Developing Software to Support Metacognition in Autism Spectrum Disorder

Background: Metacognition comprises of at least two components - metacognitive knowledge (e.g. ‘I am better at multiplication than division’) and metacognitive monitoring (e.g. ‘I do not understanding this question’). Metacognition has been found to be a powerful predictor of learning performance. For example, research has highlighted that metacognition predicts mathematical performance more powerfully than intellectual abilities and there is extensive evidence that developing metacognition is an effective intervention for students. Research has suggested that weaknesses in metacognition in students with Autism Spectrum Disorder (ASD) can result in a failure to correctly recognise when they have made errors, which can impede learning.

We have developed a computer-based mathematics tutor which incorporates a metacognitive component called an ‘Open Learner Model (OLM)’. The OLM was designed to show students their own learning trajectories so that they were made aware of which strategies they had adopted and which were successful - thus aiding in their reflection and future learning. The OLM governs the personalisation of learning by guiding how the tutor dynamically adapts both tuition and assessment for individual students. The OLM supports metacognition by allowing students to access an external representation of what they have and have not successfully achieved and to discuss this with teachers and peers. 

Objectives:  Demonstate system with findings.

Methods:  Our project worked with students through a process called ‘Participatory Design’, which involves the end-users in the design of the tutor. Students informed how the OLM should look and change as their knowledge states changed.

We monitored the performance of 27 students (aged 11-14) with ASD who were underperforming at mathematics. The students were working at a level of 9-11 year olds and a group of thirty 10-11 year olds were recruited as controls. All participants undertook three 20-minute sessions on consecutive days. At the beginning of each session, students were asked to access the OLM and the representation explained. Each session retained the data for each student, so, for example, the OLM at the start of session two was as it had been left at the end of session one. Students undertook a series of modules within the ‘Number’ topic of the UK National Curriculum for Mathematics (e.g. multiplication questions). Students could access the OLM any time. OLM access and performance upon mathematics questions were monitored by the system. 

Results:  The technology demonstration will highlight how the OLM was designed by students with ASD. On average, 139 questions were answered correctly and the OLM was accessed 6 times. The difference in the number of correct responses between session one and session three significantly correlated with the number of times the OLM was accessed. This was the case for students with and without ASD.

Conclusions:  

Metacognition can be effectively supported by this software. Although designed by students with ASD the benefits extend to those without ASD. The most revealing data is illustrated by individual learning progression diagrams (e.g. showing getting questions wrong, accessing the OLM and then getting questions correct). These diagrams will be presented alongside the technology demonstration.