The control of goal-directed, instrumental actions by primary motivational states, such as hunger and thirst, involves two processes. The first is a Pavlovian process where contextual or discriminative stimuli are associated with outcomes, influencing performance based on the relevance of the outcome to the agent's current motivational state. The second is an instrumental process involving knowledge of the contingency between actions and outcomes, which determines the value of the outcome. Motivational states influence the instrumental process indirectly, requiring the agent to learn the value of an outcome through experience in that state. Goal-directed action is defined by the interaction of two representations: (1) knowledge of the instrumental contingency between the action and the outcome, and (2) the outcome being represented as a goal. The instrumental criterion distinguishes actions based on their sensitivity to instrumental contingencies, while the goal criterion assesses whether the outcome is currently a goal for the agent. Free-operant lever-pressing by rats is identified as goal-directed, while other activities, such as spatially directed locomotion, are ambiguous. The outcome devaluation test, where the value of an outcome is reduced, demonstrates that performance depends on whether the outcome is currently a goal. This test shows that animals learn the value of an outcome through experience, a process similar to Tolman's concept of "cathexes." Tolman's cathexes theory suggests that motivational states determine the agent's representation of an outcome as a goal through connections between the motivational state and the outcome. Aversion conditioning, such as flavor aversion, forms a negative cathexis, which reduces the value of the outcome. This process is mediated by the formation of a connection between the outcome and a negative affective state. Incentive learning is the process by which agents learn the value of an outcome through direct experience, which is distinct from the learning processes that control instrumental performance. Motivational states determine the assigned value of an outcome at the time of contact, and the current motivational state can acquire control over this value, making performance sensitive to shifts in state. The determination of outcome value is influenced by feedback consequences and hedonic properties of the outcome. Incentive learning involves learning about the value of an outcome based on affective responses rather than knowledge of its consequences.The control of goal-directed, instrumental actions by primary motivational states, such as hunger and thirst, involves two processes. The first is a Pavlovian process where contextual or discriminative stimuli are associated with outcomes, influencing performance based on the relevance of the outcome to the agent's current motivational state. The second is an instrumental process involving knowledge of the contingency between actions and outcomes, which determines the value of the outcome. Motivational states influence the instrumental process indirectly, requiring the agent to learn the value of an outcome through experience in that state. Goal-directed action is defined by the interaction of two representations: (1) knowledge of the instrumental contingency between the action and the outcome, and (2) the outcome being represented as a goal. The instrumental criterion distinguishes actions based on their sensitivity to instrumental contingencies, while the goal criterion assesses whether the outcome is currently a goal for the agent. Free-operant lever-pressing by rats is identified as goal-directed, while other activities, such as spatially directed locomotion, are ambiguous. The outcome devaluation test, where the value of an outcome is reduced, demonstrates that performance depends on whether the outcome is currently a goal. This test shows that animals learn the value of an outcome through experience, a process similar to Tolman's concept of "cathexes." Tolman's cathexes theory suggests that motivational states determine the agent's representation of an outcome as a goal through connections between the motivational state and the outcome. Aversion conditioning, such as flavor aversion, forms a negative cathexis, which reduces the value of the outcome. This process is mediated by the formation of a connection between the outcome and a negative affective state. Incentive learning is the process by which agents learn the value of an outcome through direct experience, which is distinct from the learning processes that control instrumental performance. Motivational states determine the assigned value of an outcome at the time of contact, and the current motivational state can acquire control over this value, making performance sensitive to shifts in state. The determination of outcome value is influenced by feedback consequences and hedonic properties of the outcome. Incentive learning involves learning about the value of an outcome based on affective responses rather than knowledge of its consequences.