Shared neural activation and co-fluctuations underlie auditory and visual sustained attention
Poster No:
2451
Submission Type:
Abstract Submission
Authors:
Anna Corriveau1, Jin Ke1, Monica Rosenberg1,2
Institutions:
1Department of Psychology, The University of Chicago, Chicago, IL, 2Neuroscience Institute, The University of Chicago, Chicago, IL
First Author:
Co-Author(s):
Monica Rosenberg
Department of Psychology, The University of Chicago|Neuroscience Institute, The University of Chicago
Chicago, IL|Chicago, IL
Department of Psychology, The University of Chicago|Neuroscience Institute, The University of Chicago
Chicago, IL|Chicago, IL
Introduction:
Maintenance of attention over time to auditory and visual information relies, to some extent, on shared neural mechanisms [6,8]. Sustained attention ability is well-predicted by relationships between a distributed set of brain regions [5]. However, attention fluctuations occur over a matter of seconds, a time scale that is better-captured by dynamic brain measures, such as co-fluctuations in BOLD activity [2,4,10]. We compare the neural mechanisms–both activation and activity co-fluctuations–involved in maintaining attention to sounds and images. We identify a set of brain regions and connections that underlie sustained attention to information regardless of perceptual modality.
Methods:
Participants performed a continuous performance task in which streams of trial-unique sounds and images were presented simultaneously during fMRI. Across two scan sessions, participants were tasked with pressing a button when relevant stimuli (either sounds or images) belonged to a frequent category (90%) and withholding response to infrequent stimuli (10%). Visual and auditory task analyses included 55 participants each. Of these, 51 participants are shared between analyses. Attention lapses were operationalized as incorrect responses to infrequent-category stimuli and attentional state was quantified using the reaction time variance time course [VTC; 1], which characterizes moments of in-the-zone (low variance) out-of-the-zone (high variance) pressing.
BOLD activity was averaged within 268 functionally-defined regions of interest [ROIs; 7], yielding 268 ROI activation time series. Co-fluctuation times series between all pairs of regions (edges) were calculated as the element-wise product of z-scored time series (Figure 1). Activation and edge time series were convolved with a canonical HRF.
To identify regions and pairs of regions involved in task performance, we contrasted trial-evoked activation and co-fluctuations to correct vs. incorrect responses to infrequent stimuli. To identify regions related to fluctuations in attentional state, activity and co-fluctuations were related to a parametric VTC regressor. Group-level analyses were tested within auditory and visual sessions separately. To correct for family-wise error rates, max-T and network-based statistic correction [9] were used for activation and co-fluctuation analyses, respectively.
BOLD activity was averaged within 268 functionally-defined regions of interest [ROIs; 7], yielding 268 ROI activation time series. Co-fluctuation times series between all pairs of regions (edges) were calculated as the element-wise product of z-scored time series (Figure 1). Activation and edge time series were convolved with a canonical HRF.
To identify regions and pairs of regions involved in task performance, we contrasted trial-evoked activation and co-fluctuations to correct vs. incorrect responses to infrequent stimuli. To identify regions related to fluctuations in attentional state, activity and co-fluctuations were related to a parametric VTC regressor. Group-level analyses were tested within auditory and visual sessions separately. To correct for family-wise error rates, max-T and network-based statistic correction [9] were used for activation and co-fluctuation analyses, respectively.
Results:
Contrasting activation on infrequent trials (correctly withheld responses - incorrect presses) revealed distributed regions more active on correct trials in both auditory and visual sessions (Figure 2). Co-fluctuation difference contrasts revealed no significant edges in auditory sessions and therefore no overlapping edges between auditory and visual sessions.
Activity in ROIs largely within the ventral attention network were positively related to the VTC in auditory and visual sessions (Figure 2), in line with previous work [3]. Auditory and visual sessions also shared more overlapping edges than chance (p<.001) whose co-fluctuation was significantly related (47 positive, 12 negative) to changes in VTC. Further, 22 of 47 edges positively related to response variance were not predicted by ROI time courses, suggesting edges carry additional information missed by activation alone.
Activity in ROIs largely within the ventral attention network were positively related to the VTC in auditory and visual sessions (Figure 2), in line with previous work [3]. Auditory and visual sessions also shared more overlapping edges than chance (p<.001) whose co-fluctuation was significantly related (47 positive, 12 negative) to changes in VTC. Further, 22 of 47 edges positively related to response variance were not predicted by ROI time courses, suggesting edges carry additional information missed by activation alone.
Conclusions:
Activation and co-fluctuation analyses identify regions and edges associated with sustained attention to auditory and visual information. A number of regions showed significant differences between correctly-withheld responses and attentional lapses in both auditory and visual sessions, suggesting a shared mechanism for performance on target trials during a continuous performance task. Additionally, shared activations and co-fluctuations tracked the VTC, a continuous measure of attentional state, during auditory and visual sessions, suggesting a common set of regions and edges related to attention fluctuations. These results identify perceptual modality-agnostic predictors of sustained attention performance.
Modeling and Analysis Methods:
Activation (eg. BOLD task-fMRI)
fMRI Connectivity and Network Modeling
Perception, Attention and Motor Behavior:
Attention: Auditory/Tactile/Motor 2
Attention: Visual 1
Keywords:
Cognition
FUNCTIONAL MRI
Other - Edge connectivity
1|2Indicates the priority used for review
Provide references using author date format
[2] Faskowitz, J. (2020). Edge-centric functional network representations of human cerebral cortex reveal overlapping system-level architecture. Nature Neuroscience, 23(12), Article 12.
[3] Fortenbaugh, F. C. (2018). Tracking behavioral and neural fluctuations during sustained attention: A robust replication and extension. NeuroImage, 171, 148–164.
[4] Jones, H. M. (2023). Edge-based general linear models capture high-frequency fluctuations in attention (p. 2023.07.06.547966). bioRxiv.
[5] Rosenberg, M. D. (2016). A neuromarker of sustained attention from whole-brain functional connectivity. Nature Neuroscience, 19(1), Article 1.
[6] Seli, P. (2012). Consistency of sustained attention across modalities: Comparing visual and auditory versions of the SART. Canadian Journal of Experimental Psychology/Revue Canadienne de Psychologie Expérimentale, 66(1), 44–50.
[7] Shen, X. (2013). Groupwise whole-brain parcellation from resting-state fMRI data for network node identification. NeuroImage, 82, 403–415.
[8] Terashima, H. (2020). Common principles underlie the fluctuation of auditory and visual sustained attention. Quarterly Journal of Experimental Psychology, 1747021820972255.
[9] Zalesky, A. (2010). Network-based statistic: Identifying differences in brain networks. NeuroImage, 53(4), 1197–1207.
[10] Zamani Esfahlani, F. (2020). High-amplitude cofluctuations in cortical activity drive functional connectivity. Proceedings of the National Academy of Sciences, 117(45), 28393–28401.