Posterior probability judgements: Distinguishing numerical outputs from their underlying reasoning processes.
|
 |
|
Post a Comment
|
|
|
|
|
CONTRIBUTORS:
|
|
|
UNIVERSITY / COLLEGE:
|
|
|
YEAR:
|
2003
|
|
PUB TYPE:
|
Thesis/Dissertation
|
|
PAGES:
|
|
|
SUBJECT(S):
|
posterior probability judgements; reasoning processes; heuristics and biases; cognition; algorithmic accounts; computational accounts.
|
|
DISCIPLINE:
|
Psychology
|
|
HTTP:
|
|
|
LANGUAGE:
|
English
|
|
PUB ID:
|
103-409-461
(Last edited on
2004/11/07 10:45:26 US/Mountain)
|
|
SPONSOR(S):
|
|
|
ABSTRACT:
This thesis aims to further our understanding of reasoning processes that underlie posterior probability judgements, i.e., judgements of the probability that a hypothesis is true, given new relevant information. Previous research has been output-focused and compared actual judgement outputs with normative Bayesian values. The thesis research has developed a new framework of analysis, the process-focused approach. This framework, paralleling work by Marr (1982), defines a new research agenda that entails i) the identification of what judgement methods underlie observed outputs (computational analysis), and ii) the explanation of how individuals’ processing elicits the application of a particular judgement method (algorithmic analysis). The empirical work encompasses the three main posterior probability judgment paradigms and is aimed at distinguishing between competing current computational accounts. Within the Social-judgement paradigm, although aggregated posterior judgements are found to be identical to inverse likelihood judgements, further results show that no existing computational account reliably predicts individual behaviour. Within the Textbook-problem paradigm, outputs are commonly underpinned by the application of the inverse fallacy. This method is also found to be less prevalent when only one of two complementary posterior probability judgements is elicited. In the Set-based paradigm, most judgement outputs are Bayesian, but Bayesian outputs do not necessarily entail successful Bayesian reasoning. This computational analysis of judgements together with the appraisal of existing algorithmic accounts led to the development of the Sample Space Representation theory. SSR theory provides a thorough yet general account (i) explaining how Textbook problems and Set-based tasks are represented, and (ii) characterising the cognitive processes underpinning the application of the judgement methods previously identified.
|
|
|
|
STATISTICS
|
|
Click on # to view
|
|
Citations
|
|
0
|
|
References
|
|
0
|
|
Comments
|
|
0
|
|
Quality
|
|
1/7.00
|
|
Interest
|
|
1/7.00
|
|
View(er)s
|
|
1/120
|
|
|
|
|
|
|
| Prev |
Next |
|