No abstract

Prominent theories in cognitive science propose that humans understand and represent the knowledge of the world through causal relationships. In making sense of the world, we build causal models in our mind to encode cause-effect relations of events and use these to explain why new events happen by referring to counterfactuals — things that did not happen. In this paper, we use causal models to derive causal explanations of the behaviour of model-free reinforcement learning agents. We present an approach that learns a structural causal model during reinforcement learning and encodes causal relationships between variables of interest. This model is then used to generate explanations of behaviour based on counterfactual analysis of the causal model. We computationally evaluate the model in 6 domains and measure performance and task prediction accuracy. We report on a study with 120 participants who observe agents playing a real-time strategy game (Starcraft II) and then receive explanations of the agents' behaviour. We investigate: 1) participants' understanding gained by explanations through task prediction; 2) explanation satisfaction and 3) trust. Our results show that causal model explanations perform better on these measures compared to two other baseline explanation models.

A growing body of research is examining the way that virtual reality (VR) technology might enrich the lives of older adults. However, no studies have yet examined how this technology-combining head mounted displays, motion tracking, avatars, and virtual environments-might contribute to older adult wellbeing by facilitating greater social participation (social VR). To address this gap, we conducted three workshops in which 25 older adults aged 70 to 81 explored the utility of social VR as a medium for communicating with other older adults. Participants first created embodied avatars that were controlled through natural gestures, and subsequently used these avatars in two high-fidelity social VR prototypes. Findings from the workshops provide insight into older adults' design motivations when creating embodied avatars for social VR; their acceptance of social VR as a communication tool; and their views on how social VR might play a beneficial role in their lives. Outcomes from the workshops also illustrate the critical importance our participants placed on behavioural anthropomorphism-the embodied avatars' ability to speak, move, and act in a human-like manner-alongside translational factors, which encapsulate issues relating to the way physical movements are mapped to the embodied avatar and the way in which errors in these mappings may invoke ageing stereotypes. Findings demonstrate the critical role that these characteristics might play in the success of future social VR applications targeting older users. We translate our findings into a set of design considerations for developing social VR systems for older adults, and we reflect on how our participants' experiences can inform future research on social virtual reality. CCS Concepts: • Human-centered computing → Virtual reality; Collaborative and social computing; Empirical studies in collaborative and social computing.

Results suggest a need for studies that examine new and innovative forms of technology, evaluated with rigorous methodologies, and drawing on clear definitions about how these technologies address social isolation/participation.

Older adults are normally characterized as consumers, rather than producers, of digital content. Current research concerning the design of technologies for older adults typically focuses on providing access to digital resources. Access is important, but is often insufficient, especially when establishing new social relationships. This paper investigates the nature and role of digital content that has been created by older adults, for the purpose of forging new relationships. We present a unique field study in which seven older adults (aged 71-92 years), who did not know each other, used a prototype iPad application (Enmesh) to create and share photographs and messages. The findings demonstrate that older adults, even those in the "oldest old" age group, embraced opportunities to express themselves creatively through digital content production. We show that self-expression and social engagement with peers can be realized when sociotechnical systems are suitably designed to allow older adults to create and share their own digital content.