Neuroscience is in big trouble

Summary

Key Points

• Multiple recent studies from independent groups indicate that a substantial number of published neuroscience findings regarding the human brain and behavior are fundamentally flawed. 0:00
• A study in Nature Neuroscience found that about 40% of fMRI signals, which measure changes in blood flow, do not accurately reflect neuronal activity as previously understood, potentially rendering thousands of past studies misleading. 0:46
• This fMRI inaccuracy stems from the discovery that neurons can increase activity by consuming more oxygen without a corresponding increase in blood flow, challenging a core assumption of the technique. 1:00
• Another Nature Neuroscience paper revealed that lesion network mapping, used to identify brain systems responsible for symptoms in patients with brain damage, often identifies non-specific circuits that appear highly similar across various disorders. 1:51
• Further compounding the issue, functional brain network maps of individuals are not replicable when using fMRI scanners from different manufacturers, indicating significant inconsistencies in measurement. 3:03
• These methodological problems are not new, with similar issues identified over 15 years ago, yet many published studies continue to disregard known flaws in data collection, interpretation, and analysis. 4:06
• The current crisis in neuroscience is highly similar to the replication crisis experienced in psychology, where publication incentives led to the widespread use of flawed statistical techniques and non-replicable foundational studies. 4:44
• Strong incentives for scientists to publish papers and for scientific publishers to accept them contribute significantly to the perpetuation of these known methodological problems. 5:10
• While not all neuroscience is affected, these issues primarily impact specific fields focused on identifying particular brain regions or connections, potentially delaying the development of technologies like brain interfaces. 5:49