Principles of Synthetic Intelligence PSI: An Architecture of Motivated Cognition

by Bach, Joscha

• ISBN13: 9780195370676
• ISBN10: 0195370678
• Edition: 1st
• Format: Hardcover
• Copyright: 2009-04-06
• Publisher: Oxford University Press

Summary

From the Foreword: "In this book Joscha Bach introduces Dietrich Dorner's PSI architecture and Joscha's implementation of the MicroPSI architecture. These architectures and their implementation have several lessons for other architectures and models. Most notably, the PSI architecture includes drives and thus directly addresses questions of emotional behavior. An architecture including drives helps clarify how emotions could arise. It also changes the way that the architecture works on a fundamental level, providing an architecture more suited for behaving autonomously in a simulated world. PSI includes three types of drives, physiological (e.g., hunger), social (i.e., affiliation needs), and cognitive (i.e., reduction of uncertainty and expression of competency). These drives routinely influence goal formation and knowledge selection and application. The resulting architecture generates new kinds of behaviors, including context dependent memories, socially motivated behavior, and internally motivated task switching. This architecture illustrates how emotions and physical drives can be included in an embodied cognitive architecture. The PSI architecture, while including perceptual, motor, learning, and cognitive processing components, also includes several novel knowledge representations: temporal structures, spatial memories, and several new information processing mechanisms and behaviors, including progress through types of knowledge sources when problem solving (the Rasmussen ladder), and knowledge-based hierarchical active vision. These mechanisms and representations suggest ways for making other architectures more realistic, more accurate, and easier to use. The architecture is demonstrated in the Island simulated environment. While it may look like a simple game, it was carefully designed to allow multiple tasks to be pursued and provides ways to satisfy the multiple drives. It would be useful in its own right for developing other architectures interested in multi-tasking, long-term learning, social interaction, embodied architectures, and related aspects of behavior that arise in a complex but tractable real-time environment. The resulting models are not presented as validated cognitive models, but as theoretical explorations in the space of architectures for generating behavior. The sweep of the architecture can thus be larger-it presents a new cognitive architecture attempting to provide a unified theory of cognition. It attempts to cover perhaps the largest number of phenomena to date. This is not a typical cognitive modeling work, but one that I believe that we can learn much from." --Frank E. Ritter, Series Editor Although computational models of cognition have become very popular, these models are relatively limited in their coverage of cognition-- they usually only emphasize problem solving and reasoning, or treat perception and motivation as isolated modules. The first architecture to cover cognition more broadly is PSI theory, developed by Dietrich Dorner. By integrating motivation and emotion with perception and reasoning, and including grounded neuro-symbolic representations, PSI contributes significantly to an integrated understanding of the mind. It provides a conceptual framework that highlights the relationships between perception and memory, language and mental representation, reasoning and motivation, emotion and cognition, autonomy and social behavior. It is, however, unfortunate that PSI's origin in psychology, its methodology, and its lack of documentation have limited its impact. The proposed book adapts Psi theory to cognitive science and artificial intelligence, by elucidating both its theoretical and technical frameworks, and clarifying its contribution to how we have come to understand cognition.

Table of Contents

Machines to explain the mind	p. 3
From psychology to computational modeling	p. 6
Classes of cognitive models	p. 16
Symbolic systems and the Language of Thought Hypothesis	p. 19
Cognition without representation?	p. 24
Machines of cognition	p. 26
Cognitive science and the computational theory of mind	p. 26
Classical (symbolic) architectures: Soar and ACT-R	p. 31
Hybrid architectures	p. 37
Alternatives to symbolic systems: Distributed architectures	p. 38
Agent architectures	p. 42
Cognition and Affect-A conceptual analysis of cognitive systems	p. 45
Dorner's "blueprint for a mind"	p. 53
Terminological remarks	p. 55
An overview of the PSI theory and PSI agents	p. 57
A simple autonomous vehicle	p. 64
An outline of the PSI agent architecture	p. 68
Representation of and for mental processes	p. 75
Neural representations	p. 75
Associators and dissociators	p. 77
Cortex fields, activators, inhibitors and registers	p. 78
Sensor neurons and motor neurons	p. 78
Sensors specific to cortex fields	p. 79
Quads	p. 79
Partonomies	p. 81
Alternatives and subjunctions	p. 83
Sensory schemas	p. 84
Effector/action schemas	p. 85
Triplets	p. 86
Space and time	p. 87
Basic relationships	p. 89
Memory organization	p. 92
Episodic schemas	p. 93
Behavior programs	p. 93
Protocol memory	p. 95
Abstraction and analogical reasoning	p. 98
Taxonomies	p. 101
Perception	p. 102
Expectation horizon	p. 103
Orientation behavior	p. 104
HyPercept	p. 104
How HyPercept works	p. 105
Modification of HyPercept according to the Resolution Level	p. 108
Generalization and specialization	p. 109
Treating occlusions	p. 110
Assimilation of new objects into schemas	p. 110
Situation image	p. 111
Mental stage	p. 113
Managing knowledge	p. 113
Reflection	p. 114
Categorization ("What is it and what does it do?")	p. 115
Symbol grounding	p. 116
Behavior control and action selection	p. 119
Appetence and aversion	p. 120
Motivation	p. 121
Urges	p. 122
Motives	p. 122
Demands	p. 123
Fuel and water	p. 123
Intactness ("Integritäet;t", integrity, pain avoidance)	p. 124
Certainty ("Bestimmtheit", uncertainty reduction)	p. 124
Competence ("Kompetenz", efficiency, control)	p. 126
Affiliation ("Okayness", legitimacy)	p. 128
Motive selection	p. 129
Intentions	p. 132
Action	p. 133
Automatisms	p. 134
Simple Planning	p. 134
"What can be done?"-the Trial-and-error strategy	p. 136
Modulators	p. 137
Activation/Arousal	p. 138
Selection threshold.	p. 139
Resolution level	p. 139
Sampling rate/securing behavior	p. 140
The dynamics of modulation	p. 141
Emotion	p. 143
Classifying the PSI theory's emotion model	p. 145
Emotion as a continuous multidimensional space	p. 147
Emotion and motivation	p. 151
Emotional phenomena that are modeled by the PSI theory	p. 152
Language and future avenues	p. 157
Language comprehension	p. 158
Matching language symbols and schems	p. 159
Parsing grammatical language	p. 159
Handling ambiguity	p. 162
Learning language	p. 163
Communication	p. 164
Problem solving with language	p. 166
"General Problem Solver"	p. 167
Araskam	p. 167
Antagonistic dialogue	p. 168
Language and consciousness	p. 169
Directions for future development	p. 171
Döet;rner's PSI agent implementation	p. 173
The Island simulation	p. 173
PSI agents	p. 178
Perception	p. 180
Motive generation (Genlnt)	p. 181
Intention selection (Selectlnt)	p. 182
Intention execution	p. 183
Events and situations in EmoRegul and Island agents	p. 183
Modulators	p. 185
Pleasure and displeasure	p. 186
The behavior cycle of the PSI agent	p. 188
Emotional expression	p. 192
From PSI to MicroPSI: Representations in the PSI model	p. 195
Properties of the existing PSI model	p. 197
A formal look at PSI's world	p. 199
Modeling the environment	p. 202
Analyzing basic relations	p. 204
The missing "is-a" relation	p. 207
Unlimited storage-limited retrieval	p. 209
The mechanics of representation	p. 210
Solving the Symbol Grounding Problem	p. 211
Localism and distributedness	p. 219
Missing links: technical deficits	p. 222
Missing powers: Conceptual shortcomings	p. 226
The passage of time	p. 226
The difference between causality and succession	p. 226
Individuals and identity	p. 227
Semantic roles	p. 229
The MicroPSI architecture	p. 233
A framework for cognitive agents	p. 234
Towards MicroPSI agents	p. 237
Architectural overview	p. 238
Components	p. 240
Representations in MicroPSI: Executable compositional hierarchies	p. 246
Definition of basic elements	p. 247
Representation using compositional hierarchies	p. 254
Execution	p. 258
Execution of hierarchical scripts	p. 260
Script execution with chunk nodes	p. 263
The MicroPSI Framework	p. 265
Components	p. 266
The node net editor and simulator	p. 268
Creation of agents	p. 270
Creation of entities	p. 271
Manipulation of entities	p. 272
Running an agent	p. 273
Monitoring an agent	p. 273
Providing an environment for agent simulation	p. 274
The world simulator	p. 276
Setting up a world	p. 278
Objects in the world	p. 279
Connecting agents	p. 280
Special display options	p. 280
Controlling agents with node nets: an example	p. 282
Implementing a PSI agent in the MicroPSI framework	p. 286
The world of the SimpleAgent	p. 288
The main control structures of the SimpleAgent	p. 289
The motivational system	p. 292
Perception	p. 295
Simple hypothesis based perception (HyPercept)	p. 296
Integration of low-level visual perception	p. 297
Navigation	p. 300
Summary: The PSI theory as a model of cognition	p. 303
Main assumptions	p. 304
Parsimony in the PSI theory	p. 312
What makes Döet;rner's agents emotional?	p. 314
Is the PSI theory a theory of human cognition?	p. 318
Tackling the "Hard Problem"	p. 321
References	p. 325
Author Index	p. 358
Subject Index	p. 363
Table of Contents provided by Ingram. All Rights Reserved.