General Information:
The package includes data from a series of experiments where rats performed a time-perception task (peak-interval procedure) inside operant chambers. Experiment- and group-specific details are provided in their respective subfolders. However, in general, training progressed through three phases. During an initial training phase, trials began with the presentation of a cue that predicted reward availability after a certain duration elapsed. Different cues were used, each associated with a different delay to reward (e.g., tone-8s / light-16s). Rats were free to respond at any point when the cue was presented. However, reward was only delivered for a response after the cues duration had passed. When a rewarded response occurred, the cue terminated and reward was delivered. Probe trials (30% of trials within a session for each cue) were also included in which the cue was presented for much longer than normal (3-4 times the length of the longest duration in effect), and no reward was provided. Once trained, a change phase was implemented. During this phase, the duration associated with one of the cues was changed, and one of the cues was no longer presented. Finally, a test phase was instituted. This was identical to the change phase. However, we reintroduced the cue that was omitted in the previous phase (20% of trials within a session). No feedback was given during unchanged cue trials. 

Data from the last training and critical testing sessions for each experiment/group are provided in a compiled manner in two different formats. The initial columns for all files include dummy-coded information that can be used for logical indexing by data analysis software (for further details see individual experiment data READMEs). However, the files are optimized for analysis with matlab. 

The BinnedData files include the primary data used in our analyses. These files contain response rates across time during individual probe trials. Responses are grouped into 1 second bins and stored in a time X trial manner. The number of trials rats received during a session differed to some degree. Therefore, the n-th column of a rats time X trial matrix includes the value NaN (standing for not-a-number) in cases where the number of columns exceeds the number of trials the rat received. These values should be disregarded when averaging across trials to compute average probe-trial response rate across time. If done properly, mean response rate across time should resemble a gaussian-like curve. 

The second TimeEventData files contain raw timestamps that can also be used to create trial X time matrices for each rat/session. The data is stored in time-event-format (for further reading see: http://www.brown.edu/Research/Timelab/archive/users.htm), which is a two column vector. The first column shows the time at which a given event (e.g., trial start, response, trial end) occurred, expressed in seconds relative to the start of a session. Dummy-coded event labels are contained in the second column, next to the event’s timestamp. Details regarding the identity of each event are described further in each subfolder. However, across experiments, a response, trial-end, and session-end are dummy-coded using the values 4, 5, and 6, respectively. Again, logical indexing can be used to analyze the data.  
