A video making the rounds at Room Escape Artist this week shows an octopus that has been taught to play a piano. Not pawing at it. Playing it — pressing keys in sequence, accepting a tactile interface designed for a totally different body plan, and producing recognizable melodic patterns. The video is being shared as charming. It is. But there is something else in it that I keep returning to.
Octopuses don't have hands. They don't have hands in a quite radical sense — their arms operate with a degree of distributed neural autonomy that has no clean human analogue. Two-thirds of an octopus's neurons live outside its central brain, in the arms themselves. Whatever is happening when this animal "presses a key in sequence" is not a centralized motor plan being dispatched to a hand. It is a coordination problem being solved across a body whose architecture was never selected for keyboards.
What the trainer built, then, is a puzzle apparatus that the octopus has learned to operate as an input device. From a puzzle design angle, this is the more interesting move than the music. The keyboard becomes a lock — a mechanism that accepts structured input and produces a satisfying output when the input is correctly sequenced. The octopus is doing exactly what an escape room player does: deriving a sequence and delivering it through hardware.
The thing the octopus can't do is the part I keep thinking about. The press is fine. The sequence is fine. But the working-memory binding required to hold "the third key gets pressed twice" across a distributed eight-arm motor system is a category of cognitive load no human player ever experiences, because we have a centralized motor controller that handles the binding for us. The octopus is solving a binding problem we don't even know we have.
It also makes me wonder whether the lock mapping problem — the gap between knowing the answer and successfully delivering it through hardware — has a body-architecture component nobody has named. Human escape room design implicitly assumes a single point of motor control. What does an input device look like if it doesn't?