Project Summary Pathologically altered affective learning is central to accounts of nearly every psychiatric disorder including depression and anxiety disorders. In humans, non-human primates, and rodents accurately learning which stimuli in our environment predict rewards or punishments is dependent on parts of frontal cortex, striatum, and limbic system, such as amygdala. A considerable amount is known about the neural mechanisms that are associated with adaptive patterns of learning within the circuits connecting these areas. What happens to these patterns of neural activity and the specific pathways involved when learning is enhanced or diminished by psychological processes is much less clear. Obtaining this knowledge is important as it would begin to reveal the specific mechanisms through which learning can be altered, information that is essential for identifying biomarkers in and aiding therapies for individuals with pathologically altered learning. Consequently, our aim here is to begin to establish how bottom-up and top-down processes impact stimulus-reward learning at the level of single neurons and circuit-level interactions. We specifically focus on the ventrolateral prefrontal cortex (PFC) and amygdala as prior work in macaque monkeys has shown that these areas, as opposed to other parts of frontal cortex and striatum, are required for efficient probabilistic stimulus-reward learning. These two areas are also reciprocally connected and, based on neuroimaging investigations functionally interact during learning further suggesting that they form part of a functional circuit essential for stimulus-reward learning. Our hypothesis is that bottom-up and top-down influences on learning impact neural activity within and communication between ventrolateral PFC and the basolateral nucleus of the amygdala but that bottom-up and top-down learning do so through different mechanisms and pathways. We will test our hypothesis by recording activity in ventrolateral PFC and basolateral amygdala as well as interconnected parts of orbital PFC and striatum in macaques learning in a probabilistic stimulus-reward task. We will assess functional interaction between areas using recurrent neural network models and measures of coherence when learning is altered by either bottom-up (aim 1) or top-down (aim 2) processes. To test the causal role of pathways linking amygdala and ventrolateral PFC we will also use chemogenetic approaches to selectively inhibit activity in these circuits. Thus, using an innovative combination of behavioral tasks, neural recordings, chemogenetic neuromodulation, and computational approaches we will establish the patterns of neural activity within and causal importance of PFC-amygdala pathways to bottom-up and top-down influences on learning. Completing these experiments will shed light on the specific neural mechanisms and pathways associated with altered learning, information essential for determining the processes that go awry in p...