Reading focus
Syllabus: How Do We Reason and Solve Problems?
Identify how representation defines the available problem space and how knowledge guides search. Examine the article's recognition and representation examples alongside its search account.
Why you're reading it. Build an explicit process explanation that the later analogy and insight papers can extend or challenge.
A question to consider. For a puzzle with many possible moves, how would you tell whether improvement came from searching more efficiently or from representing the problem differently?
Why this reading
A problem becomes more tractable when we specify a starting state, a goal, and possible moves. Newell and Simon explain how searching such a space can produce intelligent performance, and why representation and prior knowledge can matter as much as the amount of search. They also make a much broader claim about the requirements for intelligence.
Before you read
A problem space describes possible states and the operations that connect them. A heuristic guides which possibilities to explore without guaranteeing the best answer. Means–ends analysis uses differences between the current state and a goal to choose operations or subgoals. A physical symbol system manipulates patterns that can designate things and processes; the proposal concerns an implemented system, not disembodied symbols.
Read for
- Separate the physical-symbol-system hypothesis from the more specific account of heuristic search. Identify what evidence would be needed for each.
- Compare blind enumeration with search guided by knowledge. Follow how a program's success becomes a hypothesis about human performance rather than direct evidence of the same internal process.
- Examine the roles of recognition and problem representation in reducing search. Ask where the framework assumes that a useful space and its operators have already been supplied.
Consider afterwards
- A traveler must reach a station before it closes. Specify states, legal moves, a goal test, and one heuristic. How would the account change if the traveler discovered a previously overlooked transport option?
Editorial context
This Turing Award lecture is a foundational statement of symbolic cognitive science and artificial intelligence. The claim that physical symbol systems are necessary and sufficient for general intelligent action is stronger than the claim that heuristic search explains particular tasks. It should be evaluated as a research hypothesis, not treated as settled by successful programs. The article also recognizes the importance of selecting representations, a useful limit on accounts that begin with a fixed problem space.