Task-invariant networks interfere with and task-specific networks support memory formation: An fMRI meta-analysis

Why do some moments imprint themselves in memory while others vanish without a trace? This meta-analysis identifies a dissociation in large-scale brain networks during encoding: networks associated with impairing encoding are task-invariant, whereas those supporting it are task-specific. Drawing on 56 fMRI studies employing the subsequent memory paradigm, the analysis contrasted neural activity for later-remembered versus later-forgotten trials across verbal and pictorial tasks. Using Yeo et al.’s 17-network parcellation, the results show that encoding-impairing effects consistently recruit specific subsystems within the default mode, frontoparietal, and ventral attention networks—a pattern consistent with distraction or mind-wandering. Conversely, encoding-supporting effects diverge by task: verbal encoding engages language-related networks, whereas pictorial encoding activates visuo-perceptual systems. This asymmetry suggests that encoding failure may arise from similar attentional lapses across contexts, whereas successful encoding requires precise, context-sensitive neural engagement. Taken together, these findings provide a network-level perspective on how the brain shifts between states conducive to remembering and states conducive to forgetting.