATED WORK 3.1 Social Play Technologies
Using technologies to support social play has been shown to improve social play outcomes for both autistic and neurotypical children [42,43,47,50,61,62].For example, Topobo, a digital construction system similar to Lego, has shown promise in increasing social play, particularly collaborative play among groups of autistic children [18,20].Similar results were reported for an augmented knights castle toy and a tabletop puzzle game [4,17].However, these technologies were not tested or developed explicitly for neurodiverse groups 5 , who may have a more diverse and different range of needs during social play.
Additionally, much research on social play technology for autistic children often adopts an interventionist approach, aimed at encouraging "pro-social" behaviors and developing social skills [2,7,10,13,48,63].Often, these desired behaviors and skills are modeled on non-autistic norms and these technologies may not recognize the importance of supporting the potential different interaction and communication styles of autistic children.One such technology designed to deliver social skill interventions is Auti, which was created to promote social play interactions among autistic children by using positive reinforcement to encourage 'positive' social play behaviors [2].However, this technology was developed based on a controversial therapy designed to 'correct undesired behaviors in autistic children' [49].Additionally, it requires adult support, deviating from natural play driven solely by the children's choices.
3.1.1Social Play Technologies for Neurodiverse Groups.Some social play technologies designed explicitly for neurodiverse groups have also been shown to improve social play outcomes [38,54].However, they often focus on structured or semi-structured play activities rather than unstructured play scenarios.One such example is Incloodle, a tablet application designed to support collaboration between neurodiverse pairs of both neurotypical and neurodivergent children through structured picture-taking prompts to explore facial expressions [54].Sobel et al. have identified a gap in social play technologies that are unstructured and designed for autistic and non-autistic children together [53].
The work of Frauenberger et al. has strived to contribute to this need by employing participatory design (PD) to create unstructured social play technologies with and for neurodiverse groups [21].Several design insights from their study can be applied to designing with and for neurodiverse groups of children, e.g., balancing openness and structure in PD sessions, providing space for solitary time to regulate emotionally, and ensuring the children's control over and complexity of the technology is appropriate to prevent overwhelm.However, the groups in their study were primarily neurodivergent or composed of children attending special educational provisions due to differences in social and cognitive styles [21].We wish to expand upon this research further and focus on neurodiverse groups of children, with an emphasis on both autistic and non-autistic children, in mainstream educational settings.

Considering the Double Empathy Problem in Designing for Neurodiverse Social Play
A decade ago, Milton introduced the 'double empathy problem' to recognize and investigate the interpersonal challenges between autistic and non-autistic people [36].Essentially, this notion highlights that barriers that can arise in social interactions between different groups of individuals, which are not inherent to any one group but are rooted in mutual interpersonal issues [35].For instance, when an autistic child uses body language instead of verbal communication, the barrier to understanding does not solely stem from the child's mode of communication but also from the nonautistic peer's lack of comprehension or different reliance on verbal means of communication.A recent panel discussion involving fifteen researchers emphasized that the double empathy problem could serve as a valuable framework to bridge the gap between autistic and non-autistic peers and should be integrated into future autism research [35].In this context, we propose using the double empathy problem as a conceptual tool to inform our research approach and methods and to enhance the comprehension of mutual interpersonal challenges and interactions within neurodiverse social play.It is worth noting that the double empathy problem does not currently offer specific guidelines or parameters for applying to research findings.In employing this notion, we thus aim to contribute to enhancing its potential as a guiding framework for contextual understanding and design for interaction between neurodiverse groups in HCI.

SCOPE
We build on literature to address a gap in our understanding of how to design technologies that foster inclusive interaction between autistic and non-autistic children during unstructured social play.As a starting point, the work presented in this paper is driven by two broad research questions: RQ1) What are the challenges and barriers to facilitating neurodiverse social play that includes both autistic and non-autistic children?RQ2) How should we understand and characterize the different experiences of autistic and non-autistic children when they engage in neurodiverse social play?

STUDY 1: INTERVIEWS 5.1 Methods
To address RQ1, we conducted semi-structured interviews with professionals working with both autistic and non-autistic children across various settings, including mainstream primary schools and city councils.The interviews were conducted remotely via Microsoft Teams, with each session lasting approximately 45 minutes.All interviews were audio-recorded and subsequently transcribed.The questions posed to the participants covered topics such as their experiences working with neurodiverse groups, the various forms and dynamics of inclusive social play, the challenges encountered by both autistic and non-autistic children during social play, and the strategies employed to address these challenges.

Participants
We recruited 6 participants, initially through the Bristol City Council mailing list and registered primary schools and thereafter by snowball sampling technique to expand our participant pool.These professionals possess a substantial range of experience in assisting neurodiverse groups of children (M=19.33,STD=9.95, in years).The institutions they work in cater to varying numbers of children (M=528.60,STD=239.90)within the age range of 4 to 12.Among these children, there are typically 10 to 24 (M=17.67,STD=6.44)who either have received an autism diagnosis or are in the process of being diagnosed.The interview study was reviewed and approved by the authors' institutional ethics board.

Analysis
Our analysis of the interview transcripts employed thematic analysis, following the method outlined by [14], with the initial coding process involving separate inductive coding by the first two authors.Subsequently, all authors engaged in discussions to establish a consensus on the resulting codes.Thereafter, the first two authors collaboratively coded all transcripts, resulting in the identification of initial themes.These themes were further refined and solidified through comprehensive discussions involving all authors, aligning with a standard peer validation process [23].

Findings
It's essential to acknowledge that the interview findings presented in this section are based on the interpretations of professionals' experiences and do not directly reflect the perspectives of the children themselves.While the questions posed to participants were openended and intended to encompass barriers for both autistic and non-autistic children, we found that participants predominantly focused on the challenges they believed autistic children encountered in neurodiverse social play.Given this emphasis, we have made efforts to explore additional barriers to neurodiverse social play that extend beyond those specifically identified for autistic children.
In doing so, we have employed the 'double empathy problem' as a framework to interpretively articulate aspects and challenges of interactions for all participants within neurodiverse social play activities.
In the following, we present four themes (T) we generated for understanding neurodiverse social play, which includes perceived barriers faced by autistic children.We also produced four additional themes (DE) based on our own interpretation of how such barriers could be articulated from the perspective of non-autistic children.The double empathy themes were achieved by employing a further deductive thematic analysis on the four themes, which focused on looking for the notions from the double empathy problem, such as the mutual interpersonal issues between the actors, and incorporating these in the analysis (Figure 1).

T1: Differences in Communication Styles.
Participants discussed the various methods through which autistic children convey their thoughts, noting that due to the diversity of these communication approaches, it can be challenging for autistic children to express their needs effectively.They observed that autistic children's communication might manifest through behaviors, such as hitting or poking, and that this mode of expression is unique to each child.Participants suggested that these communication disparities could hinder autistic children's efforts to initiate play.
To address these communication differences, some participants suggested finding a common ground for communication, often utilizing signs or visual aids.P6 noted that many children at their school demonstrated an understanding of these differences and were aware that some children might communicate differently.However, two participants expressed concerns that these communication differences could pose difficulties for autistic children when trying to initiate play.Participant P4, for instance, observed that one child's behavior of prodding other children seemed to be a way of seeking attention and expressing a desire to play together, but this was not always comprehended by other children: "For some [autistic] children, it can be something like they hit but they don't hit because they want to be violent, they're hitting because that is their way of communicating." -P2.These past experiences may be rooted in interactions primarily with non-autistic individuals and may not encompass those who employ different methods or modes of communication.During interviews, participants also noted instances where they observed how non-autistic children might possess preconceived notions about the meanings of certain behaviors, such as viewing hitting as negative, consistent with their prior interactions with non-autistic individuals.This, in turn, can lead to misunderstandings when neurodiverse groups interact.Imposing these pre-established, normalized communication approaches onto interactions with autistic individuals could result in a mismatch in communication: "I think from a non-disabled child's point of view, it can get quite frustrating when they're speaking to nonverbal Jimmy and they don't understand why Jimmy won't speak back to them." -P2 5.4.3T2: Strong Desire to Play on Own Terms.Participants noted that they observed autistic children may have specific preferences for how they like to do things, which may differ from the preferences of others.This divergence in preferences sometimes led to challenges in neurodiverse social play: "[autistic children] don't want to sit and wait their turn, they just wanna go off and do what they wanna do." -P3.P1 noted that in neurodiverse play, the choice of topic and setting was often influenced by the autistic child, as they believed "[autistic children] find it hard to understand what others may need from a situation".Participants mentioned that conflicts could arise when a situation didn't align exactly with the autistic child's preferences or if they felt their opinions were not being considered.According to P6, finding compromises in such situations can be particularly challenging.Furthermore, participants observed that some autistic children may be hesitant to play with their peers, with P2 describing instances where certain children become distressed when they are in close proximity to others.To encourage interaction, participants often act as role models by engaging in activities that the child enjoys, thus maintaining their interest, and gradually inviting other children into the play space.This approach ensures that the child feels comfortable with their peers in the environment.
5.4.4DE2: Expectation to Act as a Cohesive Group.Similarly, non-autistic children may have limited exposure to autistic individuals and may tend to apply their understanding of non-autistic interactions to situations involving neurodiverse groups.This tendency can lead to misunderstandings regarding the needs and preferences of autistic individuals.Non-autistic children may have a specific way they prefer to engage in social play, which they assume is universally shared because other non-autistic peers often manifest similar preferences.However, within the group, there may be autistic individuals with differing notions of what constitutes social play and friendship.These divergent perspectives can give rise to misunderstandings and communication breakdowns.One example P1 discussed is how often non-autistic children in mixed group scenarios found it difficult to understand why autistic children may not follow the shared rules of a game.5.4.5 T3: Understanding Ambiguous Rules or Norms.Several participants indicated that play situations with unclear or undefined rules, such as the social norms governing shared games, could lead to misunderstandings between autistic and non-autistic children.Most participants noted that autistic children often excelled in well-defined settings but encountered difficulties when attempting to apply social play skills to less structured, spontaneous play scenarios: "Quite a lot of [autistic children] will want to fit in.So they'll learn the [social] rules ... but the rules only apply to a certain situation, don't they?" -P1.One participant pointed out that autistic children tended to enforce the rules they believed to be correct rather than adhering to the actual rules.To address these differences in interpretation, one suggested method was to collaboratively establish the play rules with all the children before initiating the play, ensuring that each child's perspective receives equal consideration.5.4.6 DE3: Diverging from 'Correct' Rules.Non-autistic children might face challenges in comprehending that not everyone adheres to or shares the same concept of social norms and friendship, especially if their exposure to social interactions predominantly involves non-autistic individuals.Consequently, they may assume that their grasp of social rules and norms is the definitive one, given that the majority of people in their social circles behave in a similar manner.This can lead to misunderstandings, as autistic children may not conform to the behavioral expectations of a non-autistic individual, and vice versa.
5.4.7 T4: Emotional and Sensory Regulation.All participants emphasized the impact of sensory processing differences on autistic children's ability to adapt and self-regulate during social play, where numerous external stimuli originate from the play environment (e.g., hot weather) or other playmates (e.g., noise).For instance, P4 noted instances where autistic children became overwhelmed by sensory input when they struggled to regulate themselves effectively: "Their [autistic children's] ability to regulate those emotions and to recognize those signals or to know when they're having difficulties ... they need to be able to learn regulation skills." -P4.Most participants mentioned that they closely collaborate with the children, constantly monitoring their emotional well-being and providing support, even if it interrupts social play.To address these sensory challenges, participants employ what they refer to as a 'sensory diet', which involves regulating the sensory stimuli that autistic children encounter.Nevertheless, the majority of participants emphasized that these strategies should be tailored to align with the individual interests and requirements of the children, which are highly nuanced and unique to each child.

DE4: Empathizing with Sensory Perception Differences.
Non-autistic children may find it challenging to grasp the notion of hypo-or hypersensitivity to sensory stimuli, as they might not have encountered it themselves.Consequently, they may struggle to comprehend how stimuli they consider neutral or non-disruptive can be uncomfortable or distressing for others.Moreover, the actions of nonautistic children, often unintentional, can inadvertently contribute to the presence of stimuli that become uncomfortable for individuals with sensory sensitivities.This lack of awareness can lead to issues during social play, highlighting the challenge of accommodating and regulating sensory needs in a neurodiverse social setting.

STUDY 2: SOCIAL PLAY OBSERVATIONS 6.1 Method
To address RQ2 -how to understand and characterize the different experiences of autistic and non-autistic children when they engage in neurodiverse social play -we conducted 3 observation sessions, 1 each for 3 different year groups (Year 2 (age 6-7); Year 3 (age 7-8); and Year 4 (age 8-9) in a mainstream primary school where one of the professional interviewees worked.For these observations, we video-recorded unstructured play sessions on the open playground, in which children were given materials to play with that they would have access to during typical playtimes, such as crates, wheelie bins, small mats, and large building blocks (Figure 2).Each session was approximately 45 minutes in length.

Participants
A total of 36 children participated in these observation sessions.Nine of these children either had formal diagnoses of autism, were in the process of diagnosis, or showed autistic traits according to the professional but did not have an official diagnosis.Each session involved 3 autistic children (represented by the prefix A in our data) and between 7 to 9 non-autistic and neurotypical children (represented by the prefix N in our data.Some groups included neurodivergent children who were non-autistic.The Year 3 group included a child with unspecified additional needs, and the Year 4 group included one child who displayed traits of ADHD and another with unspecified additional needs, all of which are included in our 'non-autistic' category.These children have been represented in our data using the prefix D. Autistic participants are described in more detail in Table 1).We refrained from adding non-autistic participants 6 to the table because we do not have any relevant information on them other than their school year and gender, and due to the large size of our participant pool.The observation study was reviewed and approved by the authors' institutional ethics board.

Analysis
We focused our analysis primarily on the participating autistic children and their primary play partners to better understand the characteristics of neurodiverse social play.We first performed deductive qualitative analysis to categorize each instance of play using Parten's social play definitions: unoccupied, solitary, onlooker, parallel, associative, and cooperative/social [46].During the first pass-through, we concluded that not all episodes of interest during the play session could be accurately captured using these definitions.To account for this, we created the following new categories: solitary detour, social interaction, failed social play, interaction with adult, and social conflict (Table 2).Our 'solitary detour' definition builds upon the findings of Frauenberger et al. [21] to include exceptional instances of social play where the activity of one child is not related to the main goal being completed by the rest of the group but remains under the overall shared play theme.We should note that in this paper, we consider solitary detours to be an additional form of social play.
To further understand the dynamics of the instances of play observed, we performed deductive qualitative analysis to categorize the size of groups that formed during the play session (Alone, Pair, Group) and the roles of each of the children during each episode (Leader, Follower, Equal, Copying).We additionally categorized each social episode as either positive or negative, depending on whether or not the episode was enjoyable for all children involved.We determined this by analyzing speech and using any visual cues to indicate the motivation/reluctance of participants to engage.Occasionally, this could be difficult to determine, so an 'unknown'  Social interaction or play with an adult Table 2: Table depicting codes which represent social play behavior vs non-social play behavior.The full list of codes and their extended definitions can be found in Table 1 in the supplementary materials code was also used.We additionally recorded which children or adults each child interacted with by assigning participant codes. 7he full list of deductive codes can be found in the supplementary materials.
To further contextualize the play episodes, we performed inductive qualitative analysis to describe how shared play episodes were initiated and further dissolved and how participants interacted with other children and objects throughout the session.
We initially performed the first pass-through individually.The new codes described above were generated during our discussion sessions, where we reached a common consensus on the final codes for each episode of interest during the observation sessions.Then, we created the initial themes together and further refined these until we reached our final themes.We performed this analysis using ELAN [41].
We present these findings as graphs within our paper, which build upon a similar presentation style as seen in Francis et al. [20].The graphs within this paper focus on particular findings at certain time stamps within the play sessions.The full graphs for all autistic participants can be found in the supplementary materials.

Findings
In the following, we present our remarks on neurodiverse social play based on our observations.An overall summary of the observation session can be found in the supplementary materials.
6.4.1 Initiating Social Play.We observed that all autistic children mostly used various physical actions over verbal engagements to initiate social play with others, although we did witness a few instances of participants verbally asking others to play.Physical actions used to initiate social play included throwing and hitting objects (A4, A6, A7, A8, A9), copying and imitating a peer the child wanted to play with (A2, A3, A6), showing others certain objects, running towards another child, or using the affordances of an object (A1, A2, A4, A6, A7, A8, A9) to initiate social play with that object.For instance, we observed almost all of the autistic children played with the crates.The crates had ropes tied to one of their handles, affording the crates to be pulled.Sometimes, the autistic children would sit in a crate and wait for one of their peers to approach them and drag the crate around the playground.They often switched roles with that peer afterward.6.4.2Sustaining Social Play.We observed that the most successful strategy to prevent social play from dissolving was moving to be in closer proximity to a play partner and waiting for them to reengage (A1, A2, A4, A6, A7, A9).This often allowed the children to find more interesting ideas to play with by either adding those ideas to their play or switching their play to those.For example, A1 and N1 often watched other children playing and added their play ideas to the other children's (27:34-29:35).
We also observed that changing the roles and copying others involved in the play activity helped the children sustain social play.For instance, when A6 chased A4, A4 often talked to the cameras that recorded the session about how A6 was about to 'get him.' We noticed that later in the play session, A6 copied this behavior to gain the attention of A4, which was successful.A4 then switched roles and began chasing A6, continuing the social play episode (36:05-37:20).
Another successful strategy we noticed was repeating what initiated the social play.Initially, A7 was involved in the social play of A8 and A9 by pretending to be a sleeping princess, where A9 tried to kill her while A7 tried to save her.However, A8 and A9 frequently forgot about her and ran off to continue their play.We observed that A7 repeatedly pretended to sleep on her princess bed when she realized A8 and A9 were coming back, which re-initiated their play (09:04-09:36).

Dissolving Social Play.
We witnessed that the most common strategy participants used to dissolve or end social play was to walk away from their play partners.For instance, A1 decided to play with the crate instead of the ball and walked away from N1, or A9 walked away from A7 to run around and air-slash the pretend monsters with a pretend sword (27:31-27:41).
Occasionally, a social conflict happened between the children that also dissolved social play.For example, when A6 was chasing A4 with a ball, she switched to a dinosaur toy and started hitting A4 with it.A4 appeared to be hurt, and the teacher observing mediated the play.This caused the pair to stop playing together for a while (26:53-27:04).On a few occasions, we witnessed a non-autistic child getting involved in social play with a pair that included autistic children, which resulted in the pushing out of the autistic child/children from the shared play.For instance, one of the very few successful social play experiences of A3 was when he was playing a chase game with N4.After chasing N4 for a few seconds, N2 came into the play and started chasing N4, which left A3 to go back to solitary play (22:42-22:53).
On only one occasion, we observed social play being dissolved verbally when A8 threw a Frisbee with N22 and told her that she wanted to stop playing (29:24-33:12).

Solitary Detours.
We observed some autistic children (A1, A4, A7, A8, A9) taking time away from shared tasks during episodes of social play with others.Although these instances appeared to be solitary, we observed that the activities completed were still thematically related to the shared social play episode (Table 2).
During the first play session, we observed that A1 was involved in a pretend play game of kings and queens with N1 and N4 but often enjoyed time away from the group performing construction play to make a 'castle' with large building blocks (Figure 3).N1 and N4 often would wait for A1 to return to the group to continue their shared activity of building thrones.A1 would often move between his solitary detour space and the shared game space (10:13-16:15) (Figure 4).
In the second play session, A4 took an interest in the cameras we had set up to observe the play and enjoyed tapping the camera and talking to it, asking questions such as 'Is this thing on?' in a playful manner.After witnessing this behavior, A6 approached A4 and initiated a shared game of chase, which mostly involved A6 chasing after A4 with a ball.A4 both wanted to play with the cameras and play with A6, so he integrated the cameras into his play by taking solitary detours to report to the cameras about the game of chase with A6, making statements like 'Did you see that?' and 'Oh no she's coming.' In the third play session, during the group fantasy play of knights and princesses (A7, A8, and A9), all members would occasionally perform solitary detours away from the main group play.
A7 enjoyed taking time away from the group to perform selfstimulation, or to 'stim, ' by interacting with the texture of the walls, spinning a fan toy, and tapping a tube (35:10-35:21).
A9 often ran off from the group to pretend to air slash 'giants' using another tube as a sword.Once a member had returned from their solitary detour, play resumed as before or occasionally dissolved into social play in a pair (27:31-27:41).
A8 would also often perform solitary detours, although some of these detours were caused by her play partners choosing to perform pair social play away from her for a period of time (00:00-23:42).A8 instead would often organize her 'princess bed,' which consisted of mats on top of a wheelie bin.She would pretend to fall asleep to indicate she had finished building the bed, and then the other members of the group would return to play with her (04:12-04:23).Later, during the play session, she enjoyed performing solitary detours to collect play materials to bring back to the group (34:42-37:53 and Figure 5).In the example seen in Figure 5, the solitary detour taken by A8 resulted in A9 swapping to pair social play with A7.However, when A8 returned, she struggled to re-enter the group play as they were too engrossed in pair play to notice her.Later in the play session, she brought more objects back to A7 and A9, who then involved her in their shared play again.
Reflecting back on these solitary detour examples, we did not observe any non-autistic children engaging in solitary detours.Therefore, we believe that solitary detours are primarily performed by autistic children.We also observed that solitary detours were thematically linked to social play episodes that they were a part of (kings and queens -building a castle, chasereporting the chase, knights -slashing giants, knights -pretending to be a princess).
Additionally, we observed that the solitary detours almost exclusively happened between social play episodes (Figure 4).Therefore, we propose solitary detours as solitary actions that are thematically linked to social play.6.4.5 Failed Attempts at Social Play.During our observation sessions, we witnessed multiple instances of two autistic children (A3&A5) failing to initiate or join in social play with others.
A3 did not often engage in social play but instead spent most of the observation session either watching the other children play or attempting to engage with them.A3 mostly utilized snatching or stealing of objects to gain the attention of others, which was often perceived as a negative action by other children on the playground.During the observation session, this included taking crates and mats from N2, who would either fight for them back or go straight to the teacher to resolve the issue (21:41-22:08).Additionally, A3 attempted to engage with others by offering play equipment to others, such as balls or large building blocks, that the other participant(s) desired for their play.One example of this equipment offering was when A3 obtained mats for A1, N4, and N1 without their knowledge and then dumped them on top of their play space.The group did not acknowledge him but were happy to have received mats for their play scenario (15:30-15:50).
A3 would attempt to interact and play with other children by performing actions related to the others' shared game, but it was often misaligned in both time and space.These actions were either performed after the rest of the group had moved on in the scenario or in an area outside of the shared play zone.A3 followed around a group of non-autistic girls, one of whom (N3) was a previously established friend of his, and mimicked their behavior to attempt to join in with their game of chase.However, this was often performed too late and physically too far away than expected during the play.This meant that the girls did not appear to understand this as a request to join their play or chose to ignore his attempts to join (01:41-01:54).
For A5, at the start of the play session, a group of girls pushed a wheelie bin around with one participant inside.Initially, A5 assisted the girls in pushing the bin and successfully engaged in social play with them.However, she wanted to include objects of her choice (a box, a mat, a ball) in the play by putting them inside the wheelie bin.The girls repeatedly pushed away the objects and stated they did not want the objects A5 wanted to include in the play.The girls repeatedly took the object used from A5 and told her, 'Don't do that, ' or 'We don't like that.' However, A5 persisted by placing mats, balls, and boxes into the wheelie bin until she found interest in another object to play with alone (00:55-07:55).
Reflecting on these examples, we observed that the method of communication and engagement is important to be accepted in social play with other peers.A3 often snatched objects and offered them to others to communicate that he wanted to play, and A5 forced objects into others' play.However, their behavior was mostly received negatively by others who ignored and did not involve them in their play.Based on our observations, the more successful attempts at social play happened more organically, when all play partners showed interest in social play.On the contrary, the successful methods for other autistic children, such as following and copying others, did not work for A3 and A5.This suggests that despite their importance these methods are to be treated as unique to each child and each context.6.4.6 Social Conflict.We often observed social conflict between A5 and her peers on the playground due to A5 demanding certain objects for her solitary play scenarios that others were already using.However, during social play instances, we witnessed two conflicts.
One social conflict arose due to A5's repeated attempts to include certain objects into play by putting them into a wheelie bin the group of girls were playing with (as explained in section 6.4.5).A5 appeared to misunderstand the ambiguous rules of the game, and the rest of the group did not appear to want to consider her ideas for the shared play, instead wanting to continue with the game they had created.
Figure 4: Graph displaying play observation information for A1 and N1 between timestamps 10:00 and 16:30.The rows represent our qualitative description of each play segment, the play type of the participant, group size of the play episode, and play partners respectively.We chose to represent group size during solitary detours as both alone and group/pair due to its unique quality of including both (see Table 2).This figure aims to show A1's movement between solitary detour and shared play (as depicted in the bottom rows, when A1's play is still aligned with N1's overall social play theme at times when N1 is not attentive to A1).
Figure 5: Graph displays play observation information for A8 between timestamps 24:00 and 44:00.The rows represent our qualitative description of each play segment, the play type of the participant, group size of the play episode, and play partners respectively.This figure aims to show the difficulties A8 faced reintegrating back into social play with A7 and A9 after performing multiple solitary detours Another social conflict arose due to a miscommunication of play space boundaries.A5 approached D1 and N9 to play in parallel with the items in her crate.A5 declared, 'This is my driveway' while standing in a small corner of the playground, and N9 responded, "Okay, but this is where I play with my bugs' while running to the area next to A5. N9 and D1 walked into the 'bug space,' and A5 shouted in frustration at N9 and D1, 'Can you stop going in my driveway,' resulting in confusion for N9 and D1 as they had expressed to A5 previously that this area was where they were playing with their bugs and they believed a different space was her driveway (41:02-41:14).
We believe these social conflicts seen during social play happened due to what appeared to be misunderstandings and miscommunication during play.

DISCUSSION
In this paper, we presented answers to two research questions by conducting two studies consisting of: (1) Semi-structured interviews with 6 professionals who support neurodiverse social play.(2) Observation sessions with 36 children during unstructured social play.
RQ1) What are the challenges and barriers to facilitating neurodiverse social play that includes both autistic and nonautistic children?(Section 5) Autistic children might experience barriers to neurodiverse social play due to their differences in communication styles, strong desire to play on their own terms, understanding of ambiguous rules and norms, and emotional and sensory regulation (Sections 5.4.1, 5.4.3, 5.4.5 and 5.4.7).In counterpart, non-autistic children might expect their play partners to follow their accustomed methods of communication, expect to act as a cohesive group, have trouble diverging from the correct rules, and empathizing with sensory perception differences in neurodiverse social play (Sections 5.4.2, 5.4.4,5.4.6 and 5.4.8).Finding a mutual solution to these interpersonal barriers is key to facilitating successful neurodiverse social play.
RQ2) How should we understand and characterize the different experiences of autistic and non-autistic children when they engage in neurodiverse social play?(Section 6) We characterized the different experiences of autistic and nonautistic children in neurodiverse social play by analyzing how they initiate, sustain, and dissolve social play.Autistic children might choose physical actions to initiate play over non-autistic children's preference for verbal engagement, and more successful attempts at social play might happen more organically.Autistic children might fail to initiate social play if their non-autistic play partners do not accept their method of communication and engagement or if they are misaligned with their play partners in both time and space (Sections 6.4.1, 6.4.6 and 6.4.5).Autistic children might sustain social play by being in close proximity to their play partners, copying their play partners, and repeating what initiated the social play (Section 6.4.2).They might take solitary detours to incorporate their interests in social play to keep it going, whereas non-autistic children might stick to overall group interests (Section 6.4.4).Autistic children might dissolve social play by walking away from their play partners or unintentionally creating a social conflict and nonautistic children might unintentionally dissolve autistic children's play by getting involved without an invitation (Sections 6.4.3 and 6.4.6).

Design Insights for Neurodiverse Social Play
In the following sections, we discuss underlying aspects of neurodiverse social play, while presenting the connections between both studies by combining and comparing our themes and double empathy reflections from the interviews to what we observed in neurodiverse social play.We primarily focus on the concepts of solitary detours and communication to address design insights for future neurodiverse social play research and technologies under each concept.Consequently, we demonstrate how the double empathy problem could be utilized to reflect on and understand instances of neurodiverse social play.
We should note that while we were able to characterize the experiences and behavior of autistic and non-autistic children, our methodology consisting of interviews and observations is limited in providing concrete design recommendations, considering every child has unique experiences.Thus, we are only able to provide broad perspectives on design directions.In the future, we plan to employ more comprehensive methods, such as co-design with neurodiverse children, which are more likely to yield more concrete design recommendations.We should also emphasize that some of the insights presented could be addressed without the use of any technology as we observed the children or the professionals naturally implementing these strategies.However, we believe the use of technology could be helpful to support when potential breakdowns in natural behavior occur, such as what we observed in social conflict or failed attempts to engage in social play episodes.

Solitary Detours and Playing on Own Terms in Neurodiverse
Social Play.The interview participants expressed the belief that many autistic children tend to think in black-and-white terms and may struggle to understand what others require in a given situation or insist on their own viewpoint.Additionally, participants suggested that non-autistic children may follow group dynamics rather than individual preferences during social play.They might not comprehend why play should revolve around one child's preferences and may expect everyone to engage in the same activity.During the observation sessions, we witnessed some of the autistic children periodically engaging in different tasks from the rest of their social playgroup.These 'solitary detours' appear similar to parallel play but differ because the overarching theme of their play remains the same as their play partners while their individual task is undertaken alone and may not be shared or made apparent to their play partners.We believe that solitary detours can be a way for autistic children, in particular, to meet both their own motivations to play on their own terms and the expectation of other children to play as part of a cohesive group.Solitary detours can support children in undertaking play activities that others in the group may not wish to engage in but, as characterized above, still allow those children to contribute to be involved in the overarching shared play scenario.
The solitary detour we witnessed expands upon Frauenberger et al. definition of solitary detours [21], which focuses on children taking space away from group play to emotionally regulate through self-stimulation (also known as 'stimming').Although we did witness one child (A7) taking solitary detours to perform stimming, we mostly witnessed children performing solitary detours with the apparent aim of achieving unfulfilled play motivations.However, it could be argued that the solitary detours we witnessed also allowed for emotional regulation as taking space and time alone away from others may be calming for some autistic children.
We believe that solitary detours are primarily performed by autistic children, as we did not observe any non-autistic children engaging in solitary detours.Non-autistic children usually participated in the main task of the game with others.The theme of each solitary detour was often thematically linked to the main theme of the shared play, as other children may perceive unrelated activities as wishing to dissolve social play.Due to this, the objects used by the children during solitary detours also appeared to fit these themes, potentially indicating that a related object choice for the solitary detours may be important to ensure the overarching social play does not dissolve.We argue that our understanding of solitary detours moves beyond the definitions of previous literature [21] in terms of: • Solitary detours are primarily performed by autistic children • Solitary detours are thematically linked to social play episodes • Solitary detours almost exclusively happen between social play episodes • Solitary detours serve to fulfill individual play motivations that differ from other play partners We suggest neurodiverse social play technologies and research should utilize and support a more nuanced understanding of solitary detours to align individual and group motivations in neurodiverse social play.One design strategy to address this was suggested by P5, a collaborative game where "there will be multiple different scenarios and different paths you could take, so it's not just a linear progression, but they could choose which way they want to go and they could split off into individual teams as well as just go off as one big team."We also believe solitary detours would require the support of neurodiverse social play partners while they are on their individual paths.Supporting solitary detours through the use of social play technology may aid in ensuring children can easily leave and re-enter the shared play space when required.Nonnis et al. have explored the importance of including different entry and access points within a design to ensure fluid transitions between entering and exiting interaction with a shared social play technology for autistic children [27,43].We argue that integrating multiple entry and access points to a social play technology could support transitions from solitary to social play, and then also between solitary detours and the main shared play scenario for neurodiverse groups.
We also argue that neurodiverse social play technologies should be designed to allow children to easily indicate to play partners their intentions to exit and also return to the shared scenario, before and after performing a solitary detour.This may also help mitigate issues with rejoining social play, such as those we saw with A8 during the play episode (see Figure 5).Additionally, allowing the neurodiverse play partners to know explicitly when autistic children are taking solitary detours and how that impacts the overall shared play scenario might support the group social play [21].This may aid in others' continuation of the shared play scenario during solitary detours, and thus help prevent both the remaining members from pausing social play (as seen with N1 and N4 when A1 is away) and also allow autistic children to be aware of the status of the shared play episode while they are away.Thus, technologies should help to connect play partners during solitary detours and create means for them to return to neurodiverse social play, if and when they so wish.
7.1.2Communication in Neurodiverse Social Play.The interviews revealed that autistic children often communicate in ways that may be perceived as ambiguous, carrying different or unclear meanings from more typical contexts, such as using hitting as a means to engage.However, non-autistic children may tend to rely on other and more familiar methods of communication, typically verbal, and may encounter difficulties in understanding divergent communication styles.
We found that in some instances the primarily physical methods of social play initiation used by autistic children (offering objects etc.) were misunderstood or ignored by the rest of the group.Additionally, the misalignment of initiation techniques, as seen in A3, in both time and space caused further difficulties as his nonautistic peers did not perceive his copying behavior as a request to join play.Mismatches of communication styles also caused misunderstandings between some non-autistic and autistic children on the playground.In most instances of social conflict that we observed, the non-autistic children often resorted to reporting what had occurred to the teacher immediately for support rather than attempting to negotiate the situation themselves first.
Therefore, we can argue that an important point of interest for neurodiverse social play technologies and research is bridging the mismatch in communication styles and supporting children in sharing emotions, play ideas, and resolving conflict, without adult intervention.The current focus in social play technologies is social skill intervention for autistic children to better match neurotypical communication styles, rather than considering the possibility in which non-autistic children are taught through technology to better understand autistic communication styles [2,7,39,48].However, we argue that this burden of deciphering interactions should not be placed on either group.Building on the double empathy problem, the shared responsibility of communication and learning the different communication styles of play partners could be addressed by neurodiverse play technologies.This could be achieved by creating a common medium of communication through the use of technology that doesn't rely on participants themselves interpreting the behavior or language of others.Finding this common ground in a conversation between adult neurodiverse dyads has been explored by Zolyomi and Snyder and could be extended to neurodiverse play contexts [64].
Alternatively, the burden of interpretation of communicative behavior could be placed on a shared technology rather than on any one play partner.Wilson et al. have previously examined supporting self-expression for autistic children through different modalities, which could be extended for use with all children involved in neurodiverse play scenarios [59].Future neurodiverse technologies could encourage children to express their emotions, play ideas, and intentions in a way that is unique to them through any chosen sensory modality, rather than using pre-defined language or symbols [34,44,58,59].This could then translate this representation of information shared by one child to another representation that is clear and compatible with another child's unique communication style, through a sensory modality they prefer.The interview participants noted that shared social play rules could sometimes be misinterpreted by autistic children.However, many non-autistic children may not be aware of the struggles their autistic peers face in comprehending and adhering to these rules.We observed multiple instances of autistic children misinterpreting the social play rules, which caused them to be ignored by the nonautistic children or led to a social conflict between them.We believe this mismatch of understanding of rules could be addressed by promoting independence for all children when negotiating shared play rules.For example, P6 recommended that "getting the group together to establish the rules so everyone agrees on the rules [has] been quite fundamental in moving forward", which could be facilitated by future neurodiverse social play technologies and research.
The interviews highlighted that autistic children often have different sensory processing experiences compared to their nonautistic peers, and their emotional states can be particularly sensitive to sensory stimuli [19].Social play frequently involves actions with high sensory output, such as shouting and touching, by nonautistic children, which may discomfort their autistic peers.However, we did not notice any obvious difficulties with sensory regulation during our observations.Some children already employed methods of sensory regulation, such as wearing ear defenders (A7) or carrying materials (A4 carrying a stick) to stim with, which may have mitigated these issues.Many current social play technologies and research strongly focus on how the technology can support regulation strategies, but this may not be a significant issue for neurodiverse social play [2,24], particularly where children's strategies are seamlessly integrated into the play.Nonnis et al. have explored directly integrating tools for sensory regulation into tangible social play technologies for autistic children, which may also benefit neurodiverse groups [42,43].Additionally, from our observations, we noticed seamless integration of regulation strategies can happen when other children accept these needs and do not try to stop those behaviors or instead work around them.

Double Empathy Problem in Neurodiverse Social Play Research
The notion of the double empathy problem has been particularly useful in this examination of neurodiverse social play because it starts from an aim and position of developing shared and mutual understanding among all play participants.This notion supported us as researchers to consider not just the social play experience and activity of autistic children but that of all children, grounding our reflection and analysis through this lens and beyond a deficit-or intervention-based model of social play.This has further prompted reflexivity on our own assumptions of what it means to play and communicate in diverse ways.Furthermore, its worth in HCI discussions of designing technologies to support social play lies in its conceptual framing of research and design methodology and methods.In relation to this project, the double empathy problem generates consideration of mechanisms that enable mutual understanding and shared communication and opportunities for differentiated modes of social play (i.e., solitary detours), while supporting us to create design insights for future technologies.However, the practical operationalization of the double empathy problem is not well specified by previous research and there is little practical information on how to apply it within design methodologies.Thus, we were challenged to reflect on neurodiverse perspectives and should emphasize our assumptions are based on our findings and experience as neurotypical researchers.Yet, this was an initial application where we explored and speculated upon the potential of the double empathy problem to shape our research.We believe that we need to develop a framework to both understand the double empathy problem to its full extent and guide researchers like ourselves on how it could be applied.We plan to work with this concept in the next steps of our research, by which we hope to create a methodological approach to apply the double empathy problem and discuss its underlying constructs in neurodiverse social play research.

Limitations
One of the main limitations of our observations was that all of the children present already knew each other as they were from the same class and spent most lessons together.This meant that we saw strong friendships between play partners, such as between N1 and A1, which may explain the strong understanding both N1 and A1 had of each other's needs.Additionally, we only performed one observation of each group, rather than multiple over time.Running multiple observation sessions may have enabled the participants to feel more comfortable in the presence of the researchers and thus may have been reflected in their behavior.

CONCLUSION
In this research, we aimed to understand and consider the different experiences of autistic and non-autistic children during neurodiverse social play by conducting two studies consisting of: (1) Semi-structured interviews with 6 professionals who support neurodiverse social play.(2) Observation sessions with 36 children (9 autistic children) during unstructured social play.By reflecting on the findings of both studies, our contribution is two-fold: (1) We present and discuss characterizations of neurodiverse social play, including experiences and activities undertaken by both autistic and non-autistic children, some of which move beyond the current characterizations in the literature.
To highlight and specify this contribution, we uncovered that 'solitary detours' serve autistic children to fulfill their individual play motivations while maintaining social play and that these have a specific theme according to each child, whereas previous research defines solitary detours as taking space away from group play to emotionally regulate through behavior.(2) We capture and bridge the different experiences of autistic and non-autistic children in neurodiverse social play, by building upon the notion of the double empathy problem.We emerge the mutual perspectives in neurodiverse social play by reflecting on autistic children's barriers to social play shared by the professionals and considering non-autistic children's perspectives.We then combine and compare these to our observations to examine how the double empathy problem could be utilized to reflect on and understand neurodiverse social play and point towards design insights for future neurodiverse social play research and technologies.We believe our findings and contributions could serve as novel insights to better understand the complex notion of neurodiverse play and, in doing so, assist in shaping the design of future neurodiverse social play technologies and research in HCI.

Figure 1 :
Figure 1: The summary of the interview study findings, including the themes we generated for understanding neurodiverse social play and our double empathy reflection to reflect on non-autistic children's perspectives.

Figure 2 :
Figure 2: The open playground, where the observation sessions took place, as captured by video cameras recording the sessions.

Figure 3 :
Figure 3: A1 building a castle from large building blocks as a solitary detour for their game of kings and queens with N1 and N4.

Table 1 :
Table of autistic children who participated in the observation session.Bold text shows the dominant information where the participant has multiple entries (if applicable) (e.g., A1 has two play partners, N1 and N4, with N1 being the main play partner).