Neuroscience is in big trouble

Daily Science News with Sabine

Neuroscientists have long assumed that when a brain region becomes more active, more blood flows to it. That assumption is the entire basis of functional magnetic resonance imaging. A new paper now says it’s wrong for about 40% of the signals, and it isn’t the only recent study to find that a lot of what we thought we knew about the brain rests on bad measurements. Let’s take a look.

---

Per Lanterna:

Sabine Hossenfelder’s short video “Neuroscience is in big trouble” argues that core assumptions and methods in human brain imaging and network mapping are seriously flawed, with recent papers showing large fractions of published findings may be unreliable or non-specific.

In the roughly 7.5-minute piece, she highlights a 2025–2026 Nature Neuroscience study finding that ~40% of fMRI (BOLD) signals do not track the assumed link between blood-flow increases and higher neuronal oxygen use/activity—especially in the default mode network, where neurons can ramp up by extracting more oxygen without extra flow—potentially undermining decades of results. She also covers a lesion-network-mapping critique showing many supposedly disorder-specific circuits look similar across very different conditions, plus evidence that different scanner manufacturers produce non-replicable individual brain-network maps, and frames the persistence of these issues as driven by publish-or-perish incentives (paralleling psychology’s replication crisis) rather than a lack of critical voices.