Efficient coding in working memory is adapted to the structure of the environment

Working memory (WM) relies on efficient coding strategies to overcome its limited capacity, yet how the brain adaptively organizes WM representations to maximize coding efficiency based on environmental structure remains largely unknown. In our study, participants remembered a sequence of gratings defined in a two-dimensional feature space where we manipulated directional consistency, revealing enhanced performance for structured (consistent direction) vs non-strutured (non-consistent direction) contexts, particularly for individuals with lower WM capacity. Magnetoencephalography analyses uncovered dissociable neural bases: consistent sequences engaged anterior temporal and medial frontal cortices for abstract directional representations during maintenance, while inconsistent sequences preferentially reactivated item-specific representations in parietal regions. These neural patterns predicted behavioral performance, establishing a neural efficiency principle wherein the brain adaptively switches between relational and item-based coding strategies, mitigating WM constraints. These findings advance our understanding of how structures shape WM organization, offering insights into cognitive flexibility and neural resource allocation in complex environments.