Investigating the role of a pallido-striatal projection in value-guided behavior - Project summary. To maximize chances of survival, animals must not only be able to effectively predict the values of future outcomes, but also to learn from erroneous predictions. The basal ganglia (BG) are crucial for both processes. Value prediction is thought to rely on striatum, where cells store value representations in their synaptic weights. Value updating is thought to rely on modulation of these striatal synapses by dopamine, since dopamine cells encode a key updating signal in the form of reward prediction errors (RPE; the difference between expected and unexpected reward outcomes). However, anatomical studies suggest that another strong yet understudied source of striatal modulation arises from the GABAergic pallidum. Pallidum is known to have a crucial role in reward-guided behaviors. A subpopulation known as `arkypallidal' cells (in contrast to `prototypical' cells comprising the rest of pallidum) project exclusively to striatum and form massive, extremely dense axonal arborizations, making this population the largest known exogenous source of striatal inhibition. Despite anatomical evidence suggesting that arkypallidal cells are well positioned to modulate striatal value representations, no study to date has directly tested if these cells carry the signals necessary for value updating. Whether and how this important input to striatum participates in value updating to support flexible behavior remains unclear. Importantly, little is known about how pallidal cells in general signal basic motivational variables such as value or prediction error, let alone their function in complex value-guided tasks. We hypothesize that within pallidum, arkypallidal cells uniquely signal prediction errors and integrate across multiple dimensions of motivational variables to support value updating and flexible behavior. In aim 1, I will explicitly test in mice whether and how arkypallidal and prototypical cells encode motivational variables including reward and punishment value, uncertainty, and prediction error in support of value updating. Preliminary data suggest that putative arkypallidal cells preferentially signal the positive component of RPE more strongly and quickly compared to putative prototypical cells. In aim 2, I will test how pallidal cells in the non-human primate (NHP) encode and integrate across reward attributes in support of multi-attribute decision-making. Pallidum is crucial for reward-seeking behaviors, but it is not clear how it signals information when subjects must choose between reward options with multiple varying attributes. Preliminary data suggest that pallidal cells encode a wide range of attributes that subjects use to guide their choices. Further preliminary analyses suggest that putative arkypallidal cells preferentially encode reward value PE, and integrate across more decision offer attributes than putative prototypical cells. Subsequent analyses will test the hypothesis that putative arkypallidal cells compute PE signals that reflect subjects' overall value estimates of options, integrated over multiple option attributes. Subsequent experiments aim to verify these results using cell type-targeted approaches to identify arkypallidal cells in monkeys performing this multi-attribute decision-making task. These aims will elucidate the role of a poorly understood yet anatomically significant pallido-striatal projection in value-guided behavior. They will also bridge functional studies of BG circuitry across species, allowing for a more generalized understanding of BG to help guide human health.
