Causal role of human hippocampus in relational structure learning and inference
In the authors' words
Abstract When meeting new individuals or known individuals in new circumstances, we intuitively map out their relationships—not merely through direct experience, but by inferring new connections on the basis of prior relational knowledge. How such relational knowledge is represented and used in the brain remains a long-standing question in cognitive neuroscience. Here, using a behavioural task and computational modelling, we show that participants ( N = 36) leveraged existing knowledge of relational structures to infer new relationships, facilitating adaptation in changing environments. Whole-brain neuroimaging identified consistent representations of relational structure in the hippocampus. To investigate the causal link between hippocampal representations and inference, we used neural network simulations to predict the behavioural effects of disrupting such representations. Consistent with this prediction, transcranial ultrasound stimulation of the human hippocampus ( N = 20), transiently modulating its activity, disrupted structure-based inference ( P < 0.001, Cohen’s d = −0.89, 95% confidence interval −1.41 to −0.36), providing empirical evidence for a causal role of hippocampus in relational-structure learning and structure-based inference.
Appeared: Wednesday, November 18. Nature Human Behaviour. Peer-reviewed journal.