Input-dependent directionality of interactions between cortical areas

Directionality between cortical areas in multi-area recurrent network models is determined by alignment between common inputs and recurrent connectivity, shaped by internal activity timescales.

• Why it matters: Cross-covariance studies show directionality can shift with behavior or task demands, but the circuit-level mechanisms producing these rapid changes are unclear.
• What they did: The authors built recurrent multi-area cortical circuit models and analyzed how directional interactions depend on fixed connectivity, common-input alignment, and internal timescales, including locally balanced excitation–inhibition architectures.
• The result: In locally balanced circuits, inputs to excitatory over inhibitory populations predominantly control directionality of latent shared signals, matching cross-covariances from primate V1 and V2 and suggesting mechanisms for observed shifts.

AI summary of the abstract · may contain errors · not medical advice · read the paper.