Object recognition along the ventral stream: effects of visual noise
Subramanian, A.; Tüncok, E.; Kurzawski, J. W.; Majaj, N. J.; Pelli, D. G.; Winawer, J.
Show abstract
Psychophysical studies using critical-band masking have established that human object recognition is most impaired by stimulus noise in a narrow, 1.5 octave band of spatial frequency. This shows that visual recognition relies on this narrow band. Here, we investigate the biological basis of this recognition band. We used fMRI to measure BOLD responses along the ventral stream (V1 [->] V2 [->] V3 [->] V4 [->] ventral temporal cortex, VTC) to band-pass noise and to natural images perturbed by that noise. The response to noise alone widens from 2 octaves in V1 to 5 octaves in VTC, diverging more and more from the behavioural 1.5 octave recognition bandwidth. In contrast, a correlation classifier applied to the BOLD response to noisy images shows that the recognition bandwidth is conserved at 2 octaves. This contrast shows that the recognition band is more like a readout constraint than an early sensory filter. Furthermore, this 2-octave band becomes increasingly tolerant to noise, approaching behavioral tolerance in VTC. Thus, it seems that recognition improves along the ventral stream by maintaining a narrow object channel while increasing noise tolerance.
Matching journals
The top 3 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Constructing and Forgetting Temporal Context in the Human Cerebral Cortex 97%
- When the ventral visual stream is not enough: A deep learning account of medial temporal lobe involvement in perception 96%
- Fast recurrent processing via ventral prefrontal cortex is needed by the primate ventral stream for robust core visual object recognition 95%
Similar papers in this journal
- The relative coding strength of object identity and nonidentity features in human occipito-temporal cortex and convolutional neural networks 96%
- Temporal dynamics of neural responses in human visual cortex 95%
- Probabilistic representation in human visual cortex reflects uncertainty in serial decisions 95%
Similar papers in this journal
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.