The thesis discusses an architecture called TalaMind for design of AI systems. The architecture has three levels: the linguistic, archetype, and associative levels. At the linguistic level, the architecture includes the Tala language, a conceptual framework for managing concepts expressed in Tala, and conceptual processes that operate on concepts in the conceptual framework to produce intelligent behaviors and new concepts. The thesis discusses how the architecture at the linguistic level could support higher-level mentalities in human-level AI.
The archetype level is where cognitive categories are represented using methods such as conceptual spaces, image schemas, radial categories, etc. The associative level would typically interface with a real-world environment and supports connectionism, Bayesian processing, etc. In general, the thesis is agnostic about research choices at the archetype and associative levels.
The architecture is open at the three conceptual levels, for instance permitting predicate calculus, conceptual graphs, and other symbolisms in addition to the Tala language at the linguistic level, and permitting integration across the three levels, e.g. potential use of deep neural nets at the linguistic and archetype levels. The TalaMind architecture is actually a broad class of architectures, because it is open to design choices at each level.
The thesis hypotheses do not require a ‘society of mind’ architecture in which subagents communicate using the Tala conceptual language as an interlingua, yet it is consistent with the hypotheses and natural to implement a society of mind at the linguistic level of the TalaMind architecture. This is illustrated in the TalaMind prototype software.
Note that the term ‘society of mind’ is used in a broader sense than the approach described by (Minsky 1986), to refer to a multiagent system open to methods for organizing and communication between agents other than the methods specified by Minsky, e.g. including languages of thought. This broader, generalized sense corresponds more to a paper by Doyle (1983). See section 126.96.36.199.1 of (Jackson 2014) for further information.
The TalaMind Architecture for Human-Level AI