S0080 · chase_simon_chess · registry identity
Perception in chess
External resolution
Resolution status: not_found
Bears on
Canonical propositions this source was matched as evidence for.
How Hirsch uses it
Every occurrence, grouped by book and chapter, with the claim it was linked to support and (where the extraction regenerated one) the underlying warrant.
1987 · Cultural Literacy
AI chess programs that defeat human players by using knowledge-based procedures rather than checking every possible variation.
direct Effective artificial intelligence programs succeed by imitating the knowledge-based procedures of human experts rather than using brute-force calculation.
Warrant (implicit): A system that achieves the same high-level output as a human expert by employing similar heuristic shortcuts is fundamentally mimicking the cognitive processes of that expert.
could fail ifThe AI might be using statistical probability models on a scale that humans cannot perform, achieving the same result through a mechanism that is functionally different from human knowledge retrieval.
Simon and Barenfeld's estimate that a chess master recognizes approximately 50,000 positional patterns.
direct There is an upper limit of approximately 50,000 rapidly accessible schemata for any domain of human activity.
Warrant (explicit): If high-level mastery in a complex field consistently correlates with the recognition of a specific number of patterns, that number represents a structural threshold for human search and retrieval efficiency.
could fail ifThe 50,000 figure may be an artifact of the specific complexity of chess or language rather than a biological limit of the human brain's retrieval system.
The statistical coincidence between the number of positional patterns a chess master knows (50,000) and the vocabulary size of a literate person (50,000).
direct A basic vocabulary of 50,000 words or idioms serves as a functional index for a much larger volume of knowledge in long-term memory.
Warrant (implicit): When two distinct domains of high-level human performance show an identical numerical limit for accessible units, those units likely serve the same functional purpose in indexing deeper knowledge.
could fail ifThe numerical coincidence between chess patterns and vocabulary may be a spurious correlation, as the cognitive effort to maintain a word is not necessarily equivalent to maintaining a chess position.
2020 · How to Educate a Citizen
Simon and Chase's follow-up experiment using random chess piece arrangements, showing that all skill levels performed equally poorly (reproducing about six pieces).
direct When chess pieces are placed at random rather than in actual game positions, the memory advantage of grandmasters over novices disappears.
Warrant (implicit): If an expert's advantage is eliminated when the stimuli are stripped of their meaningful structure, then the expertise is based on pattern recognition of known structures rather than superior raw memory.
could fail ifExperts might rely on different heuristics for random data that are less efficient, but this does not prove that their normal performance is solely due to 'erudition' rather than superior processing speed.
Estimate that chess expertise is derived from the stored memory of approximately 50,000 specific game positions.
direct A grandmaster's expertise depends on 'erudition,' or an encyclopedic knowledge of past games that allows for the organization of perceptions into meaningful groupings.
Warrant (implicit): The quantitative volume of stored data required to achieve expert performance determines the qualitative nature of the expertise.
could fail ifThe quantity of stored positions might be a byproduct of expertise rather than its source; expertise could instead reside in the superior logical-deductive strategies used to choose moves.
Raw source_reference strings seen
As they appeared before normalization/consolidation into this source identity.
Herbert Simon and William ChaseSimon and BarenfeldSimon and Barenfeld (1969)Simon and Chase