The deep brain shrinkage blamed on depression failed to show up in 23,417 scans

For decades the standard account of depression's footprint in the brain has included smaller subcortical structures and abnormal functional connectivity in frontal and default mode networks. Yet recent meta-analyses have failed to find consistent, converging brain correlates of depression across the published literature, leaving the field without a conclusive map. That gap matters: if the imaging findings underpinning theories of depression do not replicate, the theories built on them wobble too. Settling the question requires samples far larger than any single study has managed.

The researchers pooled data from 23,417 participants drawn from 6 large population datasets, aiming to establish the neuroimaging correlates of depression comprehensively rather than within a single cohort. They related depression to structural brain measures, including grey matter volume and cortical surface area, across regions of the brain. Working across independent datasets allowed them to test whether any given correlate held up consistently rather than appearing in one sample and vanishing in the next. The published paper updates an earlier preprint version of the same analysis.

Depression was associated with reductions in grey matter volume and cortical surface area in the frontal cortex, anterior cingulate and insula, confirming earlier work pointing to prefrontal and default mode regions. The surprises came elsewhere. Subcortical brain regions, long reported to be smaller in depression, showed no significant depression correlates at all. Meanwhile, significant correlates did appear in somatomotor and visual regions, areas that rarely feature in standard accounts of the disorder.

The results partly confirm and partly overturn the received picture. The frontal, cingulate and insula findings support the emphasis prior studies placed on prefrontal and default mode regions. The absence of subcortical correlates, however, challenges a large body of earlier literature, and it chimes with the recent meta-analyses that could not find converging effects across studies. Open questions remain about why somatomotor and visual regions carry depression correlates, and whether these structural differences are causes, consequences or bystanders of the condition.

These are associations observed across populations, so nothing here shows that depression damages the brain or that brain structure causes depression. If you have encountered the claim that depression shrinks deep brain structures, this analysis of 23,417 people is a reason to hold that claim more loosely. It is equally a reminder that depression does have measurable brain correlates, concentrated in frontal regions, so the biology is real even where the older story looks wrong. For anyone following mental health science, this is what a field correcting itself looks like.