Model-Based Control in _Canis familiaris_ (shaggy dog story)

[Avery Andrews 950511]

Our kids had a friend called Alex, who had a dog called Patch, who liked
to be near where Alex was. One day Alex was playing in our house,
so Patch was sitting out on the porch. At some point Alex went off,
but Patch didn't notice (we presume), and, as the day begin to end,
Patch was still sitting on our porch. So we thought that maybe he
thought that Alex was in the house, so we let him in; he did a quick
tour of all the rooms (small house), and then headed to and out the door,
and then across the park towards his home.

So my conjecture would be that while sitting on the porch, Patch had a
world-model wherein Alex was located inside our house, so he was
controlling his perception of being near model-Alex. One thing that
differentiates a model-X perception from a real X perception is that
you don't settle for model-X when you can get real-X, so when we opened the
door and invited him in, he tried to satisfy his reference for being
near real-Alex. Another property of model-X perceptions is that they are
defeasible, and when shown to be false, they cease to be controlled for,
so when his house-tour revealed the falsity of his model, off he went.

Some repeatable experiments on this theme would certainly be called for,
but it still doesn't seem too outlandish for creatures who hunt
socially, and deal with things, including each other, that go in and
out of holes and burrows.

<[Bill Leach 950511.13:00 U.S. Eastern Time Zone]

[Avery Andrews 950511]

What is so impressive to me is that you "caught" with a story much of the
essence of what I was trying to say in my previous posting.

It is not that such "doesn't seem too outlandish for ..." as I believe
that such is not an issue. We actually see Patch do these things as you
described and indeed we ourselves do also.

That such might be irrational or even that great "turmoil" might occur
when we discover that we were "only controlling an illusion" is
irrelevent. What is relevent is that the behaviour occurs and the
presence of a model somehow satisfying a reference "inspite of reality"
would explain (at least in part) how a control system could do such

I would further suggest that such is not really an "extension" to HPCT
but rather only a single view of how the imagination connection might
function within the perceptual control system.