Langton's Ant:
Thinking about "Purpose"

What do we mean when we say that something has a "purpose"? That something is acting with "purpose"? That we ourselves are acting "purposefully"? Langton's Ant was created by Chris Langton in the 1980's as a way of helping to think about such questions. Before saying exactly what it is, let's just watch Langton's ant behave for a while and see what it makes us think. Let's get to know the ant a bit before trying to explain it (that's always a good idea in science; see A Feeling for the Organism).

The ant is the small dot in the center of the arena to the right. Click on start and see what happens. The ant is moving ... still moving .... STILL moving, right? And it doesn't seem to have much purpose, right? Its just sort of wandering around, aimlessly. But keep an eye on it for a bit more, until the counter gets to 10,000 or so. BINGO ... all of a sudden the ant seems to be going somewhere. To have acquired a "purpose" in life? To have decided to build a road?

What's going on? Why did the ant stop wandering aimlessly and start going somewhere? Would it do it again? Take the same time and go in the same direction? Try restarting the ant several times. And notice what happens, because its important. If we start over ... the same thing happens. Exactly the same thing. Langton's ant is therefore a "deterministic" system.

So, what's the big deal? Maybe Langton's ant just has a whole string of instructions saying do this, do that, do the other thing, and so on ... with the instructions for building a road at the end. The ant might look like it has a purpose, but that's just because someone put all those instructions in it. Perhaps, but if that were the case then what should happen if we disturb things a little? or a lot? Try putting some things in the ant's way, by clicking on "add patch" and drawing on the arena ...

Now THAT's interesting, no? The ant pretty much always ends up building a road (always?) but when it starts and what direction it goes in depends on what patches you've added to the arena (an intriguing side project: how much variation is there in each characteristic? is there any pattern to either variation that relates to where dots have been added?). So the ant can't simply be following a long set of instructions that say do this, then that, then that, and so forth. The ant does different things on different turns depending on what it encounters. And ends up building roads starting from different places and going in different directions. But usually (always?) it builds a road. Its a pretty stubborn ant, no? No matter how much you try to disturb it, it does pretty much the same thing. So maybe it really does have a purpose?

Maybe its time to look inside the ant. Let's go see what's there, and how it works.

barriers, more "purpose"
agent, instructions, environment (alterable by other things as well as agent itself)
how it works
agent deterministic/invariant, environment deterministic/changeable
behavior changes because environment changes, not because agent changes
agent behaves "purposefully" but doesn't have in instructions anything corresponding to "purpose"
"purpose" in creating instructions?

created with NetLogo

Earlier version here