A cognitive approach to modeling quantitative estimations: The development and test of the mapping model
Zusammenfassung der Projektergebnisse
We designed the project "A cognitive approach to modeling quantitative estimations: The development and test of the mapping model" to pursue two goals. First we aimed to specify under what conditions two recent models of quantitative estimation, the mapping model and an exemplar model, capture the cognitive processes in quantitative estimations. While the mapping model assumes a rule-based estimation process, the exemplar model assumes a similarity-based estimation process. Based on the models' assumptions about the cognitive process, we derived and tested two hypotheses about the conditions under which each of the models would perform well. Rule-based models, such as the mapping model, assume the abstraction of explicit task knowledge and thus should profit from prior knowledge about the cues. In contrast, according to the exemplar memory prediction, an exemplar-based process might be triggered when the stimulus material allows the accurate storage and retrieval of training exemplars. Overall, the results provided clear evidence for task-contingent cognitive processes in accordance with the knowledge abstraction prediction. The mapping model performed best when the participants were informed about the cues' directions or could learn them during training. When abstracting knowledge about the cues was difficult but exemplar memory could be used for accurate estimation, the exemplar model was best in predicting participants' estimations. However, we did not find evidence to support the exemplar memory prediction. A second goal of the project was to generalize the mapping model. The original formulation of the mapping model was limited to dichotomous cues. We developed a generalization to continuous cues and tested this model in an applied setting. We investigated how well the mapping model could describe the sentencing recommendations of prosecutors in theft, fraud, and forgery cases in comparison to four models of human judgment. Testing the various cognitive judgment models against each other revealed that the sentencing process was apparently not consistent with the judgment policy recommended by the legal literature. Instead, the mapping model captured the prosecutors' judgments best, suggesting that legal professionals such as prosecutors may sometimes rely on simple judgment strategies. In a next step, we focused on understanding how people learn to make multiplecue judgments and how the learning procedure may differ depending on the characteristics of the judgment task. We developed a learning procedure for the mapping model and tested it against a linear additive neuronal network model and an exemplar-based connectionist network model in a computer simulation and against experimental data. The computer simulation suggested that the mapping model was well suited to solving a judgment task if the environment had a multiplicative structure; that is, the criterion was a multiplicative function of the cues. However, if the task had a linear structure, the linear additive network model was best for solving the task. Similarly, the results from the experiments suggest that the mapping model describes participants' judgments during learning best in a multiplicative task, but the linear additive model does so best in a linear task. These results suggest that participants settle on a learning procedure early in the learning phase, which then influences how they solve the task.
Projektbezogene Publikationen (Auswahl)
- (2008, July). The mapping model: A cognitive theory for estimartion from multiple cues. Conference for Formal Models in Memory, Judgment, and Decision Making, Mannheim
von Helversen, B., & Rieskamp, J.
- (2009). Models of Quantitative Estimations: Rule-based and Exemplar-Based Processes Compared. Journal of Experimental Psychology: Learning, Memory and Cognition, 35, 867-88
von Helversen, B., & Rieskamp, J.
- (2009). Predicting Sentencing for Low Level Crimes: Comparing Models of Human judgment. Journal of Experimental Psychology: Applied, 15, 37S-39S
von Helversen, B., & Rieskamp, J.
- (2009, April). Heuristics in Legal Decision Making. ZiF-Tagung "Interdisziplinarität in den Rechtswissenschaften - Innen- und Außenperspektiven", Bielefeld
von Helversen, B.
- (2010). Learning in multiple-cue judgment tasks. In S. Ohlsson & R. Catrambone (Eds.), Proceedings of the 32nd Annual Conference of the Cognitive Science Society (pp. 174-179). Austin, TX: Cognitive Science Society
von Helversen, B., & Rieskamp. J.
- (July, 2010). Strategy choices across the lifespan: Strategy learning and strategy choice in multiple-cue judgment. Research colloquium of the department for psychology, Göttingen, Germany
von Helversen, B.
