Principle of Rationality

Newell’s bridge law for the knowledge level: an agent will take an action whose preconditions it knows to hold and whose effects it wants. Gives knowledge-level descriptions predictive force.

In this vault

Backlinks

Linked Pages

Knowledge Level

Newell: a level of system description above the symbol level, in which behaviour is explained by what the system knows and wants (principle of rationality) — the theoretical home of agent theory.

In this vault

Knowledge-representation lineage

graph TD
  MC1["McCarthy (1959)<br/>Programs with Common Sense"] --> AT[Advice Taker]
  AT --> MC2["McCarthy (1969)<br/>Epistemological Problems of AI"]
  MIN1["Minsky (1974)<br/>A Framework for Representing Knowledge"] --> FR[Frames]
  MC2 --> CIRC["McCarthy (1980)<br/>Circumscription"]
  CIRC --> NM[Nonmonotonic reasoning]
  FR --> SoM["Minsky (1986)<br/>Society of Mind"]
  NEW["Newell (1982)<br/>The Knowledge Level"] --> KL((Knowledge Level))
  MC2 --> KL
  KL --> BDI[[BDI]]
  KL --> INT["Dennett<br/>Intentional Stance"]
  BRO["Brooks (1991)<br/>Intelligence Without Representation"] -.critique.-> KL
  CC["Clark & Chalmers (1998)<br/>The Extended Mind"] --> SoM
  SoM --> AGT[[Agent Architecture]]
  BDI --> AGT

Papers: Programs with Common Sense · A Framework for Representing Knowledge · Circumscription - A Form of Nonmonotonic Reasoning · The Knowledge Level · The Society of Mind · Intelligence Without Representation · The Extended Mind · Epistemological Problems of Artificial Intelligence

The Knowledge Level

Reference: Allen Newell (1982). “The Knowledge Level.” Artificial Intelligence 18(1): 87-127. (AAAI Presidential Address, Stanford, 19 Aug 1980; also in AI Magazine 2(2), 1981.) URL

Summary

Newell’s AAAI-80 presidential address proposes that above the symbol level (data structures manipulated by programs) there exists a distinct knowledge level whose medium is knowledge and whose laws are rationality. At the knowledge level an agent is characterised by what it knows and what goals it has; its behaviour is predicted by the principle of rationality — “if an agent knows that one of its actions will lead to a goal, it will select that action” — without commitment to any particular representation or processing mechanism.

The paper motivates the proposal by diagnosing three tensions in early-1980s AI: the mystification of “representation,” the residue of the 1960s-70s theorem-proving controversy (Planner/Microplanner/Conniver’s “uniform procedures will not work” reaction against resolution), and the diversity of answers in the SIGART Special Issue on Knowledge Representation (Brachman and Smith’s survey showing no consensus on quantification, quotation, self-description, or evaluation). Newell’s resolution: treat knowledge as a competence-level notion analogous to physicists’ treatment of temperature — a level with its own laws, reduced to but not replaced by the symbol level. Representations are then understood as symbol-level implementations of knowledge, and we gain a principled way to talk about what an AI system knows without confusing it with how that knowledge is stored.

The Knowledge Level is foundational for agent-oriented programming (Shoham’s “mental state” framing), for the BDI tradition, for KQML’s and FIPA’s claim that ACL semantics should be defined at the level of beliefs/intentions rather than byte-strings, and indirectly for McCarthy’s Elephant 2000, whose speech-act semantics presupposes that one can talk coherently about what a program knows and is committed to, independently of implementation.

Key Ideas

Computer systems are describable at multiple levels (device, circuit, logic, register-transfer, program/symbol); the knowledge level sits above the symbol level.
Medium of the knowledge level: knowledge (distinct from the data structures that represent it).
Law of behaviour at the knowledge level: the principle of rationality — actions selected to achieve goals given what is known.
The knowledge level is a competence-level account: it predicts behaviour without specifying mechanism, analogously to economic rationality.
Representation is the symbol-level realisation of knowledge; mystifying representation conflates the levels.
Diagnostic of the theorem-proving backlash: “uniform procedures will not work” was a symbol-level reaction to a knowledge-level question.

Connections

Agent-Oriented Programming — Shoham explicitly situates AOP at the knowledge level; mental states are Newell-level predicates over agents.
Elephant 2000 - A Programming Language Based on Speech Acts — Elephant commits to knowledge-level program variables (propositions, commitments, past speech acts), making knowledge-level description the native semantics of programs.
Ascribing Mental Qualities to Machines — McCarthy’s ascriptivism is the philosophical cousin of Newell’s levels argument.
Knowledge Representation
Agent Communication Languages — ACL semantics in the KQML/FIPA tradition are knowledge-level (beliefs/intentions) rather than symbol-level (bit patterns).
KIF — intended as a lingua franca at the symbol level that preserves knowledge-level content across heterogeneous systems.
Common Sense Reasoning

Conceptual Contribution

Claim: AI systems have a level of description above the symbol level — the knowledge level — whose medium is knowledge and whose law is rationality; this level is reducible to but not replaceable by the symbol level, and recognising it clarifies decades of representation debates.
Mechanism: Levels-of-description methodology imported from computer architecture; principle of rationality as the behaviour law; knowledge characterised competence-theoretically (what follows from what an agent knows and wants, not how); representations reconceived as symbol-level implementations of knowledge-level content; diagnostic application to the theorem-proving controversy and the SIGART Special Issue.
Concepts introduced/used: Knowledge Level, Symbol Level, Principle of Rationality, Competence vs Performance, Knowledge Representation, Levels of Description.
Stance: foundational
Relates to: The conceptual ground for agent-oriented and BDI programming; provides the “what an agent knows” vocabulary that KQML, Agent Communication Languages, and Elephant 2000 - A Programming Language Based on Speech Acts rely on; complements McCarthy’s ascriptivist argument in Ascribing Mental Qualities to Machines.