Loose Ends

This chapter presents some ideas about Ersatz Brain Theory, which generalizes models presented in the book. It is based on three equal components: computation, cognition, and neuroscience. In the Ersatz Brain, the basic computing elements are locally interconnected groups of neurons, for example, cortical columns, and not single neurons. Columns are more powerful than neurons alone because of the potential for selectivity and reliability. A “network of networks” modular architecture is formed from interconnected groups. Response selection emerges from the stability properties of dynamical systems. Traveling waves and interference patterns also grow naturally out of dynamics and local connections. The resulting systems operate using similar rules at multiple spatial scales for different levels of integration.