Overall, magazine training and shaping Mitzy were both
faster and more enjoyable processes than magazine training and shaping her
virtual counterpart, Sniffy. For Sniffy, magazine training took approximately
50 minutes, and shaping required a little more than an hour. For Mitzy,
magazine training took very little time (approximately one 30-minute session),
and shaping was completed in two full training sessions (so, after the 3rd day).
However, one difference that I must note in comparing magazine and
shaping durations between the two rats is that in Mitzy’s case, I more or less
combined the two processes in our first training session. That is to say, by
the end of Day 1, I had already begun to shape Mitzy to touch the bar, and I
stopped reinforcing behaviors having to do with the hopper. I am glad that
Devon recommended this approach, because I believe that it accelerated Mitzy’s rate
of learning, and it also allowed both associations (i.e. sound/food and bar
press/food) to become strengthened simultaneously.
Shaping the two rats was very similar in terms of the
successive approximations that I reinforced. That is to say, after “practicing”
with Sniffy, I used the same procedure to shape Mitzy: reward for sniffing the
back corner, reward for rearing above the bar, reward for touching the bar with
hand or nose, reward for touching the bar with both hands, etc. Both rats emitted
similar behaviors of the response class (i.e. group of movements similar to bar
pressing), even though they varied in the time required to form a strong
association.
One thing that I could not learn from shaping Sniffy
was how to deal with unwanted or “extra” behaviors. Although Sniffy would rear
up or sniff the bar every now and then in his VR training, he would only
receive a reward for a normal bar press. On the other hand, Mitzy demonstrated
a large amount of such behaviors, and they would be reinforced as long as the
bar was depressed when she was doing them.
Additionally, Sniffy very rarely became distracted for
more than ~20 seconds at a time, so training him did not shed any light on what
I should do when Mitzy decided to wander off for 3 minutes at a time. Mitzy’s
cumulative records, therefore, display somewhat less steady rates of response
than do Sniffy’s. (However, we are also comparing FR and VR schedules, so
Sniffy’s should appear more steady).
Related to this is the fact that Sniffy could be
trained for 2 hours in a row, whereas Mitzy often lost motivation after ~20
minutes. This, however, did not necessarily impair what I observed in training
the rats or what I learned from the process, but it is an important difference
in the two processes.
Lastly, a major difference between the two extinction
processes was that Sniffy does not feel frustrated as a live rat likely feels
after she is no longer rewarded for her instrumental responses. Although I saw
somewhat of an extinction burst from Sniffy, his overall extinction results were
not caused by any element of frustration as were Mitzy’s.
Overall, I appreciated having the opportunity to
practice on Sniffy before training my live rat. The comments in the book and in
the program were very helpful, and I learned a great deal from shaping in
particular. Seeing Sniffy’s rate of learning in the shaping process allowed me
to modify my own technique—how “picky” I should be in reinforcing behaviors and
at what rate, how quickly I should administer a food pellet, and so on.
In conclusion, what it really comes down to is this: Mitzy is way cuter than Sniffy.
The End.
No comments:
Post a Comment