Deep neural networks are not a single hypothesis but a language for expressing computational hypotheses

An ideal vision model accounts for behavior and neurophysiology in both naturalistic conditions and designed lab experiments. Unlike psychological theories, artificial neural networks (ANNs) actually perform visual tasks and generate testable predictions for arbitrary inputs. These advantages enable...

Full description

Saved in:
Bibliographic Details
Published in:The Behavioral and brain sciences 2023-12, Vol.46, Article e392
Main Authors: Golan, Tal, Taylor, JohnMark, Schütt, Heiko, Peters, Benjamin, Sommers, Rowan P., Seeliger, Katja, Doerig, Adrien, Linton, Paul, Konkle, Talia, van Gerven, Marcel, Kording, Konrad, Richards, Blake, Kietzmann, Tim C., Lindsay, Grace W., Kriegeskorte, Nikolaus
Format: Article
Language:English
Subjects:
Benchmarks
Experiments
Human performance
Hypotheses
Neural networks
Open Peer Commentary
Vision
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:An ideal vision model accounts for behavior and neurophysiology in both naturalistic conditions and designed lab experiments. Unlike psychological theories, artificial neural networks (ANNs) actually perform visual tasks and generate testable predictions for arbitrary inputs. These advantages enable ANNs to engage the entire spectrum of the evidence. Failures of particular models drive progress in a vibrant ANN research program of human vision.
ISSN:0140-525X
1469-1825
DOI:10.1017/S0140525X23001553