Professional Documents
Culture Documents
OPERANT DISCRIMINATION
ROBERT M. HERRICK, JEROME L. MYERS,1 AND ARTHUR L. KOROTKIN2
Z7. 5. jVatw/ /!«• Development Center, Aviation Medical Acceleration Laboratory, Jo/nisville, Pennsylvania
The first stage of an experiment (Herrick, was perhaps typical of the operant discrimina-
Myers, & Burke, 1958) which was designed to tion process. The data of the present experi-
study the effects of acceleration on behavior ment permit a test of this hypothesis of
was the development of an operant discrimina- constancy.
tion. The principal purpose of the present
report, which describes the formation and METHOD
maintenance of the discrimination, is to present Discrimination Technique
an operant discrimination technique. A second-
A short discussion is necessary at this point in order
ary purpose of the report is to provide data to make explicit some of the reasoning behind the design
which allow an appraisal of a recently formu- of the discrimination procedure.
lated hypothesis which describes changes in The discrimination technique used consisted of
operant response rate during the formation of variable-interval (VI) reinforcement for responses
a discrimination. made in the presence of SD and of no reinforcement for
responses made in the presence of SA, Several durations
The discrimination procedure was designed of the SD and S4 periods were used. Since some of the
(a) to provide response-rate measures both S" periods were as short as 5 sec., a premium was placed
for SD periods and for S* periods and, inci- on responding immediately after the onset of SD. On
dentally, to encourage high respons'e rates in the other hand, the minimum duration of an S* period
was 30 sec. Shorter S* periods were not included on the
SD; (b) to insure that the SD and 5^ provided assumption that they would increase the chances for
by E were the. relevant cues for S, To achieve a response made in the presence of S* to be followed by
this latter goal, several precautions were SD, and therefore, perhaps, appear to cause (.he onset
taken to eliminate extraneous cues either for of SD, Another consideration affecting the selection of
responding or for not responding in either SD the durations was the assumption that, within limits, a
fixed period of discrimination training time could be
periods and/or SA periods. used more efficiently by providing a greater, rather
The hypothesis to be evaluated on the basis than a smaller, number of SD and S* periods.
of the discrimination data was presented by The VI reinforcement schedule used in SD was
Smith and Hoy (1954). They found that selected because it seemed to have fewer objectionable
characteristics than the other common reinforcement
during the development of a discrimination schedules, with the possible exception of a variable-
the total daily number of responses remained ratio schedule. For example, a regular reinforcement
about the same; i.e., the total number of schedule would be undesirable for several reasons. Not
responses emitte'd during the first daily dis- only would much of the lime in SD be given over to
crimination session was approximately equal eating behavior, but difficulty in interpretation of the
data would arise because of satiation effects. Of perhaps
to the total number of responses emitted in more importance, however, is the possibility that the
each of the subsequent daily discrimination discrimination could develop to some extent simply as
sessions. This constancy obtained because the a result of repeated conditionings (in SD periods) and
decrease in the number of responses emitted extinctions (in 5 d periods) in which the cue for not
responding would be a response not followed by the
in the presence of the negative stimulus (5A) delivery of a reinforcement (Bullock & Smith, 1953).
was compensated for by a corresponding in- At predetermined times during each SD period, the
crease in the number of responses emitted in reinforcement circuit was automatically set so that the
the presence of the positive stimulus (S"). next appropriate response (i.e., lever press) would result
In discussing their data, Smith and Hoy sug- in the delivery of a reinforcement. In order to achieve
a VI reinforcement schedule, these times were varied,
gested that this finding of constancy in the not only within each SD period, but also from one SD
total number of responses emitted each day period to another. In addition, in order to decrease the
during the development of a discrimination chances that 5 would stop responding during an SD
period after obtaining a fixed number of reinforcements,
1
Now at the University of Massachusetts. the number of reinforcements which could be obtained
2
Now at Mayview State Hospital, Mayview, Pa. during different SD periods was varied.
359
360 ROBERT M. HERRICK, JEROME L. MYERS, AND ARTHUR L. KOROTKIN
400 60 RESP,V M I N
LJ
(/I
UJ
O .200
Q_
if)
UJ
cc 0 25
TIME (WIN)
LjJ
DAY #1
TIME
FIG. 1. Sample daily records of one rat during 40 days of discrimination training. The small marks on the
SD curves indicate when food pellets were delivered. The records for the higher SD rates underestimate the number
of responses by a variable amount up to about 8%.
made by one rat during all the SD (or SA) SA rate had decreased to 12.9 responses/minute.
time of the session. Thus, the rates are mean With further training, the SD rate continued
rates which do not take into account any to increase, but at a slower rate. The 5A rate
changes in rate during the SD (or 5A) periods dropped until about Day 16, after which there
of a daily session. No correction was made was little change. The rank-order coefficient
in the SD rates for the time taken to eat the of correlation, relating the median SD rate
food pellets. Such a correction would, of and the median 5A rate for the first 15 days of
course, raise the SD rates. discrimination training, was found to be
Figure 1 presents cumulative response —0.81. The rate over the last 5 days of the
curves for one of the rats on selected days of experiment was about 65 responses/minute
discrimination training. Each SD (or 5A) for SD and about 5.5 responses/minute for 5A.
curve of Figure 1 is a composite of the several A comparison of these rates with those of the
S!> (or 5A) periods of a daily experimental initial day of discrimination training indicates
session. For the higher rates shown in Figure 1 that the SD rate increased to about three and
the cumulative response curves underesti- one-half times its initial value, while the
mated the number of responses by a variable SA rate decreased to about three-tenths of
amount, averaging about 6 to 8%. its initial value.
Figure 2 summarizes the data of the experi- The uppermost curve of Figure 2 summarizes
ment. On the first day of discrimination the discrimination data in terms of an index
training the median rates for the rats were computed by dividing the rate in SD by the
almost exactly the same for SD and for 5A, SD rate plus the S* rate, and multiplying
viz., 18.1 and 18.9 responses/minute, re- by 100. The median rates plotted in Figure 2
spectively. The rate curves soon drew apart, were used to compute the index values. With
however, so that after 10 days of discrimina- equal response rates in SD and 5A, a situation
tion training, for example, the median SD which was closely approximated on the first -
rate had increased to 45.2, while the median day of discrimination training, the index
362 ROBERT M. HERRICK, JEROME L. MYERS, AND ARTHUR L. KOROTKIN
Measures of behavior following repeated exposure Aiiiat. Med. Acceleration Lab. (Jolmsmlle) Rep.,
to negative acceleration. /. aviat. Med., 1958, 29, 1957, No. NADC-MA-S711.
343-349. SMITH, M. H., JR., & HOY, W. J. Rate of response
HEBEICK, R. M., MYERS, J. L., KOROTKIN, A. L., & during operant discrimination. /. exp. Psychol.,
1954 48
BURKE, R. E. Body weight and food intake > ' 259-264.
changes during instrumental learning. USN ADC, Received June 27, 1958.