# VI Schedules: Collection-time effect

[From Bruce Abbott (960215.1955 EST)]

After speculating in my last post about the effect of collection time on
reinforcement rate in VI schedules, I thought it might be well to add
collection time to the simulation and see what happens. I tried two values:
0.0 seconds and 5.0 seconds. Nominal VI schedule values ranged from 5 s to
120 s and response rate was always 1.0 resp/s average using a
constant-probability response distribution. The following is based on
averages of 5 simulated 30-min sessions.
Resp/Session
VI CT=0 CT=5 %B %Rft
5 1803.4 659.0 36.5 76.0
15 1804.2 1232.6 68.3 93.4
30 1790.0 1527.6 85.3 99.5
60 1781.4 1588.6 89.2 97.1
120 1822.0 1714.0 95.2 100.0

At 1 rsp/s there would be 1800 responses in a 30-m session. The variation
in the column labeled CT=0 is due to the use of a probability distribution
to generate responses. The same variability affects the numbers in the CT=5
column, and this produces even more sampling variability in the percentage
reduction column (%B). Even so, it is quite apparent that a 5-s collection
time can have a major impact on average response rates over a session, and
that this effect diminishes as the VI interval length increases. If the VI
sec/rft is converted to rft/min, the percentages decrease approximately
linearly from 100% as a function of reinforcement rate.

In contrast to these results for response rate, reinforcement rate was much
less seriously affected by the inclusion of the 5-s collection time. Even
on VI 5-s, the 5-s collection time only reduced obtained reinforcement rate
to 76% of its value when collection time was 0 s. Thus the greater
difference between programmed and obtained reinforcement rate at VI 15-s
than at VI 30-s (and when comparing VI 30-s to VI 60-s) seen in Fran
McSweeney's data cannot be explained away as the effect of a constant
collection time on obtained reinforcement rates. I'm getting anxious to
find out just what really IS happening on these schedules, information that
can only come from the lab.

Regards,

Bruce