The Piagetian Autonomous Modeler - An architecture for a cognitive system that constructs an internal representation of a real or simulated environment based on its interaction with the environment. The novel aspects of the modeler are: (1) how it spreads activation; (2) its use of schemes to connect the representational elements (monads); (3) its use of multi-strategy inference to extend the internal model; (4) its use of a consolidation component to provide automaticity and forgetting; and (5) its evolution of successful solutions through equilibration (using genetic adaptation).
The system is called Piagetian because it employs the notion of Monads (fundamental representational elements), Schemes (propositions that identify relatonships among internal and external elements), Assimilation (forming schemes by incorporating elements) and Accommodation (cloning and mutating schemes) which are essential to the theories of Human Cognitive Development espoused by Jean Piaget in his 1977 book The Development of Thought: The Equilibration of Cognitive Structures.
The architecture is compatible with Global Workspace Theory and Society of Mind Theory, and is primarily inspired by
- Interaction Cycles and Equilibration (Jean Piaget),
- Observables, Coordinations, Perturbations (Jean Piaget),
- Behavioral Schemas (Gary Drescher),
- Ontology Formation (Bipin Indurkhya; Marc Pickett & Tim Oates),
- Dynamic Interlaced Hierarchies (Ryszard Michalski),
- Neural Propositions (Michael Miller),
- Database Semantics (Roland Hausser),
- Microtheories (Doug Lenat),
- Impediment Ascription / Failure Prediction (Kristian Hammond)
- Temporal Activation (Michael Miller),
- Cognitive System Patterns (Michael Miller),
- Mind Servers (Michael Miller)
Proceedings of Artificial Intelligence and Simulation of Behavior 2011 (AISB-11) Symposium on Computational Models of Cognitive Development - York, United Kingdom | 2011
Proceedings of The Sixth International Conference on Complex, Intelligent and Software Intensive Systems (CISIS-12) - Palermo, Italy | 2012 - video
The Neural Proposition: Structures for Cognitive Systems
AAAI Press - AAAI 2013 Spring Symposium on Creativity and (Early) Cognitive Development (SSS-13) - Stanford, CA, USA | 2013 - video
The Construction of Reality in a Cognitive System
Poster for AAAI 2013 Workshop on Learning Rich Representations from Low Level Sensors (RepLearn-13) - Bellevue, WA, USA | 2013 (in press)
May 2013 - Coordination and Reflection
Ideas to detail and expand the inferential components.
Ideas on Attention, Regulation, Compensation, Simulation, and Consolidation.
Apr 2013 - Let’s Build a Mind Server
Meetup group to build the Mind Server platform.
See Discuss Mind Servers.
Mar 2013 - Solvers
Formed some ideas about integrating multiple subconscious problem solvers into PAM-P2 (after re-reading Minsky’s “Society of Mind”).
Feb 2013 - Premise Language REPL
Defining the PREMISE language and implementing a REPL to interpret and store propositions in scheme memory.
See Premise REPL.
Jan 2013 - Scheme Memory
Looking for a good implementation for propositional memory.
(Update) Version 1.1.7 of scheme memory is defined. as of Feb 2013.
See Scheme Memory
Oct 2012 - Prototyping phase.
Looking for development collaborators (Java, Clojure, Sockets, JADE, SQL).
- Need to build the PREMISE interpreter for a new domain specific language.
- Need help to build a MIND SERVER on Amazon Cloud or Open Stack Nova.
(If interested please contact.)
Sept 2012 - Mind Servers
Formed some ideas about integrating PAM-P2 with devices. (Shopping a paper to various conferences).
July 2012 - CISIS-12
Presented paper on “Patterns for Cognitive Systems” at CISIS-12, In Palermo, Italy. This paper answered the question posed in April 2011 on finding a good sub-architecture. Some of the patterns presented…
Oct 2011 - Re-architected PAM-P2 using Cognitive System Patterns.
An improved architecture for the modeler has been created which satisfies the augmented system requirements (see picture above).
Apr 2011 - Attention Management Architectures
New ideas ~ Looking for a good sub-architecture for attention management: i.e., model switching between prospection (the forward model), retrospection (the episode model), and the perception-response (the current model).
Apr 2011 - AISB-11
Comments and Inquiries welcome: PiagetModeler@Hotmail.com
Also on Facebook: Piaget Modeler
Copyright (c) 2010-2012 PAM-P2 Group. All Rights Reserved. Patents Pending.