The paper proposes a hierarchical Bayesian inference framework to explain the processing of visual information in the early visual cortex. Traditional views suggest that early visual neurons in areas V1 and V2 extract local features from the visual scene, which are then processed through a feedforward chain of modules. However, recent electrophysiological recordings from awake behaving monkeys reveal that the information processing in the early visual cortex is more complex and interactive. The authors propose that the activity in the early visual cortex is tightly coupled and highly interactive with the rest of the visual system, suggesting a new theoretical setting based on hierarchical Bayesian inference. This framework integrates top-down contextual priors and bottom-up observations through recurrent feedforward/feedback loops to implement concurrent probabilistic inference along the visual hierarchy. The algorithms of particle filtering and Bayesian-belief propagation are suggested to model these interactive cortical computations. The paper reviews neurophysiological evidence supporting these ideas, including timing studies, scale of analysis, and experiments on contour completion and shape-from-shading. These findings suggest that V1 plays a crucial role as a high-resolution buffer for geometric calculations, and that the early visual areas participate in all levels of perceptual computations, interacting concurrently and continuously with higher-level areas.The paper proposes a hierarchical Bayesian inference framework to explain the processing of visual information in the early visual cortex. Traditional views suggest that early visual neurons in areas V1 and V2 extract local features from the visual scene, which are then processed through a feedforward chain of modules. However, recent electrophysiological recordings from awake behaving monkeys reveal that the information processing in the early visual cortex is more complex and interactive. The authors propose that the activity in the early visual cortex is tightly coupled and highly interactive with the rest of the visual system, suggesting a new theoretical setting based on hierarchical Bayesian inference. This framework integrates top-down contextual priors and bottom-up observations through recurrent feedforward/feedback loops to implement concurrent probabilistic inference along the visual hierarchy. The algorithms of particle filtering and Bayesian-belief propagation are suggested to model these interactive cortical computations. The paper reviews neurophysiological evidence supporting these ideas, including timing studies, scale of analysis, and experiments on contour completion and shape-from-shading. These findings suggest that V1 plays a crucial role as a high-resolution buffer for geometric calculations, and that the early visual areas participate in all levels of perceptual computations, interacting concurrently and continuously with higher-level areas.