Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Videos

Fluctuation-dissipation theorem and models of learning.

Ilya Nemenman¹

¹Kavli Institute for Theoretical Physics, University of California, Santa Barbara, CA 93106, USA. ilya@nemenman@columbia.edu

Neural Computation

|July 5, 2005

Summary

Determining how brains learn is complex. Researchers found that observing learning curves alone is insufficient to identify the specific computational learning theory used by an organism.

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

Random-with-constraints: Constructing minimal models for high-dimensional biology.

Proceedings of the National Academy of Sciences of the United States of America·2026

Same author

The hierarchical timescale hypothesis: Functional and structural convergence of biological networks and artificial neural nets.

Cell systems·2026

Same author

Distribution of singular values in large sample cross-covariance matrices.

Physical review. E·2025

Same author

Randomness with constraints: constructing minimal models for high-dimensional biology.

ArXiv·2025

Same author

Physics-tailored machine learning reveals unexpected physics in dusty plasmas.

Proceedings of the National Academy of Sciences of the United States of America·2025

Same author

A mathematical model for ketosis-prone diabetes suggests the existence of multiple pancreatic β-cell inactivation mechanisms.

eLife·2025

Area of Science:

Computational neuroscience
Statistical learning theory
Theoretical neuroscience

Background:

Statistical learning theory offers many models of learning machines.
Understanding which models biological systems, like brains, implement is a key challenge.

Purpose of the Study:

To investigate the challenges in identifying the specific learning-theoretic computations used by biological information processors.
To propose an alternative experimental approach to address this challenge.

Main Methods:

Analysis of abstract Bayesian learners' performance on diverse datasets.
Evaluation of learning curves for stationary targets.

Main Results:

Performance on stationary target learning (learning curves) is insufficient to distinguish between different abstract Bayesian learning models.
Existing performance metrics may not reveal the underlying computational mechanisms.

Conclusions:

Distinguishing between various learning-theoretic computations in biological systems requires more than just analyzing learning curves.
A novel experimental setup, inspired by the fluctuation-dissipation relation from statistical physics, may offer a solution.

Related Experiment Videos