Posters
« Back
Rhythmic sampling of visual features in the brain during object recognition
EP32719
Poster Title: Rhythmic sampling of visual features in the brain during object recognition
Submitted on 19 Jun 2020
Author(s): Laurent Caplette, Karim Jerbi & Frédéric Gosselin
Affiliations: Université de Montréal
This poster was presented at OHBM 2020
Poster Views: 89
View poster »


Poster Information
Abstract: Introduction: During the fixation of an object, information is not only processed in the brain through time: it is also received on the retina continuously across time. Because of several factors including ongoing brain oscillations, information is likely to be processed differently depending on when it is received on the retina (Caplette et al., submitted). Previous studies have shown periodicities in the sampling of visual features (e.g., Landau & Fries, 2012; VanRullen, 2016), and have linked these with ongoing brain oscillations (e.g., Fiebelkorn et al., 2018). However, how different brain regions sample the visual world across time has not been investigated. Different brain areas may sample information differently and in a way that is not always correlated with behavior. Moreover, rhythmic sampling of specific information has not been investigated in the context of an object recognition task during a typical fixation.

Methods: Five neurotypical adults participated in a MEG study: each subject performed 5500 trials over the course of 5 days. On each trial, we randomly revealed random parts of a face image at random moments across 200 ms (see http://mapageweb.umontreal.ca/gosselif/meg_face2_slow.avi for an example stimulus slowed down 10x). Subjects had to categorize either the gender or the facial expression (happy vs neutral) of the face, in alternating blocks. MEG data was preprocessed and source-level activity was reconstructed using Minimum Norm Imaging. For each subject and task, ridge regressions were performed between the stimulus samples (indicating the visibility of each face feature at each moment) and MEG activity (for each source and time point), across correct trials. This allowed us to uncover, for each source, the processing (across time) of specific information received on the retina at specific moments (Fig. 1a). By visualizing both sampling and processing dimensions simultaneously, we can distinguish between oscillations in sampling (i.e. rhythmic sampling) and oscillations in processing (i.e. stimulus-related brain oscillations), and examine the properties of sampling rhythms in different brain areas.

Results: We performed a clustering analysis (Rodriguez & Laio, 2014) across sources to identify different kinds of activations while excluding noise. For some sources, information incoming at any moment on the retina was similarly processed (Fig. 1b, bottom left); for others, activity was greater for information received at specific unique moments (Fig. 1b, middle left and top left); for others, an oscillation in sampling was visible (Fig. 1b, middle center and bottom center); for yet others, oscillations in both sampling and processing were visible (Fig. 1b, top center). On several sources across occipital, parietal and temporal lobes, activity was significantly different depending on when information was received within the fixation (variance across stimulus moments; p < .05, FWER-corrected). We further uncovered significant oscillations in sampling (7-30 Hz) across the occipital lobes (FFT of regression coefficients across stimulus moments; p < .05, FWER-corrected; Fig. 2). Notably, different face features were largely sampled at different frequencies (feature x frequency interactions; p < .05, FWER-corrected; Fig. 2d).

Conclusions: We revealed that information is processed in a significantly different way depending on when it is received on the retina during a fixation and that it is in fact rhythmically sampled on multiple sources. Such oscillations in sampling are likely caused by underlying brain oscillations and illustrate that successive cycles of brain oscillations are allocated to information received at successive moments. We also showed that different features are sampled at different frequencies (i.e. frequency multiplexing). There is preliminary evidence for such multiplexing in the processing of different features (e.g., Schyns et al., 2011); we extend these findings to information sampling.
Summary: We disentangle sampling and processing in the brain. We observe different types of sampling in distinct brain areas, including rhythmic sampling at different frequencies for different features.Report abuse »
Questions
Ask the author a question about this poster.
Ask a Question »

Creative Commons