What is life?
Not to get too metaphysical in the run-up to Christmas, but it’s an interesting question to consider when you watch this video of a Lego robot with the brain of a worm bumping into a wall:
Since 2012, coders and computer scientists have been collaborating on the OpenWorm project to simulate the entire biological existence of a specific, very small worm known as Caenorhabditis elegans. It’s about capturing the life of this worm in software: 959 cells and 302 neurons, running in silicon. And a cool milestone was reached a month ago (but which I’ve only just been made aware of), as reported by IProgrammer‘s Lucy Black – sticking a simulation of the worm’s neurons into a Lego robot made it act like a C. elegans worm:
It is claimed that the robot behaved in ways that are similar to observed C. elegans. Stimulation of the nose stopped forward motion. Touching the anterior and posterior touch sensors made the robot move forward and back accordingly. Stimulating the food sensor made the robot move forward.”
Since C. elegans is so small (about a millimetre long) and so neurologically-simple compared to more complex animals (like humans, with 86 billion neurons), modelling its neurons has proven relatively straightforward, relative to the difficulty of trying to create proper artificial intelligence from scratch. The Lego robot in question has two wheels and sensors that detect touch or food – make the sensor at the front think there’s something there, like a wall, and it will move backwards. There isn’t any training involved, or memory, or learning. This is just how a C. elegans is “programmed” in real life, as it spends its entire life living in soil and manure, seeking out digestible organic matter.
It’s also the only organism to have had its entire neural network – known as its connectome – mapped out. The popular conception of how brains work has, for a long time, been that specific cells do specific things, but recent research has found that this is an oversimplification. What really happens in brains is that connectomes act as coherent wholes, and it’s not possible to say for certain how an organism will act by only looking at a part of what its brain is structured like. You’ve got to comprehend how everything works as an interconnected mechanism – the connections between the cells are just as important. It makes connectomes, even simple ones, surprisingly intensive to simulate in software.
What’s particularly worth contemplating here is the question of what this means for our notion of “life”. That’s not the same as intelligence, which usually takes up the meat of the debate on robots, but more a general issue regarding our ability as humans to instill inanimate objects with the qualities that we think of as natural. And linking 285 million of these software models together isn’t the same as building one human brain, because the connectome will be structurally insufficient to represent how a human brain works.
This Lego robot is stupid, and will remain stupid because it can’t learn, and that’s OK. And it’s far inferior to many of the best attempts at machine intelligence – like IBM’s Watson, for example – but that’s also OK. But what it is is an accurate copy of a worm’s brain put inside a machine body. (Albeit a machine made of children’s building blocks.)
When we consider something to be “alive” we often look to a set of key behavioural attributes that distinguish, say, a tree from the mountain that the tree grows on. It doesn’t have cells, but it does respond to stimuli, and if it had a solar panel to provide energy then it wouldn’t be too different from a plant. Reproduction isn’t that insurmountable a challenge, either, as there are plenty of machines out there which can make their own replaceable parts, or even of making superior parts to the ones that they themselves use (which is effectively growth and evolution, of a sort).
Compare that against the Merriam-Webster definition of “life” and we’re almost there. The sticking point is, fundamentally, metaphysical – “a principle or force that is considered to underlie the distinctive quality of animate beings”, as Merriam-Webster puts it. These kinds of experiments get right at the old arguments over whether we are merely the sum total of our physical bodies (and our connectomes) or if there’s some transcendent life force above and beyond all this. In other words: if you were in an accident and your brain was kept alive in a mechanical body (and let’s hope it’s made of something more sophisticated than Lego), would you be you, or not? Or would you be a soulless robot, but clever enough to know not to walk into walls?