TY - JOUR
T1 - The psychology of knights and knaves
AU - Rips, Lance J.
N1 - Funding Information:
*I thank David Cortner and Rita Walter for their help and advice on this project. NIMH Grant MH39633 supported the research. Experiment 1 was conducted at the University of Arizona with support from the Committee for Cognitive Science. Correspondence should be addressed to Lance Rips, Behavioral Sciences Department, University of Chicago, 5848 South University, Chicago, IL 60637. USA.
PY - 1989/3
Y1 - 1989/3
N2 - Knight-knave brain teasers are about a realm in which some people, knights, tell only truths, whereas all others, knaves, tell only lies. For example, suppose person A says, "I am a knight and B is a knight," and person B says, "A is a knave." Is A a knight or a knave? Is B a knight or a knave? In a pilot study, we asked subjects to think aloud while solving problems like these. Their statements suggested that they were making assumptions about the knight/knave status of the characters and drawing deductive inferences from these assumptions to test their consistency. This encouraged us to model the process by means of a simulation based on an earlier natural-deduction theory of reasoning. The model contains a set of deduction rules in the form of productions and a working memory that holds a proof of the correct answer. The greater the number of steps (assumptions and inferences) in the proof, the greater the predicted difficulty of the puzzle. The experiments reported here confirmed this prediction by showing that subjects were more likely to make mistakes (Experiment 1) and take longer to solve (Experiment 2) puzzles associated with a larger number of proof steps.
AB - Knight-knave brain teasers are about a realm in which some people, knights, tell only truths, whereas all others, knaves, tell only lies. For example, suppose person A says, "I am a knight and B is a knight," and person B says, "A is a knave." Is A a knight or a knave? Is B a knight or a knave? In a pilot study, we asked subjects to think aloud while solving problems like these. Their statements suggested that they were making assumptions about the knight/knave status of the characters and drawing deductive inferences from these assumptions to test their consistency. This encouraged us to model the process by means of a simulation based on an earlier natural-deduction theory of reasoning. The model contains a set of deduction rules in the form of productions and a working memory that holds a proof of the correct answer. The greater the number of steps (assumptions and inferences) in the proof, the greater the predicted difficulty of the puzzle. The experiments reported here confirmed this prediction by showing that subjects were more likely to make mistakes (Experiment 1) and take longer to solve (Experiment 2) puzzles associated with a larger number of proof steps.
UR - http://www.scopus.com/inward/record.url?scp=0024637258&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=0024637258&partnerID=8YFLogxK
U2 - 10.1016/0010-0277(89)90019-X
DO - 10.1016/0010-0277(89)90019-X
M3 - Article
C2 - 2486296
AN - SCOPUS:0024637258
SN - 0010-0277
VL - 31
SP - 85
EP - 116
JO - Cognition
JF - Cognition
IS - 2
ER -