Between 14-30% of school-aged children in the United Kingdom and the United States require additional support for learning (Department for Education, 2018; National Center for Education Statistics, 2019). The standard approach adopted by educational systems and researchers alike is to classify children and tailor support according to the particular area of the curriculum in which they are experiencing problems, or to use diagnostic criteria to label symptoms and inform intervention choices (e.g. for ADHD or dyslexia). This pragmatic, classification-based approach has many strengths. However, many children with relatively mild problems go undetected, and the needs of others with complex learning difficulties and co-occurring disorders often go unaddressed. To better capture the needs of the common struggling learner, our research adopts a transdiagnostic approach to identify dimensions rather than categories of disorder (e.g. Casey, Oliveri & Insel, 2014).
Within our group we use innovative data-driven methods challenge the diagnostic orthodoxy that has dominated the study of learning difficulties. Much of this work has been made possible through data collected in the Centre for Attention Learning and Memory, a research clinic led by group leader Dr Joni Holmes and run in collaboration with Professor Susan Gathercole, Dr Duncan Astle and Dr Rogier Kievit and their research teams.
We have used network science methods to identify both clusters of children with common behavioural profiles and clusters of symptoms that cross-cut traditional diagnostic criteria (Bathelt et al. 2018; Mareva et al., 2019). This novel approach provides converging evidence for a dimension of symptoms that includes pragmatic communication problems, social difficulties and behavioural issues, and which extends beyond traditional diagnostic boundaries (Hawkins et al., 2016).
We have also used contemporary factor analytic and measurement invariance models to identify dimensions of cognitive abilities that differentiate struggling learners, approach to understanding the cognitive difficulties associated with struggling to learn. These methods reveal the cognitive dimensions critical for learning extend across children with and without learning difficulties, providing minimal evidence for any compensatory cognitive mechanisms (Holmes et al., 2019).
Together with Duncan Astle’s lab we have used machine learning methods to reveal that children who are struggling at school can be distinguished by their cognitive and neural profiles, but that clinical or educational diagnoses do not map on to these groupings (Astle et al., 2019; Suizdgate et al., 2020). This challenges the notion that diagnostic labels provide a window in to the causes of learning problems and is having a substantial impact on the field. Crucially for practitioners, these data suggest children who present with similar learning difficulties in the classroom may be struggling for very different reasons: their cognitive (and in turn neural) profiles could be markedly distinct. This finding has been reported in an invited editorial in the Times Education Supplement (Holmes et al., 2019).