Papale, Paolo and Leo, Andrea and Cecchetti, Luca and Handjaras, Giacomo and Kay, Kendrick and Pietrini, Pietro and Ricciardi, Emiliano Foreground Enhancement and Background Suppression in Human Early Visual System During Passive Perception of Natural Images. bioRxiv. (Submitted) (2017)
|
PDF
- Submitted Version
Available under License Creative Commons Attribution Non-commercial. Download (6MB) | Preview |
Abstract
One of the major challenges in visual neuroscience is represented by foreground-background segmentation, a process that is supposed to rely on computations in cortical modules, as information progresses from V1 to V4. Data from nonhuman primates (Poort et al., 2016) showed that segmentation leads to two distinct, but associated processes: the enhancement of cortical activity associated to figure processing (i.e., foreground enhancement) and the suppression of ground-related cortical activity (i.e., background suppression). To characterize foreground-background segmentation of natural stimuli in humans, we parametrically modulated low-level properties of 334 images and their behaviorally segmented counterparts. A model based on simple visual features was then adopted to describe the filtered and intact images, and to evaluate their resemblance with fMRI activity in different visual cortices (V1, V2, V3, V3A, V3B, V4, LOC). Results from representational similarity analysis (Kriegeskorte et al., 2008) showed that the correspondence between behaviorally segmented natural images and brain activity increases throughout the visual processing stream. We found evidence of foreground enhancement for all the tested visual regions, while background suppression occurs in V3B, V4 and LOC. Our results suggest that foreground-background segmentation is an automatic process that occurs during natural viewing, and cannot be merely ascribed to differences in objects size or location. Finally, neural images reconstructed from V4 and LOC fMRI activity revealed a preserved spatial resolution of foreground textures, indicating a richer representation of the salient part of natural images, rather than a simplistic model of objects shape.
| Item Type: | Article |
|---|---|
| Identification Number: | https://doi.org/10.1101/109496 |
| Subjects: | R Medicine > RC Internal medicine > RC0321 Neuroscience. Biological psychiatry. Neuropsychiatry |
| Research Area: | Computer Science and Applications |
| Depositing User: | Caterina Tangheroni |
| Date Deposited: | 21 Mar 2017 13:32 |
| Last Modified: | 04 Aug 2017 10:14 |
| URI: | http://eprints.imtlucca.it/id/eprint/3677 |
Actions (login required)
 |
Edit Item |
