They're arguing that we have no evidence that humans have access to our underlying thoughts any more than the models do.