This post is about conditioned response, triggered by an unrelated video game design discussion (about using sound effects as triggers for data collection for analytics because “[…] all the semantically important moments in a video game have unique sound effects (or patterns of sound effects)” – Marc LeBlanc Facebook post July 16 2012).  One responder I won’t identify replied “I think it says gamers salivate when a bell is rung.”

When games, and their designers, are accused of manipulating players, akin to deceit, Pavlov’s famous salivating dogs are often cited.  I think this isn’t a fair summary of operant conditioning in games. This made me want to explore my own understanding of conditioned response, and how game designers use it.  Hence this post.

A more powerful example is in Jesse Schell’s Gamepocolypse talk, he gives many horrific examples of blatant manipulation of gamer behavior using a “bling!” sound effect for each. A s I watched, I said “no!” but felt “cool!” The sound effect told me I should feel satisfaction of getting points. That “bling!” sound was spending my conditioning capital, earned through game play.  At the end of the lecture, even after some reflection, I felt confused at an intuitive level. Is the Gamepocolypse good or bad? It’s bad, but it’s fun!    It’s good to reward toothbrushing!  It’s horrible if game designers place products in my dreams!  This confused, complex feeling is correct: operant conditioning is not simple. the drooling dogs of Pavlov are not the whole story.

I want to start at the research.  In achieving the conditioned response, Pavlov first identified an inherently rewarding activity (dogs eating).  He then connected the ringing bell to that event.  The final salivating-for-bell behaviour was “borrowing” from the inherently rewarding activity of eating. If he had fed the dogs dirt, would they learn to salivate with the bell alone?  I think not, because there was no inherently rewarding activity.

BF Skinner’s Law of Effect says “behavior which is reinforced tends to be repeated (i.e. strengthened); behavior which is not reinforced tends to die out-or be extinguished (i.e. weakened)” (source).

To measure this I use the metaphor of ‘conditioning capital’:  designers earn ‘conditioning capital’ by providing inherently rewarding activity.  This ‘conditioning capital’ can be spent for various purposes.

For example, when a designer places a “ka-ching” sound effect, they are either predicting the player feels game mastery at that moment (which earns conditioning capital), or rewarding player for doing what the designer wants (spending capital).

In my play experience, games often build conditioning capital with rewards. For example, when I earn a powerup, the game plays a “bling” sound. This is a tiny reward.  It feels like the game is recognizing, externalizing, and acknowledging my tiny mastery of navigation.

Collecting a powerup is inherently rewarding: it improves my options for gameplay.  The “bling” sound is extra. It’s not just audio feedback. “Bling!” is a happy sound in my culture. I enjoy hearing it on its own. It’s a tiny reward. It is also being reinforced. I get something good, I hear “bling”, thus “bling” means good.  The game is building a conditioned response.

If you doubt that effect sounds are more than pure feedback, do this thought experiment: Play a generic 2D platformer in your mind, replacing the powerup sound swith a culturally meaningless ‘tick’ sound.  Would it be just as satisfying as Super Mario’s sounds?  Is it clear how important positive sounds are for effects, and how much conditioned capital is on tap through these sounds?

Of course, designers can use operant conditioning to manipulate the player.  Jon Blow criticizes this (Blow 2011) and asks designers: ““Are you trying to take advantage of your players and exploit them? Or are you trying to give them something?”As a designer, I give these tiny rewards for many reasons.

* guide the player towards designer-desired gameplay (e.g. collect coins)

* to amplify and validate the internal reward (yes, player, that was good!).

Perhaps one could define “bad” game design as spending more behaviorial ‘capital’ than it generates.  For example, if a math-memorization game uses “Boss Killer” achievements and players’ pleasure is primarily a conditioned response accumulated from past experience of killing bosses, then that capital will run out.  It may take a while – a lifetime of gaming has built a lot of capital around certain reward signals – and but it could be a clear basis for judgement of game design.

This ‘conditioned capital’ is so large for certain video game sounds that popular media mocks our helpless responses, even while using it effectively. In the animation “Scott Pilgrim vs” (youtube), at 2:52 Scott hits the boss, earning a “KO” victory sound effect. At 3:13 Scott kisses a girl and gets a “Powerup” sound effect.  It’s not just sound effects. There are many visual effects with cultural capital. In the Scott Pilgrim comic book series, Scott gets a “level up” graphic when he achieves anything major.

I view the conditioning capital around these iconic sounds as a shared resource: a Commons, in the Tragedy of the Commons sense.  Low quality gamification can be seen as a pure “capital spend” of video game reward conditioning. For an extreme example, consider low-quality ‘gamification’ of a classroom: convert grades to “points” and assignments to “quests”, with no other changes.  It is a ‘pure spend’ – it is borrowing the ‘cool’ that kids associate with games – and obviously manipulative.  BK Skinner’s theory suggests it will quickly become less effective, unless there are some genuinely new experiences behind it.

Of course, operant condition isn’t the whole story of player motivation – far from it. Player motivation is complex and multivariant, and behaviorism is only one small part of the picture. For example, in-game achievements are sought for social status signaling reasons (Medler 2011, “Validating Motives” section). These powerful motivations could well trump any motivation from intrinsic reward or conditioned response.  But operant conditioning is a important topic in any reflective game designer’s mind.

To summarize, I believe games designers should build ‘conditioning capital’ by providing players with inherently rewarding experiences, and be aware when they are spending it.

I’ve been trying to put my finger on what’s so wrong with poor old Math Blaster’s oft-maligned core mechanic (solve an arithmetic problem to progress the game).

Learning game researchers often critique this type of design, using arguments common from anti-gamification (my favorite is “chocolate covered broccoli”). But like gamification advocates, defenders ask: What’s wrong with that? It works, adding motivation to rote learning. While some question the value of extrinsic motivation, let alone rote learning (see wolfram’s TED talk), it does work.  No, there is something else. Something more frustrating.

Nintendo’s Help Cat (see Danc’s insanely detailed critique) runs away from the cursor, making it harder, not easier, to get hints . Since users assume winning yields valuable information, Help Cat winners are predisposed to value the factual knowledge delivered (hints about using the hardware).

After noting how Help Cat exposed hidden assumptions in UX and instructional design, Help Cat implies a question of learning games:

Is game play activity the reward, or the task?

Math Blaster’s design implies that game activity is simple, easy entertainment — a reward for real work of calculation. The “fun” in games is “simple, easy entertainment.”  But games are a form of play, and it may seem strange to say, but video game play is rarely simple or easy. In fact, it has a lot more in common with the frustrations and satisfactions of learning than “simple, easy entertainment”, as James Paul Gee convincinly explains in some detail.

Maybe playful design can help learning game designers disrupt the fallacious tension between fun and learning.  This line of thinking is leading back to one of my favorite design principles: “Find the Fun in the Learning.”  (“Moving Learning Games Forward”, MIT education arcade).

Can we achieve flow states by using real-time mechanics in physical prototypes of video games?  I did a very quick investigation and convinced at least myself that the answer is “yes”.

Background

In “Game Design Workshop,” Fullerton et al acknowledge the problem:

People who are not used to physical prototyping might argue that this method does not accurately represent the player experience on a computer. They might think a pen and paper prototype might work for a turn-based game, but not for an action-based shooter because gameplay is integrally tied to the 3D environment and the ability of the players to act in real time. 

Their answer, however, didn’t explain why it’s not often done:

“We are not arguing that physical prototyping replaces those things. What we are saying is the overall gaming system can benefit tremendously in its early stages by building a physical prototype.”

While, for many types of design questions, turn-based is sufficient; however, Csikszentmihalyi notes the player’s skewed perception of time during flow states:

“…in general, most people report time seems to pass much faster. But occasionally the reverse occurs: Ballet dancers describe how a difficult turn that takes less than a second in real time stretches out for what seems like minutes.”

If we could extend the range of physical prototyping to include flow states, I believe it would be very useful to video game designers.

I’m convinced it’s not too hard to achieve flow states with physical prototypes.

A brief search for real-time board games found only these:

1. Counter – Selfpublished in 2011, it’s a real-time, hex-grid miniature battle – sounds ideal for RTS prototyping. Rules here.  Students and I will build, playtest and report back.

1. Call of Duty RT is a real-time card game with a map. The video around 8:00 shows a “semi-realtime” play experience. Lots of pauses and rules execution.

These sources are related and useful to real-time prototypers:

1. There are some wargames that have the player issuing troop controls in a somewhat real-time mechanic (cite and more research needed)
2. Real-Time Chess (RTC) offers design ideas that could apply to common physical real-time problems (Chaboissier 2009 video)
3. Hungry Hungry Hippos (video). It has a territory, it’s realtime, and while there’s no real strategy , it’s fun. for kids. for a short while anyway. (OK, I admit it’s not really relevant– I just love it!)

I would appreciate any suggestions for this list, provided they:

1. have some strategic or tactical element;
2. have a map or territory;
3. give a flow experience, like a real-time video game.

I’ve excluded mapless real-time card games. e.g. Jab Frenzy Brawl Fightball; Slapjack, Spit, etc.  I believe the spatial design element is crucial to the RTS video game genre, which these lack.

Method

This project has been done on stolen time, so I used autoethnography as my method. I assessed the flow state by simply noting that I met the conditions for flow and strongly felt flow while playing.  I freely acknowledge that from a scientific point of view, my method is highly prone to bias and is weak on many obvious fronts (sample size, for a start), but I believe it is sufficient to justify further work.

My students and I created two real-time versions of Checkers.  The first version was “standard Checkers rules, except no turns.”  Four students watched and took notes as G and I played.  I estimate that we made 1 to 3 moves per second, and about 90% of our moves had some kind of purpose or plan (neither accidental, nor random / stalling). I recall two “breathers” – five-second periods where neither moved in about 5 minutes of play.

With Version 2 we added two rules:

1. You must touch your nose after moving a piece one space.  This prevented long, continuous zigzag slides from one end to the other of the board, giving the other player the opportunity to break a run by jumping them, while still keeping most of the flow feeling of a run.
2. collide=slapjack: If two pieces collide (ie we tried to move to the same place at the same time), then a “slapjack” minigame determined the winner: with the other hand, each player touched a certain spot on the table next to the board. Whoever’s finger was at the bottom, gets the spot.  The other player takes their move back.  Play does not stop during this.

I observed, but did not play, Version 2.  Players appeared to be as deeply engaged as I was.

Data (Realtime Checkers – player experience summary)

As I played version 1, I felt flow from the first move, and continued to feel it all the way until we stopped playing.  The 5 minute playtest felt like 20. My heart was pounding. My brain was in full game-strategy mode.

However, Checkers ruleset is too simple for RTS.  It felt more like novice multiplayer FPS experience: a pretty close match on decisions per second, and complexity of decision-making, to a beginner practicing against NPCs in TF2 or a friendly Unreal Tournament deathmatch.

I can imagine the reader might be snorting at the idea that Checkers could  “feel like multiplayer FPS.”  Here’s what I mean:

I felt the familar burden of low-grade performance stress, trying to keep an eye on the shifting sands of opportunities and threats, peppered with little sharp stabs of joy when I found a good move, and micro-dispair when I missed something important.

Because I was new to the game, I trialed, discarded or improved tactics, all in real time flow state, with little regard to long-term strategy (of which I doubt there is much).  I thought ahead two moves which enabled me to create crude move-jump combos, rather than move-by-move. I tried lame-duck tactics (to tempt her out when she was turtling).

I tried an online single-player version (“extreme checkers”  here …warning: lots of popup ads), but the unnatural speed of piece motion and mouseclicking made the first few minutes frustrating made the experience pretty weak.  The physical experience actually brings the game closer to FPS than the digital version.

Conclusions

I strongly believe flow states can be achieved with physical prototypes and believe there is reason to hope that physical prototypes might be able to model flow states in video games.  This study is far too limited to say much beyond that.

Returning to Fullerton’s observation on the gap between physical prototypes and video games: Obviously, much was missing between a FPS and realtime checkers:

1. able to see the entire territory (birds-eye view)
2. no range attack
3. no theme
4. no music

Some of these are easily testable. I look forward to future playtests with the stereo cranked and Warhammer figures.

Future research directions

There are two directions that this research suggests.

The first question is: is it possible to achieve a flow state in a map-based tabletop RTS?  While Checkers is too limited in strategic options to offer much insight on this question, I think games like CODRS and Boxing show that realtime play in phyiscal games with deeper strategy are combinable, and I feel there was enough promise to try it.

A larger and more crucial question is:  Does a flow experience in a physical prototype have any relation to a flow experience in the subsequent video game design?   In other words, if a prototype has flow, does this predict anything useful for the digital game?   I would dearly love to find out, though it might requires a much larger effort than a quick investigation such as this. As academics are so fond of saying, this is clearly a question for future research.