Hierarchy and feedback in network API

Hi everybody! On Subutai´s video of Network API explanation it was shown this visualization which I found very interesting and I had some questions to do.

So for the first scheme it is representing a hierarchy between 2 cortical layers and I was wondering which meaning the output of the second TM have… You get active and depolarized cells from the input of the second SP which is already a neural representation of the image sensor; mind-blowing

The second one I do believe it is a biological representation of how different sensory organs are related in the brain but if we speak in terms of nupic applications, data relation is done in the encoder region where semantic meaning can be related between different input. Am I right?

And in the third one I don´t see how feedback of a second layer output can be introduced with first layer output as the input of the second layer… I mean, feedback is already done internally when the active cell state is updated in the TM.

Any opinion is welcome!