Jove
Visualize
Contact Us

Related Experiment Videos

When more is less: negative exposure effects in unsupervised learning.

John P Clapper1

  • 1Department of Psychology, California State University, 5500 University Parkway, San Bernardino, CA 92407-2397, USA. jclappe@csusb.edu

Memory & Cognition
|October 27, 2006
PubMed
Summary
This summary is machine-generated.

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

Graded similarity in free categorization.

Cognition·2019
Same author

Alignability-based free categorization.

Cognition·2017
Same author

The impact of training sequence and between-category similarity on unsupervised induction.

Quarterly journal of experimental psychology (2006)·2014
Same author

The effects of prior knowledge on incidental category learning.

Journal of experimental psychology. Learning, memory, and cognition·2012
Same author

Prior knowledge and correlational structure in unsupervised learning.

Canadian journal of experimental psychology = Revue canadienne de psychologie experimentale·2007
Same author

Adaptive categorization in unsupervised learning.

Journal of experimental psychology. Learning, memory, and cognition·2002
Same journal

Music enhances associative generalization: Evidence from a memory integration task.

Memory & cognition·2026
Same journal

Video, text, and memory: An emotional verbal overshadowing effect.

Memory & cognition·2026
Same journal

Limited protective effects of multilingualism against age-related cognitive decline.

Memory & cognition·2026
Same journal

Validation of illustrated texts: Can pictures raise awareness of inconsistencies?

Memory & cognition·2026
Same journal

4I remember (and forget) your happy smiling face: Directed forgetting of emotionally expressive faces of in-group and out-group members.

Memory & cognition·2026
Same journal

Identity in the spotlight: Matching faces without overlapping features.

Memory & cognition·2026
See all related articles
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

Unsupervised learning may involve discrete category invention, not just correlation tracking. Experiments showed increased training instances negatively impacted learning, supporting models that allow for such non-monotonic effects.

Area of Science:

  • Cognitive Science
  • Computational Neuroscience
  • Machine Learning

Background:

  • Unsupervised learning models are broadly categorized into correlation tracking and category invention.
  • Correlation tracking models predict monotonic increases in learning with exposure.
  • Category invention models allow for non-monotonic effects, such as negative exposure effects.

Purpose of the Study:

  • To compare correlation tracking and category invention models of unsupervised learning.
  • To investigate the impact of training instance exposure on unsupervised learning.
  • To determine if human unsupervised learning aligns with discrete category invention.

Main Methods:

  • Two experiments were conducted to assess unsupervised learning.
  • The number of training instances was manipulated to observe effects on learning.

Related Experiment Videos

  • Human data results were compared against predictions from computational models.

    • Main Results:

    • Increasing training instances negatively affected unsupervised learning, violating monotonicity.
    • A category invention computational model successfully reproduced the human data patterns.
    • A correlation tracking model failed to replicate the observed human data.

    Conclusions:

    • Human unsupervised learning exhibits non-monotonic effects, challenging correlation tracking models.
    • Results provide strong evidence for a discrete category invention process in learning.
    • Category invention models offer a more accurate framework for understanding certain unsupervised learning phenomena.