| Autor: |
Jeffrey S Bowers, Gaurav Malhotra, Marin Dujmović, Milton Llera Montero, Christian Tsvetkov, Guillermo Puebla, Federico G Adolfi, John Hummel, Rachel Flood Heaton, Benjamin D. Evans, Jeffrey Mitchell, Ryan Blything |
| Rok vydání: |
2023 |
| Popis: |
We are pleased there is widespread agreement that psychology has an important role to play in building better models of vision. But there are important confusions and disagreements that we discuss in our response. Our key claim is not that we should reject image-computable DNNs, but rather, image-computable DNNs should be used alongside alternative modelling frameworks to improve our understanding of vision. And when using DNNs to study human vision, a change of focus from prediction-based studies to controlled experiments that test hypotheses is needed. We also discuss a current bias in the field to focus on DNN-human similarities rather than differences. This not only contributes to a false characterization of DNN-human similarities in the domain of vision (and language and other domains), but it also impacts on research practices in ways that delays progress in building better models of mind. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
