methodology - control as a special case of surprise

[Avery.Andrews 940723.1120]

PCT says that control is the thing to look for when figuring out how
living systems managed to stay alive, but I think there's a more
general formulation that might be more useful at the outset, which
is to look for `surprises': things that are true when you would expect
them to be otherwise. There are for example certain australian tadpoles
that grow up in rather shallow and impermanent water, and frequently
bask in the sun. As the water level slowly goes down after rain,
you would expect these creatures to frequently get stranded in local
depressions (noting that they're not very smart, and presumably have
no direct way of sensing the water level several meters away), but this
happens suprisingly rarely, providing a phenomenon worthy of
investigation (maybe somebody has already done it - at this point I
don't know).

PCT leads to the expectation that there is some perception they are
controlling that is keeping them of trouble (up to a point), but before
you can get motivated to try to find out what that perception is, you
have to judge that something worth understanding is going on, and I
think the `surprise' criterion provides a valid basis for this

In fact, `control' is a special case of the surprise criterion: when
you inject into a system a disturbance that you would expect to cause
some feature of the system to change, and it doesn't, that's a surprise.
But the generalized surprise criterion suggests things to investigate in
situations where one is far to ignorant to make a plausible proposal
about control. PCT also says that these surprises will almost always
have particular form of explanation, the existence of a controlled
perception, which is a reasonably clear, although somewhat abstract,
form of explanation