Evidence accumulation modelling offers new insights into the cognitive mechanisms that underlie linguistic and action-based training

Evidence accumulation modelling has been shown to uncover new insights into the cognitive mechanisms that underlie decision making from behavioural data. By jointly modelling reaction time and accuracy data, such decision models estimate latent variables that represent distinct computational processes, such as stimulus encoding, response caution and the quality of information processing. In this study we used an evidence accumulation model, the Linear Ballistic Accumulator (LBA), to shed new light into the mechanisms that underlie learning based on linguistic and action-based training. The LBA model was applied to behavioural data from a previously published training study where participants learned to name, tie or name and tie a set of knots. Our results showed that training is multifaceted and associated with an increase in stimulus-encoding time, a reduction in response caution, as well as an increase in the speed of information accumulation. Furthermore, the results showed that there was an added benefit to the rate of evidence accumulation when naming and tying experience were combined. This latter finding suggests that performance benefits from multi-modal training may be instantiated in computational processes that are associated with the quantity and quality of information accumulation during decision making. Overall, in applying this computational approach to accuracy and reaction time data, we uncover new insights into the mechanisms that govern experience-dependent plasticity.