Search

Accurate future predictions are essential for guiding behavior, and disruptions in this process are associated with psychiatric disorders. Here the authors show that changes in baseline dopamine levels can alter the efficacy of learning from positive versus negative reward prediction errors.

Dopamine neurons in the ventral tegmental area calculate reward prediction error by subtracting input from neighbouring GABA neurons.

Individual dopaminergic neurons encode future rewards over distinct temporal horizons.

Dopamine neurons encode reward prediction errors (RPE) that report the mismatch between expected reward and outcome for a given state. Here the authors report that when there is uncertainty about the current state, RPEs are calculated on the probabilistic representation of the current state or belief state.

During foraging with threat–reward conflicts in mice, dopamine modulates two competing neuron types in the striatum for flexible threat coping, from initial threat avoidance to eventual overcoming of the threat.

The authors show that behavior and dopamine responses track prospective, but not retrospective, contingency. These results are explained by temporal difference (TD) learning models, indicating that TD errors are a measure of contingency.

New data on brain-wide circuits centred around two interconnected hypothalamic neuron populations provide significant mechanistic insights into the emergence of social need during social isolation and the satiation of social need during social reunion.

D1- and D2-expressing striatal neurons encode separate parts of a learned reward distribution, paralleling modern approaches in machine learning.

A study finds evidence supporting co-release of glutamate and GABA, excitatory and inhibitory fast neurotransmitters, from a single axon terminal in neurons of the ventral tegmental area that project to the lateral habenula.

Experiments using conditional RIM1 and RIM2 knockout mice and acute pharmacological manipulations clarify the role of rapid dopamine dynamics, whereby these dynamics are dispensable for movement initiation but important for reward-guided conditioned behaviours.

Uchida and colleagues consider integration of information for perceptual decision making, focusing on olfactory and visual systems. They argue that there are neural mechanisms that construct discrete sensory samples from a continuous input stream to facilitate important computational functions.

A long-standing idea in modern neuroscience is that the brain computes inferences about the outside world rather than passively observing its environment. The authors record from midbrain dopamine neurons during tasks with different reward contingencies and show that responses are consistent with a learning rule that harnesses hidden-state inference.

Menegas et al. show that dopamine neurons projecting to the posterior striatum reinforce avoidance of threatening stimuli. Their results indicate that there are two axes of reinforcement learning using dopamine, the value axis and the threat axis.

Analyses of single-cell recordings from mouse ventral tegmental area are consistent with a model of reinforcement learning in which the brain represents possible future rewards not as a single mean of stochastic outcomes, as in the canonical model, but instead as a probability distribution.

The authors found that dopamine signals move gradually from time of reward to time of cue through intermediate timepoints, similarly to the evaluation signals used in temporal difference learning. These findings thereby fill a gap between computational theories and the brain.

Dopamine neurons in the ventral tegmental area are thought to signal reward prediction error. The authors show that these neurons respond with striking homogeneity during classical conditioning. All dopamine neurons appear to calculate reward prediction error similarly, enabling robust and consistent broadcasting of this signal throughout the brain.

How are goal-directed actions directed and motivated? This paper presents behavioral and electrophysiological data in rats to dissect the differential role of the ventral striatum (VS) and dorsomedial striatum (DMS) in action-selection and motivation. The work suggests that the DMS is important in encoding the net expected return, sensitizing the rats to the local average reward value in a block of trials.

Photometric recordings and optogenetic manipulation show that dopamine fluctuations in the dorsolateral striatum in mice modulate the use, sequencing and vigour of behavioural modules during spontaneous behaviour.

Odors evoke complex spatiotemporal patterns of activity in the olfactory bulb. The authors show that the spike rates of downstream piriform cortex neurons (PCNs) reflect the relative timing of activation. Posterior PCNs are more sensitive to input timing than anterior PCNs.

Galanin-expressing neurons in the medial preoptic area coordinate different aspects of motor, motivational, hormonal and social behaviour associated with parenting by projecting to different brain regions depending on the type of behaviour and sex and reproductive state of mice.

Dopamine signals are implicated in not only reporting reward prediction errors but also various probabilistic computations. In this Opinion article, Gershman and Uchida propose that these different roles for dopamine can be placed within a common reinforcement learning framework.

The AgRP-expressing neurons in the arcuate nucleus drive food-seeking behaviours during caloric restriction; a mouse study of monosynaptic retrograde rabies spread and optogenetic circuit mapping reveals that these neurons are activated by input from hypothalamic paraventricular nucleus cells and their activation or inhibition can modulate feeding behaviour.

Dopaminergic neurons in the mouse ventral tegmental area signal the difference between received and expected reward, whereas GABAergic neurons signal expected reward.

The hypothesis that dopamine reports reward prediction errors has been both influential and controversial. This Perspective characterizes the present state of evidence, indicating where it succeeds and where it falls short. A complete account of dopamine will probably need to move beyond the reward prediction error hypothesis while retaining its core explanatory power.

Szuts et al. have developed a wireless neural recording system that outperforms existing rodent telemetry systems in either channel count, weight or transmission range. They show that it can be used to record brain signals in animals outdoors and in tunnels.

Traditionally, 'metacognition' has been thought to be the purview of primates, but it is now shown that rats may compute and use estimates of their own confidence when making difficult perceptual decisions. The paper finds correlates of task difficulty and predictors of trial-by-trial outcome in the activity of orbitofrontal cortex, which may be most parsimoniously explained as a representation of subjective confidence.