Numerous cognitive scientists believe that a human-level thinking machine must be composed of potentially hundreds of distinct subsystems with different structures, reasoning and learning mechanisms, and knowledge representations —with these components and their inter-relationships defining a (or a family of) dynamic architecture(s). DARPA tasked IDA to help define a program, integrated cognition, to specify architectural strategies and develop infrastructure mechanisms (if needed) that enable computation based systems to achieve a reasonable facsimile of human cognition. IDA developed a new framework to understand the problem of integrated cognition, evaluated the current state of knowledge, and has proposed a straw man architecture to kick-start the effort. The benefits of integrated cognition for thinking machines would be immense. This is truly a DARPA hard problem, the solution of which will have huge potential payoffs, in particular, to realize intelligent C2, C4ISR, and robotic force elements.