Octopus and Edge Computing - An inspiration from biology

In his book, the neuroscientist Godfrey-Smith speculates that octopuses began to acquire their remarkable distributed intelligence when they lost their shells, and so had to rely on their wits to avoid being eaten. Their nervous system have a special structure.

They have about 500 million neurons, more than 350 million are divided up among its eight arms. Research also supports previous findings that octopus’ suckers can initiate action in response to information about their environment, coordinating with neighboring suckers along the arm. The arms then process sensory and motor information, and muster collective action in the peripheral nervous system, without waiting for commands from the brain.

Whereas we mammals are brain-heavy, the cephalopod’s nervous system is more evenly distributed around its body. The tentacles have their own sensors and controllers, even memory. Their deeply different body plan seems to have led to a different, more dispersed kind of intelligence.

Among the benefits of this distributed model are fast response times, and the ability of multiple parts of the brain to multitask, or work on separate jobs simultaneously. This is true even though the bulk of octopus brain activity still reside in the central brain. The brain acts as a coordinator of the eight other functions and a central memory unit. This sounds an awful lot like an animal version of edge computing, one of the emerging models for distributed tasks.

So, what’s the similarity between an octopus and edge computing which is also a kind of distributed intelligence system

The advantage of distributed smarts

Among the benefits of this distributed model are fast response times, and the ability of multiple parts of the brain to multitask, or work on separate jobs simultaneously. This is true even though the bulk of octopus brain activity still reside in the central brain. The brain acts as a coordinator of the eight other functions and a central memory unit.

This sounds an awful lot like an animal version of edge computing, an emerging model in the tech industry where a central, typically massive, cloud-based “brain” connects to multiple external units that have more limited processing capability but whose value comes from their proximity to the task at hand.

Location based sensing and actuation logic can use edge computing. These applications require some immediate decision making by the local devices. Then, those edge devices funnel data back to cloud/central brain, where it is collected and analyzed. Resulting lessons are then returned to edge devices for response improvement.