Looking for new ways to model a GAME of catch in an interesting way...

Dear Matlab community,

At the moment I'm pondering about a question some of you will likely find to be an interesting problem and I'm curious about your stance.

Check out this game!

Rules are simple: Throw the ball around. No score, no winner, no loser.

Backgound: This game, in which you play with two or three computer players, is used in social psychology for research into ostracism: In the variant you see here, the random guys on the left (player 1: "some other guy") and on the right (player 3: "yet another guy") exclude the player above (player 4: "POOR VICTIM") from the game of catch. Do you (player 2) notice? (Play around with the URL to change settings=4plos, or 4plall. or look here for more info.)

The research design goes like this:

1. you convinve a bunch of people that they are actually playing REAL people and
2. you can manipulate the identity of the poor excluded guy (aka target). (For example you can make him obese.)
3. Let them play a few rounds.
4. All gameplay is recorded, so that we end up with a list of all events and look at how people behaved.

In previous studies, people have mainly focused on simple analysis approaches, such as

• when the ball is first thrown to the target,
• how long people hesitate (reaction time) before throwing to the target or
• how often they throw to the target compared to everyone else.

Check out my dropbox folder for my code and example data if you like.

load example_set.mat

for loading some data (each person played four rounds, differing in names and ostracism.)

[log rounddata data] = readlog('example_set.mat')

• where "log" is a small summary, calculating reaction times and such,
• "rounddata" are single games and
• "data" is the complete set as found in the .mat file

I have specific ideas that you want to test and for example like to evaluate whether people played according to some strategies (for example: "throw to a player who hasn't had the ball during the last 2 throws" or "keep fairness of ball throw opportunities") and how often they deviate from them.

What would be your way of representing a game's record in Matlab?

How would you approach a Monte Carlo Simulation of this game? (Given that the play behavior of the computer players is semi-random only: in a fair game they randomly play from blocks say player 1 (left) (1 2 3, 1 2 3, 1 2 3... in a fair game or (1 2, 1 2, 1 2

Can you point me to resources about machine learning or graph theory that I should know about? Any Matlab examples that relate to this problem?

I've been through my data in any ordinary way, what I'm really looking for is something EXCITING...