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

Simple, individually unique, and context-dependent learning methods for models of human category learning.

Toshihiko Matsuka1

  • 1Rutgers University, Newark, New Jersey, USA. tmatsuka@stevens.edu

Behavior Research Methods
|September 21, 2005
PubMed
Summary

Gradient descent is a standard but limited model for category learning. A new stochastic optimization model, SCODEL, offers a more descriptive, context-dependent, and individually unique approach to human learning.

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

The psychological link between decisions and communicative behavior based on verbal probabilities.

Cognition·2025
Same author

The Observer's Lens: The Impact of Personality Traits and Gaze on Facial Impression Inferences.

Journal of eye movement research·2024
Same author

Conscious observational behavior in recognizing landmarks in facial expressions.

PloS one·2023
Same author

Familiarity-Matching: An Ecologically Rational Heuristic for the Relationships-Comparison Task.

Cognitive science·2020
Same author

Do People Explicitly Make a Frame Choice Based on the Reference Point?

Frontiers in psychology·2019
Same author

Typicality or Fluency?

Experimental psychology·2018

Area of Science:

  • Cognitive Science
  • Computational Neuroscience
  • Machine Learning

Background:

  • Gradient descent is a widely used algorithm in computational models of category learning.
  • However, it functions as a normative rather than a descriptive model of human learning.
  • Existing models face challenges with complexity, regularity, and context independence.

Purpose of the Study:

  • To introduce a novel, hypothesis-testing-like learning algorithm as an alternative to gradient descent.
  • To address the limitations of existing models by proposing a more descriptive and flexible approach.
  • To develop a model that captures individually unique and context-dependent learning processes.

Main Methods:

  • Development of SCODEL, a new learning algorithm based on stochastic optimization.

Related Experiment Videos

  • Conducting four simulation studies to evaluate the model's performance.
  • Comparing SCODEL's capabilities against the limitations of traditional gradient descent models.

    • Main Results:

    • SCODEL provides qualitatively simple interpretations of category learning processes.
    • The model demonstrates the ability to depict individually unique learning patterns.
    • SCODEL successfully captures context-dependent learning dynamics.
    • Simulation results indicate SCODEL can function as diverse learner types in various scenarios.

    Conclusions:

    • SCODEL offers a promising alternative to gradient descent for modeling human category learning.
    • The model's flexibility allows for more realistic and nuanced representations of individual and contextual influences.
    • Future research can explore SCODEL's application in more complex cognitive tasks and real-world learning environments.