First, the authors claim that it's agreed on that early vision "follows a mostly feedforward hierarchy." The impression I get is that this is far from agreed upon, that there are many feedback connections to V1 and the LGN, and that there's little agreement about how these feedback connections work.