I'd like to get some suggestions for my PhD project. In short, I've got a bunch of creatures, organized into tribes, that can use signals to command each other. The creatures can only understand commands from other members of their tribes. Part of my experiment is to determine what kinds of social structures are most efficient, and in this case I'm trying out various obedience criterias.

For example, in structure type A, every creature obeys every command it hears, no matter which other creature in its tribe is giving the command. In type B, creatures only obey commands given by those who are older. In type C, creatures obey commands from those who are healthier. In type D, creatures test the commands against their own preferences before deciding whether or not to obey. I've got some more ideas, but I'd love to get some suggestions from you guys. Just assume that the creatures can use whatever criteria or knowledge you think is important, because if I don't have the capability for already I can add it.

10 Comments

Dennis said:

One variant of what you
already have would be to
impose a multi-level hierarchy
and have creatures only follow
orders from another that is
higher in the hierarchy.

Wacky Hermit said:

I use a system called "yield points" to decide who has the right-of-way at a stop sign (the law notwithstanding, it's NOT actually who gets there first). Sports cars have more "yield points" than minivans, red cars have more "yield points" than tan cars, and bigger vehicles have more than smaller vehicles. I started doing it as a joke, but tallying up yield points turns out to predict pretty well who will go first at a stop sign.

Maybe you could use a "yield point"-like system to decide whose orders your creatures will obey. That way you wouldn't have to limit yourself to one category of hierarchy, like health or age.

You might also try a random element in behavior. Sometimes people do things for random reasons.

adamj said:

How about setting some criteria where an individual is more likely to follow the commands of one who has followed their commands in the past?




A way to make this more sophisticated would be to assign each individual a base likelihood for following any random command, but modify that likelihood by the command giver's track record (or that of the command giver's group) of obeying the commands of the current command receiver. For instance, the receiver may refuse to obey anyone who disobeyed the last command given to them, or has ever disobeyed a command given to them, or has never disobeyed a command given to them, etc.


Megan said:

What about proximity. I think that most people would listen to a person near them who is speaking rather than a person from far away who is yelling. So, perhaps your little creatures would only take orders from an order-giver near them as opposed to one further away.

BogardanZero said:

What about something like clan alliegence? Would it be possible to make each individual only follow orders from those within their own family?

Marty said:

I like adamj's suggestion, some sort of herd behavior, but you would want to include some way to guage the positive/negative benefits of the obedience. Send one lemming over a cliff, they all will follow. Such is the problem with lemmings -- not neccesarily the problem with "herding" per se...


My own suggestions, or concerns, are:


If a unit is already busy performing an act of obedience, must it drop everything and obey the next command? (disobeying the earlier instruction?), or are commands queued and followed in sequence? If the latter, is there a timeout period, where commands no longer apply (perhaps the task was completed by another unit who was given similar instructions). Also, can a unit still give commands, while engaged in obedience to another? At what point do the number of commands overwhelm the populations ability to respond to them?


Good questions, sounds like fun -- i wish i were smart enough to program such things.


Reminds me of DigialSexLife, a fun little javascript sim a buddy of mine did years ago...

DeoDuce said:

Maybe they could learn how to communicate with bettas in a bathtub. That would make for an interesting dissertation.

Adam Greenwood said:

Perhaps the creatures would be more likely to obey a command the more sources they heard it from.

caltechgirl said:

What about a structure where an individual does whatever one subgroup commands, but the opposite of another subgroup. Say, older creatures' commands would be obayed, but younger creatures' commands would be thwarted.

I do like the reinforcement idea, following commands from individuals who gave commands previously, though.

Crash said:

Perhaps you could institute a "floating" leadership role for each tribe. Your creatures would have a leader chosen for them on a random basis. They would have to obey that leader for a predetermined amount if time. (A day or two, depending on the size of the tribe.) Then, after a single term the tribe could chose to stay with the same leader or reject their leadership, in which case they would be assigned a new leader on a random basis. I imagine, greater or lesser success in the tribes endevours would be the deciding factor. There would be no term limits, so the best leader could remain the leader for an extended period of time.

Leave a comment

The comment login system is acting strange. If you get an error message saying you aren't logged in when you are, just reload the comment page and try again. I'm trying to track this bug down, but it's not easy.

Supporters

Email plasticATgmailDOTcom for text link and key word rates.

Site Info

Support