 The study found that neural responses to natural scenes are organized into sequences, which can be decoded using either rate vectors or sequence order. These sequences are stimulus specific and can be maintained even when the absolute timing of the responses is changed. Furthermore, the study showed that the decoding performance of rate vectors is inferior compared to decoding based on sequence order. Finally, the study proposed that recurrent processing transforms signals from stationary visual scenes into sequential responses, which are then decoded using a Bayesian matching operation. This article was authored by Yang Yiling, Catherine Shapcott, Alina Peter, and others.