Project

Back to overview

Architectural Process Models of Decision Making

English title Architectural Process Models of Decision Making
Applicant Marewski Julian
Number 146702
Funding scheme Project funding
Research institution Département de comportement organisationnel Faculté des HEC Université de Lausanne
Institution of higher education University of Lausanne - LA
Main discipline Psychology
Start/End 01.01.2014 - 31.12.2016
Approved amount 188'425.00
Show all

Keywords (3)

Memory models; Decision Strategies; ACT-R

Lay Summary (German)

Lead
Welches ist der beste Parkplatz für mein Auto? Welches Land hat die größere Fläche? Jeden Tag schließen wir aus bekanntem Wissen auf unbekannte Sachverhalte. Ein großer Teil der Entscheidungsforschung beschäftigt sich mit der Frage, wie Menschen solche Inferenzen treffen, wobei eine Vielzahl an unterschiedliche Theorien hervorgebracht wurden. Unsere Forschung zielt darauf ab, die Vorhersagen dieser Theorien vergleichbar zu machen, indem wir sie auf demselben Spezifikationsniveau beschreiben.
Lay summary
Um zwischen den Vorhersagen verschiedener Theorien quantitativ differenzieren zu können, ist es zunächst notwendig, diese Theorien auf derselben, hoch-detaillierten Beschreibungsebene zu spezifizieren, die neben Entscheidungsprozessen auch Annahmen über visuelle, manuelle und Gedächtnisprozesse beinhaltet. Unser erstes Ziel ist es deswegen zunächst verschiedene Entscheidungstheorien in der kognitiven Architektur ACT-R zu implementieren, ein System, welches unterschiedliche kognitive Prozesse wie Wahrnehmung, Motorik und Gedächtnis vereint. Wir beginnen mit einer Programmierung mehrerer möglicher Versionen von zehn verschiedenen Entscheidungstheorien aus der Literatur in ACT-R. In einem zweiten Schritt identifizieren wir die plausibelste Version der Implementation der Theorien anhand der Daten von Versuchspersonen, welche in einem Instruktionsparadigma angeleitet werden, genau einer der Theorien zu folgen Anschließend testen wir, welche der Entscheidungstheorien das Verhalten von Versuchspersonen in einer künstlichen sowie in einer natürlichen Entscheidungsumwelt am besten beschreiben. In einem vierten Projekt reanalysieren wir mit Hilfe unserer Modelle die Daten vorangegangener Inferenzstudien. Unser Projekt wird nicht nur zur Beantwortung der Frage beitragen, anhand welcher Theorien der Entscheidungsprozess in verschiedenen Situationen am besten beschrieben werden kann, sondern auch eine höhere Präzision von Vorhersagen in den Entscheidungswissenschaften ermöglichen.
Direct link to Lay Summary Last update: 02.05.2013

Responsible applicant and co-applicants

Employees

Publications

Publication
Connections between ACT-R’s declarative memory system and MINERVA2
Dimov Cvetomir (2016), Connections between ACT-R’s declarative memory system and MINERVA2, in Proceedings of the Cognitive Science Society, Texas, Austin.Not applicable., Not applicable..
Modeling and Aiding Intuition: Commentary Section
Hoffrage Ulrich (ed.) (2016), Modeling and Aiding Intuition: Commentary Section, Journal of Applied Research in Memory and Cognition, This is a special issue..
Modeling and Aiding Intuition: Introduction to the Commentary Section
Marewski Julian, Hoffrage Ulrich, Fisher Ronald (2016), Modeling and Aiding Intuition: Introduction to the Commentary Section, in Journal of Applied Research in Memory and Cognition, 5, 318-321.
Modeling and Aiding Intuition in Organizational Decision Making
Marewski Julian (ed.) (2015), Modeling and Aiding Intuition in Organizational Decision Making, Journal of Applied Research in Memory and Cognition, This is a special issue of a journal.
Unveiling the Lady in Black: Modeling and aiding intuition
Hoffrage Ulrich, Marewski Julian (2015), Unveiling the Lady in Black: Modeling and aiding intuition, in Journal of Applied Research in Memory and Cognition, 4(3), 145-163.
Cognitive Architectures as a Scaffolding for Risky Choice Models
Dimov Cvetomir, Marewski Julian, Cognitive Architectures as a Scaffolding for Risky Choice Models, in Raue M. Lermer I. Streicher B. (ed.), Springer, New York, NY.
Do people order cues by retrieval fluency when making probabilistic inferences?
Dimov Cvetomir, Link Daniela, Do people order cues by retrieval fluency when making probabilistic inferences?, in Journal of Behavioral Decision Making, BDM2002.
The fast-and-frugal heuristics program
Hoffrage Ulrich, Hafenbrädl Sebastian, Marweski Ulrich, The fast-and-frugal heuristics program, in Ball Linden (ed.), Routledge, Basingstoke, UK.

Collaboration

Group / person Country
Types of collaboration
Univeristy of Syracuse United States of America (North America)
- Publication
Brandenburg Medical School Theodor Fontane Germany (Europe)
- Publication
University of Lausanne Switzerland (Europe)
- Publication
- Research Infrastructure
Max Planck Institute for Human Development Germany (Europe)
- Publication
- Research Infrastructure

Scientific events

Active participation

Title Type of contribution Title of article or contribution Date Place Persons involved
38th Annual Meeting of the Cognitive Science Society Poster Connections between ACT-R’s declarative memory system and Minerva2 11.08.2016 Philadelphia, United States of America Dimov Cvetomir;
7th Thurgau Experimental Economics Meeting Talk given at a conference Validating architectural process models of decision making 07.04.2016 Kreuzlingen, Switzerland Dimov Cvetomir; Marewski Julian;
7th Thurgau Experimental Economics Meeting Talk given at a conference The Time Has Come: Cognitive Architectures meet Behavioral Economics 07.04.2016 Kreuzlingen, Switzerland Marewski Julian;
Invited talk at Max Plank Institute for Human Development, Berlin, Germany Individual talk Predicting Behavioral and Neural Responses During Decision Making With a Cognitive Architecture 09.02.2016 Berlin, Germany Dimov Cvetomir;
25th Subjective Probability, Utility and Decision Making Conference Talk given at a conference Does memory accessibility affect the order in which information about decision alternatives is considered? 17.08.2015 Budapest, Hungary Dimov Cvetomir;
TEAP – Conference of Experimental Psychologists Talk given at a conference Architectural models in decision making: modeling cue-based heuristics in ACT-R 02.04.2014 Giessen, Germany Dimov Cvetomir;


Associated projects

Number Title Start Funding scheme
144413 Strategy Selection 01.04.2013 Project funding
140503 Simple Heuristics for Human Inferences 01.04.2013 Project funding

Abstract

One of the central goals of many theories in the cognitive and decision sciences is to describe the cognitive processes underlying decision making behavior. Yet, there are relatively few theories of decision making that allow making detailed quantitative predictions about process data, such as the pattern of information search, reaction times or eye-fixations. Moreover, the vast majority of theories do not model how decision making interplays with memorial, perceptual, and motor processes, and there are also hardly any connections between the various theories. Finally, the theories are specified at varying levels of description, making it difficult to compare their ability to account for process data. These theoretical and methodological problems fuel important debates about which decisional processes-ranging from simple heuristic ones to complex integration processes-describe behavior best (e.g., Goldstein & Gigerenzer, 2002; Dougherty, Franco-Watkins, & Thomas, 2008; Newell & Bröder, 2008; Scheibehenne, Wagenmakers, & Rieskamp, in press). With this proposal, we aim at remedying these problems by lending quantitative precision to competing models of decision strategies, including heuristics, sequential sampling strategies, connectionist parallel constraint satisfaction mechanisms, exemplar models, and linear integration rules. We do so by implementing these models in the ACT-R architecture (e.g., Anderson, 2007). ACT-R is a computational framework that allows one to integrate formal models of memory, perception, action, and other aspects of cognition. In Project 1, we will develop a data-base of ACT-R implementations of decision theories, thereby focusing on a paradigm for memory-based decisions. In this paradigm, people infer unknown quantities exclusively by extrapolating from knowledge stored in their memories. In Project 2, we will use the ACT-R framework to model decision processes of subjects who are instructed (in Exp. 1) to actually follow the steps formulated in various models of decision strategies. In Project 3, we will examine (Exp. 2) which ACT-R implementation predicts process data best if subjects are not instructed to use a particular strategy.. While Exps. 1-2 will rely on learning paradigms in which subjects first acquire the relevant memory contents prior to making decisions (cf., Bröder & Schiffer, 2003a; Hoffrage, Hertwig, & Gigerenzer, 2000), in Exp. 3, we will test the models in a more naturalistic experimental setting, modeling subjects’ memory contents as they have been acquired outside of the lab, and-just as in Exp. 2-letting them choose freely how to decide (cf. Marewski & Schooler, 2011). Finally, in Project 4, we will further test the ACT-R implementations by re-analyzing existing process data from the literature, this way aiding to resolve the aforementioned controversies. The main target journal is Psychological Review. Moreover, we plan to make all ACT-R model codes available to the public, so that they can also be used by other researchers, aiding to sustainably improve the aforementioned theoretical and methodological problems in the field.
-