Prediction-error, the discrepancy between real and expected outcomes, drives associative learning. It is best exemplified in the blocking paradigm. In blocking, impairment in learning about the predictive relation between a cue (e.g., a clicker) and an outcome (e.g. footshock) is observed when this learning takes place in the presence of a good predictor (e.g. a light) for the same outcome. Small prediction-error generated by the light leads to impairment in learning about the clicker-footshock relationship. The mere presentation of the two stimuli in compound in the absence of pre-training of one of those stimuli does not yield blocking. That is, in the so-called overshadowing control condition, the clicker is presented in compound with the light that was not previously associated with the footshock. This arrangement leads to robust learning about the clicker due to the presence of a maximum prediction-error. Dopamine (DA) in the ventral tegmental area (VTA) has been implicated in reward prediction-error (RPE). Evidence suggests an opposing role of DA in fear and reward. Here we undertook several experiments aimed at elucidating the role of VTA DA neurons in aversive prediction-error (APE). We used a powerful behavioural and theory-driven approach by combining blocking and the corresponding overshadowing control in the context of aversive (fear) learning along with optogenetics. We used the Th-cre+/- rats in order to exercise fine temporal control over VTA DA neurons during aversive learning. Taken together, our results provide evidence that optical stimulation of VTA DA neurons and their terminals in the nucleus accumbens (NAc) at the time of expected shock augmented the blocking effect by attenuating APE and further impaired learning about the blocked cue. We did not observe such an effect in the overshadowing control nor many neural control groups.