Experiment 3:
First, we trained rats to associate two cues with an 8- and 16-second duration, respectively. Then, we instituted a change phase, in which we stopped presenting the 16-second cue. In one ‘no-change group’, we kept presenting the 8-second cue, maintaining its duration. Critically, we also introduced a novel cue that predicted reward availability at 4-seconds. This changed the reward structure, without introducing a cue-duration relationship change. In contrast, in the ‘change group’, we decreased the 8-second cue’s duration to 4 seconds and associated the novel cue with an 8-second duration.

The initial columns of all data files contain dummy-coded information needed to index the data. Description below.

Binned data files:
Column 1: subject number, Column 2: phase (1 = last training session, 2 = test session), Column 3: cue (1 = short cue, 2 = novel cue, 3 = long cue), Column 4: time-values relative to trial start that correspond to binned data for each probe trial a rat completed. The binned data start at column 5 and are oriented in a time X trial matrix format. As described in the general README, if a rat did not receive an Nth trial during a session, the Nth column will contain the value NaN and should be disregarded when computing mean response rate across trials.

Time event data files:
Columns 1 to 3 are same as above. The final two columns contain time-stamps for events (as seconds from session start) and the corresponding values in the adjacent column contain dummy-coded values for the identity of the event (1 = short cue on, 2 = novel cue on, 3 = long cue on, response = 4, trialEnd = 5, sessionOver = 6). 

