Print  Close

Top-down vs bottom-up origins of sensory uncertainty in early visual cortex

Poster No:

2550 

Submission Type:

Abstract Submission 

Authors:

Joshua Corbett1, Jacob Paul1, Ruben van Bergen2, Wouter Schellekens2, Linzhi Tao1, Floris de Lange2, Marta Garrido3

Institutions:

1University of Melbourne, Melbourne, Australia, 2Donders Institute for Brain, Cognition and Behaviour, Radboud University, Nijmegen, Netherlands, 3The University of Melbourne, Melbourne, Australia

First Author:

Joshua Corbett  
University of Melbourne
Melbourne, Australia

Co-Author(s):

Jacob Paul  
University of Melbourne
Melbourne, Australia
Ruben van Bergen  
Donders Institute for Brain, Cognition and Behaviour, Radboud University
Nijmegen, Netherlands
Wouter Schellekens  
Donders Institute for Brain, Cognition and Behaviour, Radboud University
Nijmegen, Netherlands
Linzhi Tao  
University of Melbourne
Melbourne, Australia
Floris de Lange  
Donders Institute for Brain, Cognition and Behaviour, Radboud University
Nijmegen, Netherlands
Marta Garrido, PhD  
The University of Melbourne
Melbourne, Australia

Introduction:

Humans make better decisions by weighting sensory information according to its uncertainty. Despite behavioural evidence indicating the importance of sensory uncertainty in perception and action, little is known about how this information is encoded in the brain. We considered whether changes in sensory uncertainty are best explained by variability in bottom-up (feedforward) sensory processing (i.e., neural noise), or fluctuations in top-down (feedback) signals (i.e., attention), using a novel layer-specific fMRI approach.

Methods:

We recorded BOLD activity from superficial, middle, and deep cortical layers of primary visual cortex while participants performed an orientation judgement task. Using this data in conjunction with a model-based decoding algorithm (TAFKAP; Van Bergen & Jehee, 2021), we gathered estimates of orientation and uncertainty for each layer separately, on each experimental trial. Given anatomical knowledge that bottom-up signals target middle layers and top-down signals target both superficial and deep layers (Felleman & Van Essen, 1991), we considered whether the encoding of uncertainty is implemented via feedforward or feedback processes.
Supporting Image: layer_segs.png
 

Results:

In a sub-sample of 13 participants (out of a full sample of 30 participants that has already been collected), we found above chance orientation decoding from superficial (p=0.005), middle (p=0.002), and deep (p=0.002) layers. However, we did not find a relationship between uncertainty and behavioural variability for any layer in this sub-sample (ps > 0.05).
Supporting Image: ori_decoding.png
   ·Layer specific decoding of orientation. Dots represent decoding error for individual participants.
 

Conclusions:

Our findings excitingly demonstrate that orientation information is decodable at the laminar level. While we did not find a relationship between uncertainty and behavioural variability in this sub-sample, the next step will be to look at this relationship in the full (already collected) sample of N=30.

Modeling and Analysis Methods:

Multivariate Approaches

Novel Imaging Acquisition Methods:

BOLD fMRI 2

Perception, Attention and Motor Behavior:

Perception: Visual 1

Keywords:

Cognition
Cortical Layers
Multivariate
Perception
Vision

1|2Indicates the priority used for review

Provide references using author date format

Felleman, D. J. (1991), 'Distributed hierarchical processing in the primate cerebral cortex', Cerebral cortex
Van Bergen, R. S. (2021), 'TAFKAP: An improved method for probabilistic decoding of cortical activity', bioRxiv