A remarkable feature of human cognitive behavior is the ability to act in an intelligent, goal-directed manner. Such goal-directed actions are thought to be the result of a variety of cognitive control processes that enable the formation, maintenance, and updating of goal representations, as well as top-down biasing mechanisms that enhance the processing of goal-relevant information, inhibit goal-irrelevant information, and detect conflicts between them. An important aspect of these control mechanisms, that is nevertheless often overlooked, concerns the role of non-cognitive factors such as affect and motivation. In particular, the pursuit of goals must be prioritized by the value of the outcomes attached to them. Indeed, it seems clear that a primary function of affect and motivational systems is to provide just these sorts of value and prioritization signals. Thus, the cognitive control processes that regulate and coordinate goal-directed behaviors are likely to interact with brain systems that determine the potential affective/motivational value of possible responses 
. The current study addresses this issue by focusing on the effect of different motivational variables on brain activity during cognitive task performance.
It is not surprising that there has been a growing interest within cognitive neuroscience on how executive or cognitive control processes might interact with motivational states and the processing of reward information. A few studies have begun to investigate how manipulations of reward and motivational states modulate neural activity during the performance of different cognitive tasks 
. A common finding in these studies has been that reward/motivational manipulations increase activation in a network of brain regions that are typically engaged by executive control demands during task performance, such as the lateral prefrontal cortex (PFC) and parietal cortex. Additionally, this work has revealed that motivational manipulations also tend to engage regions that are considered to be more closely linked with reward and affective processing, such as the striatum, orbitofrontal cortex, and amygdala 
However, an important question that still has not been well addressed in the previous literature is the temporal dynamics by which motivational effects influence executive control. Executive control processes can be sustained in nature, reflecting maintained goal states or expectancies, or transient, reflecting moment-to-moment fluctuations in environmental demands, or internal processes (e.g., error detection). Likewise, motivational signals can also fluctuate transiently (e.g., high value vs. low value rewards available) or in a more state-like manner (e.g., thirst vs. satiation). Thus, it is important to implement experimental manipulations and methods that permit dissociation of executive and motivational processes according to their temporal dynamics. In fMRI studies, it is possible to decompose brain activation signals into transient and sustained effects through the use of mixed blocked/event-related designs 
. Such designs have been used to selectively dissociate distinct executive control components in a wide-range of domains, including task-switching, episodic memory, prospective memory, and decision-making. However, the mixed design has only rarely been used in studies of motivation and executive control 
In Locke & Braver 
, it was found that task performance under reward conditions was primarily associated with a sustained rather than transient increase in activation within executive regions such as PFC and parietal cortex, when compared to performance under baseline, non-reward conditions. However, a limitation of this study was that the manipulation of reward occurred only in a blocked rather than trial-by-trial manner. Thus, it was not clear whether the sustained effects reflected non-specific processes such as arousal or attention, or even the requirement to maintain the increased reward value of the task across trials (since reward information was not explicitly provided to participants on each trial). Conversely, in prior studies that have manipulated reward only in a trial-by-trial fashion, it has not been possible to establish whether the motivational effects of potential rewards produce sustained changes in brain activity that are independent of any trial-related effects. Thus, an optimal experimental design for identifying and potentially dissociating sustained from transient motivational effects is one in which rewards are manipulated in both a block-based and trial-by-trial manner.
In the current study, we utilized such a design to examine the effects of motivation on cognitive control during a working memory task. Participants performed a standard item recognition paradigm 
, in both reward and non-reward blocks. Furthermore, within reward blocks, trials were randomly varied among three types: no-reward, low-reward and high-reward. Because the trial-by-trial manipulation of reward value was indicated to participants at the start of each trial, this value could be encoded and represented in a fully transient manner. Thus, there was no a priori
requirement for reward information to be maintained across trials (since these were continually changing), or for the motivational effects to be tonic in nature. That is, because the reward value changed on a trial-by-trial basis, a tonic state change might be a less efficient way of adapting cognitive processing and performance. Nevertheless, we hypothesized that sustained effects on cognitive control processes might still be present, and reflect the increased incentive salience of the reward block (as a whole) relative to the no-reward block.
A second goal of the study was to address questions related to whether the type of reward available influences the nature of motivational effects. For example, motivational incentives can be categorized as involving primary rewards, such as food or liquid, or secondary rewards, such as money. Human studies examining the neural correlates of motivational effects in cognitive tasks have typically employed monetary reinforcers 
, whereas in animal studies motivational effects are standardly studied with primary reinforcers, such as food or liquid. In principle, there is no reason why primary reinforcers cannot be examined in human motivation-cognition studies. Indeed, there is a growing neuroimaging literature that has employed primary rewards within the context of classical and instrumental conditioning 
. One benefit of employing primary rewards to examine motivational effects in human studies is the ability to draw closer links to the animal work. But more importantly, such studies would provide the ability to directly test whether motivational incentives exert their effects in a domain-general or category-specific manner 
. There seems to be a basic, but implicit assumption in the literature that motivational incentives effects are category-independent 
, but this assumption has not yet been directly tested. To our knowledge, there have not been any studies directly comparing the impact of different incentive categories on brain activity and behavior during performance of cognitive tasks (but see 
. for conditioning studies of this type). There are reasons to predict that there may be interesting effects of incentive type. Specifically, primary and secondary incentives might influence behavior through different neuronal routes or circuits, based on the way the reward information is encoded (e.g., primary incentives may operate primarily through sensory and subcortical pathways while secondary incentives may operate through multi-modal cortical ones).
In the current study we directly compared the effects on cognitive control processes and associated brain activity dynamics associated with reward incentives, when the rewards were monetary bonuses received after the session (Money condition) or drops of juice delivered directly on each trial (Liquid condition). A within-subjects design was employed with incentive manipulated across blocks, but all other aspects of the design and task held constant. Furthermore, in the Liquid block as well as the Money block, reward value was also manipulated on a trial-by-trial basis. Thus, our design permitted not only a direct comparison of Liquid vs. Money rewards on neural activity, but also whether the incentive category effects differentially affected sustained vs. transient activation components. Because this was the first study of its type, we did not have any strong a priori predictions regarding whether common or selective effects of incentive type would be observed.