News:

Forum changes: Editing of posts has been turned off until further notice.

Main Menu

Operant Conditioning & RPG System

Started by Zak Arntson, January 23, 2002, 07:17:53 AM

Previous topic - Next topic

Zak Arntson

Okay, my wife has a degree in Psych, so I've got a layman's view of the field. I've often thought about how gambling psychology plays into video games: The most addictive offer an intermittent reward system.

Concrete Example: In Castlevania Symphony of the Night, killed monsters occasionally toss out items for the Player. This is random, but occurs enough to keep me playing for hours, just to see what weird stuff I can get.

How does this tie into roleplaying? An rpg's System can reinforce certain play behaviors. So, my ignorant view of conditioning:

Positive Reinforcement - Giving something good to reward.
Negative Reinforcement - Taking something bad away to reward.
Positive Punishment - Giving something bad to punish.
Negative Punishment - Taking away something good to punish.

I may be wrong here, but the meat of it is:

To reinforce behaviors, a System can Give or Take Away something Good or Bad. Examples:

Real World: Slot Machines, where people lose more money than they get out, because at times they win a big bunch of coins. So here the Negative Punishment (removing money) is far outweighed by the prospect of Positive Reinforcement (winning money)

D&D's Experience Point system is most often Positive Reinforcement because it is constantly increasing (with intermittent bursts of levelling-up) with good results. The Negative Punishment aspect (level-draining) absolutely terrifies PCs, but in a purely Gamist sense.

D&D's Hit Point system is a Negative Punishment because you lose points that keep your PC alive.

---

So how does this translate into Game Design? Well, any reward/punishment mechanics you have in your System should take into account the benefits and qualities of the studied psychological affects.

I'm asking any Psych folks out there (I'll talk to my wife as well) what they know about the different conditioning methods. And to correct my inevitable misinformation! All I know beyond what I've written above is that my wife tells me Positive Reinforcement is typically the most effective.

I do know there's something to tying Conditioning into a System! I'll probably combine this with my Unknown Factor thoughts and produce some neat Zak-method of game design.

Zak Arntson

I know I'm replying to my own post. Yikes! Anyhow, I was looking http://chiron.valdosta.edu/whuitt/col/behsys/operant.html">here at an explanation of Operant Conditioning, and there were four questions that sound like good ones when determining a reward/punishment mechanic:


1. What is the behavior you want to change?
2. Will the behavior be strengthened or weakened?
3. What is the consequence?
4. Is the consequence added or subtracted?


So here's a few rpg mechanic examples:

1. Behavior to change: Attention payed to Relationships.
2. Strengthened/Weakened: Strengthened
3. Consequence: Control of PC's fate
4. Added/Subtracted: Added

This would be Positive Reinforcement, and look! I'm concretely outlining a System (and gameplay) goal!

1. Behavior to change: Killing Monsters.
2. Strengthened/Weakened: Weakened
3. Consequence: Empathy & Humanity
4. Added/Subtracted: Added.

I've concretely outlined by goal to keep PCs from killing monsters by enhancing their Empathy!

---

I know this might seem like a "duh," since most of us design games with thoughts like, "How can I support XYZ." I do think it's important to work through these things and figure out why we design the way we do.

hardcoremoose

Zak,

Check out http://www.indie-rpgs.com/forum/viewtopic.php?t=349&highlight=reward+systems+making+players+behave">this thread, where I sort of started hinting at this sort of thing.  It may be of some interest to you.

From a game designer's stand point, understanding why players do what they do, and then finding ways to encourage and/or discourage that behavior, seems pretty important to me.  

- Moose

Zak Arntson

Good thread! Only it seems to discuss fixing dysfunctional groups more than getting to my point. Which is, given a functional group, game design can be helped with a basic understanding of psychological conditioning.

There's that great mentioning of Group Psychology in that thread. As in "fear of failure." So I wonder if, on an individual level Positive Reinforcement works best. But perhaps as a group, fear of Positive Punishment/Negative Reinforcement can encourage individual actions?

Rather than exploring dysfunctional gaming, I'd like for this thread to stay on the assumption of a functional group and run from there, with an emphasis on the merits and flaws of the different Operant Conditioning methods.

I also think that exploring the Group vs. Individual results you get from different methods is very intriguing, since most rpgs involve two realms: Individual (Player concentrating on her PC) and Group (Player working with Players and PC working with PC).

There's also the GM vs. Player aspect. It'd be interesting to see thoughts on Conditioning mechanics applied to the GM, rather than the common "GM is god" approach.

Bankuei

This was something I immediately picked up about the mechanics in the Pool... I thought it was interesting that your character could NOT grow unless you chose to gamble to get more dice.  Awesome reinforcement there.

The random payoff factor is the best part about the D&D random charts, in that players are praying that they'll get lucky and get the +5 holy vorpal sword of wishes.   It's also the same logic behind the random stat rolls and any gambling mechanic.

As far as personally, two things I'm working on where the reward system/psychology factor plays a massive role.  

First is the Persona/Forgotten Fist game, where the only reward is story points, which you get only for affirming or revising your character's beliefs, not for succeeding, failing or just showing up.  These points give you control over the story, but not for improving your character.  Likewise, death/removal from play is also in your hands, so the usual reinforcement of death/powermongering for the sake of avoiding death is removed.  Story points also stay with you even if you decide to change characters.  In other words, I took out anything that would reinforce anything other than telling a story.  All the reward is based on, and reinforces the idea of telling the story.

The second thing, is influenced by a variety of things, D&D, Exalted, Final Fantasy 10, and the LOTR system idea.  First, I realized that any game that gives me a preview of my character's growth is like a carrot with a stick on it, leaving my inner gamist adrool.  It's a perfect reinforcement if I've ever saw one.  D&D and Exalted do this with classes or Charm paths.  

Final Fantasy 10 has a massive skill map which really is just a giant class flowchart, but it looks cool, and most importantly highlights the tough choices gamists love, like do I specialize, focus more on this, or that?  I was trying to figure out what it was about most point based systems that left me feeling unsatisfied, and it was that there was no reinforcement for you to not specialize and max your character in what you want.  when you take a class, or a skill/power path, you're forced to take some abilities you don't care about, and you also have to make choices based on investment(should I single class or multiclass?)

So I've been looking at the LOTR system and building a hack/slash & blast system with tons of extra abilities that I want to fit into a massive flowchart of fun for pure gamist entertainment.  Right now I've got somewhere around 50 abilities written and I'm trying to figure out how to set up the paths for maximum reinforcement and balance... arggh!  I wonder if there's some sort of sick reinforcement that keeps people like us designing these games :P

Chris

J B Bell

OK, this is from my Psych 102 class, many years ago, so I may be using the Nether Mouth on this one, but:

One crucial consideration in reward/punishment systems is scheduling.  That is, how often and how consitently is the stimulus generated?

What makes gambling addictive is *not* that the payout is given a higher value by the player than the far more consistent (and measurable) "negative punishment" of losing money.  It is that the payout is at random intervals.

This is nicely exemplified in training my recalcitrant cats to, say, not jump all over the kitchen when I'm trying to prepare their dinner.  I used to just squirt them.  This made them fear the God Bottle.  If they observed (and they are sharp observers) that I didn't have it handy, why, they ignored me.

Next (in my training and theirs) comes squirting them and saying "No!".  Their attention is still on the bottle at this point.  What works to achieve what I want (getting obedience from the cats for my voice, instead of some gadget) is saying "No!", and squirting them.  But only sometimes.  Now the cats are taking a negative gamble if they do something I don't like--the powerful conditioning factor here isn't the punishment, which isn't so bad really, but not knowing if you'll get punished.

Similarly, with the arcade game where baddies throw out goodies that are random in character, and in timing, you get a nice addictive response.  If the goodies were all the same, or if you knew they'd come out every fourth baddie, it would be a bore.

For especially cruel GMs & designers, there's a marvellously evil thing called the "superstitious reinforcement" (or something like that).  This is a reward or punishment produced totally randomly.  When inflicted on pigeons, you end up with their doing very odd behaviors--pecking at a partiuclar spot, bobbing their heads in peculiar ways, etc.

Anyway, to go back to gaming in the normal world, we can see why whiffing--not merely negative punishment, but nicely random--is so de-protagonizing.  Well, I mean, we can see a different perspective on it that confirms what's known from bitter experience.

This would suggest to me that the way to get a "good player" of whatever kind, you do not set up a schedule of rewards that's regular a la D&D.  You would decide how often to reward someone (say, 1 in 3 cool role-playing events, or monster-whackings, or story-advancings, or whatever) and then randomize which particular events get rewarded somehow.

All in all, though, I think I agree with the folks in the above-mentioned thread that feel that trying to condition players to a particular style is a bit unwholesome.  However, behavioral psych can provide a good tool to see when a system rewards weakly, or, worse, rewards behavior that goes against its stated goals, sometimes very powerfully.

--TQuid
"Have mechanics that focus on what the game is about. Then gloss the rest." --Mike Holmes

Le Joueur

Quote from: TQuidOne crucial consideration in reward/punishment systems is scheduling.  That is, how often and how consitently is the stimulus generated?

What makes gambling addictive is *not* that the payout is given a higher value by the player than the far more consistent (and measurable) "negative punishment" of losing money.  It is that the payout is at random intervals.
I've heard the same works for collectible card games.  Unlike baseball cards, collectible card games deliberately create scarcity and then package one 'scarce' card per pack.  As demonstrated in the previous example, when you first begin collecting them, you are rewarded nearly every time you open a package (your supply of 'scarce' cards is low so every one is likely to be one you don't have yet).  As time goes on, and your collection grows, these rewards become fewer and fewer, and more random.

The way the article I read put it in terms of the works of B. F. Skinner.  And it made us all sound like rats!

Fang Langford
Fang Langford is the creator of Scattershot presents: Universe 6 - The World of the Modern Fantastic.  Please stop by and help!

Zak Arntson

Quote from: TQuid
...It is that the payout is at random intervals

--SNIP!--

This would suggest to me that the way to get a "good player" of whatever kind, you do not set up a schedule of rewards that's regular a la D&D.  

And this is the dilemma ... I am assuming a functional gaming group. So we don't need to promote a "good player!" All our players are good!!

Random reinforcement would be effective, but not my ideal method. Random rewards seem to make the assumption of training the Players. I want to use Operant Conditioning to support my Premise. It will encourage behavior, but certainly not train them.

I'll state my goal for this thread, 'cause I think I forgot an important notion:

Assuming a functional gaming group, how can we use Operant Conditioning with our Mechanics to support our Premise/Design?

So from your good examples of random rewards, it sounds like there are better ways for rpgs than random rewards. I especially like your treatment of "whiffing." It's a randomly allocated Negative Punishment! So our reaction to whiffing with Fortune in the Middle is a conscious decision to ease the punishment and even turn it into a reward!

And yeah, the random intervals are deprotagonizing. Which is why the mechanic we use to tie a conditioning method into the Unknown Factor is so important! In most cases we want our Players to feel like protagonists. In addition, armed with this idea that random intervals are a way to deprotagonize: If our Premise calls for deprotagonizing our Players, it sounds like we've discussed a good way to do it.

And where else does a random interval come in with a functional System? Bankuei brought up the Pool, which is an excellent example. I've used gambling mechanics in my own games, and twisted them so that instead of deprotagonizing, they encourage Player involvement and serve the Premise! (check out the Pool and my own the Jon Morris Sketchbuk Roleplaying Game).

I'm loving this discussion (as if you couldn't tell).

Laurel

Keeping with the assumption of a functioning player group, I would say Positive Discipline interrelates into this.  I'm going to suggest that players, like children *s* respond better to being told what to do rather than being told what not to do.   Other ways of encouraging specific behaviors with Positive Discipline are:

"Do As I Do" aka teaching by example
Respectful tone & good eye contact
Set necessary, clear, reasonable limits & consistently enforce them
Offer the opportunity/freedom of choice whenever possible
Positive behavior increases when it receives positive attention

According to the theories of positive discipline,  new and annoying but not harmful behavior can be discouraged by effective ignoring- not talking to, looking at the player or using any body language that indicates attention or interest.   Other ways to discourage something would be redirection and allowing consequences.

J B Bell

Quote from: Zak
Assuming a functional gaming group, how can we use Operant Conditioning with our Mechanics to support our Premise/Design?

It sounds like you've got the idea pretty well.  Random positive reward is the way to go.

The only missing element I see is good ol' communication.  That is, you can approach your players and talk explicitly about how they want to be rewarded.  If you've got informed consent all around, ethical considerations of "manipulating" players go away, and you can tweak your social contract and possibly repair deficiencies in the game system or augment it to get the behavior you (all!) want.  It's been my experience that knowing one is being conditioned doesn't somehow ruin the mystery or make one feel like one's free will is taken away (though some die-hards who seem to equate "freedom" with "maximum options for behavior" might)--rather, it just makes it clear that one has a program for change, and this can be very positive.

Pardon my psycho-babble.  You get the idea.  

--TQuid
"Have mechanics that focus on what the game is about. Then gloss the rest." --Mike Holmes

Zak Arntson

The past two posts are good advice for modifying group behavior, but I want to concentrate on design. If the design is done well, the group shouldn't have to resort to "fixing" the game. (remember: Assume functional Players).

I'm going to number your list, Laurel, to make it easier for me to address.
Quote from: Laurel
1. "Do As I Do" aka teaching by example
2. Respectful tone & good eye contact
3. Set necessary, clear, reasonable limits & consistently enforce them
4. Offer the opportunity/freedom of choice whenever possible
5. Positive behavior increases when it receives positive attention

1. Is a more in-play action. In design, this would be the inclusion of examples. "Do as we do in the pages of the game."

2. I would argue that this is included in the assumption of functional group. (if the group prefers respect and eye contact :)

3 & 4. Perfect! These are things that we can explicitly put in our System! With #4, though, our System can deny opportunity or freedom if that supports our Premise.

5. Again, this is more an in-play matter. But it can be explicitly addressed in the design.

---

Quote from: TQuid
It's been my experience that knowing one is being conditioned
-- SNIP! --
makes it clear that one has a program for change, and this can be very positive.

TQuid, you're drifting towards fixing dysfunctional gaming. Your quote here ties into discussion of Design. The best part: "knowing one is being conditioned," is something I see equivalent to "everyone has read the rules." So how do you feel we can design our games to enforce the Premise?

J B Bell

Quote from: Zak
TQuid, you're drifting towards fixing dysfunctional gaming. Your quote here ties into discussion of Design. The best part: "knowing one is being conditioned," is something I see equivalent to "everyone has read the rules." So how do you feel we can design our games to enforce the Premise?

D'oh!  You caught me.

OK, as a design principle, "gambling mechanics" is one cool idea.  I also think that Scattershot's (I think it's in Scattershot) idea of having a mechanical reward that follows the player rather than a particular PC is really brilliant.  So there's another principle to point up:

Don't be limited by the idea of only rewarding play behaviors through PC's.

Sorcerer mechanically provides a very nice carrot & stick of the GM's ability to make the majority of a PC's effectiveness come from good role-play on the part of the player.  Obviously this is only appropriate for a Narrativist setting; Simulationist games have certainly done the most with this, with one of the main behavioral conditionings being how players are expected to treat combat (from "realistically deadly" to "cinematic").

I'd like to remark on the earlier recommendation of Laurel's to "offer the opportunity/freedom of choice whenever possible".  I think this should be modified to "freedom of approriate choice".  Maximizing options is, I think, a Simulationist urge, and not always good for a Narrativist game.  Note the varying reactions to Sorcerer's very simple character types:  those with magic, and those without.  It seems very limiting, and yet spontaneous creativity around this simple design consideration has been truly extraordinary.  Mechanics must allow & encourage choice-making that supports your game design goals, and restrict or skip (in the interests of avoiding boredom, usually) injecting "choicefulness" into situations that do not support the design goals.

I recall reading a while ago in a post giving advice on plot creation that one of the phases was called the "squeeze" or something like that--options were limited to force interesting decisions at this point.

I digress again.  Let's back up and look at where in the game the designer has opportunities to insert conditioning:


    At character creation (in Effectiveness, Resources, and Metagame)
    In Play (bonuses and penalties to rolls, expansion and constriction of choices, even potentially mechanics for commemorating "good" play)
    In Between Play (character advancement, "de-briefing" {this could use an explicit mechanic, I've never seen one}, changing a character "roster" in games like Ars Magica)
    End of Game ("ultimate rewards" for PC's, voting on Most Valuable Player, etc.--this is really a variation on In Between, probably)
    [/list:u]

    Now, what types of rewards are there?


      The traditional PC-oriented rewards:  Effectiveness, Resources, & Meta-Game (I think I might tentatively add "aesthetic"--no, I'm not sure what I mean yet)
      The above three, applied to the player and applicable by her or him to any of their own (or maybe other players') PCs
      Good old-fashioned praise by GM and players--making this into a mechanic is probably questionable, but something proper to mention in a book of rules, I think
      Reputation
      Fully out-of-game stuff like not having to pay into the pizza fund (I've heard of this kind of thing but never seen it in action--Ron cautions against it, wisely I think)
      And probably many others
      [/list:u]

      Whew.  That helps break it down a bit for me; I hope this is helpful.

      --TQuid
"Have mechanics that focus on what the game is about. Then gloss the rest." --Mike Holmes

Laurel

Bankeui already mentioned Forgotten Fist in his own post, but since I've had a chance to see it in action, I'll mention it too as a system that uses a random but self-perpetuating, positive reward system to encourage and it seems extremely effective.

I think that's the key... a random element that nevertheless builds upon itself.  So that even if players don't get the "goodie" each time, the goodie itself perpetuates the behavior.  In the case of Forgotten Fist, its Story Points- and when a player earns a Story Point for imaginative roleplay, they are given an opportunity to affect the game at a later time, creating an opportunity to potentially earn another Story Point by being very clever in how they spend the original.  There's no guarantee this will happen, however; the GM controls the output of Story Points into the game.

Blake Hutchins

While it's a piss-poor example of roleplaying in the computer realm, at least in the single-player game, Diablo II is the most Pavlovian game I can imagine: press a button, get a treat, press a button, get a treat... ad infinitum.  Any negative reinforcers are pretty minimal (lose gold carried, possibility of having to re-do the level, loss of followers).

A simple way to look at this issue is:  What buttons do you want to offer the player?  What treats?

Best,

Blake

Zak Arntson

Quote from: Blake Hutchins
A simple way to look at this issue is:  What buttons do you want to offer the player?  What treats?

It's a little more than that, though. Think about the Behavior & Consequence model. Treats can be taken or given. And spankings can be given or removed.

And TQuid, good lists!