[MUD-Dev2] [DESIGN] Rewards

[David Love]

On Mon, 2007-04-16 at 10:42 -0400, John Buehler wrote:

> > "John Buehler" <johnbue at msn.com> wrote:
> 
> As with all participants in Word Wars here, we have distinct contexts in
> which we operate, and the way that various words are wired into our brains
> differ.  As a result, Word Wars invariably just go round and round in
> circles.  There's just no way to distill concepts to anything more than
> references to other concepts - which are themselves dissimilar.  Because of
> that, I pass on trying to debate what the word "reward" should mean.

I've been catching up on old email, so my apologies if I'm retreading
ground that's covered a bit later in the thread.

Since operant conditioning (and the psychology of game design) have
already been brought up, here's a few more definitions (I just yanked
them from Wikipedia - they're pretty accurate, though you'd fine them
much more precisely defined in a psych text, assuming it was a good
one :) ):

      * Positive reinforcement is an increase in the likelihood of a
        behavior due to the addition of a reinforcer after a behavior.
        Giving (or adding) food to a dog for sitting is an example of
        positive reinforcement (if this increases the likelihood of the
        dog sitting in the future). The behavior of taking illicit drugs
        may be reinforced by the feelings of euphoria produced by the
        drug.

      * Negative reinforcement is an increase in the likelihood of a
        behavior when the consequence is the removal of an aversive
        stimulus. Turning off (or removing) a shock when a rat presses a
        bar is an example of negative reinforcement (if this increases
        the likelihood of the rat pressing the bar in the future).
        Scratching an itch is another example of negative reinforcement.

      * Positive punishment changes the surroundings by adding an
        aversive stimulus following a behaviour in order to decrease the
        likelihood of the behaviour occurring in the future. An example
        would be spanking a child for making a mess.

      * Negative punishment changes the surroundings by removing a
        stimulus that is a reinforcer. An example would be taking candy
        away from a child whenever the child behaves inappropriately.

End of Wikipedia quoting.

Both intrinsic and extrinsic rewards are a form of positive reinforcer
(though you can make a neurological argument that what intrinsic reward
really is involves the release of whatever "neurotransmitter of the
week" is currently the cause of/solution to all of life's problems -
this also helps explain the "intrinsic reward is better than extrinsic
reward" evidence).

Food for thought, especially when considering the "framing" issues
brought up before.  Positive reinforcement is not the same as negative
punishment - this can be shown both qualitatively and quantitatively.

Good game designers know this (either experientially, or from study) as
well as knowing about variable reinforcement schedules (I forget who
brought that up, but thanks) - that's why uber l33t loot doesn't always
drop.  The game is far more addictive with the variable reinforcement.

As has been mentioned before, things become disruptive when the behavior
the designer thinks that they are reinforcing is different than the
behavior they are actually reinforcing (or punishing).  And that's not
just with humans.  Talk to any animal trainer and they'll tell you that
they run into the same problem.  It's not easy to get it right (and that
may very well be what separates a good game from a great game).

-dlove