Monday, September 17, 2007

Colonial Expansion

To understand how intelligence can emerge from groups of simple cells, it helps to have examples of other emergent behaviors.

Hive animals, such as ant and bees, work in ways similar to the ways in which cells in our body work, but because they are "bigger", they're easier to study.

Let's look at how bees manage to figure out the best place to go get nectar every day.

Bees look for nectar that they then take back to the hive to make into honey. The higher the sugar content of the nectar, the more efficient the bees are at making the honey. Bees are pretty good at finding the best sources of nectar, even when those sources change. How is this done? Is there some guiding intelligence that takes the reports back from the worker bees and decides to send all the other workers to the best spots?

In a sense there is a guiding intelligence, but it is an emergent intelligence, based upon statistics and feedback loops in a system. Here's how it works for bees. The worker bees head out in random directions looking for nectar. When a bee finds nectar, it collects some then comes back to the hive. It then does a little "dance", which other worker bees at the hive can watch to learn the directions to the nectar.

Different bee species have variations on the dance, but basically the orientation of the dance correlates to the relative position of the sun or the direction of the nectar relative to the hive, and the length of the "waggle" portion of the dance is correlated to the distance from the hive. The worker bees in the audience can head out in the correct direction and the approximate distance, then look for the nectar source.

Ok, neat, this is how other bees can find nectar more efficiently than just every bee randomly searching each time. But when you have lots of different nectar sources, how do most of the bees go to the best nectar source? If they're all coming back and dancing about random nectar locations, then the audience bees should also be spread out among the random sites.

But...the time spent on the dance is driven by the richness of the nectar. The bee gets a buzz on, and the better the buzz, the longer he can dance. With bees coming and going all day, those bees that dance longer will have a larger audience of bees - more bees will catch their act if they perform it ten times rather than just once. So all those bees go find the better buzz, and they too come back and perform longer dances, which grows exponentially to send even more bees to the better source.

Each bee just follows a few simple rules. Step one - get ready to leave the hive. If you see a bee dance while leaving the hive, stop and watch it, then orient yourself in the right direction and fly ten waggles thataway. If you don't see a recital going on, just leave in a random direction. Remember to count how many waggles you fly, and in what direction relative to the hive (or sun).

Now look for nectar. If you find some, take it back to the hive. Pass the nectar to another type of bee, then go perform. Perform as long as you feel the energy to perform.

Go back to step one.

If you model this, you see that no matter where the best source of nectar is in a given day, most the bees will end up heading for it, thus optimizing the nectar collection for the hive.

No single bee made a decision. Nobody passed along the information to the (nonexistent) decision makers. The dance says nothing (directly) about how good the nectar is. Nor is there any "debate" among the bees about the relative merits of the different nectar locations. It was purely the inherent mechanism that the better the nectar, the longer the bee could dance, and the statistics inherent in that fact that drove the "intelligent" behavior of the hive.

"That is real intelligence," you say. Well, it's a smart move on the part of the bees, certainly, but it isn't how humans make decisions. (Or is it? I'll come back to how groups of humans make decisions in a later post).

What this example highlights is how separate "cells", each just doing what comes naturally, can create a higher order "emergent" behavior when the cells are in a group. Studies of human organs indicate similar sorts of statistical emergent behaviors in blood cells, liver cells, and even brain cells, where the normal biological functions of the cells and the chemical byproducts of these functions create a form of communication between cells. Groups of cells (an organ) also work with other organs for even higher order emergent behaviors. Our organs use chemical signaling (such as nitric oxide, hormones and neurotransmitters) and electrical signaling (such as potassium and sodium ion channels).

Understand that all of these mechanisms develop over time through the evolution of genetics. Those mutations that provide better survival and breeding in a given environment become standard in the later generations. Bees that could dance longer given good nectar statistically attracted more bees to the right place, so the hive outperformed other hives that didn't have workers with this trait. Over time, only those hives who had workers with this trait were the only ones around.

Evolutionary theory isn't just for biology, although it's pretty amazingly useful to help understand how and why organisms are the way they are. This powerful concept can be applied at any level of organization - including brain organization, and even human social organizations. The exact same principals apply.

I'll try to put together examples of each of these in the following weeks.

Labels: , , ,


Post a Comment

Links to this post:

Create a Link

<< Home