Representing Emotion in a Mathematics Tutor Designed for People with ASD by People with ASD

Background:  Recognizing and responding to the emotions of other humans has been argued to be a core deficit in Autism Spectrum Disorder (ASD), although evidence is mixed. For example, this deficit is not evident in simplified, distinctive animated representations of human-like emotion. Screen-based characters (digital personas) are often used within computer-based tutors to support learning. Representing emotionally-laden elements (e.g. a smiling face) within digital personas is typically motivating and beneficial to the user experience, resulting in enhanced learning. However, how emotion is best represented within user interfaces and the impact of this emotion upon learning for people with ASD is not well understood. Firstly we involved adolescents with ASD in the design of the mathematics tutor, including a digital persona, using a process called ‘Participatory Design’. This data accompanying the demonstration explored how representing emotion within a digital persona providing feedback within a mathematics tutor impacted upon adolescents with ASD.

Objectives:  There were three objectives: Firstly to identify whether adolescents with ASD demonstrated a preference for feedback from a digital persona compared to text-based feedback; Secondly to identify if a digital persona enhanced learning compared to text-based feedback; and thirdly to explore whether the representation of emotion within the digital persona conferred any additional preference or learning benefits in adolescents with ASD.

Methods:  This demonstration presents a technology that was used to assess user preferences and learning in twenty six adolescents with ASD under three conditions: 1) with a digital persona displaying emotion; 2) with a digital persona not displaying emotion; and 3) text only, with no digital persona present. All participants undertook all conditions, with the order of conditions counter-balanced across participants. Specifically, each condition involved users learning mathematics using a computer-based tutor designed by adolescents with ASD. The three conditions varied the nature of the delivery of feedback to users. Before and after each condition an evaluation of user preferences was conducted. The mathematics tutor also monitored performance in each condition, so that the number of correct responses to mathematic problems per condition could be identified. Finally, a paper-based pre and post mathematics assessment was conducted to evaluate if learning had generalized (over the three conditions combined).

Results:  Adolescents with ASD preferred a digital persona and performed better (got more mathematics questions correct) when a digital persona was present, compared to when no digital persona was present. However, the two conditions with a digital persona present were the same. Overall, learning did generalize to a paper and pencil assessment of mathematics.

Conclusions:  Representations of emotion within the digital persona did not confer any additional benefits for adolescents with ASD. If the provision of emotionally-laden stimuli does not enhance performance, what aspects of the digital persona do facilitate performance (when compared to text-based feedback) in those with ASD? Overall the technology demonstrated did enhance learning of mathematics in adolescents with ASD that generalized to a pencil-and-paper based assessment.

Here we demonstrate the whole mathematics tutor system with the digital persona providing feedback upon mathematics performance.