Functionality of arousal-regulating brain circuitry at rest predicts human cognitive abilities

Abstract

Arousal state is regulated by subcortical neuromodulatory nuclei, such as locus coeruleus, which send wide-reaching projections to cortex. Whether higher-order cortical regions have the capacity to recruit neuromodulatory systems to aid cognition is unclear. Here, we hypothesized that select cortical regions activate the arousal system, which in turn modulates large-scale brain activity, creating a functional circuit predicting cognitive ability. We utilized the Human Connectome Project 7T functional magnetic resonance imaging dataset (N=149), acquired at rest with simultaneous eye tracking, along with extensive cognitive assessment for each subject. First, we discovered select frontoparietal cortical regions that drive large-scale spontaneous brain activity specifically via engaging the arousal system. Second, we show that the functionality of the arousal circuit driven by bilateral posterior cingulate cortex (associated with the default mode network) predicts subjects' cognitive abilities. This suggests that a cortical region that is typically associated with self-referential processing supports cognition by regulating the arousal system.

Keywords: arousal; cognition; default mode network; posterior cingulate cortex; resting state; spontaneous brain activity.

Figure 1.. Hypothesized large-scale circuit regulating tonic arousal, based on temporal lag and sign of time series cross-correlations.

A. Simplified wiring diagram of a large-scale brain circuit for recruitment of the brainstem arousal system. Select cortical regions (X1) signal LC via fast myelinated projections (with ${\tau }_1$ signifying temporal delay), whereas LC activation leads to slow NE release in cortex, which in turn modulates cortical sustained activity (with a delay of ${\tau }_2$). In resting state, NE release suppresses spontaneous brain activity. Non-light-mediated pupil size fluctuations reflect LC activity at rest (with a delay of ${\tau }_3$). B. Neural activity cascade expected from circuit depicted in A during resting state. Cortical regions X1 cause increase in LC tonic activity. X1 regions may also cause network-level activation in regions X2 (not depicted). LC tonic firing in turn results in large-scale slow release of NE, resulting in a decrease of cortical spontaneous activity with a delay of multiple seconds. C. Expected cross-correlation (a measure of signal similarity as a function of temporal lag) between X1/X2 type of brain regions and the arousal signal, whereas $t_1$ and $t_2$ depict the relative timing of the cross-correlation peaks/troughs. A positive peak is expected to appear earlier than negative trough $\left(t_1<t_2\right)$, irrespective of whether the pupil size or LC activity is used to indicate the arousal state.

Figure 2.. Spatiotemporal interactive dynamics of arousal state and spontaneous brain activity.

A. Cross-correlation between pupil size predictor and fMRI cortical parcel-level activity (Glasser parcellation) compared with a null distribution calculated via reversal of pupil size time course (t-statistic). The horizontal axis depicts the temporal delay at which the correlation was calculated. The parcels are grouped according to large-scale resting-state brain networks (Ji et al. 2018) and sorted according to peak/trough time delay (black dots). B. Peaks/trough amplitude of pupil-brain cross-correlation (same color scale as A) presented on the cortical surface. C. Illustration of gain modulation depicting the potential suppression of low brain activity such as during resting state (blue arrow) and amplification of high brain activity (red arrow). D. Distribution of cross-correlation peak/trough time lags across brain regions. E. Large-scale resting-state networks presented on the cortical surface. F. Averaged cross-correlation between pupil size and brain activity within each resting-state network (Vis1:Visual 1, Vis2: Visual 2, SMN:somatomotor, CON:cingulo-opercular, DAN:dorsal attention, Lang:language, FPN: frontoparietal, Aud:Auditory, DMN: default mode, PMN:posterior multimodal, VMN: ventral multimodal, OAN: orbito-affective). Networks that on average show larger absolute value of positive peak than negative trough are colored in warm colors (FPN and DMN), same color code as (E).

Figure 3.. Regions identified as regulating large-scale spontaneous activity via pupil-linked arousal.

A. Brain regions (Glasser parcellation) predictive of pupil-linked arousal in agreement with “independent effect” criterion. These regions were identified by considering network-level and cross-network activity correlations via multiple regression linear mixed models. The legend provides commonly used names for Glasser brain regions. B. Effects of brain regions presented in A on large-scale network and subcortical activity as mediated by arousal. Top: illustration of causal mediation analysis conducted for each region presented in A. Bottom: Each cell of the matrix shows ACME (Average Causal Mediation Effect), representing the level of influence of brain activity in the arousal-driver region on subsequent activity in the cortical network or subcortical structure that is specifically routed via arousal. The white circles indicate p-value < 0.05 (FDR corrected). The bar plots show the ACME averaged across the arousal-driver regions (horizontal bars) and across the arousal-driven networks/subcortical areas (vertical bars). The network abbreviation is the same as Figure 2.

Figure 4.. Functionality of arousal-regulating brain circuit predicts individual differences in cognitive ability.

A. Distribution of cognitive ability scores across subjects. B. Distribution of global ACME (Averaged Causal Mediation Effect), that is, ACME averaged across the driving and the affected brain areas. C. Distribution of age across subjects. D. The relationship between global ACME and cognitive ability with linear model fit result depicted and inferred parameters specified. E. The relationship between participant age and cognitive ability. F. Main effect of ACME, signifying the functionality of the large-scale arousal circuit driven by each candidate arousal-driving region on cognitive ability. P-values are specified after FDR correction and the values below 0.05 are marked in red. G. Same as F but for interaction effect of ACME and age on cognitive ability. The main effect of age was not significant for any of these regions.