The effect of knowledge and context in the elicitation of inferences in class-inclusion reasoning
Titel:
The effect of knowledge and context in the elicitation of inferences in class-inclusion reasoning
Auteur:
Richard, J. F. Leynet, M. E.
Verschenen in:
European journal of cognitive psychology
Paginering:
Jaargang 5 (1993) nr. 2 pagina's 149-171
Jaar:
1993-06-01
Inhoud:
Two experiments were carried out with adult subjects to investigate the conditions of production of inferences in a problem-solving task involving reasoning about class inclusion. We tested the hypothesis that the production of appropriate inferences is facilitated whenever pieces of information are stored in memory that can be matched at least in part with the conclusion resulting from the inference. In Experiment 1, problems involving the same inferential pattern but belonging to three different domains were compared. The inclusion relations were of the type “is a kind of” (semantic domain), “more than” (being 8 years old is being more than 7 years old: logico-mathematical domain) or “what I have received on a given day is a part of what I have in the evening” (possession domain). In the semantic domain, the critical inference may be facilitated by the semantic network of concepts stored in memory. In the logico-mathematical domain, there is no information that corresponds to the critical inference, so that inferences have to be elaborated merely by use of general knowledge about the meaning of the relation. The possession domain is an intermediate case, since the inference corresponds to knowledge about relations in the possession domain. The results clearly support a memory-cueing hypothesis: There is a marked order of difficulty ranging from the logico-mathematical domain (most difficult) to the semantic domain (least difficult). In Experiment 2, equivalent problems using the same conceptual information were compared: The critical inference involved scanning a series of numbers either in direct order (more than 7 is 8, 9, 10) or in reverse order (8, 9, 10 is more than 7). The number of correct solutions was higher and the response latency much shorter for those problems in which the critical inference was of the type “more than 7 is 8, 9 or 10''.
Tijdschrift beschrijving