Berkeley and Heiserman: A Foundation for Embodied AI Architecture

This paper re-evaluates the foundational work of Edmund C. Berkeley and David L. Heiserman, suggesting their ideas represent a largely untapped architectural framework for embodied machine intelligence. Berkeley, often recognized for connecting symbolic logic to computing, is presented as having a broader vision: designing machines that acquire, retain, and respond to information while organizing their behavior over time. Berkeley's program, particularly in "Symbolic Logic and Intelligent Machines," emphasizes the operational definition of intelligent machines, linking logical forms directly to hardware realization and describing behavior through the coordinated interaction of inputs, outputs, memory, and control. This perspective moves beyond abstract symbol manipulation to consider temporally extended, environment-coupled behavior in robots. Heiserman's work further extends this by focusing on adaptive creature architectures built around memory, confidence, and generalization. Together, their contributions are framed not as historical relics but as a robust, under-explored architectural blueprint for developing intelligent, embodied robotic systems.