Professional Documents
Culture Documents
stimulus; properties which might be first session to 125 in later sessions. The
expected to include frequency as well SD and SA periods were presented in a
random order. Daily training sessions were
as intensity, rate of interruption, and continuedD until the average rate of response
so on. in the S periodsAwas at least four times
greater than in S periods in each of two
EXPERIMENT I successive sessions. When testing the dis-
crimination against this criterion, the sessions
Method consisted of 25 SD periods and 50 SA periods.
Subjects. The 5s were eight male, white Five 5s received differential training. From
Carneau pigeons maintained by restricted six to nine sessions were required to reach
feeding at 80% of their body weight under the criterion of discrimination.
free feeding. In testing for generalization, the stimuli
Apparatus. A Skinner automatic key- were presented for 33-sec. periods separated
pecking apparatus was used. The general by 7-sec. blackouts as in training. Eight test
features of the apparatus have been described stimuli were used. They consisted of seven
elsewhere (Ferster & Skinner, 1957). The tones approximately equally spaced along a
reinforcement was a 4-sec. period of access logarithmic scale of frequency. Their fre-
to a tray of mixed grain. The occurrence of quencies were 300 cps, 450 cps, 670 cps, 1500
a reinforcement was signaled by turning off cps, 2250 cps, 3500 cps, and 1000 cps (SD in
the illumination of the response-key, and by training). The eighth stimulus was no tone.
turning on illumination for the food tray. In a test session, the eight stimuli were pre-
The auditory stimuli were pure tones sented in eight blocks, making a total of 64
sounded by a 5-in. speaker suspended 3 in. presentations. The order of presentation
behind the response-key. The tone was was governed by 8 X 8 latin squares so that
interrupted for 0.25 sec. in each second in each stimulus appeared once in every ordinal
order to reduce possible adaptation to a position in the block.
steady tone. The amplification of tones at Tests for generalization were conducted
different frequencies was adjusted to give without reinforcement except where noted.
a sound pressure level of 70 decibels re 0.0002 Three tests for generalization separated by
dynes/cm.2 as measured with the aid of a single retraining sessions were usually made.
calibrated microphone. The procedure during retraining was the
Preliminary training. Four sessions of same as that used during training.
preliminary training were given. The key-
response was conditioned using a method of Results
successive approximations and then placed
on a variable interval (VI) schedule of rein- Nondifferential training. The per-
forcement. The tone which served as the centage of responses made to each
reinforced stimulus, or SD, was presented as test stimulus following nondifferential
soon as several responses occurred. The
third and fourth sessions were divided into training is shown in Fig. 1. Prior to
25 working periods of 33-sec. duration by the the generalization tests, S?o received
introduction of 7-sec. blackout intervals. In 10 training sessions, 5?i received 20,
each working period, the SD was presented
and responses were reinforced on the VI
schedule which delivered from 1 to 3 reinforce- o NO. 70
ments with an average of 1.7 reinforcements. • NO. 71
Procedure. In the nondifferential training V NO. 72
method, sessions consisted of 25 working
periods in which a tone at 1000 cps (SD) was
presented and responses were reinforced on
the VI schedule. Three 5s received 10, 20,
or 40 daily sessions of nondifferential training. 300 450 670 IOOO I50O 2250 350O NO
In the differential training sessions, the FREQUENCY IN CYCLES PER SECOND TONE
and more rapid extinction in succes- and no tone. Generalization tests were
sive tests. Significant interactions conducted without reinforcement. Three
tests were made with a single retraining
were obtained for the effects of stimuli session between tests.
with blocks (F = 1.80; 49/273 df,
P < .01), and of stimuli with tests Results
(F = 2.28; 14/273 df, P < .01). The
interaction effects were due in part Nondifferential training. The gra-
to a progressive sharpening of the dients obtained following nondiffer-
generalization gradient both from the tial training are shown in Fig. 3.
beginning to the end of a test, and in Prior to the generalization tests, S&
successive tests. Similar trends are received 10 training sessions, Sao re-
examined in more detail with the ceived 20, and S38 received 40. The
results of Exp. II. gradients are similar to those obtained
following nondifferential training with
EXPERIMENT II respect to a single frequency in that
they show a nearly equal percentage
Experiment II was designed to of response to all test stimuli, includ-
extend the comparison of generaliza- ing no tone. Again, it is apparent
tion gradients obtained following dif- that increasing the amount of training
ferential or nondifferential training did not greatly alter the result. As
to a case in which responses are rein- in Exp. I, there was some evidence
forced to tones at two frequencies for a shallow gradient; each 5 had a
(cf. Kalish & Guttman, 1957). local peak in response strength at 450
cps, although the curves over all test
Method stimuli were irregular.
The same method was used as in Exp. I
An analysis of variance for this
except for the details noted below.. condition shows effects similar to
Procedure. Nondifferential training ses- those for the nondifferential condi-
sions consisted of 30 SD periods. A 450-cps tion in Exp. I. The effect of stimuli
tone was presented in IS of these periods; a was again significant (F = 3.23; 7/448
2500-cps tone was presented in the remaining
15 periods. The tones were presented in a df, P < .01), and the simple inter-
random order. Three 5s received 10, 20, or actions of stimuli with blocks or tests
40 daily sessions of nondifferential training. were not significant.
The differential training procedure in- Differential training. Generaliza-
volved the same SD periods, and in addition, tion gradients following differential
A
S periods in which no tone was presented
and responses were nonreinforced. The training are shown in Fig. 4. Differ-
number of SA periods was increased from 30
in the early training sessions to ISO in later 30
sessions. This procedure does not involve o NO. 29
• NO. 30
explicit training to discriminate between the V NO. 36
two tones used as an SD.
Daily sessions were continued until the
criterion of discrimination was reached (4 to i 10
1 ratio of SD to SA rate of response) for each
tone with respect to no tone. Six 5s received
"• 250 450 670 1000 1500 250O 4000" NO
differential training. From 7 to 11 training FREQUENCY IN CYCLES PER SECOND TONE
sessions were required to reach the criterion
of discrimination. FIG. 3. Generalization gradients follow-
The tests for generalization were made ing noiidifferential training with a 450-cps
with the following stimuli: tones at fre- tone and a 2500-cps tone as SDs. Individual
quencies of 250 cps, 450 cps (SD), 670 cps, gradients are based on the mea.ns of three
1000 cps, 1500 cps, 2500 cps (SD), 4000 cps, generalization tests.
250 HERBERT M. JENKINS AND ROBERT H. HARRISON
a general sharpening of the gradient ential, these stimuli are repeatedly rein-
from the first to later tests, and within forced and may predominate in control
the course of a single test. Figure 5 of the response with the result that the
shows the mean percentage of re- gradient of generalization observed upon
sponses to the test stimuli in succes- varying the experimental stimulus is flat
or nearly so. When, on the other hand,
sive tests. Figure 6 plots these data training is differential, these incidental
for successive blocks within the tests. stimuli are both reinforced in the pres-
ence of SD and nonreinforced in its
DISCUSSION absence. This reduces their effective-
ness in controlling the response. Hull
The experiments provide clear evi- assumed, as does Restle (1955), that
dence that differential training can stimuli accompanying reinforced and
greatly sharpen a gradient of generaliza- nonreinforced responses adapt out, or
tion even when this training does not in some manner become nonfunctional.
involve an explicit discrimination be- The resulting increase in the relative
tween stimulus values along the dimen- weight of the experimental stimulus in
sion of the gradient. The conclusion controlling the response makes it possible
stated by Lashley and Wade (1946, to obtain a sharply sloping gradient of
p. 74): "the 'dimensions' of a stimulus generalization along some dimension of
series are determined by a comparison the experimental stimulus. It is of
of two or more stimuli and do not exist interest to note that, according to this
for the organism until established by account, differential training involving
differential training" is weakened by neighboring stimulus values would also
this result if the phrase "two or more weaken the control by incidental stimuli.
stimuli" is taken to mean stimuli having The effect of that procedure in sharpen-
values on the dimension of the gradient. ing the gradient might be largely due
It now seems likely that the absence to this factor rather than due to the
or near absence of a sloping gradient presumed development of a counter
along some dimension of a stimulus does gradient of conditioned inhibition cen-
not imply that training to discriminate tered around the value of the nonrein-
values on this dimension is required. forced stimulus. In an experiment by
The present results suggest that the. Reinhold and Perkins (1955) it was
failure to obtain such a gradient may be found that differential training on a
due to a lack of control over the occur- feature of a compound stimulus unre-
rence of the response by the experimental lated to the feature used in testing for
stimulus. When presenting or removing generalization produced a gradient with
the stimulus has little or no effect upon a steeper slope than did nondifferential
the occurrence of the response, we can training. They suggested that a set to
only conclude that the stimulus, and discriminate was developed by differen-
consequently any dimension of the tial training and that this resulted in less
stimulus, is irrelevant to the response. generalization. However, their results
The results pose a theoretical question: could also be accounted for by the reduc-
how account for the effectiveness of non- tion in effectiveness of incidental stimuli
reinforcement in the absence of the tone through differential training.
in- placing the tone in control of the A weakness in the account offered
response? A discussion of stimulus gen- by Hull and Restle lies in the specifica-
eralization by Hull (1952, p. 59 ff) is tion of the conditions for adaptation
relevant here. He pointed out that in of incidental stimuli. It is not sufficient
addition to the experimental stimulus, to state that a stimulus accompanying
incidental stimuli (e.g., stimuli arising both the reinforcement and nonreinforce-
from constant features of the apparatus) ment of a response adapts out, since
accompany the reinforcement of the in the present experiment responses to
response. When training is nondiffer- the SD were reinforced only inter-
252 HERBERT M. JENKINS AND ROBERT H. HARRISON
first approximation, the gradients were sym- GUTTMAN, N., & KALISH, H. I. Discrimin-
metrical about the SD or SDs when plotted ability and stimulus generalization. /.
on a logarithmic scale of frequency. exp. Psychol, 1956, 51, 79-88.
It is concluded that training to dis- HOVLAND, C. I. The generalization of con-
criminate the presence from the absence of a ditioned responses: I. The sensory gen-
stimulus can greatly enhance the amount of eralization of conditioned responses with
control exerted over a response by a specific varying frequencies of tone. J. gen. Psy-
property of the reinforced stimulus. chol, 1937, 17, 125-148. (a)
HOVLAND, C. I. The generalization of con-
ditioned response. IV. The effects of
REFERENCES varying amounts of reinforcement upon
BROWN, J. S. Generalized approach and the degree of generalization of conditioned
avoidance responses as a function of responses. /. exp. Psychol, 1937, 21, 161-
stimulus intensity and strength of motiva- 176. (b)
tion. /. comp. Psychol, 1942, 33, 209- HULL, C. L, A behavior system. New Haven:
226. Yale Univer. Press, 1952.
BULLOCK, D. H., & SMITH, W. C. An effect KALISH, H. I., & GUTTMAN, N. Stimulus
of repeated conditioning-extinction upon generalization after equal training on two
operant strength. /. exp. Psychol., 1953, stimuli. /. exp. Psychol, 1957, 53,139-144.
46, 349-352. LASHLEY, K. S., & WADE, M. The Pav-
lovian theory of generalization. Psychol.
FERSTER, C. B., & SKINNER, B. F. Schedules Rev., 1946, 53, 72-87.
of reinforcement. New York: Appleton- LITTMAN, R. A. Conditioned generalization
Century-Crofts, 1957.
of the galvanic skin reaction to tones.
GRANDINE, L., & HARLOW, H. F. Generali- J. exp. Psychol, 1949, 39, 868-882.
zation of the characteristics of a single REINHOLD, D. B., & PERKINS, C. C., JR.
learned stimulus by monkeys. /. comp. Stimulus generalization following different
physiol. Psychol, 1948, 41, 327-338. methods of training. J. exp. Psychol,
GRANT, D. A., & SCHILLER, J. J. Generali- 1955, 49, 423-427.
zation of the conditioned galvanic skin RESTLE, F. A theory of discrimination
responses to visual stimuli. J. exp. Psy- learning. Psychol. Rev., 1955, 62, 11-19.
chol., 1953, 46, 309-313. SCHLOSBERG, H., & SOLOMON, R. L. Latency
GRICE, G. R. The acquisition of a visual of response in a choice discrimination.
discrimination habit following responses to /. exp. Psychol, 1943, 33, 22-39.
a single stimulus. /. exp. Psychol., 1948,
38, 633-642. (Received May 11, 1959)