Reciprocal V1–LM wiring crafts cross-area consensus via line attractor dynamics
In mice, researchers recorded activity from V1 and LM during visual processing and used an iLQR-VAE–based latent circuit model to reveal that reciprocal excitatory connections between V1 and LM implement a line attractor, selectively slowing congruent activity decay while accelerating decay of incongruent patterns to dynamically align activity across areas. The resulting distinct timescales for consistent vs. inconsistent patterns support dynamic consensus as a general cortical computation principle.