Print Close

Time-dependent contextual feedback signals increase V1 responses to unpredictable scene information

Presented During:

Advanced investigations of perceptual and attention systems

Wednesday, June 26, 2024: 11:30 AM - 12:45 PM

COEX

Room: Grand Ballroom 104-105

Poster No:

2547

Submission Type:

Abstract Submission

Authors:

Zirui Zhang¹^,2, Clement Abbatecola¹^,2, Angus Paton¹^,2, Lucy Petro¹^,2, Lars Muckli¹^,2

Institutions:

¹Centre for Cognitive Neuroimaging,School of Psycholog and Neuroscience, University of Glasgow, Glasgow, United Kingdom, ²Imaging Centre for Excellence, Queen Elizabeth University Hospital, University of Glasgow, Glasgow, United Kingdom

First Author:

Zirui Zhang
Centre for Cognitive Neuroimaging,School of Psycholog and Neuroscience, University of Glasgow|Imaging Centre for Excellence, Queen Elizabeth University Hospital, University of Glasgow
Glasgow, United Kingdom|Glasgow, United Kingdom

Co-Author(s):

Clement Abbatecola
Centre for Cognitive Neuroimaging,School of Psycholog and Neuroscience, University of Glasgow|Imaging Centre for Excellence, Queen Elizabeth University Hospital, University of Glasgow
Glasgow, United Kingdom|Glasgow, United Kingdom

Angus Paton
Centre for Cognitive Neuroimaging,School of Psycholog and Neuroscience, University of Glasgow|Imaging Centre for Excellence, Queen Elizabeth University Hospital, University of Glasgow
Glasgow, United Kingdom|Glasgow, United Kingdom

Lucy Petro
Centre for Cognitive Neuroimaging,School of Psycholog and Neuroscience, University of Glasgow|Imaging Centre for Excellence, Queen Elizabeth University Hospital, University of Glasgow
Glasgow, United Kingdom|Glasgow, United Kingdom

Lars Muckli
Centre for Cognitive Neuroimaging,School of Psycholog and Neuroscience, University of Glasgow|Imaging Centre for Excellence, Queen Elizabeth University Hospital, University of Glasgow
Glasgow, United Kingdom|Glasgow, United Kingdom

Introduction:

Introduction Information processing operations in the visual cortex are tuned to the statistical regularities of sensory inputs and are crucially dependent on context. Neurobiologically inspired computational frameworks of visual processing emphasise functional interactions between higher and lower cortical areas, whereby higher areas send feedback signals that influence feedforward processing in lower areas. Using a partial visual occlusion approach in which a mask covers the lower right quadrant of natural scene images, we can isolate feedback signals in the retinotopic visual cortex that processes the occluded image portion (Muckli et al., 2015, Morgan 2019, Muckli 2023). Based on our earlier findings using apparent motion stimulation where feedback signals suppress predictable sensory inputs (Alink et al., 2010), we hypothesised that a priming contextual scene would increase the response to subsequent unpredictable sensory information, while it would reduce or stabilise the response to consistent, expected sensory information.

Methods:

Results are from 24 subjects, mean age = 26.34 years. We selected ten grey-scale images from five categories. We designed an event-related fMRI paradigm consisting of the 5 experimental conditions (consistent context before, inconsistent context before, consistent context after, inconsistent context after, and target only) and 1 baseline condition. The lower right quadrant of an image (Target) was always shown between 0.8s and 1.8s from trial start. In the target only condition, this was the whole trial. In the before conditions, contextual information was provided from 0s to 1s. In the after conditions, contextual information was provided from 1.6s to 2.6s. Contextual information was either consistent or inconsistent with target, yielding 5 conditions. We applied an ROI-wise deconvolution analysis to compare the effect of contextual consistency on the univariate BOLD response amplitude containing the peak of the signal in the Target region of V1.

·Fig 1. A) Two portions of image stimuli: the ‘Context’ (Left) and the ‘Target’ (Right). B) Experimental paradigm and example C) ROI defination: Processes of defining V1 Target area

Results:

We found that the presence of contextual feedback signals increases the BOLD response in the Target region of V1 compared to when there was no contextual information available, t = 2.697. p = 0.007. Compared to the Target only condition, we found significantly higher activity when the Target was preceded by inconsistent contextual information, t = 2.737, p = 0.006. However, we found no significant difference between the Target only and
consistent before conditions, t =0.635, p = 0.53, suggesting that bottom-up activation is not increased when there is preceding top-down information that predicts the content of the bottom-up activity. Furthermore, there was a significant difference when directly comparing consistent before vs inconsistent before, t = 2.101, p = 0.035. As a control, we also found that this unexpected later contextual information increased the response to the bottom-up Target compared to the Target only condition, which was true no matter whether this information was consistent with the Target (t = 2.139, p = 0.03) or not (t = 2.76, p =0.007). Further, there was no significant difference comparing the response to the Target when it was followed by either consistent or inconsistent contextual information, t = 0.572, p = 0.57.

Supporting Image: subcon_tar_navy_red.png

·Fig 2. BOLD signal changes: consistent before vs target (upper left), inconsistent before vs target (upper right), consistent after vs target (lower left), inconsistent after vs target (lower right)

Conclusions:

We found that top-down contextual modulation from other retinotopic regions contributes to the V1 BOLD signal in response to visual stimulation in a target region. This modulation causes attenuation when consistent and amplification when inconsistent, but only when context presented before the target region. Contextual top-down activity is 'explained-away' if presented with consistent stimulation. Contextual stimulation presented afterwards is unaffected by consistency. There is an interaction between bottom-up and top-down information related to natural scene processing in the early visual cortex, evidenced by BOLD response changes. This interaction depends on the consistency and timing of top-down and bottom-up information.

Higher Cognitive Functions:

Higher Cognitive Functions Other ²

Perception, Attention and Motor Behavior:

Perception: Visual ¹

Keywords:

FUNCTIONAL MRI

Perception

Vision

^1|2Indicates the priority used for review

Provide references using author date format

Alink, A., Schwiedrzik, C. M., Kohler, A., Singer, W., & Muckli, L. (2010). "Stimulus predictability reduces responses in primary visual cortex." The Journal of Neuroscience, 30(8), 2960-2966. https://doi.org/10.1523/JNEUROSCI.3730-10.2010
Muckli, L., De Martino, F., Vizioli, L., Petro, L., Smith, F., Ugurbil, K., Goebel, R., & Yacoub, E. (2015). "Contextual feedback to superficial layers of V1." Current Biology, 25(20), 2690-2695. https://doi.org/10.1016/j.cub.2015.08.057
Morgan, Andrew T., Lucy S. Petro, and Lars Muckli. 2019. “Scene Representations Conveyed by Cortical Feedback to Early Visual Cortex Can Be Described by Line Drawings.” The Journal of Neuroscience: The Official Journal of the Society for Neuroscience 39 (47): 9410–23.
Muckli*, Lars, Lucy S. Petro*, Clement Abbatecola, Ahsan Adeel, Johanna Bergmann, Nicolas Deperrois, Alain Destexhe, et al. 2023. “The Cortical Microcircuitry of Predictions and Context – a Multi-Scale Perspective.” Zenodo. https://doi.org/10.5281/ZENODO.8380093.