We will see that the involvement of neuromodulation in computations to do with Alectinib utility illuminates all these issues and also highlights a number of other general properties. One important complexity about utility is the parallel involvement of two different instrumental systems and also Pavlovian influences. These systems are subject to neuromodulation in partially different ways, and so are discussed individually below. The goal-directed, or model-based, instrumental system (Dickinson and Balleine, 2002), which involves frontal regions and the dorsomedial striatum

(Balleine, 2005; Valentin et al., 2007), is believed to construct a model of the task and to use that model prospectively to predict LY294002 price outcomes consequent on choices (Tolman, 1948). One central mark of goal-directed control is its sensitivity to motivational state—predicted outcomes are evaluated under current (or possibly predicted; Raby et al., 2007) motivational states. The second instrumental control system is habitual, or model free (Dickinson and Balleine, 2002), and is more closely associated with a different set of regions that includes the dorsolateral striatum (Balleine, 2005;

Tricomi et al., 2009). This learns what to do from direct experience of past actions and reward and so plans retrospectively (Thorndike, 1911). That planning is retrospective implies that it is the motivational state that pertained during learning that is important, and so model-free actions may be inappropriate for the current motivational state. Finally, for instrumental systems, choices are ultimately contingent on the delivery of suitable outcomes. Conversely, under Pavlovian control, elicitation of preparatory and consummatory actions associated with predictions of,

or the actual presence of, biologically significant reinforcers, appears to be automatic. Evidence for this is that the actions are still elicited even if they have deleterious consequences in terms of actually getting or preventing good or bad outcomes (Williams and Williams, 1969; Hershberger, 1986; Dayan et al., 2006). One interpretation is that Pavlovian actions are the result of evolutionary preprogramming, providing heuristic choices that are typically, though not always, appropriate. The predictions underlying Edoxaban Pavlovian control may be made in model-based or model-free ways. Appetitive and aversive utilities act in rather distinct ways, a fact that is better understood for model-free control. Thus, reward and punishment are considered separately in the latter. Dopamine is a key ascending neuromodulator. There is ample evidence that the phasic activity of DA neurons and the phasic release of DA in macaques (Bayer and Glimcher, 2005; Schultz et al., 1997; Morris et al., 2006; Satoh et al., 2003; Nakahara et al., 2004), rodents (Hyland et al., 2002; Roesch et al.