Magazine Training, Shaping, and Extinction with Sniffy

Sniffy is a virtual rat program were the behavior of a live rat is simulated on your computer to prepare for actual experiments and training on a actual rat.  What Sniffy looks like is shown below:

(Image of Sniffy, and the virtual operant box)

In order to shape sniffy to bar press, we needed to magazine train first.  The magazine training creates a secondary reinforcer, which was the sound of the hopper when food was presented. This was first done by releasing a pellet of food every time Sniffy was near the hopper.  Once this became a consistent association, I begun the shaping process with only rewarding Sniffy for rearing up on the back wall of the operant box. Gradually Sniffy was only reinforced for rearing up near the bar to encourage pressing the bar. Eventually Sniffy was only reinforced for pressing the bar when the association with the bar pressing and presentation of food was strong enough. Sniffy would press the bar in succession 6 to 10 times, consistently every minute.  At that time is when Sniffy was shaped to press the bar.  The magazine training took roughly 20 minutes, and shaping took roughly 30 minutes.

(Cumulative record of magazine training, and each dot represents when Sniffy was rewarded)

(Cumulative Record of shaping Sniffy to bar press FR 1)

Sniffy was put on one more schedule of reinforcement before extinction from bar pressing, and this was VR 5.  This means that Sniffy would only be fed after pressing the bar 5 times in succession. As the cumulative record shows below, Sniffy did not have a distinct pattern of pressing the bar 5 times in succession to receive food.  Eventually Sniffy recognized the pattern of behavior necessary to be rewarded, which was 5 bar presses in succession.

(Cumulative record of VR 5 reinforcement schedule)

The last step of the shaping process Sniffy needed to complete was extinction.  This is when the learned behavior of bar pressing is terminated and forgotten.  This happens simply by not reinforcing any behavior at all, until the learned behavior is not being performed any longer.  The cumulative record of extinction is shown below.

(Cumulative record of extinction)

Eventually Sniffy's associations between the sound of the hopper and food disappeared and the bar pressing Sniffy learned vanished from the lack of reinforcement.  This disappearance of behavior is less likely to happen with a live rat (Lil Sebastian) because it will still be a underlying learned behavior, which will be very easy to bring back if need be.

This program did help me feel more prepared about shaping Lil Sebastian, but I am not expecting it to be exactly the same.  There were some differences between the live rat and Sniffy.  This exercise definitely made me more excited about shaping Lil Sebastian.

More to come soon!