Linking Attentional States and Neuronal Dynamics in the Locus Coeruleus During a Decision-Making Task

Abstract

Our brains perform a remarkable number of computations continuously to interpret our external environments and respond appropriately. While performing a cognitive task, we may direct attention to the stimuli specific to the current task, alternative stimuli in the environment, past knowledge and experiences, or our internal biases when making decisions. This selective attention is an essential property of our brains, allowing the allocation of cognitive resources to relevant information, filtering out distractions and optimising cognitive performance. Locus coeruleus, the primary noradrenaline (NA) releasing brainstem nucleus, impacts diverse cognitive functions like attention, arousal, sensory processing and goal-directed behaviour (Poe et al., 2020). The LC is also a vulnerable brain structure, being one of the first areas that undergo degeneration during the pathology of neurodegenerative diseases such as Parkinson’s and Alzheimer’s. Additionally, while the early-stage symptoms of these diseases can be variable, they often affect attention (Muslimovic, Post, Speelman et al, 2005; Sharpe, 1992). However, the complete role of the LC-NA network in attention during health and disease remains elusive. In this study, we modelled attentional states exhibited by mice during a vision-based perceptual decision-making task (International Brain Laboratory et al., 2021) using a Bernoulli generalised linear model - hidden Markov model (GLM-HMM: Ashwood et al., 2022). Each identified attentional state has a different dependency on variables like task-relevant stimuli, internal bias, and history, such as previous response and reward, describing degrees of engagement. We explored the calcium dynamics of LC neurons during these attentional states through fibre photometry. We found that the mean calcium activity of the LC neurons just after stimulus presentation seems to predict the attentional state of mice during this task and, thus, the related accuracy/performance and task-relevant stimulus sensitivity. This framework, presented here, sets the groundwork for predicting task engagement and performance from LC activity.