Utilize este identificador para referenciar este registo:
|Título:||The contextual nature of conditional reasoning|
|Autor:||Byrne, Ruth M. J.|
|Data de Defesa:||1986|
|Editora:||University of Dublin|
|Resumo:||In this thesis, it is proposed that the products of human rea-soning- are a function of the context in which that reasoning occurs. Consequently, the effects of two kinds of contexts are addressed, in two series of experiments.
In the first series, the environmental context, in which conditional reasoning is embedded, was explored. In Experiment 1 subjects were given conditional arguments, accompanied by contextual information consisting of a second conditional. The antecedent in the second conditional was either an alternative condition or an additional condition. It was found that alternative conditions suppress the inferences traditionally considered fallacious, while additional conditions suppress the inferences sanctioned as logically valid. It was also found that when these alternative or additional conditions are conjoined or disjoined In the minor premises that accompany the con¬ditional arguments, this suppression no longer occurs,
In Experiment 2, groups of subjects were again given conditional
arguments that were either accompanied by extra contextual information or not. Further groups were given conditional arguments that were prefaced either by Information about the relevance of contextual Information, or by tasks requiring the retrieval of contextual Information. It was found that neither recognizing the relevance of contextual Information nor retrieving that Information in isolation, is enough to suppress the inferences. It was found once more, that
contextual Information, which embodies both of these features, suppresses the inferences, replicating the primary results of Experiment 1.
In Experiment 3, groups of subjects were given conditional inducements. One group received inducements unaccompanied by any contextual Information. Other groups received them accompanied by information about the situation in which the utterance was made, or by information about the duration of the utterance (either that it was long or short), or accompanied by information about both the situation and the duration. It was found that information about both the situation and that it was a long duration suppresses the invalid inferences, while information about the situation and that it was a short duration suppresses the valid inferences. Neither situational information, which was hypothesized to enable the retrieval of relevant information, nor durational information, which was hypothesized to enable the recognition of the relevance of other information, were singly sufficient to suppress the inferences. A model of inference-testing, based on the conjoined operation of the processes of recognizing the relevance of other information and retrieving specific instances of the relevant information, is described.
A second way that the context in which conditional reasoning is embedded can affect reasoning was investigated, In Experiment 4, the frequency of inferences made from sequences of conditional premises
was compared to the frequency of inferences made from those premises in isolation from each other. Subjects were given conditional prem¬ises of different forms. It was found that fewer inferences are made on sequences of premises than from those premises in isolation from each other, for certain forms of premises.
In Experiment 5, two features, identified as potentially responsible for this, were manipulated. Some subjects received the sequences of conditional premises as before, while others received the premises in isolation, again as before, both of the form for which differences were observed in Experiment 4. Other subjects received arguments constructed to ensure that either the joint presentation of the premises was preserved, or that the uncertainty of using an intermediate conclusion as a minor premise from which to make an inference was preserved, to establish the respective roles of these factors. It was found that the source of differences in the frequency of inferences between sequences of premises and those prem¬ises in isolation can be attributed to the joint representation of the premise information.
Experiment 6 determined the role of some of the features of the form of argument for which differences were observed. Groups of sub¬jects were presented with two different forms of sequences of premise, and comparable premises in isolation. One of these new forms possessed one of the features, that of a negative in the first condi¬tional, while the second sequence possessed two of the features that of a negative in the first conditional with that negative located at the consequent of the conditional. Neither possessed the third feature, of the atransitivity of the argument. It was found that neither of these sequences of premises differed from comparable premises in isolation, in the frequency of inferences made. This indicates that all three features are responsible for the joint representation of arguments of certain sequences leading to fewer inferences. A model is proposed that shows how the representation of sequences of premises differs from the representation of premises in isolation. Finally, the implications of the effects of context on conditional reasoning found in these experiments, for the four theories of reasoning described in the introductory review, and for cognitive psychology in general, are explored in the final chapter.|
|Descrição:||Tese de Doutoramento apresentada à University of London|
|Aparece nas colecções:||PCOG - Tese de doutoramento|
Todos os registos no repositório estão protegidos por leis de copyright, com todos os direitos reservados.