INDUCT · Investigating Normative and Descriptive Uses of Confirmation Theory
7РП — „Хора“ (Действия „Мария Кюри“)
- Период
- 2009-06-22 → 2010-06-21
- Финансиране от ЕС
- 53 632 €
- Участници
- 1
- Схема
- MC-IEF
Линиите свързват координатора с партньорите.
Накратко на български
Байесовите модели на потвърждението изследват как хората оценяват доказателствата за една хипотеза, например защо често грешим при изчисляване на вероятности. Резултатите помагат за по-доброто разбиране на индуктивното мислене в епистемологията и експерименталната психология.
Кратко обяснение, генерирано от езиков модел по текста на CORDIS. Оригиналът е по-долу.
Резултати накратко
Investigating normative and descriptive uses of confirmation theory
The main objectives of the INDUCT project were to clarify the normative and descriptive status of the Bayesian models of confirmation. During this year, I carried out two experimental studies and relevant theoretical analysis. The main experimental achievements of the project are provided by the results of two empirical studies which show that: - confirmation judgments are more accurate and reliable than probability judgments; - intuitive assessments of confirmation relations are crucially involved in major probabilistic fallacies, such as the conjunction fallacy. The two manuscripts reporting these studies are currently under preparation. In addition to the two main experimental studies, I co-authored other articles concerning related topics. In particular: - how second-order probability affects hypothesis confirmation; - how to solve the paradox of irrelevant conjunction. The results obtained from these empirical and theoretical studies provide a new perspective on the inductive reasoning topic, and could have significant implications for both epistemology and experimental psychology, as well as for specific applied domains. Concerning the training objectives, I have greatly benefited from attending Prof. Chater's laboratory, and a summer school in cognitive science and machine learning. I have also been invited as a speaker at three workshops/conferences, and at a seminar which took place at the University of Leuven. These meetings provided me an important opportunity to exchange ideas and embark on new projects with other researchers working on the same topics in different universities. Finally, as far as it concerns the contribution to my career development, I successfully applied for an Associate Professor position which was opened at the University of Trento. Relevant contact details: katya.tentori@unitn.it
Текст от CORDIS, на английски · Данни: CORDIS, © Европейски съюз
Цел на проекта
Inductive reasoning concerns how evidence affects the credibility of hypotheses, thus playing a crucial role in many cognitive processes (e.g. learning, diagnosis, prediction, etc.) and in several applied settings (e.g. law, medicine, etc.). It includes two related but distinct issues: (i) the assessment of degrees of belief in hypotheses in light of available evidence, and (ii) the assessment of the net impact of a piece of evidence on a hypothesis, i.e., the degree of confirmation provided by the former to the latter. Issue (i) has been subject of thorough investigation. Standard probability calculus is seen as the appropriate normative benchmark, especially by contemporary Bayesian theorists, on the basis of Dutch-book arguments. Also, inquiries in cognitive psychology have extensively tested the empirical validity of the normative account, identified its limitations and suggested various descriptive models to account for them. Issue (ii), however, has not attained comparable attention and consensus. A plurality of non-equivalent formal models of confirmation have been put forward, whose normative adequacy is still disputed. Confirmation has been experimentally investigated by cognitive psychologists in rather limited domains, and discussions of the connection between the normative and descriptive level have been sporadic. The present research project aims at filling this gap by pursuing the following goals in an integrated fashion: - extending preliminary theoretical and experimental results on both the normative and descriptive virtues and limitations of Bayesian models of confirmation; - providing a behavioral basis for the comparison of alternative models of confirmation in terms of a new analysis of the value of information; - exploring implications of confirmation theory in the psychology of thinking at large and in applied settings, as the estimation of probative value of evidence when deciding its admissibility in judicial proceedings.
Оригинален текст от CORDIS (на английски).
Участници
- UNIVERSITY COLLEGE LONDON · LONDONКоординаторнепознат регион
Връзки
Данни: CORDIS, © Европейски съюз