Our recent studies have identified a group of neurons in the basal forebrain (BF) region that encodes motivational salience using robust bursting responses (Lin & Nicolelis, 2008). Such bursting responses lead to faster decision speed (Avila & Lin, 2014) and generate an event-related potential response in the frontal cortex (Nguyen & Lin, 2014). During the current reporting period, our research effort focused on two main directions: (1) how does BF encoding of motivational salience emerge during learning? (2) establish the platform to conduct optogenetic experiments in awaking behaving mice, with the goal of determining the neurochemical identity of salience-encoding BF neurons. (1) In the first research direction, we investigated how does the encoding of motivational salience emerge in BF neurons during learning, with regard to reinforcement learning (RL) theories. Reward prediction error (RPE) signals have long been proposed in RL theories as the driving force for new learning. A key prediction of RL theories that has rarely been observed experimentally is that, during learning, RPE signals should backpropagate from the unexpected outcome to the cue predicting that outcome through gradual shifts in latency. Here we provide evidence for such a temporal backpropagation in a novel RPE signal in the basal forebrain (BF). We first established that BF neuronal activity reflected RPE by encoding the difference between predicted and received reward. In rats exposed to a new stimulus-reward association, the receipt of unexpected reward following the new stimulus triggered a cascade of continuous backpropagation of BF activity to earlier events leading up to reward, which quickly emerged within a few trials and stabilized over several sessions. Backpropagated BF activity was tightly coupled with behavioral performance in single trials throughout learning. Furthermore, BF activity at an intermediate stage of RPE backpropagation, which was activated even in the absence of the new stimulus, underlied exploratory reward-seeking behaviors during the early learning phase in both stimulus-absent trials as well as in new stimulus trials. The evolution of BF RPE signals therefore reveals the underlying temporal dynamics of new learning, and also provides a rare glimpse into the internal models animals created to predict reward and guide reward-seeking behaviors. (2) We have previously suggested that salience-encoding basal forebrain neurons are non-cholinergic neurons based on the lack of firing rate modulation across wake-sleep cycles. However, the neurochemical identity of salience-encoding neurons remains unclear. Determining their neurochemical identity represents a key step in further understanding their functions especially because BF is a neuroanatomically complex region comprised of multiple macrosystems and multiple populations of projection neurons. A powerful approach to determine the neurochemical identity of salience-encoding BF neurons is to optogenetically label specific neuronal populations in the BF, and simultaneously record single unit activity and photo-stimulate this region in awake behaving mice. As a first step toward this goal, we established an experimental platform in transgenic Cre mice combining operant behavior, electrophysiology and optogenetics. We first established an experimental platform in three strains of transgenic Cre mice (PV-, ChAT- and VGluT2-Cre) that combines operant behavior, electrophysiology and optogenetics. Mice were trained to maintain fixation in a nosepoke port and then respond quickly to a reward-predicting tone to collect reward in an adjacent port. Mice were able to achieve high hit rates (>90%) and high trial numbers in a session (>150 rewarded trials) with fast reaction times (<300ms). Neuronal recording in combination with photo-stimulation of the BF show that: 1) The majority of recorded BF neurons in mice showed robust bursting responses to the motivationally salient tone similar to salience-encoding BF neurons in rats and 2) photo-stimulation of the three major cortically-projecting BF neuronal populations can directly or indirectly modulate (excite or inhibit) bursting responses to the salient tone. Taken together, these results suggest that the three main populations of neurons participate in concert, directly or indirectly, in the circuit underlying motivational salience in the BF.
