A plant that turns toward the light, or a worm that writhes after severation, doesn’t do so out of free will.
Their internal biochemistry mechanically responds in a deterministic (if stochastic) way. They don’t make choices.
In Vehicles, Valentino Braitenberg asks if we humans aren’t the same way.
Even I could come up with the easy arguments for stimulus→response:
- something annoys me → bad mood → don’t pay attention to the car that’s pulling out → accident
- physics says so
- I read a compelling book about robots → inspired to go to graduate school and dedicate my life to synthetic consciousness → entrenched in a career with no prospects
- people predictably respond to stimuli: we avoid people & situations we don’t like and gravitate to what we do like (subject to feasibility constraints).
But Braitenberg does something much more convincing. He builds robots to prove his point.
He starts by resolving the problem of Burridan’s Ass stochastically. A phototropic robot might be stuck at
θ = 0° between two light sources, but since we can’t get it to exactly 0° the robot—without free will or choice—heads toward one of the “bales of hay”.
That problem is resolved with two wires connecting two stimuli to two engines. As the book progresses Braitenberg builds more lifelike robots using more connections—complex networks that reroute external stimuli to mechanistic, deterministic robotic response.
Braitenberg doesn’t get all the way to the dramatic complexity of "I love you! … I know." but given what’s possible with a few tens of connections, what could be possible with hundreds of trillions of connections?