Scale-free alpha dynamics bias intrinsic over extrinsic values in free choice

Scale-free temporal structure in resting-state intrinsic neural activity has been proposed to support the interaction between internal and external cognition, but its role in decision making remains unclear. We examined whether individual differences in the dynamics of resting-state alpha activity are related to how much people rely on externally reinforced options versus internally generated preferences during free choice. Participants first performed an externally guided reward-learning task, in which they learned through feedback which abstract shapes tended to yield rewards. They then completed an internally guided preference-judgment task, repeatedly choosing between these previously rewarded shapes and new shapes without feedback. Resting-state electroencephalography was recorded before, between, and after the two tasks. A computational model provided trial-by-trial estimates of stimulus value in preference-judgment task, and long-range temporal correlations in the alpha-band amplitude envelope were quantified using detrended fluctuation analysis. We replicated the superiority of intrinsically learned value: the most preferred new shape acquired a higher final value than the previously rewarded shapes, despite their higher initial values derived from external reinforcement. Stronger scale-free temporal correlations in resting-state alpha activity were associated with a larger superiority of intrinsically learned value, primarily reflecting lower final values for previously rewarded stimuli and relatively higher values distributed across multiple intrinsically learned options. These findings suggest that scale-free temporal dynamics in resting-state alpha activity reflect an intrinsic neural bias that favours internally generated preferences over extrinsically reinforced values in free choice.