The recognition of critical patterns in one's history causes changes toward a learning goal.

This work is currently ongoing. Here we will first attempt to describe very simple models that implement the basic concepts and cognitive architectural design decisions that will hopefully facilitate the building of control systems that are robust through adaptation. Our current plan for building adaptation primitives is based on simple forms of pattern recognition and reaction. The novelty of our approach will be focusing our reactive learning algorithms on the run-time causal forms of processes, synthesizing and subsequenty recognizing novel critical steady states in these causal process histories. These forms of reflection will further allow for novel forms of layered learning, reasoning, and other forms of reflectively layered adaptation. We believe that creating new forms of reflective process modularity, such as reflective adaptation layers, are a necessary and inherent component for designing powerful new models for understanding learning and reasoning in Humans and otherwise.

All prerequisites for the code listed in this document is currently compiled into the Funk2 default system image.

This is an example of a simple list structure containing symbolic content representing an English sentence.

This is a second example of a list structure containing symbolic content:

This is an example of a variable list structure. Note that lists (denoted by braces [ and ]) that begin with ? are symbolic variables.

This is a second example of a variable list structure.

Knowledge of steady states allows our system to track current steady states and also have knowledge of goal states, anti-goal states, and other forms of knowledge about how to react to different forms of steady states. Knowledge of transitions between steady states allows our system to predict the effects of actions based on previously recognized temporal transitions.

λ