We also found two distinct early-visual populations. Suppression populations carried denser cross-modal semantic information and supported finer semantic readout; enhancement populations were more…