Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Exemplars, prototypes, and the flexibility of classification models.

Henrik Olsson1, Pia Wennerholm, Urban Lyxzèn

  • 1Department of Psychology, Uppsala University, Box 1225, SE-751, Uppsala, Sweden. henrik.olsson@psyk.uu.se

Journal of Experimental Psychology. Learning, Memory, and Cognition
|July 9, 2004
PubMed
Summary

This study investigated model flexibility in cognitive categorization. Findings show that both prototype and exemplar models can overfit data, highlighting the need to distinguish true cognitive processes from error variance.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Social networks affect redistribution decisions and polarization.

PNAS nexus·2025
Same author

Development and retrospective validation of an artificial intelligence system for diagnostic assessment of prostate biopsies: study protocol.

BMJ open·2025
Same author

Physical Color Calibration of Digital Pathology Scanners for Robust Artificial Intelligence-Assisted Cancer Diagnosis.

Modern pathology : an official journal of the United States and Canadian Academy of Pathology, Inc·2025
Same author

Networks of beliefs: An integrative theory of individual- and social-level belief dynamics.

Psychological review·2024
Same author

Analogies for modeling belief dynamics.

Trends in cognitive sciences·2024
Same author

Effectiveness and Cost-effectiveness of Artificial Intelligence-assisted Pathology for Prostate Cancer Diagnosis in Sweden: A Microsimulation Study.

European urology oncology·2024

Area of Science:

  • Cognitive Psychology
  • Computational Neuroscience
  • Machine Learning in Cognitive Science

Background:

  • Previous research (Minda & Smith, 2001; Nosofsky & Zaki, 2002) presented conflicting findings on prototype vs. exemplar model performance in categorization tasks.
  • Discrepancies in model performance were potentially attributable to differences in category complexity, size, or response mechanisms.

Purpose of the Study:

  • To investigate whether excessive model flexibility, specifically over-fitting, could explain the divergent results reported for prototype and exemplar models.
  • To evaluate the generalization performance of prototype and exemplar models under conditions of linearly separable category structures.

Main Methods:

  • Employed cross-validation techniques to assess model performance and generalization.

Related Experiment Videos

  • Tested both a standard prototype model and an exemplar model incorporating a response-scaling mechanism.
  • Utilized linearly separable category structures to probe model fitting and error variance.

    • Main Results:

    • Demonstrated that both the prototype model and the exemplar model with response-scaling exhibited over-fitting (over-filling) within the tested linearly separable category structure.
    • Indicated that model performance metrics can be inflated by fitting to error variance rather than genuine cognitive processes.

    Conclusions:

    • Over-fitting poses a significant challenge for both prototype and exemplar models, potentially explaining prior conflicting findings.
    • Emphasizes the critical need for robust model evaluation methods to ensure that selected models accurately reflect underlying cognitive mechanisms and avoid capitalizing on random error.