Dynamics of brainstem arousal systems and pupil size predict cortical interactions for flexible decision-making

Ruud L. van den Brink, Keno Hagena, N. Wilming, Peter R. Murphy, Joshua Calder-Travis, Jürgen Finsterbusch, C. Büchel, Tobias H. DonnerbioRxiv 2023 · DOI

TL;DR

This study reveals how brainstem arousal systems, tracked by pupil size, dynamically influence our ability to flexibly switch between different decision-making strategies. By combining fMRI, pupillometry, and computational modeling, the researchers found that brainstem activity and pupil dilation precede changes in neural communication within the brain, suggesting a key role for arousal in orchestrating flexible cognitive control.

What Problem Does This Paper Address?

Making decisions often requires us to adapt our responses based on changing rules or contexts. For example, if you're driving, you might need to switch from accelerating to braking based on traffic signals. This ability to flexibly map sensory input to motor output is crucial for adaptive behavior. While we know that the brain's cortical networks are involved in this process, the precise mechanisms that enable this flexibility, especially when rules change unexpectedly, remain unclear. Specifically, how does the brain reconfigure its internal pathways to support these rapid shifts in decision-making strategies? This paper investigates the role of brainstem arousal systems, which are known to influence global brain states, in mediating these flexible cognitive adjustments.

Methods

The researchers reanalyzed data from a previous study where participants performed a visual decision-making task. This task involved inferring a hidden rule that dictated the correct response to a visual stimulus. The rule could change unpredictably, requiring participants to adapt their strategy.

The key methods employed were:

  • Brainstem fMRI: To measure activity in specific neuromodulatory centers in the brainstem.
  • Pupillometry: To track pupil diameter, a known indicator of arousal and brainstem activity.
  • Time-resolved assessment of cortical correlations: Using fMRI data, they analyzed how activity patterns in visual and motor cortices (representing stimulus and action, respectively) were correlated over time. This "stimulus-action coupling" reflects the strength of the mapping between what is perceived and how it is acted upon.
  • Computational Modeling: A Bayesian inference model was used to quantify participants' internal "belief" about the active rule, as well as computational variables like "change point probability" (CPP) and "uncertainty."

The novel aspect here is the integration of these techniques to specifically link brainstem arousal signals to dynamic changes in cortical processing during flexible decision-making.

Datasets

The study reanalyzed data from a previously published experiment involving 18 human participants. No new datasets were introduced.

Key Insights and Ideas

  • Brainstem Arousal as a "Context Switcher": The study strongly suggests that brainstem arousal systems, signaled by pupil dilation, are not just passive indicators of attention but actively participate in preparing the brain for rule changes.
  • Pupil Size as a Window into Brainstem Activity: The pupil's size closely mirrored the activity patterns observed in key brainstem neuromodulatory nuclei, reinforcing its utility as a non-invasive proxy for arousal.
  • Arousal Precedes Cortical Reorganization: Crucially, brainstem activity and pupil dilation were observed to precede the strengthening of connections (correlated variability) between stimulus-processing and action-planning areas in the cortex. This temporal relationship is key to understanding the causal role of arousal.
  • Computational Variables Drive Arousal: The arousal response (and pupil dilation) was not just triggered by the rule switch itself, but also by latent computational variables like "change point probability" and "uncertainty," which signal the likelihood of a rule change.

Key Results

  • Rule Switches Elicit Brainstem and Pupil Responses: When the hidden rule changed, there were significant evoked responses in catecholaminergic brainstem nuclei (SN, VTA, LC) and a corresponding dilation of the pupil.
  • Brainstem and Pupil Activity Predict Cortical Coupling: Cross-correlation analyses revealed that activity in brainstem arousal centers (particularly the basal forebrain and catecholaminergic nuclei) and pupil dilation occurred before increases in the strength of stimulus-action coupling in cortical networks. This lead time ranged from approximately 1 to several seconds.
  • Pupil Dynamics Encode Inference Variables: The derivative of pupil diameter was correlated with "change point probability" (CPP), while pupil diameter itself was correlated with "uncertainty," demonstrating that pupil dynamics reflect key computational aspects of rule inference.

Limitations and Future Work

  • fMRI Signal Limitations: The authors acknowledge that fMRI signals can be noisy and difficult to precisely localize within small brainstem nuclei. They employed several strategies to mitigate this, but it remains a general challenge.
  • Distinguishing Neuromodulatory Signals: It's difficult to definitively distinguish between neuromodulatory signals and other sources of brainstem activity using fMRI alone.
  • Causal Manipulations Needed: While the study shows strong correlational evidence and temporal precedence, direct causal manipulation of specific neuron types in the brainstem and concurrent electrophysiological recordings in the cortex would be needed to definitively establish causality.
  • Future Directions: The authors suggest future studies could use designs with larger lags and jitter to better disentangle the precise timing of pupil responses and computational variables. Further research could also explore the role of specific neuron types within the basal forebrain and their interaction with catecholaminergic nuclei.

Summary updated

Abstract

Abstract Most perceptual decisions entail a flexible mapping from sensory input to motor output. Flexible input-output mapping is reflected in correlated variability within the cortical network involved in perceptual decisions. Here, we tested the idea that brainstem arousal systems are involved in flexible input-output mapping and changes in cortical correlated variability. We combined brainstem fMRI, pupillometry, and time-resolved assessment of the intrinsic correlations between cortical population codes for stimulus and action. Human participants reported the orientation of visual stimuli by button presses, while the required stimulus-response mapping rule could undergo hidden and unpredictable changes. Rule switches evoked brainstem and pupil responses as well as changes in computational model-inferred, latent variables. These variables governed participants’ rule-switching behavior and pupil responses. Brainstem activity and pupil dilation preceded increases in the strength of correlations between cortical stimulus and action codes. Brainstem arousal systems may promote the reorganization of sensorimotor cortical pathways for flexible decisions.
BibTeX
@misc{brink2023dynamics,
  author = {Ruud L. van den Brink and Keno Hagena and N. Wilming and Peter R. Murphy and Joshua Calder-Travis and Jürgen Finsterbusch and C. Büchel and Tobias H. Donner},
  title = {{Dynamics of brainstem arousal systems and pupil size predict cortical interactions for flexible decision-making}},
  year = {2023},
  doi = {10.1101/2023.12.05.570327},
  howpublished = {bioRxiv},
}
Figure