partititioning into f's and d's

[From Bill Powers (960206.1430 MST)]

I'll be off on a quick trip East for my father's birthday, starting
Thursday Feb. 8 and returning Monday, Feb. 12. He'll be 96.

···

-----------------------------------------------------------------------
Martin Taylor 960206 12:20 --

     In your usual description of a perfaction loop--a diagram--you show
     one CEV that is affected by one disturbance input and one influence
     from the output function of the perfactor. You recognize verbally
     that there are many _physical_ influences that contribute, but
     nevertheless the CEV has to be scalar, because the perceptual
     variable derived from the function that defines the CEV is scalar.

Yes, you're right about that.

If we have a perceptual variable p that is an input- function i of some
set of environmental quantities x1 ..xn, we have p = i(x1..xn). Out of
the n variables, m are affected by action, so the environmental feedback
function works through variables x[1].. x[n-m], and the variables x[m]
..x[n] are the aggregate disturbance. On this much, we agree.

My preference is to characterize the perceptual input function in terms
of _all_ the input variables that affect it, leaving the distinction
between feedback variables and disturbances to be determined when the
effects of action on the environment are known. Two perfaction systems
having identical input functions with identical sets of x's as arguments
may use different actions to perfact the perception, which would change
the partitioning of the x's into f's and d's.

Your point, made several posts ago, that the effective environment has
to be defined relative to the perfacting system in question is still
valid. I merely point out that the output connections also have to be
taken into account. Two systems perfacting the same CEV are not
necessarily doing so through the same environmental variables, and what
is a disturbance to one system is not necessarily a disturbance to the
other.
-----------------------------------------------------------------------
Best,

Bill P.

[Martin Taylor 960206 15:45]

Bill Powers (960206.1430 MST)

I merely point out that the output connections also have to be
taken into account. Two systems perfacting the same CEV are not
necessarily doing so through the same environmental variables, and what
is a disturbance to one system is not necessarily a disturbance to the
other.

I think we are, for once, in complete agreement. That's what I was trying
to get across through my f-d notation. The f's represent the output
connections of the perfacting system being analyzed, the d's disturbances
to that system. In another perfaction system with the same CEV, it could
be that all the f's and d's are interchanged--each system's output connections
being only disturbances to the other. Come to think of it, that's the normal
case, isn't it, if we let some of the d's be d's for both systems?

My preference is to characterize the perceptual input function in terms
of _all_ the input variables that affect it, leaving the distinction
between feedback variables and disturbances to be determined when the
effects of action on the environment are known.

Yes, if you haven't characterized the output connections, but have
characterized the CEV, you _must_ do it that way. And if you are treating
a reorganizing problem in which new output connections may change some
d's into f's (or vice-versa), you probably want to start by treating all
the variables symmetrically.

I think that how you notate it depends on which aspects you want to emphasize,
the unity of the function or the split between the variables the perfacting
system can and cannot influence.

···

---------------

I'm coming to like "perfaction," I'm sorry to say. The notion of "per" =
"by means of" or "through", the pun with "perfection", the "action"
and the "faction" both have useful imagery, and it's not too long. It
seems to have the connotation both of bringing something "to perfection"
and of acting in a "perfecting" way, so it has something of both "sponte-"
and "retro-" faction.

I'd prefer to stick with "control", but if a new word is necessary, I'd
not be unhappy to use "perfaction."

---------------
Happy birthday, Father Powers. May we wish his son an equally long and
hopefully healthful life.

Martin