controlling constructed perceptions (fwd)

I sent this message to Gary Cziko and Tom Bourbon and meant to send it to
Bill P but didn't have the specific address so I'm sending it to the net.
Please send replies to me directly.

I've been thinking lately about the idea that not only do we Construct
reality out of sense data (something lots of people don't understand) but
that we even Construct the world "without" the sense data and use the
sense data only to constrain the constructions. Such an idea accounts,
for instance, for release hallucinations in which people see objects in
blind spots in their visual field that exist as a result of cortical
damage. The idea here is that objects are constructed and only eliminated
if their is conflicting info (which there isn't). There is alot to say
about this but its not really the thing I want to get at here....What
seems interesting, is how this seems to change how we make our PCT models.
Actually, I can't tell if it actually does change the models or
not--maybe such a detail is taken care of in the Input function. Is this
the case or does such an idea require some redrawing?

Mark Olson
Neuroscience Program
University of Illinois