Mathematics anxiety and cognition : a computational modelling study

Abstract

Anxiety about performing numerical calculations is becoming an increasingly important issue. Termed mathematics anxiety, this condition negatively impacts performance in numerical tasks which can affect education outcomes and future employment prospects. The disruption account proposes this poor performance is from the anxiety and its worrying thoughts disrupting the limited resources of working memory (specifically the attentional and inhibitory functions) leaving less cognitive resources available for the current task. There are many behavioural studies on mathematics anxiety. However, its underlying cognitive and neural mechanisms remain unclear. This thesis examines the relationship between mathematics anxiety and attentional control using neural network modelling, there are no neural network models simulating mathematics anxiety. The numerical Stroop task and the symbolic number comparison task were modelled with a single neural network model architecture examining the effect of modifications to both tasks. Different model modifications were used to simulate high and low math-anxious conditions by modifying attentional processes and learning. The model simulations suggest that mathematics anxiety is associated with reduced attention to numerical stimuli. These results are consistent with attentional control theory where anxiety decreases the influence of the goal-directed attentional system and increases the influence of the stimulus-driven attentional system. Notably, when simulating the numerical Stroop task, the high math-anxious model with reduced attention to numerical stimuli experienced less neural activation in the response layer for the inhibitory condition than the low math-anxious model, suggesting an under activation of working memory resources when experiencing conflict. Furthermore, the model was able to account for several other cognitive conditions, including reduced learning, the physical Stroop task across learning, and the speed-accuracy trade-off.

Date of Award	2022
Original language	English