A Boolean net trainable as a computing robot control

To implement a Hebbian rule, the authors organize a Boolean net, in control of a robot executing long task sequences, as a set of "node assemblies". The resulting system, the robot plus the neural net, adapts automatically to a changing environment. As an example, the authors exhibit a net learning to call, in the due order and with the right temporal cadence, the robot actuators, so to have them write the sequences of binary digits, every time the numeral name is entered as the the robot input.