部門公開セミナー

Cognitive processes have been attributed to interactions between distributed brain areas coordinated by large-scale neuronal networks, termed neurocognitive networks. However, these processes are typically simultaneous and interdependent, and how their underlying neural structures are organized during behavior remains unknown. We hypothesize that the neurocognitive network consists of structured subcomponents that are regulated by discrete brain-wide activity patterns, and aim to characterize the spatiotemporal dynamics of these mesoscopic network structures. We use an electrocorticography array covering a cortical hemisphere in macaque to collect high-fidelity broadband neuronal signals during various cognitive behaviors, ranging from visual perception to social cognition, and use multivariate connectivity analyses to quantitate the content and directionality of information transfer within the network. With an unbiased deconvolution analysis, we show that the neurocognitive network consists of a battery of latent mesoscopic components, each a cognitive fingerprint of spatio- spectro-temporal dynamics discharging each other to propel a cognitive chain. For example, we show that a bottom-up network could coordinate multiple remote cortical areas to encode the content of a specific social context, and that later retrieval of the same context could be reactivated by bottom-up sensory signals, allowing top-down signals to align with ongoing sensory perception. Our findings show how brain networks remodel at each step to process cognitive information, and reveal how apparently seamless cognition is constructed from a dynamic chain of network communication modules. The technology we describe may be used for organizing whole brain function into constituent network patterns, including cognitive impairments in diseases of network coordination.