Research on autism, which is defined as a life-long developmental disability affecting social interaction, has focussed predominantly on how autistic individuals perceive and interact with others with less emphasis on the perspectives of their interactional partners. Yet autistic viewpoints have highlighted how other people are part of a two-way breakdown in interaction originating from differences between people rather than the deficit of any one individual, a phenomenon known as the double empathy problem. A gap therefore exists in the literature in terms of understanding how autistic sociality (i.e. the range of social opportunities possible for a given individual on the spectrum) is shaped by different interactional partners. This thesis examines the double empathy problem in three interactional contexts. Study 1 examines relationships between autistic people and their family members through focussing on perspective-taking, the ability to impute mental states to others. In light of prior research where autistic abilities have been assessed using abstract scenarios, Study 1 implements a two-way measure of perspective-taking which considers both sides of 22 real-life relationships (n=44) consisting of autistic adults and their family members, to understand how autistic people are seen by familiar others as well as vice versa. It uses a mixed-methods approach, where members of each dyad were individually asked about 12 topics, providing quantitative scores and qualitative explanation of their rating of Self, their rating of their partner, and their predicted rating by their partner. Comparison of perspectives provided a means for detecting misunderstandings and their underlying rationale. The contribution of Study 1 is that it shows perspective-taking is two-sided: family members can be biased in underestimating the perspective-taking of their autistic relatives, while autistic adults are aware of being negatively viewed despite disagreeing with such views. Study 2 examines interactions between autistic adults (n=30) partaking in a naturally occurring activity of video-gaming at a charity. It is a qualitative study using participant observation, with each conversational turn systematically rated in terms of coherence, affect and symmetry to identify the key features of neurodivergent intersubjectivity, the process through which autistic people build shared understanding in their own non-normative ways. The contribution of Study 2 is to identify two forms of neurodivergent intersubjectivity which enable shared understanding to be achieved, but which have traditionally been viewed as undesirable from a normative social viewpoint: a generous assumption of common ground that, when understood, lead to rapid rapport, and, when not understood, resulted in potentially disruptive utterances; and a low demand for coordination that ameliorated many challenges associated with disruptive turns. Study 3 examines interactions involving lay people (n=256) who believe they are interacting with an autistic partner through an online collaborative game, when in fact they are playing with an intelligent virtual agent (IVA) who behaves the same way for all participants. Its contribution is methodological as it develops a new application for simulating interactions in experimental research called Dyad3D. Study 3 uses Dyad3D to explore how disclosure of an autism diagnosis by the IVA affects social perception and social behaviour in comparison to a disclosure of dyslexia and a condition where there is no diagnostic disclosure. Combined with a post-game questionnaire, Study 3 triangulates self-reported (quantitative rating scales and qualitative explanation) and behavioural measures (quantitative scores of actions within the game) to understand the interplay of positive and negative discrimination elicited through using the label of autism. It highlights that diagnostic disclosure of autism leads to significant positive bias in social perception when compared to a disclosure of dyslexia or a no disclosure condition; yet participants are not as helpful towards the autistic IVA as they think they are, indicating a potential bias in helping behaviour. The thesis takes an abductive methodological approach which integrates with a wider call for a more participatory model of research in the study of autism. Abduction is a form of reasoning which involves the iterative development of a hypothesis that holds the best explanatory scope for the underlying phenomena observed. It is inherently aligned with a participatory model of research because abduction involves the ongoing exploration of ideas that may originate from multiple sources (i.e. interactions with autistic people as well as research outputs). Taking a more holistic approach to the development of knowledge with autistic people which recognises the legitimacy of different claims to knowledge is important, because prior research in the field has often failed to critically reflect on researcherparticipant positionality and the principals underlying the development of research agenda. For this reason, the thesis details the participatory activities which surround and interconnect with the development of the three empirical studies. Overall the thesis contributes to understanding autistic sociality as a dynamic, interactionally shaped process. It reasons that autistic people have unrealised social potential, both in terms of imagining other perspectives (Study 1) and coordinating with others (Study 2). However, such social potential may not be easily recognised by other non-autistic people who may be biased in their assumptions about autism (Study 1 and Study 3). Consequently, the evidence presented in this thesis helps to explain some of the processes that underscore the double empathy problems reported in literature, including poor mental health (because autistic people are aware that they are misunderstood by others, see Study 1), employment prospects (because autistic social potential is under-recognised by others, see Study 1 and 3), and quality of life (because neurotypical standards of communication are not compatible with neurodivergent forms of intersubjectivity, see Study 2). The thesis therefore makes suggestions for how we design enabling environments which are sensitive to the dynamic factors that can enable autistic sociality to flourish.