When taking a computer-style approach to the question of how information might be managed inside the brain, fundamental architectural conflicts emerge. To avoid those, the brain needs to operate with about 100 or fewer “straight-line� neurons between thought and muscle activation. Parallel processing needs to happen so computer-style complexity and addressing and administration challenges do not exist. This points toward a switching architecture as opposed to a processing architecture. In addition, the complex nature of the brain suggests an integrated feedback structure is needed to make sense out of complex information coming from different sensory inputs.

In line with these criteria, the proposed session introduces the Neural Network Switching Model. This model aligns with the emerging view of the brain operating in a pattern-forming, self-organizing way (and with mini-columns). The session proposes how, with a multi-sensory feedback structure embedded in the model, the brain is able to make sense of highly complex information such as understanding the meaning of a sentence in which the letters are mixed up at word level.