Are you alive? How do you know? Sure, you can move. But so can a rock that rolls downhill. You've got a regular heartbeat. But ocean waves and streetlights vary rhythmically too. You make a variety of sounds, but so do battery-operated toys. And your molecular makeup isn't appreciably different from a compost heap.

In fact the only way you can make a convincing case that you're alive is by the way you behave.You act independently, you respond to changing situations and do things that couldn't be imagined in advance. And that is, finally, the most conclusive evidence that you're probably an authentic life form.

But this brings us to a profoundly unsettling notion: Being alive is essentially a matter of patterns and processes, and has surprisingly little to do with the particular kind of protein wetware driving those behaviors.

In a conceptual nutshell this is the theory behind the rapidly growing field of "artificial life," or AL.

AL researchers try to identify the distinctive behaviors of living things and then use them to devise software simulations that move, eat, mate, fight and cooperate without being told what to do.

So far most AL creatures consist of nothing more palpable than a few lines of program code and live only on landscapes made of pixels and data sets. Yet silicon-based AL fauna have proven so eerily similar to their meat-based forebears that some researchers are beginning to argue that they're just as "alive" as beavers or Bostonians.

If so, computer scientists have done what it took this amiable blue planet 4 billion years to accomplish: build living things from scratch.

Playing God

Each individual "agent" is built to act independently, sort of like molecules or ants or humans. Think you're too sophisticated to be represented by a computer model? Then think about this: Human beings spend an appalling amount of their time obeying relatively simple, and even downright moronic, rules. If the light is green, then go. If another employer offers more money, then change jobs. If it's too cold, turn up the heat or go to a warmer place.

From that perspective, life is a kind of game in which each person copes with the muddle of chance and necessity by applying a set of basic "if-then" principles. Such rules are easily translated into computer code. And to the astonishment of the first investigators who did so, it turns out that it takes very few rules to generate amazingly complex and animated patterns.

Bytes of a Feather

It's a phenomenon that occurs often in nature and can be so weird that it seems to border on the mystical. Microdroplets of water arrange themselves into six-pointed snowflakes. Millions of living cells, each groping toward its own tiny goal, collectively form the mathematically magnificent shape of the Nautilus shell. Ants build and maintain a colony.

All of these are "emergent" properties, meaning they arise spontaneously from dynamic systems. They're not dictated by some external authority, as is the symmetrical stomp of a drill team. They're orderly, but their order -- like the rise or fall of the stock market -- emerges from the aggregate of thousands or millions of individuals acting alone.

Artificial life behaves just this way. In a classic case each digital critter is given just three rules on how it should move relative to its neighbors. When let loose their collective motion eerily resembles the flocking behavior of birds.

In another case biologist Tom Ray created a passle of "agent" programs in his laptop. Each agent had a simple job: make a copy of itself in memory. (A reaper killed old ones after a while.) Ray left the programs running all night and woke to a startling sight -- his agents were engaging in digital equivalents of competition, fraud and sex.

They had evolved. Unknown to Ray a computer glitch occasionally altered lines of code when the program-agents copied themselves. Most mutations were fatal, and individual programs "crashed" and died. But some changes let an agent do its job better, such as running with fewer instructions -- a good thing in their crowded memory patch. As the shorter versions replicated, they soon outnumbered their larger cousins. Darwin would have felt right at home.

In this way, after hundreds of thousands of generations, some versions had developed ways to "trick" other agents to do their dirty work, allowing the tricksters to become even smaller. Others evolved ways to share code to form new, more robust agents -- essentially combining genes like birds and bees.

Looking into the future
-- and the past

A complex creature like a mammal begins life as a fertilized egg that divides into a cluster of a few cells that are all exactly identical. Yet as the animal grows, each cell becomes exquisitely specialized and differentiated. How does a cell determine whether to become a bit of kidney or a piece of bone? Watching artificial agents divide might give us clues.

Insights about our biology aren't the only applications of AL research. Electronic societies that evolve into patterns resembling migration, war and segregation provide new experimental tools for social scientists. "Perhaps one day," Brookings Institution investigators Joshua M. Epstein and Robert Axtell write, "people will interpret the question, 'Can you explain it?' as asking 'Can you grow it?'"

As artificial societies become more reliable predictors of real behavior, they could become what author M. Mitchell Waldrop calls "flight simulators for policy" that would, for example, "allow politicians to practice crash-landing the economy without taking 250 million people along for the ride."

AL pioneer Doyne Farmer predicts that this kind of research will eventually "enact another major change in the global rate of evolution" by blurring the distinction between artificial and natural.

"We may," says Farmer, "be the first species to create its own successors."