# Subtype-specific enhancement of implicit statistical learning in migraine: insights from BOLD signal variability

## TL;DR

The study finds that migraine patients show enhanced implicit learning linked to brain signal variability, suggesting a potential biomarker for understanding cognitive changes in migraine.

## Contribution

The study introduces a novel connection between BOLD signal variability and implicit statistical learning in migraine subtypes.

## Key findings

- Migraine without aura patients showed better statistical learning than healthy controls.
- Migraine patients had lower uptime in high variability BOLD states compared to controls.
- Learning indices correlated differently with BOLD variability states and network connectivity in migraine patients versus controls.

## Abstract

Migraine is associated with distinct cognitive alterations even during the interictal phase, yet the underlying mechanisms of implicit learning processes remain unexplored. Previous research has shown that the temporal variability of the blood-oxygen-level-dependent (BOLD) signal reliably predicts clinical symptoms in migraine. Building on these findings, the present study aims to investigate how implicit statistical learning processes relate to this resting-state functional imaging marker in episodic migraine. We hypothesized that implicit statistical learning would differ between migraine and non-migraine participants, and that these behavioral differences would be determined by group-specific patterns of resting state BOLD signal variability.

This study employed a cross-sectional, case-control design. A total of 28 migraine patients (14 with aura, 14 without aura) and 22 healthy controls were enrolled in this study. Resting-state functional images were acquired during the interictal phase, and all participants accomplished the Alternating Serial Reaction Time task. Statistical learning performance was compared across the three groups using mixed-design ANOVAs. BOLD variability states were calculated based on their time-varying measures, and non-parametric permutation tests were used to examine group-level differences in functional network connectivity within these states. Subsequently, we analyzed group differences in the descriptive metrics of BOLD variability states. Finally, correlation analyses were performed to investigate how learning indices were associated with BOLD state descriptor and functional connectivity strength in each group.

The migraine without aura group showed significantly better performance on statistical learning compared to healthy controls. Based on the clusterability, low and high variability BOLD states were identified. Lower uptime in high variability states was found in migraine patients compared to healthy controls (p < .05). Differing correlation strength between the learning indices and the temporal state descriptors was found across groups. In addition, varying functional network connectivity strength in the low variability state correlated with the learning indices differently in migraine patients and healthy controls.

Our results indicate that low variability BOLD states are associated with enhanced statistical learning in migraineurs. Furthermore, subtype-specific improvements in implicit pattern learning were observed, which may reflect adaptive network dynamics. These findings suggest that BOLD variability is a reliable functional imaging biomarker to better understand memory-related alterations in migraine.