Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Episode 27: The Invisible Rules of Roleplaying
#21
I did my PhD on the topic of how infants learn to understand the goals of others. So it was an interesting thing to hear in this episode all of the discussion around efficiency, constraints, and goals in the context of games, rather than in the context of human development.

When we talk about these things in psychology, a lot of it stems out of a paper from 1995 by a pair of researchers (Gergely & Csibra) and another one from 1998 by Amanda Woodward .  One of the only ways that preverbal babies can tell us what they are know is by watching what they look at, and for how long.  Babies tend to watch something when they don't understand what's happening, and then as they start to make some sense of it, their attention drifts to new things (this is called habituation). So, if you want to get a sense of how an infant understands things as a scientist, you can create a scene for them to watch, and then, after they have habituated to what you are showing them, you change something about the scene.  If this tends to recapture the interest of infants, this tells you that you have changed the rules in some way, that they are going back to the drawing board to figure out what's happening so to speak
  If they don't find the change interesting, then you haven't.  So you can mess around with this type of thing to get a sense of how infants understand the actions of others and why people do what they do.

One of the early things that babies learn is that people have goals.  This applies to robots and animated figures, too, if you give the infant's a sense that these things are acting of their own accord.  Gergely and Csibra did this experiment where they showed babies this video, of a little ball rolling around (self propelled, to make it seem like an agent rather than an object).  The ball would approach this other shape, but there is a wall in the way, so it hops over the wall and rolls over the other shape.  After the infant's started to lose interest, they would change the scene.  Now the wall is still there, but it's behind the other shape, it's not in the way anymore.  And the ball would either roll directly to the triangle, or it would keep the same path of motion and do a little jump (over nothing) on the way.  The action is more similar to what the infant's saw, but there's no reason for it anymore because the obstacle is gone.  Infants looked longer at this, despite the fact that the video was more similar in terms of what was happening to the previous video.  

It was a violation of the expectation that an agent will behave rationally, efficiently to the constraints of their situation, towards their goal.  Woodward did a similar experiment, this time with a person's hand reaching out and grabbing an one object (out of two available objects) repeatedly.  Once the babies had habituated, she changed the scene, swapping the object positions.  The hand now reaches out and grabs the same object (new position) or the new object (same position).  Babies look longer when she grabs the new object, signaling that they understood the action to be about the object itself.  If you replace the human arm with a grasping rod, though, now babies expect the rod to keep going to the same place after the swap, not the same object.  

This all sets of a lot of research about what makes a goal, and what signals a thing to be a goal.  It's not just that a person is doing it (you can change the conditions a bit and suddenly the expectation of consistency goes away).  I was thinking about this a lot during the discussion of efficiency, because I have argued with people who would say that without some element of efficiency to a constraint, infants do not interpret an action as goal directed at all, that it requires the constraints to even be recognizable as a goal.  Whereas I would argue that it shows the goal to be something other than the end point.  I think in the episode, the hosts used an example of climbing a mountain efficiently.  Under this understanding, there isn't a way to climb a mountain efficiently unless you need to get to something and climbing a mountain is the only way to get there.  I think, if you are climbing a mountain and it isn't fully required to do so, you are demonstrating that the goal is, at least in part, the experience of climbing the mountain rather than just getting to the top.  Which, I think, speaks to what makes a game by the definition discussed.  The goal has to be, in part, the experience, not just the action itself.

It makes me think about animals, as well.  When they play with each other, it's often in the form of play fighting.  Which, to a person, often resembles real fighting, and can quickly turn into real fighting if someone isn't following the rules (very like human games, haha).  But the way that animals seem to be able to tell that it's play fighting and not real fighting is, in some cases through signals (like the "play bow" that dogs do when they want to play with each other), but also through inefficiency.  They get to have the practice of real fighting, which helps them get better at it, but they do it in a way that wouldn't work in a real fight, where they don't go for the throat but for other parts of their body.  The inefficiency is a cue to the fact that it's not the real deal.
Reply


Messages In This Thread
RE: Episode 27: The Invisible Rules of Roleplaying - by Dr_Colossus - 11-12-2026, 03:59 AM

Forum Jump:


Users browsing this thread: 1 Guest(s)