Cognitive Synergy between Procedural and Declarative Learning in the Control of Animated and Robotic Agents Using the OpenCogPrime AGI Architecture

The hypothesis is presented that ”cognitive synergy” – proactive and mutually-assistive feedback between different cognitive processes associated with different types of memory – may serve as a foundation for advanced artificial general intelligence. A specific AI architecture founded on this idea, OpenCogPrime, is described, in the context of its application to control virtual agents and robots. The manifestations of cognitive synergy in OpenCogPrime’s procedural and declarative learning algorithms are discussed in some detail. Supposing one agrees that a broadly ”integrative” approach is an appropriate way to achieve advanced AI functionality. One must then specify what ”integrative” really means. At one extreme, it could mean merely connecting together different software components that solve problems via highly independent internal processes, in such a way that they occasionally pass each other problems to solve and receive each others’ answers. At the other extreme, it could mean binding various components together in tight feedback loops so that the internal dynamics of each component can be effectively understood only in the internal context of the others. While many approaches to integration may be workable, here we are specifically concerned with AI of the ”very tightly integrated” type. Specifically, we describe some theoretical elaborations and practical applications of the concept of cognitive synergy introduced in (Goertzel 2009a), defined roughly as: the fitting-together of different intelligent components into an appropriate cognitive architecture, in such a way that the components proactively and mutually assist each other’s internal operations, regularly coming to each others rescue ”mid thought process” in situations where ineffective cognition is observed or anticipated. We describe some aspects of a research program designed to explore (and leverage the potential truth of) the hypothesis that the cognitive synergy ensuing from integrating multiple symbolic and subsymbolic learning and memory components in an appropriate cognitive architecture and environment, can ultimately produce artificial general intelligence (AGI) at the human level or beyond. For discussion of what is meant Copyright c © 2011, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved. by the term AGI, see (Goertzel and Pennachin 2005); in brief, it is intended to refer to the broadly human-like capability to solve problems in a variety of domains without domain-specific training, including potentially domains unknown to the system or its programmers at the time of system creation, and the ability to generalize knowledge from a set of domains to dramatically different domains. This approach fits neatly with what is known about human neurobiology. The human brain is an integration of an assemblage of diverse structures and dynamics, built using common components and arranged according to a sensible cognitive architecture. Its algorithms and structures have been honed by evolution to work closely together – they are very tightly inter-adapted, in the same way that the different organs of the body are adapted to work together. Due their close interoperation they give rise to the systemic behaviors that characterize human-like general intelligence. At the broadest level, there are four primary challenges in constructing a cognitive synergy based AGI system: 1. choosing an overall cognitive architecture that possesses adequate richness and flexibility for the task of achieving advanced cognition 2. choosing appropriate AI algorithms and data structures to fulfill each of the functions identified in the cognitive architecture (e.g. visual perception, audition, episodic memory, language generation, analogy,...) 3. ensuring that these algorithms and structures, within the chosen cognitive architecture, are able to cooperate in such a way as to provide appropriate coordinated, synergetic intelligent behavior ( critical since advanced cognition is an integrated functional response to the world, not a loosely coupled collection of capabilities) 4. embedding one’s system in an environment that provides sufficiently rich stimuli and interactions to enable the system to use this cooperation to ongoingly create an intelligent internal world-model and self-model We describe here an AGI-oriented architecture called OpenCogPrime (OCP), based on the open-source OpenCog project http://opencog.org, which is fundamentally reliant on the cognitive synergy concept, and which is currently being used (in research projects) to control animated agents in video game worlds, and (via the OpenCogBot Proceedings of the Twenty-Fifth AAAI Conference on Artificial Intelligence