Figure 1.

Conceptual approach and experimental design. (A) A cartoon representation of applying concepts from behavioral science, such as classical conditioning, towards controlling a biological network. Knowledge is not needed of how a dog’s brain changes during the training session that pairs the sound of a bell (stimulus) to a drooling response, but it is possible to form such a memory. In the same way, we need not understand the intricacies of a biological network in order to train a stimulus to a response. (B) Given a biological network model, such as the example on the far right, potential memory circuits are built that test the model with every possible pairing of stimulus (UCS, violet), response (R, blue), and potential conditioned stimulus (CS, yellow). The number of potential memory circuits is therefore the number permutations of N nodes with three choices (i.e., six in this cartoon). (C) Cartoon of analytical pipeline. Thirty-five biological models are used to build both potential memory circuits as well as 500 random networks. All networks are tested for memory, habituation, and sensitization (not shown). (A) is adapted from [67], with permission.