Project

Back to overview

Strategy Selection

English title Strategy Selection
Applicant Marewski Julian
Number 144413
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 01.04.2013 - 31.03.2016
Approved amount 176'876.00
Show all

Keywords (3)

ACT-R; Strategy Selection; memory models

Lay Summary (German)

Lead
Beim Treffen von Urteilen oder Entscheidungen können Menschen häufig auf mehrere Strategien zurückgreifen, welche sich zum Beispiel darin unterscheiden, wie viel Wissen in der Entscheidung berücksichtigt wird. Mit unserem Forschungsprojekt untersuchen wir die Frage der Strategieselektion: Wie wählen Menschen zwischen verschiedenen Strategien wenn sie mit einem bestimmten Entscheidungsproblem konfrontiert sind?
Lay summary

Ein ökologischer Ansatz der Strategieselektion (siehe Marewski & Schooler, 2011) geht davon aus, dass die Anzahl anwendbarer Strategien in einer konkreten Situation von den Eigenschaften des kognitiven Systems begrenzt wird, was die Strategieselektion durch die Reduktion auf die Menge von anwendbaren Strategien vereinfacht. Diese Begrenzung ergibt sich daraus wie Strategien mit kognitiven Kapazitäten, wie dem Gedächtnis, zusammenspielen und Regelmäßigkeiten in der Umwelt abbilden. Das Ziel unseres Forschungsprojekts ist es, die Anwendungsbereiche verschiedener Strategien in Abhängigkeit von Regelmäßigkeiten in der Umwelt, wie der Häufigkeit der von der Strategie benötigten Informationen, zu modellieren.

Im ersten Projekt nutzen wir das in ACT-R (Anderson et al., 2004) implementierte Gedächtnismodell um auf Basis der Antreffenswahrscheinlichkeit von Informationen in der Umwelt detailliertes Wissen über die Attribute von Objekten vorherzusagen. Im zweiten Projekt modellieren wir auf Basis dieses Gedächtnismodells die Anwendungsbereiche verschiedener Strategien welche verschiedene Mengen an Informationen benötigen. Um weiterhin die Strategieselektion in Situationen zu modellieren in denen mehrere Strategien anwendbar sind, integrieren wir in einem dritten Projekt unser Gedächtnismodell mit einem komplementären Modell der Strategieselektion, mit Rieskamp und Otto’s (2006) Modell des Verstärkungslernens, welches annimmt dass Personen lernen, Strategien in Abhängigkeit von deren Treffsicherheit auszuwählen. Mit unserem Vorhaben hoffen wir einerseits das Strategieselektionsproblem zu adressieren, und weiterhin zur Entwicklung übergreifender Modelle der Kognition beizutragen (vgl. Newell, 1990).

REFERENZEN

Anderson, J. R., Bothell, D., Byrne, M. D., Douglass, S., Lebiere, C., & Qin, Y. (2004). An integrated theory of the mind. Psychological Review, 111, 1036–1060.

Marewski, J. N., & Schooler, L. J. (2011). Cognitive Niches: An ecological model of strategy selection. Psychological Review, 118, 393-437.

Newell, A. (1990). Unified Theories of Cognition. Cambridge, MA: Harvard University Press.

Rieskamp, J., & Otto, P. E. (2006). SSL: A theory of how people learn to select strategies. Journal of Experimental Psychology: General, 135, 207–236.

Direct link to Lay Summary Last update: 21.08.2013

Lay Summary (English)

Lead
When making inferences, one can often choose between several strategies that differ, for example, in the amount of knowledge they need to come up with a decision. Our research aims at implementing different strategies and their interaction with basic cognitive capacities such as memory and time perception into the cognitive architecture ACT-R. This will allow us to model the strategies’ applicability and accuracy and predict which of these strategies people will select in different situations.
Lay summary

An ecological approach to strategy selection (see Marewski & Schooler, 2011) suggests the number of executable strategies to be limited by the workings of the cognitive system, simplifying strategy selection by reducing the consideration set of applicable strategies. This limitation arises from how the strategies and cognitive capacities, such as memory, interact and represent regularities in the environment. The goal of our research project is to model the areas of applicability (or cognitive niches) for different knowledge-based strategies as a function of the environmental frequency of the knowledge required by these strategies. In our first project, we use the memory model embedded in the ACT-R theory (Anderson et al., 2004) to predict people’s declarative knowledge of attributes of real-world objects based on the frequency with which this information is likely to be encountered in people’s environments. In our second project, we use this knowledge to model the areas of applicability for strategies requiring different amounts of declarative knowledge. To further model how strategy selection is achieved in areas where more than one strategy is applicable, in a third project, we integrate our memory model with a complementary model of strategy selection, namely with Rieskamp and Otto’s (2006) reinforcement learning model which assumes people to learn to select strategies as a function of the strategies’ accuracy in a given decision environment. With our research, we will help to address the strategy selection problem as well as contribute to building integrative, encompassing models of cognition (cf. Newell, 1990).

 

REFERENCES:

Anderson, J. R., Bothell, D., Byrne, M. D., Douglass, S., Lebiere, C., & Qin, Y. (2004). An integrated theory of the mind. Psychological Review, 111, 1036–1060. 

Marewski, J. N., & Schooler, L. J. (2011). Cognitive Niches: An ecological model of strategy selection. Psychological Review, 118, 393-437.

Newell, A. (1990). Unified Theories of Cognition. Cambridge, MA: Harvard University Press.

Rieskamp, J., & Otto, P. E. (2006). SSL: A theory of how people learn to select strategies. Journal of Experimental Psychology: General, 135, 207–236.

Direct link to Lay Summary Last update: 21.08.2013

Responsible applicant and co-applicants

Employees

Publications

Publication
Do people order cues by retrieval fluency when making probabilistic inferences?
Dimov Cvetomir M., Link Daniela (2017), Do people order cues by retrieval fluency when making probabilistic inferences?, in Journal of Behavioral Decision Making, Advance online publication(doi: 10.10), 1.
Human-like machines: Transparency and comprehensibility
Patrzyk P. Link D. & Marewski J. (2017), Human-like machines: Transparency and comprehensibility, in Behavioral and Brain Sciences, 40, E276.
An ecological model of memory and inferences
Link Daniela, Marewski Julian N., Schooler Lael J. (2016), An ecological model of memory and inferences, in Proceedings of the 38th Annual Conference of the Cognitive Science Society, Cognitive Science Society, Austin, TX, USA.
Unpacking decision domains. Commentary on "Domain-specific preferences for intuition and deliberation in decision making"
Olds Justin, Link Daniela (2016), Unpacking decision domains. Commentary on "Domain-specific preferences for intuition and deliberation in decision making", in Journal of Applied Research in Memory and Cognition, 5(3), 325-328.
Populating ACT-R's declarative memory with internet statistics
Link Daniela, Marewski Julian N. (2015), Populating ACT-R's declarative memory with internet statistics, in Proceedings of the 13th International Conference on Cognitive Modeling.
Strategy selection: An introduction to the modeling challenge
Marewski Julian, Link Daniela (2014), Strategy selection: An introduction to the modeling challenge, in Wiley Interdisciplinary Reviews: Cognitive Science, 5(1), 39-59.
Strategy selection: A theoretical and methodological challenge
Marewski Julian N., Bröder Arndt, Glöckner Andreas, Strategy selection: A theoretical and methodological challenge, in Journal of Behavioral Decision Making, (Special Is).

Collaboration

Group / person Country
Types of collaboration
Max Planck Institute for Human Development Germany (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Publication
- Exchange of personnel
Syracuse University United States of America (North America)
- in-depth/constructive exchanges on approaches, methods or results

Scientific events

Active participation

Title Type of contribution Title of article or contribution Date Place Persons involved
38th Annual Conference of the Cognitive Science Society Talk given at a conference An ecological model of memory and inference 10.08.2016 Philadelphia, United States of America Link Daniela; 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;
25th Subjective Probability, Utility, and Decision Making Conference Individual talk Does memory accessibility affect the order in which information about decision alternatives is considered? 16.08.2015 Budapest, Hungary Link Daniela;
Summer School on Theories and Methods in Judgment and Decision Making Research Poster Environmental frequencies and the use of take-the-best 09.08.2015 Schwarzenbruck, Germany Link Daniela;
37th Annual Conference of the Cognitive Science Society Poster The cognitive niches of knowledge-based decision strategies 22.07.2015 Pasadena, CA, United States of America Marewski Julian; Link Daniela;
13th International Conference on Cognitive Modeling Poster Populating ACT-R's declarative memory with internet statistics 09.04.2015 Gronigen, Netherlands Marewski Julian; Link Daniela;
3rd European Summer School on Computational Modeling of Cognition with Applications to Society Poster Can fluency approximate validity? 27.07.2014 Laufen, Germany Link Daniela;
Right patient, right place, right time. When and ho to use the emergency department: health literacy, decision-making, and communication Individual talk Heuristic decision making in medicine. Can simple strategies lead to better decisions? 11.04.2014 Università Cattolica del Sacro Cuore, Milan, Italy Link Daniela;
24th Subjective Probability, Utility, and Decision Making Conference Talk given at a conference Strategy selection: an unresolved modeling challenge 18.08.2013 Barcelona, Spain Marewski Julian;
3rd ACT-R Spring School and Master Class Talk given at a conference Towards modeling strategy selection based on internet statistics 12.04.2013 Groningen, Netherlands Link Daniela;


Self-organised

Title Date Place

Associated projects

Number Title Start Funding scheme
146702 Architectural Process Models of Decision Making 01.01.2014 Project funding

Abstract

The thesis that people possess a repertoire of decision strategies to choose from has been formulated in many areas, including inferential decision making, choice, social interactions, mathematical skill, categorization, and question answering. If one adopts this view, a major problem is then to determine how people select from these strategies to solve given tasks-the strategy selection problem. This problem represents a major theoretical challenge in the cognitive and decision sciences, but it is also a stumbling block for theories in other disciplines, including, for instance, economics, biology, and machine learning, where the problem presents itself in terms of the selection of actions, behaviors, algorithms, operators, routines, and production rules. In this project, I contribute to solving the strategy selection puzzle by extending an existing quantitative model of strategy selection. This model predicts how selection emerges through the interplay among the strategies, the workings of basic cognitive capacities, such as memory, and the structure of the environment in which we humans live. Specifically, in Project 1, I extend the memory component of Marewski and Schooler’s (2011) model of strategy selection to quantitatively predicting people’s detailed declarative knowledge of attributes of real-world objects (e.g., detailed knowledge about companies). This extension is necessary to tackle the strategy selection problem for a number of classic models of decision making, including noncompensatory lexicographic and compensatory (e.g., weighted-additive) integration strategies, all of which operate on such knowledge (i.e., by using the knowledge to make decisions). In Project 2, I use the extended memory model to model these knowledge-based strategies’ cognitive niches, that is, the range of situations in which the strategies are applicable. A strategy’s applicability depends, for instance, on whether sufficient knowledge about an object’s attributes can be retrieved. To further model how the interplay of the environment and cognitive capacities leads people to learn to prefer one applicable strategy over another, in Project 3, I integrate the extended model with another existing model of strategy selection, namely Rieskamp and Otto’s (2006) reinforcement learning model. Tackling Projects 1-3 will (a) help to address the strategy selection problem for classic decision strategies, providing a refined theory that allows quantitatively predicting which of these strategies people may rely on when making inferences about unknown states of the world. Along the way, I provide a theoretically-grounded method to (b) populate models of memory with realistic, detailed declarative knowledge about real-world objects as well as (c) a theory that allows remedying the experimental decision scientist’s methodological problem of identifying which of various plausible decision strategies a person may use in a decision task. Finally, by integrating the extended memory model and Rieskamp and Otto’s (2006) model, I (d) contribute to building integrative, encompassing models of cognition (cf. Anderson et al., 2004; Newell, 1990).
-