Wednesday, October 24, 2007

Constructing Artificial Emotions: A Design Experiment

My latest essay on emotion in games is up on Gamasutra. There are pretty pictures about brains. You can read it here.

It asks that simple, innocent question,"What can we do to make games evoke emotions?" The answers are more about applying the lessons of experimental psychology than the 300 hot tricks of screenwriting.

While I was looking into this topic, I read an essay in Scientific American on 'dangerous ideas' and it got me thinking about the sort of 'unthinkable' ideas in game design. This essay contains a smattering of them and I'm curious which ones you find intriguing.
  • "Games are great at causing emotions."
  • "You can replicate meaningful religious experiences with a game."
  • "Most media such as books, movies and poetry are far more about our past experiences than any inherent value of the work. "
  • "Isolating gamers from the outside world is a highly effective strategy for maintaining service contracts."
  • "In order to increase the impact of games, we must engage the body as well as the mind." The Wii Fit is just the start, baby. That slack faced hardcore couch potato experience is about to become an experience for dinosaurs (fat, emotionally stunted dinosaurs at that)
Does it hurt to say such things out loud? I have great faith in the ability of science and reality to weed out the ideas that contain no substance. Whether any of these concepts hold water will be directly up to the efforts of talented and innovative game designers. But what if one or two of them held a kernel of truth? My god, what a brilliant future lies ahead.

I am quite looking forward to your thoughts on the essay. Grab a mug of tea, find a comfy chair and dig in.


Constructing Artificial Emotions: A Design Experiment

"What's the Big Idea" by Steve Mirsky


  1. I am tired of people saying that it is hard to emulate emotions on computer. That's really easy! The difficult part is to map the feelings to something humanly meaningful. As you said yourself, past experiences create the meaning. Without past, there is no meaning.

    Therefore an artificial construct must fulfil three conditions:
    - a long term memory (something that is not reset by shutting the game off or by entering as a new player)
    - a value system (this is easy, really, and this is what I meant that it is easy to emulate emotions in computer environments)
    - a mapping system of the emotions the computer generated personality SHOULD have (and this is where multiplayer online games should provide plenty of material)

  2. I was actually a bit surprised on reading the article that there was no mention of Jenova Chen's work (the whole point of which, as I understand, is to invoke emotions other than adrenaline rush and power fantasy -- what emotion do you feel when you play Cloud or flOw?).

    Of course, having an actual Process (rather than just a few exemplars) was really nice to see, and I think that was the best part of the article.

    The other thoughts you mention seem like logical corollaries to me.

    And no, it doesn't hurt (for me, at least) to say things that challenge our fundamental way of thinking. I rather like being in such an environment; my wife pulls the intellectual rug out from under me on a regular basis. (That said, the rest of the world seems highly intolerant of such things... but to me that makes it all the more important to say what needs to be said.)

  3. Excellent article!

    I must confess that some of it began to sound a bit sinister to me, in a "how to start your own religious cult" sort of way; but on further thought, the same could be said of a discussion of any medium that uses deliberate techniques to invoke emotional responses in its audience, from theatre to film.

    (One nit to pick: I think you mean "sweetbreads", not "sweetmeats". Sweetbreads are cow thymus glands, sweetmeats are candies.)

  4. Appreciate the comments!

    Sweetmeats Sweetmeats are confectioneries, but sweetmeat is also an older term for offal, or organ meat. Here is the wikipedia page:

    Still, it seems to have fallen out of practice in the last half century.

    Sinister overtones
    That was something I intentionally left in the essay. Personally, I feel that this style of game design has extensive room for abuse. If you find yourself questioning the ethics of reality television shows, then you find yourself questioning the ethics of the broader category of emotional games. MMOs right now are a bit of a wild west atmosphere. I wonder how long before they begin hiring ethicists as part of their game design teams?

    flOw: I must admit that I'm somewhat jaded and saw flOw as yet another exercise in skill mastery with minimal 'new' emotional content. It seems that single player games have particular difficulty evoking emotions. Ultimately, they are abstract systems with no real world connections. Feeling emotion is expensive and single player games trigger enough BS sensors that we are steeled against crying when our level 4 fighter has its hitpoints reduced to 0 by the orc sprite.

    take care

  5. Interesting article, particularly re the sinister overtones and the concomitant ethical issues you start running into. I was very strongly reminded of Paul Czege's seminal "Most Dangerous Gaming Idea" post from a couple of years back (, second comment down).

    Interestingly, that was coming out of the indie (paper-and-pencil) RPG scene, which has been making major advances on this front using only game design, without any tech. They do, of course, have your "small group discussions" in spades, and human equivalents of many of the systems you describe. Which makes me wonder how much of this article is about recreating the face-to-face RPG experience with the convenience of online participation (and much better audiovisuals).

  6. Hi,

    Interesting read.
    Do you really believe, manipulating players at their inner core is a "good" thing
    and game designers should strieve for it. Do you like to manipulate eight year old kids?

    Some of your sentences, well, taken out of context, sound like a designers power-trip with
    very sinister undertones.

    The individuals that buy into the game will behave according to the standards of their dominant social group, fellow gamers dancing through life in an artificial, designer-manipulated culture.

    In this brave new world of emotional experiences, you design interactive systems that play the player like an instrument. Except instead of tunes, they are belting out tears.

    I, for myself, believe emotions towards a computer game and its entities are wasted, and we as designers should teach children to see and resist any kind of manipulation, especially on the
    emotional level.

    Well, just my 5 cent.


  7. Are you implying that you can effectively evoke emotion by creating feedback loops from the body? I believe that is false. It is as real as laughter on the background of sitcoms. How can you evoke emotion without understanding it? How can you do it repeatedly with the same construct?

    TV today does exactly that: flies blind. It takes some recipes that demonstrated financial returns and redoes them to nauseaticum. But you can (somewhat) forgive them, as television is a passive medium.

    Computer games, although they are right now in the same dark hole of recipe design, are meant to be interactive. You cannot fall in love with a computer girl that says the same things over and over again. You can't feel hate, anger of fear for an opponent that makes the same moves over and over again until you find a weakness after loading the game the fiftieth time. And certainly nothing more evolved, like compassion, empathy, complex social interaction.

  8. great essay.
    thanks for the thought share.

    btw, are you rolling into project horseshoe this year?

  9. @cosmind: I am indeed. I'll be giving a talk on Thursday.


  10. Danc, great essay, as usual. Very interesting read.

    "single player games have particular difficulty evoking emotions"

    I disagree. I've never managed to play much of Silent Hill -- not for being scary, but for being so utterly *depressing*. Doesn't that count? It gets my body to release adrenaline using standard horror movie techniques (reverse camera angles, sound cues, etc) and then feeds me emotional context (the wife's grave, a small kid sobbing, etc). I think it goes right up your essay's alley.

    Other games have succeeded in this as well, but I'm at work right now, so I'll post about it on my own blog and link back here. :)

    @anonymous: Of course we want to manipulate 8-year old kids! Don't we do it every Christmas and Halloween? It's for entertainment purposes, remember. :)


  11. "we are steeled against crying when our level 4 fighter has its hitpoints reduced to 0 by the orc sprite."

    Possibly because we don't lose anything by it. Reload from last save point and try again.
    Needs more roguelikes.

  12. I'm continually struck by similarities in our thinking, Dan! Schacter-Singer's Two Factor model was the predecessor to the work of Damasio and LeDoux on which I tend to lean. (Read Montague's work "Why Buy This Book?" fills in a lot of gaps.) At this year's OGDC I showed one of your game system diagrams as similar to a model of my own, itself an accidental variation/simplification on/of Norman (I was turned onto DOET late). In fact, next week I'm giving a talk at GDTW on games as "emotional artifacts". Craziness...

    Keep up the good work!

  13. Enjoyed reading the paper! It's great to see the psychology of emotion touched on from a design view.

    Your article made a lot of sense-- in particular, the notion of evoking salient personal memories is really a great thought. A player seeing his or her own name is often enough to evoke an emotional response; see what M. Baldwin et al. are up to over at MINDHABITS (, if you're curious.

    Specifically, I really like the idea of pairing tagged images with recalled semantic memories. This is something that could be implemented very easily, even with today's technology; I'd love to see even a simple messaging client that reacts to certain words (e.g., the WeeMees on AIM, etc.)

    Relying on the Schacter-Singer model too heavily is a little dodgy, though. Research on the theory is sparse at best, and I have extreme intuitional reservations about the accuracy of such a model. Even if a two-factor hypothesis is apt, the implementation would still prove tricky: I think it's more difficult to evoke physiological responses in game design than it would be just to manipulate emotions directly. I wouldn't recommend going about things from that direction-- unless from a supplementary perspective?

    Look forward to reading the next one!

  14. "Games are great a causing emotions."

    What the love does this mean?

  15. great article! it took me a few days worth of breaktime-reading to get through, but it's well worth it. I've really loved reading your blog, it's very inspirational! I'm in over my head as a n00b game designer having moved into the field from marketing. your essays really give me something to think about:) thanks!