Different underlying mechanisms for high and low arousal in probabilistic learning in humans

Abstract

Humans are uniquely capable of adapting to highly changing environments by updating relevant information and adjusting ongoing behaviour accordingly. Here we show how this ability -termed cognitive flexibility-is differentially modulated by high and low arousal fluctuations. We implemented a probabilistic reversal learning paradigm in healthy participants as they transitioned towards sleep or physical extenuation. The results revealed, in line with our pre-registered hypotheses, that low arousal leads to diminished behavioural performance through increased decision volatility, while performance decline under high arousal was attributed to increased perseverative behaviour. These findings provide evidence for distinct patterns of maladaptive decision-making on each side of the arousal inverted u-shaped curve, differentially affecting participants' ability to generate stable evidence-based strategies, and introduces wake-sleep and physical exercise transitions as complementary experimental models for investigating neural and cognitive dynamics..

Staying in control: characterising the mechanisms underlying cognitive control in high and low arousal states

“…As predicted, both RT and accuracy were better fitted to a quadratic model (RT: AIC = -671.33; accuracy: AIC = -1077.3) than to a linear one (RT: AIC = -611.97; accuracy: AIC = -1066.1). These results thus point to a curvilinear relationship between arousal level and general task performance (see Figure 2A-B), in line with the inverted U-shaped curve (Yerkes-Dodson, 1908), showing slower and less accurate responses as participants transitioned towards the extremes of the arousal spectrum, and convergent to previous results in another task (Ciria et al, 2021).…”

“…It has been previously shown how decreased arousal fails to fully interrupt higher order cognitive processes, such as perceptual decision-making (Bareham et al, 2014(Bareham et al, , 2015, semantic discrimination (Kouider et al, 2014) or probabilistic learning (Ciria et al, 2021). However, as reported by the entire body of literature on drowsy states (Bareham et al, 2015;Canales-Johnson et al, 2020;Ciria et al, 2021;Goupil & Bekinschtein, 2012;Jagannathan et al, 2022;Kouider et al, 2014;Noreika et al, 2020;Xu et al, 2023), decreasing the level of arousal leads to slower RT, lower accuracy and decreased sensitivity in decision-making. Our DDM analyses revealed that, in a simple conflict cognitive control task, overall RT and accuracy pattern might be presumably due to: (1) a slower rate of task-related information evidence accumulation; (2) a longer non-decision time that could be caused by a deceleration in sensory encoding, motor response execution, or both; and (3) a wider separation between decision thresholds, which means that more evidence would need to be accumulated by the system to reach a response criterion.…”

“…A total sample of 72 healthy participants (aged from 18 to 50 years old) was included in the present study. The sample size for each experiment was determined on the basis of previous experiments on behavioural indexes of performance as a function of high and low arousal levels (Bareham et al, 2014;Ciria et al, 2021;Jagannathan et al, 2022). All participants reported normal or corrected-tonormal vision and binaural hearing and had no recent history of cardiovascular, musculoskeletal, neurological or psychiatric illness.…”

“…For this reason, the interaction between arousal and cognition has been frequently approached from the experimental framework initially proposed by Yerkes-Dodson (1908). According to their well-known inverted U-shaped theory, an intermediate level of arousal will be associated with optimal cognitive performance, while levels above or below will lead to a decrease in performance, as recently found in drowsy and high-intensity exercising participants (Ciria et al, 2021) for probabilistic reversal learning, a higher order cognitive task.…”

High-intensity physiological activation disrupts the neural signatures of conflict processing

“…Fluctuations in physical arousal occur naturally throughout the day. These spontaneous fluctuations unfold in a nonlinear manner (Bernaola-Galván et al, 2017, Ogilvie andWilkinson, 1984) and become severe at extreme states such as deep sleep (Goupil and Bekinschtein, 2012) or intense physical exertion (Schmit and Brisswalter, 2020), modulating cognitive processing (Ciria et al, 2021, McMorris et al, 2011, Noreika et al, 2020, Wickens et al, 2015. In particular, it has been shown that cognitive control (i.e., the ability to adaptively adjust cognitive processes to inhibit distracting information while maintaining task relevant information) is markedly impacted as the level of arousal decreases (Cain et al, 2011, Goupil and Bekinschtein, 2012, Tucker et al, 2010.…”