Research
Our laboratory investigates neural mechanisms of brain states and perception, and develops quantitative tools to measure and manipulate neural activity.
Research Overview
We study how brain states shape perception and computation across cortical and subcortical circuits. Our work combines conceptual theory, systems neuroscience, cellular mechanisms, and tool development.
Research Topics
Traditional deep learning and predictive coding models often assume a strictly hierarchical architecture. In contrast, neurobiological evidence suggests that cortical processing is tightly intertwined with direct communication to and from subcortical regions. Based on this view, we propose the shallow brain hypothesis: hierarchical cortical processing operates together with massively parallel subcortical networks, including thalamo-cortical loops, to support fast, flexible, and efficient computation.
Our recent work showed that general anesthesia decouples apical dendrites and somata in cortical layer 5 pyramidal neurons, disrupting feedback-related signaling while leaving basic neuronal function largely intact. Building on this finding, we investigate how dendritic and circuit dynamics support conscious perception and how these processes are selectively disrupted under anesthesia. Using advanced optical, genetic, and circuit-level approaches, we aim to reveal cellular and network mechanisms underlying conscious brain states.
Dendritic Integration Theory proposes that the apical dendrites of cortical layer 5 pyramidal neurons play a central role in integrating large-scale cortico-cortical and thalamo-cortical loops that are essential for conscious perception. A key strength of this theory is that it identifies a concrete cellular mechanism rather than relying only on abstract functional descriptions.
Our theoretical work generates experimentally testable predictions, and answering them requires new ways to observe and manipulate neural circuits. We therefore develop novel micro-optical and circuit-level tools that enable previously inaccessible experiments. We see tool development not as a side project, but as a core driver of conceptual and experimental discovery.