Ventral Tegmental Area: Dopamine Neurons and Reward Learning
Author: Ukrainian Psychological Hub · Published: October 12, 2026 · Editorial Policy
The ventral tegmental area (VTA) is a small, anatomically complex region of the ventral midbrain that contributes to reward-based learning, motivation, behavioral flexibility, and responses to biologically important events. Many VTA neurons produce dopamine and communicate with the nucleus accumbens, prefrontal cortex, and other forebrain regions. Those projections help organisms learn what predicts consequential outcomes and adjust behavior as circumstances change. The VTA also contains GABA-releasing and glutamate-releasing neurons, mixed neurotransmitter phenotypes, and circuits whose functions depend on their inputs, outputs, and internal state. Morales and Margolis's circuit review.
The best-supported general account treats the VTA as a heterogeneous component of distributed brain circuits. Classical dopamine recordings established reward-prediction-error signals; newer work identifies divergent cell populations, terminal dopamine dynamics, and information about reward features beyond a single numerical value. Human neuroimaging offers complementary, spatially coarser evidence. de Jong, Fraser, and Lammel's synthesis; Kahnt and Schoenbaum's 2025 perspective.
What is the ventral tegmental area?
The VTA is a set of neighboring cell groups in the midbrain tegmentum, situated near the substantia nigra and other brainstem structures. The Latin anatomical label refers to its ventral tegmental location. It is routinely depicted as a single dot in popular illustrations, but cytoarchitectonic boundaries and human MRI delineation are less straightforward. Anatomical investigators describe multiple subdivisions with differences in cell composition and connections. Trutti and colleagues' functional neuroanatomical review.
Its clinical and psychological significance follows from its connectivity: VTA neurons receive information about sensory cues, internal needs, past outcomes, and action contexts; their downstream effects alter local neural circuits involved in learning, action, attention, and incentive motivation. An account of VTA function therefore begins with specific projections and measurements rather than assigning a whole psychological faculty to the region. For a larger circuit-level treatment, see Brain Reward System: Circuits, Reinforcement, and Motivation.
Where is the VTA located in the brain?
The VTA lies in the ventral midbrain, medial to substantial portions of the substantia nigra, in a region connected with the medial forebrain bundle and neighboring mesencephalic structures. The nearby substantia nigra pars compacta (SNc) also contains many dopamine neurons; the anatomical continuity between these groups matters because neuroimaging and older electrode studies sometimes aggregate their signals. A two-dimensional diagram is useful for orientation but loses the complex three-dimensional and species-dependent boundaries.
Histology, gene-expression profiles, tract tracing, and high-resolution atlases help map the region. Human studies often use probabilistic regions of interest rather than a universally exact VTA contour. Trutti and colleagues describe the nomenclature and delineation problem; a 7-Tesla human VTA probabilistic atlas represents an important imaging advance, while still expressing spatial uncertainty.
VTA versus substantia nigra
The VTA and SNc are adjacent midbrain dopamine-related structures with partly overlapping chemistry, yet they differ in typical projection patterns, anatomical organization, and population responses. A familiar shorthand is VTA–ventral striatum/prefrontal cortex and SNc–dorsal striatum/motor circuitry. This is a useful orientation, not a strict partition: projections vary by neuron class and subregion, and both systems integrate extensive input. Watabe-Uchida and colleagues' whole-brain input mapping.
What does the VTA do? The principal functions
Reward-guided learning
A well-established role of identified midbrain dopamine populations is to supply learning-relevant signals when actual outcomes depart from predictions. Early primate electrophysiology found that some dopamine neurons responded to unexpected reward, transferred responding to reward-predicting stimuli after learning, and reduced firing when an expected outcome failed to appear. This pattern helped connect neuronal firing with formal reinforcement learning. Schultz, Dayan, and Montague, 1997.
Learning is not identical to enjoying an outcome. A cue can gain predictive value, elicit preparation, or change choice because prior outcomes affected synapses and network states. Specific dopamine-dependent learning mechanisms operate alongside glutamatergic plasticity, attention, prior expectations, and action-state computations. The broader formal theory is developed in Reward Prediction Error: How the Brain Learns From Outcomes.
Motivation and willingness to act
VTA-related circuits influence approach, effort allocation, cue-driven seeking, and pursuit of goals. Motivation is not an invariant concentration of dopamine in the whole brain. A reward-predicting cue may recruit phasic firing; local dopamine release in an accumbens subregion may change with the current opportunity and action policy. In rodents, simultaneous measurement of VTA firing and accumbens dopamine showed that some motivational release dynamics were not simply mirrored by VTA cell spiking. Mohebi and colleagues, 2019.
Reward-driven action depends on the meaning of a cue, uncertainty, costs, physiological needs, and previous reinforcement. The psychological distinction between incentive motivation and behaviorally expressed pleasure is discussed in How Rewards Affect Motivation: Incentives, Interest, and Behavior.
Salience, novelty, and responses to aversion
Some dopamine populations are responsive to novelty, alerts, salient nonrewarding events, or aversive contingencies. A short-latency response to an important event cannot automatically be read as pleasure or positive reward value. Projection-specific studies report that different dopamine neurons contribute to reward seeking and avoidance, and that stressors can recruit distinct VTA-related mechanisms. Bromberg-Martin, Matsumoto, and Hikosaka; Lammel and colleagues; review of heterogeneous reward and aversion circuits.
The same word 'salience' can refer to perceptual prominence, motivational significance, or unexpected information, depending on the task and model. When examining a study, ask exactly how salience was operationalized and whether the recorded neurons were genetically identified and projection-defined.
Flexible decisions and state-dependent control
VTA signals can influence prefrontal and striatal computations involved in evaluating alternatives, maintaining task engagement, and updating beliefs about outcomes. The prefrontal cortex also influences VTA circuits through direct and indirect routes. These are recurrent systems: causal direction and timescale depend on the experiment. A single VTA activation map never demonstrates that the region independently decides, plans, or controls an entire behavior.
How VTA dopamine neurons communicate
Dopamine synthesis, firing, and receptor signaling
Dopamine-producing neurons synthesize dopamine through enzymatic pathways that include tyrosine hydroxylase. When these neurons fire, action potentials may trigger dopamine release from axon terminals. Released dopamine then acts on dopamine receptors and associated intracellular signaling systems in target cells, modifying neuronal excitability, plasticity, or circuit gain in ways that depend on receptor type and cellular context. Reuptake and metabolism help shape its spatiotemporal effects. A separate overview covers the neurotransmitter across brain systems: Dopamine: Reward Prediction, Learning, Motivation, and Movement.
Different measurements interrogate different stages of this process. Dopamine-neuron spikes are electrical events at the cell body; calcium signals approximate aspects of cellular activity; voltammetry and fluorescent sensors estimate extracellular dopamine dynamics near terminals; PET ligand signals reflect tracer binding and sometimes endogenous dopamine competition under specific assumptions. They are related observations, not interchangeable measures of a single 'dopamine level.' de Jong and colleagues; Cumming on receptor radioligands.
Phasic responses and slower states
Experimental literature often contrasts brief, event-related phasic firing with slower changes in baseline firing, population recruitment, and terminal release. This is a descriptive distinction rather than a complete two-mode theory. Phasic responses can convey rapid outcome-related information; longer timescale dopamine dynamics may track evolving expectations, action opportunities, or ongoing activity. Terminal release is also affected by local receptors, acetylcholine-containing interneurons, and other modulation. Mohebi and colleagues; de Jong and colleagues.
The phrase 'dopamine spike' may denote neuronal action-potential bursts or a transient increase in extracellular concentration, which are distinct experimental quantities. Method descriptions should identify which was actually measured.
The VTA contains several neuron types
Dopaminergic populations
VTA dopamine neurons differ in genetic markers, morphology, intrinsic excitability, physiological response properties, input patterns, target regions, and possibly cotransmitters. A sample of cells recorded near the VTA cannot be presumed to represent all dopamine-producing neurons in the region. Newer single-cell and single-nucleus sequencing has strengthened the evidence for molecularly differentiated populations, while the relationship between transcriptomic classes and complete behavioral functions remains under study. 2025 review of VTA molecular heterogeneity; 2025 neuronal heterogeneity review.
GABA neurons
VTA GABAergic neurons can exert local inhibitory influence over neighboring cells and can project to regions outside the VTA. They participate in behavioral timing, reward expectation, aversive responses, and pathway-specific control in experimental systems. Identifying a GABAergic cell does not specify one psychological effect: its location, partners, and task contingencies determine what stimulation or inhibition accomplishes. Cell-type-specific recordings in mice reveal response classes that differ across dopamine and GABA populations. Cohen and colleagues.
Glutamate neurons and cotransmission
The VTA also contains glutamatergic populations and cells capable of expressing molecular machinery associated with more than one transmitter. Some dopamine-related projections can use glutamatergic or GABAergic signaling under particular anatomical and developmental conditions. Whether a given axon terminal actually releases two transmitters, and with what temporal pattern, must be established experimentally. Transporter coexpression alone is insufficient to infer full behavioral function. Morales and Margolis; review of glutamate cotransmission and vesicular synergy.
Molecular discovery is increasingly precise, but a molecular identity is only one dimension of VTA classification. Contemporary work combines transcriptomics with projection tracing, spatial mapping, and functional recordings to test which neuronal subpopulations respond to particular inputs. Opto-seq study of input-specific gene induction.
VTA pathways and projection targets
Mesolimbic pathway: VTA to nucleus accumbens
The VTA–nucleus accumbens projection is one of the most studied routes for dopamine-mediated reinforcement and motivated behavior. The accumbens is part of the ventral striatum and has subregions that differ in microcircuit architecture and functional responses. Experiments show that VTA dopamine axons reaching different accumbens territories can carry distinct dynamics. Consequently, a finding in the accumbens core cannot automatically be generalized to the shell or the entire mesolimbic system. de Jong and colleagues.
The nucleus accumbens integrates dopamine modulation with excitatory information about cortical goals, hippocampal context, amygdala-linked relevance, and internal states. Dopamine helps alter how this distributed network learns from and acts upon contingencies. To place its striatal mechanisms in context, read Basal Ganglia: Action Selection, Learning, and Movement.
Mesocortical pathway: VTA and prefrontal cortex
Some VTA dopamine neurons project to prefrontal cortex, where modulatory input can influence working-memory stability, adaptation, and task-dependent control. This pathway has distinctive molecular and physiological characteristics and should not be assumed to implement the same signal as a VTA–accumbens projection. Prefrontal and VTA networks influence one another via multiple circuit routes. The evidence is strongest for defined neural and task effects, not a universal formula such as 'more cortical dopamine equals better focus.' Lammel and colleagues; Watabe-Uchida and colleagues.
Connections with amygdala, hippocampal formations, and other sites
VTA-related networks interact with the amygdala, hippocampal formations, hypothalamus, habenula-related pathways, and brainstem structures. These circuits contribute information about cues, contextual associations, bodily needs, and aversive events. Inputs from lateral hypothalamus and other structures help explain how the same environmental incentive can be processed differently depending on physiological state. A mouse whole-brain monosynaptic-tracing study revealed an unexpectedly diverse input architecture. Watabe-Uchida and colleagues.
Directional arrow diagrams should be read as simplified maps of known routes. Specific anatomical projections and their synaptic strengths differ by subregion and species; the existence of a connection alone does not identify the information it transmits.
How does reward prediction error work in VTA neurons?
In a simple reinforcement-learning model, a reward prediction error is the difference between the outcome obtained and the outcome expected. In temporal-difference models the update can additionally incorporate predictions of future outcomes. A positive prediction error can follow a better-than-expected event; an expected reward may elicit relatively little response at delivery; omission of an expected reward can be accompanied by a suppression in firing. These patterns have been reported in identified midbrain populations and remain foundational to learning theory. Schultz, Dayan, and Montague.
Consider a neutral tone followed by food in an animal task. Initially an unexpected food delivery may recruit a rapid dopamine response. With experience, activity can shift toward the predictive tone. If food is omitted after a well-learned cue, the response can dip near the expected delivery time. The precise shape depends on state representation, uncertainty, reward type, delays, biological population, and measurement. This example explains a well-supported family of findings, not an invariant response of every VTA neuron.
The computational interpretation is also evolving. Models in which dopamine carries only scalar reward-value differences capture important data, yet experiments indicate informative signals related to sensory reward identity, associative structure, salience, and task demands. Some proposals treat these signals as richer teaching information, but no single universal code has been established for all midbrain neurons. Kahnt and Schoenbaum, 2025.
Is dopamine a prediction-error signal or a motivation signal?
Both observations occur, often at different circuit sites and timescales. Firing of a population of dopamine neurons may correlate with model-derived prediction errors, while dopamine release in the accumbens may vary with current motivational opportunity. The relationship can depend on presynaptic modulation and subregion. This is an empirical question about specific datasets, not a debate resolved by assigning the transmitter exactly one role. Mohebi and colleagues; de Jong and colleagues.
What the VTA has to do with pleasure
The VTA contributes to incentive motivation and reward-related learning, yet dopamine signaling is not a direct meter of pleasure or subjective happiness. Reward can be analyzed into hedonic impact, motivation to obtain the outcome, and learning about its predictors. These components interact but show experimental dissociations. Animal behavioral responses can index some aspects of hedonic impact, while human reports are required when a claim concerns conscious pleasure. Berridge, Robinson, and Aldridge.
This distinction matters for interpreting common claims about food, social media, music, gambling cues, and romantic attraction. A neural response in a reward-related region does not establish how pleasant an experience felt, whether someone has an addiction, or which personal choice they should make. Reward-associated VTA activity is one part of a much larger neurobehavioral process.
How do scientists study the VTA?
Single-neuron and population electrophysiology
Extracellular electrodes record electrical activity near neurons at millisecond-scale resolution, and modern experiments can identify particular cell classes with optogenetic tagging, genetic strategies, or post hoc anatomy. Such experiments can test response timing and correlations with cues, choices, outcomes, and movement. Spike waveforms or anatomical placement alone do not always prove dopamine identity, and cross-species differences limit direct generalization to everyday human psychology. Cohen and colleagues' neuron-type-specific study.
Optogenetics and causal intervention
Optogenetics can make genetically targeted neurons or axons light-sensitive in experimental models. Time-specific stimulation or inhibition helps test whether activity in a selected pathway contributes causally to a defined behavior. The effects depend on targeting, light delivery, stimulation pattern, collateral projections, and behavioral design. Results from rodent circuits are important mechanistic evidence; they are not established clinical treatments for humans. Lammel and colleagues' input-specific experiments.
Dopamine release sensors and electrochemistry
Fast-scan cyclic voltammetry uses electrochemical signatures to estimate rapid concentration changes near an electrode, whereas engineered fluorescent sensors can report transmitter-linked signals in selected environments. These technologies bring temporal and anatomical resolution, but calibration, optical artifacts, selectivity, and tissue localization matter. They measure local extracellular chemistry, not a comprehensive subjective feeling. Pairing terminal measurements with identified VTA firing is especially valuable for testing mechanisms. Mohebi and colleagues.
Anatomical tracing, transcriptomics, and cell atlases
Tracer methods map routes and, under specific protocols, monosynaptic connections; single-cell sequencing maps gene-expression states; spatial transcriptomics helps preserve location. These approaches reveal circuit organization and molecular identity, but gene expression is not itself a behavioral output, and a traced anatomical input does not prove it is functionally active during a specific experience. Watabe-Uchida and colleagues; 2025 VTA molecular review; 2024 Opto-seq experiment.
Human fMRI and structural MRI
Structural MRI estimates tissue properties and anatomy. Functional MRI measures blood-oxygen-level-dependent (BOLD) contrast, a hemodynamic signal related to underlying neural and vascular processes. BOLD in a VTA region of interest is not direct measurement of dopamine concentration, transmitter release, or neuronal spikes. Small size, physiological noise, partial-volume effects, registration, and uncertain anatomical borders make midbrain imaging demanding. In a carefully designed human study, VTA BOLD responses were associated with positive prediction-error computations; this supports a task-related hemodynamic correspondence rather than chemical identification. D'Ardenne and colleagues; Trutti and colleagues.
Ultra-high-field imaging and probabilistic atlases can improve anatomical localization. They do not remove ambiguity about causal cell types or the subjective interpretation of BOLD. A dedicated methods guide explains Functional MRI: How fMRI Measures Brain Activity and Its Limits.
PET and radioligand imaging
Positron emission tomography uses radioactive tracers to measure tracer kinetics and target binding. Dopamine-related ligands can estimate receptor availability or, under validated task and pharmacological designs, infer aspects of endogenous dopamine competition. PET does not ordinarily isolate a single VTA projection in real time, and receptor-binding estimates depend on tracer, model, spatial resolution, and biology. Cumming's radioligand review.
What human research establishes—and what remains uncertain
Human studies support the relevance of midbrain and ventral-striatal circuits to reinforcement learning and incentive-related computations. Their strength is measuring intact people carrying out behavioral tasks, sometimes with sophisticated computational models. Their limitation is that most noninvasive imaging measures population-level or hemodynamic proxies rather than identified dopamine cells and their specific synapses. A model fit and VTA-associated BOLD response together strengthen a mechanistic hypothesis, but do not establish that dopamine in a named cell class caused an individual's psychological state.
Conversely, rodent experiments can directly manipulate genetically defined neural populations with high temporal specificity. Those results have strong within-model causal value and need careful translation across species, incentives, social environments, and clinical presentations. Combining human imaging, experimental animal physiology, and computational analysis provides complementary evidence; no modality subsumes the others.
How VTA research has changed
The classical phase emphasized anatomical dopamine pathways and behavioral reinforcement. Recording studies then showed systematic changes in responses to surprising and predicted reward, motivating computational accounts of temporal-difference learning. Work with cell-type identification, pathway-specific manipulations, and anatomical tracing established that the VTA includes important non-dopaminergic populations and multiple forms of input-output specificity. Schultz, Dayan, and Montague; Cohen and colleagues; Watabe-Uchida and colleagues.
Research from the late 2010s through 2025 has sharpened distinctions among neuron subtypes, dopamine firing versus terminal release, value versus sensory-feature learning, and transcriptionally differentiated cell classes. Particularly useful recent syntheses are de Jong and colleagues, 2022, Kahnt and Schoenbaum, 2025, and the 2025 VTA heterogeneity review. These developments expand the picture from a single reward signal toward experimentally testable circuit-dependent computations.
The VTA in addiction, mood research, and clinical neuroscience
Substance-related learning
Many drugs and associated cues alter dopaminergic and nondopaminergic signaling in mesocorticolimbic circuits. Repeated exposure can modify reinforcement, associative learning, cue responsiveness, and stress-related processes. The VTA is important to mechanistic models of addiction, but addiction involves distributed circuits, developmental and environmental risk factors, social context, and diagnostic behavioral criteria. No single VTA scan or dopamine assay diagnoses substance use disorder. Morales and Margolis; 2026 clinical-mechanistic VTA review.
A changed reward response is not equivalent to an individual diagnosis. Clinical assessment of problematic substance use considers behavior, impairment, time course, and other validated criteria. The VTA evidence provides biological context for that clinical process.
Anhedonia and depression
Motivational and reward-learning alterations are research interests in depression and anhedonia. Some studies associate midbrain-related measures with individual differences, but there is no clinically validated universal VTA biomarker for diagnosing depression, measuring subjective pleasure, or selecting treatment in routine practice. Neuroimaging findings are sensitive to sampling, task, medication, and analytic decisions. For symptom-focused clinical content, see Anhedonia in Depression: Loss of Interest, Pleasure, and Motivation.
Experimental neuromodulation and translation
Circuit-targeted stimulation and manipulation have identified causal contributions of VTA-related pathways in experimental animals. Translating these results into safe and effective interventions requires species-appropriate targeting, human trials, clinical endpoints, long-term follow-up, and risk–benefit assessment. The existence of a promising neural target does not establish a therapeutic protocol, and this article makes no recommendation for self-directed stimulation, supplements, or medication changes.
Common misconceptions about the VTA
Myth: The VTA is the brain's pleasure center
The phrase compresses complex circuitry into one psychological label. VTA dopamine pathways influence learning and incentive motivation, while hedonic experience recruits multiple interacting brain systems. VTA activation alone cannot prove that someone felt pleasure. Berridge and colleagues.
Myth: Every dopamine neuron in the VTA does the same thing
Distinct populations vary by molecular identity, projection target, local modulation, and task response. Average activity can conceal opposing signals in different cell classes or targets. de Jong and colleagues; 2025 VTA heterogeneity review.
Myth: Dopamine is a simple chemical reward released whenever we like something
Dopamine participates in learning, motivation, movement-related processes, and state-dependent neural computation. Stimulus timing, expectation, uncertainty, and terminal dynamics all matter. A reward prediction error can decrease when an expected outcome occurs, even when the outcome remains desirable. Schultz and colleagues; Mohebi and colleagues.
Myth: A bright VTA spot on an fMRI scan proves dopamine release
An fMRI map displays an analysis of BOLD measurements. It does not directly visualize dopamine. Small-region localization, statistical thresholding, task contrasts, and neurovascular coupling must be evaluated before biological interpretation. Trutti and colleagues.
Myth: A VTA difference predicts psychiatric diagnosis or treatment response
Group-level correlates and exploratory imaging markers are research results. Individual clinical utility requires external validation, robust prediction, established thresholds, prospective testing, and demonstrated benefit. A VTA association from a small imaging study does not meet those criteria.
Practical implications
VTA science helps explain why expectations, cue histories, reinforcement schedules, and changing opportunities can shape behavior. Learning from outcomes depends on the relationship between what was anticipated and what actually happened; behavior can therefore change without a parallel change in how enjoyable an outcome feels. These principles inform careful interpretation of habit formation, motivation, and cue-driven responding without reducing them to a transmitter slogan. See How Habits Form: Cues, Repetition, and Automaticity.
For readers evaluating claims about foods, apps, or so-called dopamine resets, a useful question is whether the claim cites direct neural measurement, a behavioral model, or speculative storytelling. Studies of substances and rewarding stimuli are context-dependent; for a concrete example, see Sugar and Dopamine: Reward, Motivation, and Common Myths.
For researchers and advanced readers, identify the VTA subregion, cell-marker strategy, projection target, species, measurement type, timescale, and behavioral task before interpreting an effect. This practice is part of broader scientific reasoning about Neural Circuits: How Brain Networks Generate Behavior and Neurotransmitters: How Chemical Messages Affect the Brain.
Frequently asked questions
Where is the ventral tegmental area located?
It lies in the ventral midbrain near the substantia nigra. Its subdivisions and borders are difficult to define precisely in human MRI, which is why many studies rely on carefully constructed anatomical atlases and probabilistic regions.
What is the main function of the VTA?
The most extensively studied functions involve reinforcement learning, prediction-related neural signals, incentive motivation, and state-dependent responses to important events. These functions arise through differentiated cell populations and their connections.
Does the VTA produce dopamine?
Yes. Many VTA neurons synthesize dopamine and project to other brain regions. The region also houses neurons using GABA, glutamate, or more complex signaling phenotypes.
What is the difference between the VTA and nucleus accumbens?
The VTA is a midbrain source of major ascending projections; the nucleus accumbens is a ventral-striatal target that integrates dopaminergic modulation with cortical, limbic, and local inputs. They interact within broader networks rather than working as isolated modules.
What happens when expected reward does not arrive?
In a classic experimental setting, a subset of dopamine neurons may show a pause or reduction in firing near the omitted expected outcome. That pattern is consistent with a negative prediction error, subject to task conditions and cell-type variation.
Does VTA activity equal happiness or pleasure?
No direct one-to-one mapping exists. VTA-related signals are implicated in learning and motivation, and pleasure involves distinct but interacting neural and psychological processes.
Is VTA dopamine always activated by positive experiences?
No. Responses depend on prediction, salience, novelty, aversion, projection type, and internal state. Certain dopamine populations can also participate in avoidance and responses to threatening events.
Can a brain scan measure VTA dopamine?
Conventional fMRI measures BOLD signals related to hemodynamics, while PET and other specialized methods can probe aspects of dopamine-related biology under model assumptions. Neither ordinary brain scan directly reads all VTA dopamine firing or every person's experience.
Is the VTA involved in addiction?
Yes, VTA-related circuits figure prominently in neurobiological accounts of drug reinforcement and learned cues. Addiction is a clinical condition defined by a broader pattern of behavior and impairment, not by one VTA measurement.
Can a damaged VTA or abnormal VTA scan diagnose depression, ADHD, or addiction?
No validated universal individual diagnostic rule follows from VTA imaging alone. Clinical conditions require their own established assessment criteria, and research measures must demonstrate individual predictive utility before use as diagnostic tests.
How is the VTA different from the entire reward system?
The VTA is one component of a distributed system encompassing several cortical and subcortical structures. Its distinctive scientific questions concern its anatomical subdivisions, neuron classes, inputs, outputs, and signaling dynamics.
Conclusion
The ventral tegmental area contributes to reward learning and motivated behavior through anatomically and molecularly diverse neuronal populations. Evidence for prediction-error signaling is strong for particular midbrain dopamine populations, while current research reveals additional roles for terminal dopamine dynamics, GABA and glutamate neurons, sensory-feature information, novelty, and aversive responses. The most accurate explanation specifies cells, projections, timescale, method, and task; this enables meaningful connections between basic neuroscience, human reward research, and behavioral science without turning a complex circuit into a single psychological slogan.