-
In addition to the vdmlab dependencies, this project also requires matplotlib, seaborn, jupyter and pandas libraries.
conda install matplotlib seaborn jupyter pandas
- Pull emi_biconditional.
- Get MedPC data from the server. Put training data files in
emi_biconditional\\cache\\datafolder.
This code contains analyses for:
- Behavior (medpc):
analyze_behavior.pyWhich analyzes and plots durations, number of entries, latency to first entry, and percent with responses. - Behavior (vdmlab operant box):
analyze_vdmlab_behavior.py
Below is a brief overview of biconditional experiment task and analysis work flow.
The main objective of this task was to determine whether subjects are able to learn a biconditional discrimination in a reasonable amount of time.
Each session contained 32 trials (8 of each type; with a total of 16 rewarded and 16 unrewarded trials). There was an average 4 min intertrial interval (range from 2.5 min to 5.5 min). Each cue (light or sound) was presented for 10 s, with a 5 s delay between cues. Two sessions were run daily during the light cycle; one beginning at 7am, the other ending at 7pm.
Cues in this experiment were:
- light1: steady cue light
- light2: flashing house light (2 flashes per s)
- sound1: pure tone (frequency: 1500, amplitude: 100. ~85dB) [note: outside LED on during sound1]
- sound2: white noise (amplitude: 85. ~85dB)
Each light-sound pairings were counterbalanced across rats (4 in each group) such that each pairing that is rewarded in one group is not in the other and vice versa.
- Trial 1: light1 -> sound2 -
- Trial 2: light1 -> sound1 +
- Trial 3: light2 -> sound1 -
- Trial 4: light2 -> sound2 +
- Trial 1: light2 -> sound2 -
- Trial 2: light2 -> sound1 +
- Trial 3: light1 -> sound1 -
- Trial 4: light1 -> sound2 +
RH01-RH06 & R103, R105 were trained in medpc boxes. RH05 & R105 underwent further training in vdmlab operant box.