The effects of low-grade inflammation and sleep loss on model-based and model-free control: Results of an experimental double hit model

Methods: In an ongoing study, we recruited 46 individuals who complete three sessions: sleep loss (2 nights of 4h in bed), normal sleep (2 nights of 9h in bed) (within-subjects), and a low-grade inflammation condition (COVID-19 vaccination) preceded by either sleep loss or normal sleep (between subjects). Blood samples were taken (not analyzed), sickness symptoms were assessed (SicknessQ with 10-point scale), and model-based and model-free control was quantified (a sequential decision task). Result(s): Sickness symptoms were highest after vaccination with sleep loss (M = 34.6), followed by vaccination with normal sleep (M = 24.3) and sleep loss (M = 23.8), and normal sleep only (M = 15.3). Model-free control increased in the vaccine as compared to the non-vaccine condition (b = 0.23, 95% CI 0.10, 0.37, p <.001), most clearly in the normal sleep condition. Model-based control decreased after sleep loss versus normal sleep (reward + common: b = -0.47, 95% CI -0.67, -0.28, p <.001, non-reward + rare: b = -0.43, 95% CI -0.63, -0.18, p <.001), which was not modulated by vaccination. Conclusion(s): These results suggest that low-grade inflammation and sleep loss independently attenuate behavioral control towards a cognitively less expensive but inflexible decision style. The potential role of sleep-immune pathways in model-based and model-free control will be discussed. Copyright © 2022

Experimental sleep loss and low-grade inflammation attenuate behavioural control towards a cognitively less expensive but inflexible decision style

Introduction: Behavioural control arises from a balance between model-based and model-free behaviour. Model-based behaviour is cognitively costly but enables adaptation to changes in the environment. In contrast, model-free control is fast, cognitively inexpensive, but inflexible. Overreliance on model-free control and/or reduced model-based control is found across various mental health conditions, suggesting that these modes of control may be influenced by common trans diagnostic processes. Since insufficient sleep and low-grade inflammation are highly common in mental ill-health, we assessed how they by themselves and in combination influence behavioural control Methods: In an ongoing study, we recruited 46 individuals who completed three sessions: Sleep loss (2 nights of 4 h in bed), normal sleep (2 nights of 9 h in bed) (within-subjects), and a low-grade inflammation condition (COVID-19 vaccination) preceded by either sleep loss or normal sleep (between subjects). Blood samples were taken (not analysed), sickness symptoms were assessed using the SicknessQ, and model-based and model-free control was quantified (using a sequential decision task). Result(s): Sickness symptoms were highest after vaccination with sleep loss (M = 34.6), followed by vaccination with normal sleep (M = 24.3) and sleep loss (M = 23.8), and normal sleep only (M = 15.3). Modelfree behaviour increased in the vaccine as compared to the nonvaccine condition (b = 0.23, 95% CI 0.10, 0.37, p < 0.001). Modelbased control decreased after sleep loss versus normal sleep (reward + common: B = -0.47, 95% CI -0.67, -0.28, p < 0.001, nonreward + rare: B = -0.43, 95% CI -0.63, -0.18, p < 0.001), which was not modulated by vaccination. Conclusion(s): These results suggest that sleep loss and low-grade inflammation independently attenuate behavioural control towards a cognitively less expensive but inflexible decision style.