Who Puts the "Social" in "Social Computing"?: Using A Neurodiversity Framing to Review Social Computing Research

TL;DR

It is argued that a neurodiversity perspective informed by critical disability theory allows us to engage with alternative forms of sociality as meaningful and desirable rather than a deficit to be compensated for.

Abstract

Human-Computer Interaction (HCI) and Computer Supported Collaborative Work (CSCW) have a longstanding tradition of interrogating the values that underlie systems in order to create novel and accessible experiences. In this work, we use a neurodiversity framing to examine how people with ways of thinking, speaking, and being that differ from normative assumptions are perceived by researchers seeking to study and design social computing systems for neurodivergent people. From a critical analysis of 84 publications systematically gathered across a decade of social computing research, we determine that research into social computing with neurodiverse participants is largely medicalized, adheres to historical stereotypes of neurodivergent children and their families, and is insensitive to the wide spectrum of neurodivergent people that are potential users of social technologies. When social computing systems designed for neurodivergent people rely upon a conception of disability that restricts expression for the sake of preserving existing norms surrounding social experience, the result is often simplistic and restrictive systems that prevent users from "being social" in a way that feels natural and enjoyable. We argue that a neurodiversity perspective informed by critical disability theory allows us to engage with alternative forms of sociality as meaningful and desirable rather than a deficit to be compensated for. We conclude by identifying opportunities for researchers to collaborate with neurodivergent users and their communities, including the creation of spectrum-conscious social systems and the embedding of double empathy into systems for more equitable design.

Figures (5)

• Figure 1: Publications per year for papers included in our corpus for the years 2013-2023. Note that 2023 only includes publications from January 1, 2023 to June 30, 2023.
• Figure 2: Publications by country, where countries with two or more publications are listed individually. Due to collaborations, bars do not sum to 84. North America, and the United States specifically, dominates discourse on ND social computing, meaning conceptions of neurodivergence that differ from those in the US, UK, and Australia are unlikely to be considered despite these systems involving users across multiple regions.
• Figure 3: Identifying characteristics presented for each of the 101 neurodivergent and 72 neurotypical samples. A sample could be identified across multiple characteristics (ex. a description of "children ages 7-17 from the United States" would be coded under "Age" and "Nationality"). ND participants are most often described by their age and binary gender (male or female) when gender is presented. NT participants are most often identified exclusively by their familial relation to a corresponding ND participant or their employment (teacher, therapist, etc). For more information on the application of identifying characteristics, see "Identifiers" in Section 8.2.8 of the Appendix.
• Figure 4: Age distribution by sample type. Note that the number of samples does not sum to 181 because multiple age groups can be present within a sample and some samples do not distinguish between ND and NT participants. NT involvement in ND social computing research is almost exclusively restricted to NT adults, except for a few cases in which ND and NT children are directly compared.
• Figure 5: Gender distribution by identity. A male majority sample is a sample of n participants where at least (n/2) + 1 participants are male, a female majority sample has (n/2) + 1 female participants, a balanced sample has the same number of male and female participants, and a sample with no majority features no group with more than n/2 participants. ND samples are much more likely to be male majority, and NT samples are much more likely to be female majority.