Pavlovian or classical conditioning is an associative learning paradigm. As a hallmark, the outcome does not depend on the actions of the performing agent. This allows the experimenter to have full control over the outcome contingencies, making this task ideal for probing the neural mechanisms underlying the processing of probabilistic outcomes (Pearce and Hall, 1980; Schultz et al., 1997). We developed a probabilistic Pavlovian conditioning protocol in mice using an open-source head-fixed setup where we use two auditory cues predict probabilistic outcomes with different contingencies. We use a one training session/day protocol, i.e., each training step should be performed on consecutive days. The probabilistic training protocol consists of two cues and associated reinforcement; one cue predicts likely reward (80% reward, 10% punishment, 10% omission) and the other cue predicts likely punishment (25% reward, 65% punishment, 10% omission) (Figure 1).

This Pavlovian training protocol allows us to study the differential behavioral and neuronal correlates of expected and surprising outcomes. It has been tested in combination with chronic in vivo electrophysiological recordings and optogenetic manipulations in ChAT-Cre and PV-Cre mouse lines.

1. Figure Schematic diagram of reinforcement contingencies during different training phases.