Questions: Reward Learning and Dopamine Circuits

Question 1

An animal has been thoroughly conditioned: a tone reliably predicts food. On a test trial, the tone sounds but no food is delivered. What happens to VTA dopamine neuron activity at the moment the food would have arrived?

• A. Dopamine activity surges, because the animal is highly motivated by hunger at that moment
• B. Dopamine activity is unchanged from baseline, because no food arrived and no prediction was made
• C. Dopamine activity dips below baseline — a negative prediction error signal
• D. Dopamine activity increases at the tone and then returns to baseline when food is omitted

Answer: (C)

After conditioning, the brain predicts food at a specific time. When food fails to arrive, dopamine neurons generate a negative prediction error: activity falls *below* baseline at the moment reward was expected but absent. This is the neural signal for 'worse than expected.' It is not merely silence — the dip is an active downward signal that weakens the value assigned to preceding events. This asymmetry (burst for better-than-expected, dip for worse-than-expected) is the key signature of prediction error coding.

Question 2

After extensive training with a tone-food pairing, when does the dopamine burst primarily occur during a successful trial (tone sounds, food delivered on schedule)?

• A. At the moment of food delivery, because dopamine codes the presence of reward
• B. At the tone onset, because it is now the earliest reliable predictor of reward
• C. Equally at both the tone and food delivery, because both events are associated with reward
• D. Only when the food is eaten, reflecting the pleasurable taste experience

Answer: (B)

This temporal shift is the hallmark of learned prediction error signaling. Early in training, dopamine fires at the unexpected food reward. As the tone reliably predicts food, the dopamine burst migrates backward to the tone — the earliest predictor. The dopamine response at food delivery itself returns to baseline (no surprise = no prediction error). The system has learned what predicts reward and signals that learning by moving the burst to the predictive cue. If the tone itself were predictable by an earlier cue, the burst would migrate back further still.

Question 3

Dopamine neurons fire most strongly in response to rewards themselves, and their firing rate at the time of reward delivery stays elevated even after extensive conditioning.

• T. True
• F. False

Answer: False

This is the central misconception about dopamine and reward. What dopamine actually encodes is prediction error — the discrepancy between what was expected and what occurred. After conditioning, the animal fully predicts the reward following the cue, so there is no positive prediction error at delivery, and dopamine activity at that moment returns to baseline. The burst now occurs at the conditioned cue. Dopamine responding to rewards themselves describes only early, unexpected encounters with reward — it is a learning signal, not a 'pleasure signal' that fires whenever something good happens.

Question 4

In addiction, drugs of abuse can sensitize the dopamine prediction error system to assign enormous value to drug-related cues, even as natural rewards like food and social connection lose their ability to activate the system effectively.

• T. True
• F. False

Answer: True

Drugs of abuse produce dopamine surges far larger than natural rewards — often 5–10× the normal signal. This massive 'prediction error' trains the system to assign enormous value to drug-related cues (paraphernalia, environments, people associated with drug use). Meanwhile, the system's baseline recalibrates such that natural rewards, which produce much smaller dopamine signals, appear to generate insufficient prediction errors to sustain motivation — the neural substrate of anhedonia and the compulsive focus on drug-seeking over natural pleasures.

Question 5

Why does dopamine fire at a conditioned stimulus rather than at the reward itself after learning is complete, and what would happen to dopamine activity if the expected reward were suddenly and permanently omitted?

Model answer: Dopamine encodes prediction error (actual − expected reward), not reward itself. Once a cue reliably predicts reward, the animal fully expects the reward at delivery — so there is no positive prediction error, and dopamine at reward delivery returns to baseline. The dopamine burst migrates to the cue because the cue is where the first 'better-than-expected' signal now occurs (the cue was unexpected before it became a predictor). If the expected reward is permanently omitted, dopamine initially dips below baseline at the expected reward time (negative prediction error). Over repeated omissions, the cue itself stops predicting reward, the conditioned response extinguishes, and the dopamine burst at the cue disappears — the system unlearns the association through repeated negative prediction errors.

This migration of the dopamine signal is not a quirk — it is exactly what a rational learning system should do. Resources and attention should be directed to the earliest reliable signal of a coming reward, not to the reward itself when it's already certain. The extinguishing of the cue response when reward is permanently omitted shows the same mechanism in reverse: negative prediction errors at omission, then at the cue, systematically downgrade the value estimate until the cue is no longer worth responding to.