Introduction
Problem solving is a goal-directed cognitive process involving transitioning from an initial state to a desired goal state when the path to that solution is not immediately obvious. Cognitive psychology distinguishes between general strategies (domain-general 'weak methods' applied across novel or unfamiliar tasks) and domain-specific procedures ('strong methods' optimized for structured expert domains).
1. General Strategies: Heuristics and Weak Methods
General strategies are cognitive shortcuts and universal heuristics that require substantial working memory resources. They are employed primarily when specialized schemas or content-specific knowledge are unavailable.
- Means-Ends Analysis: Involves assessing the discrepancy between the current state and the goal state, then creating subgoals to systematically diminish this difference, as conceptualized in Allen Newell and Herbert Simon's General Problem Solver (GPS).
- Working Backward: Beginning at the desired goal state and iteratively tracing steps back to the initial conditions, a method frequently utilized in mathematical proofs and maze navigation.
- Analogical Transfer: Mapping structural relationships from a familiar base or source domain onto an unfamiliar target problem domain.
- Empirical Demonstration: Gick and Holyoak (1980) demonstrated the boundary conditions of general heuristics using Duncker's Radiation Problem. Participants typically struggled to solve the medical challenge spontaneously unless they were explicitly directed to draw a structural analogy from an earlier military scenario involving the dispersion and convergence of troops.
2. Domain-Specific Procedures: Strong Methods
Domain-specific procedures rely on highly organized, automated, and schema-driven knowledge bases tailored to specific fields such as clinical medicine, chess, or physics.
- Empirical Demonstration: In a foundational card-sorting experiment, Chi, Feltovich, and Glaser (1981) evaluated how individuals categorized physics problems:
- Novice Problem Solvers: Categorized problems according to superficial surface features (such as pulleys, inclined planes, or rotational apparatus), relying on slow, domain-general heuristics to search through potential equations.
- Expert Problem Solvers: Disregarded surface attributes and categorized problems based on deep, fundamental structural principles (such as the Law of Conservation of Energy or Newton's Second Law), enabling instantaneous access to automated procedural routines.
3. Transition Mechanism: Anderson's ACT-R Architecture
The progression from domain-general weak heuristics to efficient domain-specific procedures is explained by John R. Anderson's Adaptive Control of Thought-Rational (ACT-R) cognitive architecture, mediated by knowledge compilation.
- Declarative Stage: The beginner encodes instructions as explicit declarative propositions held in working memory, manipulating them through slow, interpretive general heuristics.
- Knowledge Compilation: Through sustained deliberate practice, knowledge transitions into a procedural format via two concurrent operations:
- Proceduralization: Replaces declarative fact-retrieval by directly creating specialized production rules (if-then statements) that map perceptual cues directly to appropriate cognitive actions.
- Composition: Merges a sequence of multiple discrete production rules into a single integrated production step, reducing cognitive friction and attentional demands.
- Procedural Stage: Domain-specific operations execute autonomously and rapidly with minimal cognitive load, bypassing working memory capacity constraints.
Conclusion
While general strategies serve as versatile evolutionary tools providing cognitive flexibility when navigating novel environments, domain-specific procedures represent the pinnacle of cognitive efficiency and expert performance. Superior problem-solving relies on the fluid interplay of both modes depending on task novelty and domain mastery.