Prodigy:
Methodological Assumptions

1. Unified Architecture Hypothesis

   Intelligent behavior arises out of internal reasoning and is not simply a response to external stimuli. Such reasoning requires an internal world model and this hypothesis is an implicit corollary to the Physical Symbol System Hypothesis. As such, this hypothesis is consistent with most work in artificial intelligence. However, there are systems with no explicit representation such as the subsumption architecture which do not rely on this assumption.

2. Maximum Rationality Hypothesis

   Prodigy is designed so that it attempts to achieve maximum rationality consistent with its goals and knowledge within the constraints imposed by the architecture and the environment.

3. Deliberative Reasoning Hypothesis

   Prodigy learns deliberatively rather than reflexively, meaning that it may bring knowledge to bear on the decisions of when and what to learn (as well as determining if learned knowledge should be retained). These decisions are made in order to maximize the utility of learned knowledge.

4. Glass Box Hypothesis

   Prodigy uses a uniform, declarative knowledge representation which is penetrable throughout the architecture. Additionally, all knowledge may be uniformly accessed by all modules of the architecture.

5. Multiple Learning Methods Hypothesis

   Prodigy does not subscribe to a universal learning method like Soar but rather uses a number of different learning mechanisms. These methods allow Prodigy to be used as a test-bed for exploring relationships between problem-solving and learning. A central assumption to the use of these modularized learning methods is that interference between different methods either does not exist or can be completely described (and thus anticipated).

6. Environmental Consistency Hypothesis

   In Prodigy, the environment is assumed to change much slower (if at all) with respect to the speed of the reasoning and learning mechanisms. Thus, environmental consistency is assumed.

7. Not Motivated by Psychological Validity

   Unlike other architectures (e.g., Soar and Teton), Prodigy's architecture is not meant to exemplify particular aspects of human problem solving performance or learning. Instead, Prodigy is posited as a "model of an idealized rational intelligence" and offers a structure in which to explore the properties of planning and learning.

Current Location: Prodigy-Methodological Assumptions

Go to NEXT page.