The authors did, however, make an effort to model fast and slowly changing (“phasic” and “tonic”) patterns of LC activity. Whether these patterns relate to the physiology of phasic and tonic firing of LC neurons remains unclear, of course. However, what is remarkable in the present work is that LC activity is specifically modulated by unexpected uncertainty. This specific relationship was predicted by computational modeling (Yu and Dayan, 2005) and behavioral evidence from pupillometry studies (Preuschoff et al., 2011). This fascinating convergence of theory and physiology paves the road PD 332991 for future studies. There are a number emerging questions

which the current study encourages us to tackle. We would like to highlight just two here. The first relates to the exciting possibility to image functional activity in the SN/VTA and LC simultaneously and thus observe the interaction of both regions during decision making. The second relates to the role of the hippocampus in coping with

unexpected uncertainty. As careful but nevertheless inquisitive creatures we balance between drives to exploit what we know and explore Enzalutamide the unknown. In so-called “model-free” reinforcement learning, recent reward outcomes are integrated into action-value associations and exploration is undirected (Sutton and Barto, 1998). But the exploration/exploitation dilemma can also be approached from a Bayesian perspective. Decision making in dynamically changing environment improves if the statistics of the environment (model of the world) are tracked to assess the salience of new information and the beliefs about action values are updated accordingly. In such a model-based framework, uncertainty should promote exploration, science as supported by some studies (e.g., Badre et al., 2012). On the other hand, human participants tend to avoid uncertain options when ambiguity is high (reviewed by Bach and Dolan, 2012). There are probably different computational mechanisms

that bias the balance toward exploration: one mechanism detects the lack of knowledge in the face of unexpected uncertainty while another mechanism assigns a “bonus” for potential reward to the detected uncertain option or environment, thus favoring their sampling. An intriguing possibility is that these two computational processes depend on two distinct neuromodulatory systems: the noradrenergic system detecting uncertainty and the dopaminergic system assigning bonuses to the uncertain options. The current advances of fMRI now allow us to investigate such hypotheses pertaining to the interaction of the LC and SN/VTA. One remarkable finding is the involvement of the hippocampus in tracking unexpected uncertainty related to reward outcomes. Beyond its association to memory and spatial navigation, the hippocampus, especially its anterior portion, is also related to what is generally known as anxiety response (Fanselow and Dong, 2010).