Towards a “Treadmill Test” for Cognition: Improved Prediction of General Cognitive Ability from the Task Activated Brain
2bkvs0bkTop3Predictive scene contains the data shown in Figure 3. WMConsensus, SocialConsensus, RelationalConsensus, GamblingConsensus, EmotionLangMotorConsensus, and MultiTaskConsensus contains the data shown in Figure 4 and Supplemental Figure 2.
Identifying brain-based markers of general cognitive ability, i.e., general intelligence, has been a longstanding goal of cognitive and clinical neuroscience. Previous studies focused on relatively static, enduring features of the brain such as gray matter volume and white matter structure. Task-based imaging, though far less studied, is promising because tasks can potentially selectively activate the brain regions responsible for intelligent cognitive performance. In this report, we investigate prediction of intelligence based on fMRI task activation patterns during the N-back working memory task as well as six other tasks in the Human Connectome Project data set (n=967), encompassing 15 task contrasts in total. We find that whole brain task activation patterns are a highly effective basis for prediction of intelligence. The 2-back versus 0-back contrast achieved a 0.52 correlation with intelligence scores in ten-fold cross-validation analysis, while a model trained on all 15 task contrasts achieved a 0.57 cross-validated correlation. Additionally, we found that task contrasts that produce greater fronto-parietal activation and default mode network deactivation—a brain activation pattern associated with executive processing and higher cognitive demand—are more effective in intelligence prediction. These results suggest a picture analogous to treadmill testing for cardiac function: Placing the brain in a more cognitively demanding task state significantly improves brain-based prediction of intelligence.
Human Brain Mapping - DOI: 10.1002/hbm.25007
- Chandra Sripada
- Mike Angstadt
- Saige Rutherford
- Aman Taxali
- Kerby Sheddon
- University of Michigan