Fiser, Aris. Experience-dependent spatial expectations in mouse visual cortex. 2017, Doctoral Thesis, University of Basel, Faculty of Science.
|
PDF
2866Kb |
Official URL: http://edoc.unibas.ch/diss/DissB_13028
Downloads: Statistics Overview
Abstract
The presence of optical illusions tells us that what we see - at least a large fraction of it – is influenced by our expectations, which are built on our experience of the environment. This means that, at a perceptual level, our brain fills in parts of the visual world with things that aren’t there.
In this work, we have discovered the presence of neurons in the primary visual cortex (V1), the earliest cortical stage of visual processing, that predict the identity of upcoming visual stimuli that mice observed in a virtual environment. We also show that these predictions are dependent on the animals’ spatial location, suggesting that an internal representation of space can serve as a scaffold for predictions. Consistent with an influence representations of space on visual processing, we find that the activity of neurons in V1 was modulated by location in the environment, and it is thus possible to decode not only the stimulus the mouse is observing at any given time, but its location as well, in the case where an identical stimulus is presented in multiple locations.
Furthermore, we identified the anterior cingulate cortex (ACC) as a potential source of stimulus predictions to V1, as V1-projecting ACC axons carried stimulus-predictive activity. Finally, omitting an expected visual stimulus drove strong responses in V1. These results are consistent with a predictive coding framework, wherein predictions of future sensory stimuli are compared to ongoing sensory input, and mismatches between the two lead to error signals.
In this work, we have discovered the presence of neurons in the primary visual cortex (V1), the earliest cortical stage of visual processing, that predict the identity of upcoming visual stimuli that mice observed in a virtual environment. We also show that these predictions are dependent on the animals’ spatial location, suggesting that an internal representation of space can serve as a scaffold for predictions. Consistent with an influence representations of space on visual processing, we find that the activity of neurons in V1 was modulated by location in the environment, and it is thus possible to decode not only the stimulus the mouse is observing at any given time, but its location as well, in the case where an identical stimulus is presented in multiple locations.
Furthermore, we identified the anterior cingulate cortex (ACC) as a potential source of stimulus predictions to V1, as V1-projecting ACC axons carried stimulus-predictive activity. Finally, omitting an expected visual stimulus drove strong responses in V1. These results are consistent with a predictive coding framework, wherein predictions of future sensory stimuli are compared to ongoing sensory input, and mismatches between the two lead to error signals.
Advisors: | Mrsic-Flogel, Thomas and Keller, Georg and Senn, Walter |
---|---|
Faculties and Departments: | 05 Faculty of Science > Departement Biozentrum > Former Organization Units Biozentrum > Neural Networks (Mrsic-Flogel) |
UniBasel Contributors: | Mrsic-Flogel, Thomas |
Item Type: | Thesis |
Thesis Subtype: | Doctoral Thesis |
Thesis no: | 13028 |
Thesis status: | Complete |
Number of Pages: | 1 Online-Ressource (52 Seiten) |
Language: | English |
Identification Number: |
|
edoc DOI: | |
Last Modified: | 04 May 2019 04:30 |
Deposited On: | 03 May 2019 12:20 |
Repository Staff Only: item control page