|
[DPRG] Will's emotions
Subject: [DPRG] Will's emotions
From: Randy M. Dumse
rmd at newmicros.com
Date: Fri Mar 2 15:33:07 CST 2007
> The idea as I understand it is that "reactive" behaviors are
> associated closely in time, ...
>
> The level above the reactive level, which Will would like to
> call the "emotional" level, controls the weighting and
> priority of the reactive behaviors based on long term
> accumulation of inputs/states/whatever. ...
Took me a bit to revive my memories on this, but this sounds
very much like a follow on to a previous RBNO conversation I
really enjoyed.
Glad to see the use of "inputs/states/whatever" because my
favorite word is the middle one, and of course the only one that
matters. ;)
I was also very amused to see S. Rainwaters reply so full of the
word "state" as well.
So let me see if I can restate my last opinion where I left it.
All animals have emotions down to a very low level. To me
emotions are to most animals what thought is to humans. That is,
having emotions is a very important part of survival strategy.
Let me tell a joke I really liked I heard in my youth. A man was
walking along through a cemetry late at night, because it was a
short cut to his house. Unknown to him, in his path was a newly
dug empty gravesite. He fell in, and for all his jumping and
scrabbling, he could not get out. So he curled up in a corner
and slept, to await morning, and his assumed rescue. A bit
later, another man came along and fell in. The commotion woke
the first man who watched the newcomer scramble at the walls as
he did. He chuckled to himself, knowing the newcomer would have
no better luck than himself. Finally he said, "You might as well
face it, you're staying down here with me. You aren't getting
out of this grave...and just then, hearing "the devil" call to
him from down in the earth, he did.
I think emotions are meta-states we enter which modify the lower
processing machines. For instance, when we are
frightened/terrorized as in the joke, we the state of being
terrorized puts the heart rate up to full, releases adrenilin,
and prepares our muscles for maximum output for flight. This
emotional state will last a short while, depending on resolution
or continuing stimuli causing retriggering.
Likewise, too many bumps of our toes will cause a change of
emotional state, which causes us to try different strategies,
because we are annoyed.
The same should be true of our robot design, like in canyoning.
Too many bumps in too short a time, and a higher state is
triggered that modifies our behavior, to produce an escape
behavior. The escape behavior may be as simple as always turning
the same way until free. But rather than being a part of the
lower level response, I think it is a different response, and it
is enabled/disabled by the emotional state in control.
This has huge survival value. Without the concept of internal
state storage (as I've previously stated - even if often hidden
rather than explicit) our robots will keep doing the same dumb
behavior over and over again. Any robot that can choose
alternative sets of behavior looks smarter than any robot that
cannot.
Randy
www.newmicros.com
More information about the DPRG mailing list
|
