The Default-Mode Network

Through the 1990s, an embarrassing anomaly kept showing up in PET and fMRI studies across many cognitive-neuroscience labs. Whenever subjects engaged a focused task, a specific cluster of brain regions reliably deactivated relative to baseline; whenever subjects rested in the scanner with no task, those same regions activated. The pattern was so consistent that it was usually treated as noise — until Marcus Raichle named it in 2001 the default-mode network. The brain at rest was not a quiet brain. It was intensely active, generating self-referential thought, autobiographical memory, prospection, the inner monologue itself. The single most consequential finding in cognitive neuroscience that nobody had been looking for had been hiding in plain sight, in the discarded baselines of every task-based imaging study, for two decades.

The default-mode network is a small set of cortical hubs — medial prefrontal cortex, posterior cingulate and precuneus, the bilateral angular gyri, and medial temporal structures including the hippocampus — whose low-frequency BOLD fluctuations correlate strongly at rest and decouple again when attention turns outward. It is one of several large-scale resting-state networks identified in the 2000s; what made the DMN distinctive was that its functional contents proved to be unusually unified. Self-referential thought, autobiographical memory, theory of mind, moral reasoning, prospection, mind-wandering, mental simulation — all the things one does when not doing anything in particular — engage the same circuit. The constructive-simulation hypothesis articulated by Schacter and Addis takes the unification one step further: the DMN is the brain's internal-scenario generator, assembling pasts, futures, hypotheticals, and other minds with the same underlying machinery. On that reading, the DMN is the substrate of the self-as-story that humans seem to be.

The clinical signature reinforces the picture. Alzheimer's disease attacks DMN regions early and severely, with connectivity changes detectable years before symptoms and an amyloid-deposition pattern that overlaps the network's topography almost exactly. Depression presents the opposite imbalance — DMN hyperactivity and reduced suppression during external tasks, the neural form of rumination. Schizophrenia and autism each show their own characteristic DMN abnormalities, mapping plausibly onto the disordered self-experience and atypical theory-of-mind processing each disorder involves. The most striking finding has come from psychedelics: Robin Carhart-Harris and colleagues showed in 2012 that psilocybin acutely suppresses DMN connectivity in a dose-dependent way, and that the depth of ego dissolution subjects report correlates with the depth of that suppression. Meditation in experienced practitioners produces a milder version of the same effect. One network, four or five very different ways to perturb it, and a remarkably coherent map between the perturbation and the phenomenology of self.