Explaining Neural Transitions through Resource Constraints

Abstract One challenge in explaining neural evolution is the formal equivalence of different computational architectures. If a simple architecture suffices, why should more complex architectures evolve? The answer must involve intense competition for resources under which brains operate. I show how recurrent networks can be favored when increased complexity allows efficient use existing resources. Although resource constraints alone drive change, recurrence shifts landscape what later evolvable. Hence organisms on either side transition boundary may have similar cognitive capacities but very potential evolving new capacities.