Skip to content
DraconiticResearch
Search Sources API
203,203 studies31 sourcesUpdated 5 min ago
Peer-reviewedCC BYNature Communications

Higher visual areas act like domain-general filters with strong selectivity and functional specialization

  • Meenakshi Khosla1,2,3
  • Leila Wehbe4,5,6,7
7 affiliations
  1. 1Department of Cognitive Science, University of California, San Diego, CA, USA
  2. 2Department of Computer Science and Engineering, University of California, San Diego, CA, USA
  3. 3Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, MA 0239
  4. 4Machine Learning Department, Carnegie Mellon University, Pittsburgh, PA, USA. [email protected]
  5. 5Neuroscience Institute, Carnegie Mellon University, Pittsburgh, PA, USA. [email protected]
  6. 6Machine Learning Department, Carnegie Mellon University, Pittsburgh, PA 1513
  7. 7Neuroscience Institute, Carnegie Mellon University, Pittsburgh, PA 1521
Open at sourcePDFDOICite

Abstract

Neuroscientific studies rely heavily on a-priori hypotheses, which can bias results toward existing theories. Here, we use a hypothesis-neutral approach to study category selectivity in higher visual cortex. Using only stimulus images and their associated fMRI activity, we constrain randomly initialized neural networks to predict voxel activity. Despite no category-level supervision, units in the trained networks act as detectors for semantic concepts like 'faces' or 'words', providing solid empirical support for categorical selectivity. Importantly, this selectivity is mostly maintained when training the networks without images that contain the preferred category, strongly suggesting that selectivity is not domain-specific machinery, but sensitivity to generic patterns that characterize preferred categories. The ability of the models' representations to transfer to perceptual tasks further reveals the functional role of their selective responses. Finally, our models show selectivity only for a limited number of categories, all previously identified, suggesting that the essential categories are already known.

Related

  1. Functional specialization and dynamical interaction in human amygdala subregions support fearful-expression processing
    Dan Cao, Jiachen Cui, Xinyuan Yan +6
    Communications BiologyPeer-reviewedCC BY-NC-ND 4.0Full textNeuroscienceCrossref
  2. Solute selectivity and functional divergence of aquaporins in reef-building corals
    A. Nagashima, A. Kato
    bioRxivPreprintAll rights reservedFull textEvolutionary biologybioRxiv
  3. Equidistance between Primary Sensory-Motor Landmarks Shapes Functional Specialization in the Human Brain
    Lidón Marin-Marin, Xiuyi Wang, Jonathan Smallwood +2
    The Journal of NeurosciencePeer-reviewedCC BY 4.0Full textNeurosciencePMC
  4. Siphonophore genome structure and the evolution of functional specialization
    Namrata Ahuja, Darrin T. Schultz, Dalila Destanović +10
    PLoS ONEPeer-reviewedCC BYFull text6 citedEvolutionary biologyZoology & animal behaviourOpenAlex
  5. Odorant Receptors Mediating Benzaldehyde Detection in Mythimna separata: Functional Identification and Ligand Selectivity.
    Dongdong Sun, Xiaoqing Wang, Yiping Fan +4
    InsectsPeer-reviewedCC BYFull textZoology & animal behaviourEurope PMC

Cite

BibTeXRIS
BibTeX
@article{khosla2026higher,
  title = {{Higher visual areas act like domain-general filters with strong selectivity and functional specialization}},
  author = {Meenakshi Khosla and Leila Wehbe},
  journal = {Nature Communications},
  year = {2026},
  month = jun,
  doi = {10.1038/s41467-026-73938-9},
  url = {https://doi.org/10.1038/s41467-026-73938-9},
  note = {CC BY},
}
RIS
TY  - JOUR
TI  - Higher visual areas act like domain-general filters with strong selectivity and functional specialization
AU  - Meenakshi Khosla
AU  - Leila Wehbe
PY  - 2026
DA  - 2026/06/12
JO  - Nature Communications
DO  - 10.1038/s41467-026-73938-9
UR  - https://doi.org/10.1038/s41467-026-73938-9
AB  - Neuroscientific studies rely heavily on a-priori hypotheses, which can bias results toward existing theories. Here, we use a hypothesis-neutral approach to study category selectivity in higher visual cortex. Using only stimulus images and their associated fMRI activity, we constrain randomly initialized neural networks to predict voxel activity. Despite no category-level supervision, units in the trained networks act as detectors for semantic concepts like 'faces' or 'words', providing solid empirical support for categorical selectivity. Importantly, this selectivity is mostly maintained when training the networks without images that contain the preferred category, strongly suggesting that selectivity is not domain-specific machinery, but sensitivity to generic patterns that characterize preferred categories. The ability of the models' representations to transfer to perceptual tasks further reveals the functional role of their selective responses. Finally, our models show selectivity only for a limited number of categories, all previously identified, suggesting that the essential categories are already known.
SN  - 2041-1723
ER  -