Nucleus accumbens dopamine release reflects Bayesian inference during instrumental learning.

Dopamine release in the nucleus accumbens has been hypothesized to signal reward prediction error, the difference between observed and predicted reward, suggesting a biological implementation for reinforcement learning. Rigorous tests of this hypothesis require assumptions about how the brain maps sensory signals to reward predictions, yet this mapping is still poorly understood. In particular, the mapping is non-trivial when sensory signals provide ambiguous information about the hidden state of the environment. Previous work using classical conditioning tasks has suggested that reward predictions are generated conditional on probabilistic beliefs about the hidden state, such that dopamine implicitly reflects these beliefs. Here we test this hypothesis in the context of an instrumental task (a two-armed bandit), where the hidden state switches repeatedly. We measured choice behavior and recorded dLight signals reflecting dopamine release in the nucleus accumbens core. Model comparison based on the behavioral data favored models that used Bayesian updating of probabilistic beliefs. These same models also quantitatively matched the dopamine measurements better than non-Bayesian alternatives. We conclude that probabilistic belief computation plays a fundamental role in instrumental performance and associated mesolimbic dopamine signaling.

Transient food insecurity during the juvenile-adolescent period affects adult weight, cognitive flexibility, and dopamine neurobiology.

A major challenge for neuroscience, public health, and evolutionary biology is to understand the effects of scarcity and uncertainty on the developing brain. Currently, a significant fraction of children and adolescents worldwide experience insecure access to food. The goal of our work was to test in mice whether the transient experience of insecure versus secure access to food during the juvenile-adolescent period produced lasting differences in learning, decision-making, and the dopamine system in adulthood. We manipulated feeding schedules in mice from postnatal day (P)21 to P40 as food insecure or ad libitum and found that when tested in adulthood (after P60), males with different developmental feeding history showed significant differences in multiple metrics of cognitive flexibility in learning and decision-making. Adult females with different developmental feeding history showed no differences in cognitive flexibility but did show significant differences in adult weight. We next applied reinforcement learning models to these behavioral data. The best fit models suggested that in males, developmental feeding history altered how mice updated their behavior after negative outcomes. This effect was sensitive to task context and reward contingencies. Consistent with these results, in males, we found that the two feeding history groups showed significant differences in the AMPAR/NMDAR ratio of excitatory synapses on nucleus-accumbens-projecting midbrain dopamine neurons and evoked dopamine release in dorsal striatal targets. Together, these data show in a rodent model that transient differences in feeding history in the juvenile-adolescent period can have significant impacts on adult weight, learning, decision-making, and dopamine neurobiology.

A basal ganglia circuit for evaluating action outcomes.

The basal ganglia, a group of subcortical nuclei, play a crucial role in decision-making by selecting actions and evaluating their outcomes. While much is known about the function of the basal ganglia circuitry in selection, how these nuclei contribute to outcome evaluation is less clear. Here we show that neurons in the habenula-projecting globus pallidus (GPh) in mice are essential for evaluating action outcomes and are regulated by a specific set of inputs from the basal ganglia. We find in a classical conditioning task that individual mouse GPh neurons bidirectionally encode whether an outcome is better or worse than expected. Mimicking these evaluation signals with optogenetic inhibition or excitation is sufficient to reinforce or discourage actions in a decision-making task. Moreover, cell-type-specific synaptic manipulations reveal that the inhibitory and excitatory inputs to the GPh are necessary for mice to appropriately evaluate positive and negative feedback, respectively. Finally, using rabies-virus-assisted monosynaptic tracing, we show that the GPh is embedded in a basal ganglia circuit wherein it receives inhibitory input from both striosomal and matrix compartments of the striatum, and excitatory input from the 'limbic' regions of the subthalamic nucleus. Our results provide evidence that information about the selection and evaluation of actions is channelled through distinct sets of basal ganglia circuits, with the GPh representing a key locus in which information of opposing valence is integrated to determine whether action outcomes are better or worse than expected.

Rule learning enhances structural plasticity of long-range axons in frontal cortex.

Rules encompass cue-action-outcome associations used to guide decisions and strategies in a specific context. Subregions of the frontal cortex including the orbitofrontal cortex (OFC) and dorsomedial prefrontal cortex (dmPFC) are implicated in rule learning, although changes in structural connectivity underlying rule learning are poorly understood. We imaged OFC axonal projections to dmPFC during training in a multiple choice foraging task and used a reinforcement learning model to quantify explore-exploit strategy use and prediction error magnitude. Here we show that rule training, but not experience of reward alone, enhances OFC bouton plasticity. Baseline bouton density and gains during training correlate with rule exploitation, while bouton loss correlates with exploration and scales with the magnitude of experienced prediction errors. We conclude that rule learning sculpts frontal cortex interconnectivity and adjusts a thermostat for the explore-exploit balance.

Between the primate and 'reptilian' brain: Rodent models demonstrate the role of corticostriatal circuits in decision making.

Decision making can be defined as the flexible integration and transformation of information from the external world into action. Recently, the development of novel genetic tools and new behavioral paradigms has made it attractive to study behavior of all kinds in rodents. By some perspectives, rodents are not an acceptable model for the study of decision making due to their simpler behavior often attributed to their less extensive cortical development when compared to non-human primates. We argue that decision making can be approached with a common framework across species. We review insights from comparative anatomy that suggest the expansion of cortical-striatal connectivity is a key development in evolutionary increases in behavioral flexibility. We briefly review studies that establish a role for corticostriatal circuits in integrative decision making. Finally, we provide an overview of a few recent, highly complementary rodent decision making studies using genetic tools, revealing with new cellular and temporal resolution how, when and where information can be integrated and compared in striatal circuits to influence choice.

Transient stimulation of distinct subpopulations of striatal neurons mimics changes in action value.

In changing environments, animals must adaptively select actions to achieve their goals. In tasks involving goal-directed action selection, striatal neural activity has been shown to represent the value of competing actions. Striatal representations of action value could potentially bias responses toward actions of higher value. However, no study to date has demonstrated the direct effect of distinct striatal pathways in goal-directed action selection. We found that transient optogenetic stimulation of dorsal striatal dopamine D1 and D2 receptor-expressing neurons during decision-making in mice introduced opposing biases in the distribution of choices. The effect of stimulation on choice was dependent on recent reward history and mimicked an additive change in the action value. Although stimulation before and during movement initiation produced a robust bias in choice behavior, this bias was substantially diminished when stimulation was delayed after response initiation. Together, our data suggest that striatal activity is involved in goal-directed action selection.

Engaging in an auditory task suppresses responses in auditory cortex.

Although systems that are involved in attentional selection have been studied extensively, much less is known about nonselective systems. To study these preparatory mechanisms, we compared activity in auditory cortex that was elicited by sounds while rats performed an auditory task ('engaged') with activity that was elicited by identical stimuli while subjects were awake but not performing a task ('passive'). We found that engagement suppressed responses, an effect that was opposite in sign to that elicited by selective attention. In the auditory thalamus, however, engagement enhanced spontaneous firing rates but did not affect evoked responses. These results indicate that neural activity in auditory cortex cannot be viewed simply as a limited resource that is allocated in greater measure as the state of the animal passes from somnolent to passively listening to engaged and attentive. Instead, the engaged condition possesses a characteristic and distinct neural signature in which sound-evoked responses are paradoxically suppressed.