Toward Computational, Embeddable, Cognitive Models of Context

Abstract Computational models of context that are compatible with human cognitive representations could be used to augment human decision-making and to enable improved interactive and cooperative capabilities of non-human agents. If unconstrained by human cognitive limitations and biases, such cognitively-inspired models of context could also scale to problem scales and time scales that exceed human capabilities. Expanding on a recently developed theoretical model for context called the Narratively Integrate Multi-level (NIM) framework, this paper presents computational details that operationalize the concepts in the NIM framework and define a set of tools that can be used to design and build human-like models of context and embed them into larger computational systems.