The paper by Yu and Dayan explores the role of neuromodulators acetylcholine (ACh) and norepinephrine (NE) in the brain's handling of uncertainty, particularly in Bayesian statistical inference and learning. They propose that ACh signals expected uncertainty, arising from known unreliability of predictive cues, while NE signals unexpected uncertainty, indicating strongly unexpected observations due to context switches. These signals interact to enable optimal inference and learning in noisy and changing environments. The authors support their theory with physiological, pharmacological, and behavioral data, suggesting that ACh and NE play distinct but complementary roles in attentional tasks. They propose a generalized attention task that involves both forms of uncertainty and use it to interpret existing experimental data, making specific predictions about the responses of ACh and NE systems. The model predicts that ACh and NE have antagonistic and synergistic interactions, with ACh setting the threshold for NE-dependent contextual changes. Simulations show that impairments in these systems can be partially alleviated by interventions affecting the other system. The paper also discusses the limitations and extensions of their theory, emphasizing the need for further empirical and theoretical research to understand the complexity of neural representations of uncertainty.The paper by Yu and Dayan explores the role of neuromodulators acetylcholine (ACh) and norepinephrine (NE) in the brain's handling of uncertainty, particularly in Bayesian statistical inference and learning. They propose that ACh signals expected uncertainty, arising from known unreliability of predictive cues, while NE signals unexpected uncertainty, indicating strongly unexpected observations due to context switches. These signals interact to enable optimal inference and learning in noisy and changing environments. The authors support their theory with physiological, pharmacological, and behavioral data, suggesting that ACh and NE play distinct but complementary roles in attentional tasks. They propose a generalized attention task that involves both forms of uncertainty and use it to interpret existing experimental data, making specific predictions about the responses of ACh and NE systems. The model predicts that ACh and NE have antagonistic and synergistic interactions, with ACh setting the threshold for NE-dependent contextual changes. Simulations show that impairments in these systems can be partially alleviated by interventions affecting the other system. The paper also discusses the limitations and extensions of their theory, emphasizing the need for further empirical and theoretical research to understand the complexity of neural representations of uncertainty.