Abstract: Math anxiety has been consistently shown to impair performance in numerical and mathematical tasks, but its impact on specific cognitive stages—particularly the visual perception of numerical symbols and the processing of numerical magnitude—remains poorly understood. The present study investigated the neural mechanisms underlying symbolic number processing deficits in individuals with high math anxiety (HMA) using event-related potentials (ERP). A hierarchical task design was employed to dissociate distinct levels of numerical processing, including a visual baseline task (non-digit color judgment), a digit color judgment task, and a numerical magnitude judgment task. This design allowed for the separation of general visual processing, symbolic digit recognition, and abstract numerical magnitude processing.
Fifty-eight university students were selected from a larger screening sample based on their scores on the Shortened Mathematics Anxiety Rating Scale (sMARS), forming high math anxiety (HMA, n = 29) and low math anxiety (LMA, n = 29) groups. Participants completed the three experimental tasks while EEG was recorded. ERP analyses focused on the anticipatory P3 component elicited by task cues, the N170 component associated with early visual processing of digits, and the P2P component related to numerical magnitude processing.
Results showed that individuals with HMA exhibited significantly reduced N170 amplitudes during digit processing, suggesting weakened neural responses during early visual recognition of symbolic numbers. In contrast, the HMA group showed increased P2P amplitudes during numerical magnitude processing, indicating altered or less efficient neural processing of quantity representations. Although behavioral performance did not reveal robust group differences—likely due to task simplicity and ceiling effects—neural measures clearly differentiated the groups.
Multivariate pattern analysis (MVPA) further demonstrated that individuals with low math anxiety could distinguish between digit and non-digit stimuli earlier and with higher decoding accuracy than those with high math anxiety. Specifically, the LMA group showed earlier and more reliable neural decoding of digit-related information, whereas the HMA group exhibited delayed and less accurate classification, indicating reduced efficiency in symbolic number discrimination.
Importantly, mediation analyses revealed that the anticipatory P3 amplitude fully mediated the relationship between math anxiety and both N170 and P2P amplitudes in the digit color judgment task. This finding suggests that deficits in anticipatory cognitive-emotional control contribute critically to downstream impairments in symbolic number processing. Source localization using sLORETA traced the anticipatory P3 activity to the inferior parietal region (Brodmann area 40), a key node within the frontoparietal control network. Reduced activity in this region may reflect impaired emotional regulation and cognitive control during the anticipation of numerical stimuli in individuals with high math anxiety.
Taken together, these findings suggest that symbolic number processing deficits in individuals with high math anxiety arise from a multi-level disruption that begins during anticipatory preparation and propagates through early visual recognition to later numerical magnitude processing. These results highlight the role of anticipatory emotional and cognitive control mechanisms in shaping numerical cognition and provide new neurocognitive evidence for understanding and potentially intervening in math anxiety.

Key words: mathematical anxiety, symbolic number, ERP, N170, MVPA