Adaptron Cognitive Architecture

The Adaptron Cognitive Architecture and Binary Neurons as a General–Purpose Representation


Adaptron is a cognitive architecture that uses compositional hierarchies of binary neurons (binons) as its representational system. Binons are general-purpose relational and functional nodes for representing knowledge, concepts and abilities.  Adaptron satisfies many of the important requirements for artificial general intelligence. These include purposeful, grounded, autonomous, general-purpose, scalable and reliable. Its architecture consists of a behavioural and a mental layer. The behavioural layer is an integrated perception–action hierarchy that interacts with the environment via senses and action devices. It contains a mental model of the world as experienced by Adaptron. It is based on what has been perceived and done. The mental layer recalls percepts as concepts for thinking and rehearsing actions. It controls mental operations such as paying attention and selecting actions. Both layers are composed of binons and operate on the same principles. A binon is a simple deterministic artificial neural node that represents a relationship. It has two lower nodes and is reused by zero or more upper nodes. Binons are general-purpose components that interact in an object-oriented fashion. The four types of binons are name, value, entity and control binons. Name binons represent the category names for types of entities. Value binons are used to represent amodal property values such as position, size, intensity, time and duration. Entity binons represent things and their types such as properties, objects, events and actions. Control binons are used to learn, manage and repeat behavioural and mental processes. Binons are either spatial or temporal. Spatial binons represent things whose features occur simultaneously whereas temporal binons represent things whose features occur sequentially. Action and expectation binons are behavioural control binons. They are equivalent to command neurons in neuroscience, production rules in cognitive science, or the forward and inverse models in motor control. Mental operation binons focus attention, perform thinking and initiate actions. Learning takes place in the five stages of reflexes, babbling, reuse, practice and automaticity. It operates based on the dual process theory. Once binons have been learnt, practiced and are reliable they become automatic habits.