Image: Wikimedia Commons
The sea slug Pleurobranchaea californica is a simple creature. It has three basic moods: flee, feed, or fuck. This simplicity makes it a good candidate for recreating in a virtual world.Researchers at the University of Illinois College of Medicine at Urbana-Champaign developed Cyberslug, a simulation of the sea slug, to better understand how simple brains work and apply that knowledge to the broader field of artificial intelligence research. Their study, “Implementing Goal-Directed Foraging Decisions of a Simpler Nervous System in Simulation,” is published in the journal eNeuro.
Much like humans and other animals, Pleurobranchaea decide how to go about their days based on an ability to assess motivation against memory. For example, if the slug remembers that the creature it encounters stung it last time, it’s probably a flee-the-scene kind of situation.Sea slugs are also self-aware: They’re able to discern whether they’re hungry enough to take a big risk and bite the stinging thing.The researchers plugged all of these factors into the graphic modeling program NetLogo, and then watched what happened. When the “slug” encountered two types of virtual prey named “Hermi” and “Flab”—after the sea-slugs Hermissenda crassicornis and Flabellina iodinea—it would either avoid or pursue them, based on how “hungry” it was at the time. (The artificial slug recognized prey based on virtual odors they secreted.) Essentially, researchers taught the AI slug to behave like it would in nature, complete with “feeling” and self-awareness.
“The bot adds up sensation, motivation, and memory to make a cost-benefit decision of, ‘is this prey worth attacking?’” lead researcher on the study Rhanor Gillette told me in an email. “The approach-avoidance decision is the basic decision of all economic behavior. So, like us, the bot and the animal make economic decisions not on the basis of simple information by itself, but on how the information makes it ‘feel.’”Gillette told me that Cyberslug’s behavior is “the core of natural intelligence,” something contemporary AI has yet to completely nail. “Similar circuits are active in the ancient parts of our brain, and they drive our higher intellectual, economic, and social activities.”