ICCM: Architectures

The Common Model of Cognition (CMC) has been proposed as a high level framework through which functional neuroimaging data can be predicted and interpreted. Previous work has found the CMC is capable of predicting brain activity across a variety of tasks, but it has not been tested on resting state data. This paper adapts a previously used method for comparing theoretical models of brain structure, Dynamic Causal Modeling (DCM), for the task-free environment of resting state, and compares the CMC against six alternate architectural frameworks. For a large sample of subjects from the Human Connectome Project (HCP), the CMC provides the best account of resting state brain activity, suggesting the presence of a general purpose structure of connections in the brain that drives activity when at rest and when performing directed task behavior.

Autonomous systems are a new frontier for pushing sociotechnical advancement. Such systems will eventually become pervasive, involved in everything from manufacturing, healthcare, defense, and even research itself. However, proliferation is stifled by the high development costs and the resulting inflexibility of the produced systems. The current time needed to create and integrate state of the art autonomous systems that operate as team members in complex situations is a 3-15 year development period, often requiring humans to adapt to limitations in the resulting systems. A new research thrust in interactive task learning (ITL: Laird et al., 2017) has begun, calling for natural human-autonomy interaction to facilitate system flexibility and minimize users’ complexity in providing autonomous systems with new tasks. We discuss the development of an undifferentiated agent with a modular framework as a method of approaching that goal.

A common complaint levied at analyses based on cognitive architectures is their lack of connection to observed functional neuroimaging data, particularly for architectural models that rely on high level, theoretical components of cognition. Previous work has connected task-based functional MRI data to the Common Model of Cognition (CMC), using a top-down modeling approach. Here, a bottom-up method, Granger Causality Modeling (GCM), is applied to the same task-based data to infer a network of causal connections between brain regions based on their functional connectivity. The resulting network shares many connections with those proposed by the Common Model.

As the study of individual differences becomes more widespread, questions arise about the reasons that a particular individual might adopt a particular strategy. Using both the behavioral and functional neuroimaging data of healthy adults from Human Connectome Project (HCP) we examined decision making in an incentive processing task (Delgado et al. 2000). A pair of distinct ACT-R models, representing a Declarative strategy and a Procedural strategy, were used to classify subjects as either Declarative or Procedural decision makers based on their behavioral data. A machine learning Lasso analysis was performed on each subject’s resting state functional connectivity, and was able to match the ACT-R model classifications to a high degree of accuracy. This suggests that the strength of connections between brain regions may play an important role in shaping the decision making process of a given individual.