LATENT INHIBITION:

THE EFFECT OF NONREINFORCED PRE-EXPOSURE TO THE

CONDITIONAL STIMULUS1
K. E. LUBOW AND A. U. MOORE 2
Con/ell University

Experiments dealing with latent learning and, further, that it is a stimulus-stimulus

and sensory preconditioning have sought, for association that is learned. Although the first
the most part, to differentiate between two conclusion is acceptable when it refers to
conceptual trends in associative learning "patent reinforcement," the second conclusion
theory. One of these theories states that con- is not deducible from the data of these experi-
tiguity of stimulus and response plus rein- ments. In most cases where S-S association is
forcement are necessary to form an association assumed, the E has not been interested in ob-
(Hull, 1943). The other states that mere taining response measures during the course of
contiguity of stimuli is sufficient to form an the pre-exposure period. There are some ex-
association (Tolman, 1932). ceptions (e.g., Lumsdaine, 1939). A more
Early latent-learning experiments pre-ex- complete elaboration of the suggestion that
posed Ss to the multiple undefined stimuli of a latent learning and sensory preconditioning
complex maze (Blodgett, 1929; Tolman & may be response-mediated may be found in
Honzik, 1930), These stimuli were to be as- Osgood (1953). In general, it appears that
sociated in the absence of reinforcement, and there is a technical as well as conceptual
the association later was to be made manifest difficulty in finding and measuring the type
by the introduction of reward. On the other of response that is important.
hand, sensory preconditioning experiments The experiments to follow demonstrate a
pre-exposed the 6*s to two discrete stimuli design which uses pre-exposure to a single
(Brogden, 1939). These, again, were to be relevant stimulus with no "patent reinforce-
associated in the absence of reinforcement and ment." It is hypothesized that any difference
the association was to be made manifest in in rate of subsequent conditioning between
the course of subsequent training. the pre-exposed stimulus and a novel stimulus
These two classes of experiments are similar; is a function of a response learned during the
the only apparent differences are the type of nonreinforced presentation of the stimulus;
stimuli to be associated and the form of training an attempt is made to identify the mediating
used to make the otherwise latent learning response.
apparent. Sensory preconditioning experiments
have employed discrete specifiable stimuli and EXPERIMENT I
classical conditioning; latent-learning experi- Method
ments (at least the early ones) have used
unknown or vaguely identified "maze stimuli" Subjects
and instrumental learning. The data from both There were eight 5s: four goats, two males and two
types of research, when they produce positive females; and four sheep, two males and two females.
results, frequently are used as the basis for The .Ss were members of the flock kept by the Cornell
Behavior Farm Laboratory. All were experimentally
concluding, either implicitly or categorically, naive and led a pattern of life common to domestic herd
that reinforcement is not necessary for learning animals. The mean age was approximately one-half
1 year.
This study was reported, in part, at the Eastern
Psychological Association meetings, Philadelphia, 1958. A pparalus
The research was supported by a grant from the
Josiah H. Macy, Jr. Foundation and by a fellowship to The apparatus consisted of a metal head stock which
the first author from the National Institutes of Health, \vas constructed so that the animal could move its head
Public Health Service. in any direction (see Fig. 11. The stimuli were a 60-w.
2
The authors wish to thank George Boguslavsky for bulb and a rotor 2 ft. lo the right of the stock. The
his aid in preliminary investigations and in the design rotor was composed of two triangular aluminum blades
of the apparatus. with a 3j-in. height and a 5-in. base. When activated as
415
416 R. E. LUBOW AND A. U. MOORE

tinued to be alternated until a previously determined

learning criterion of ten CRs had been reached for that
signal to which conditioning was slowest. Thus, both
signals were presented the same number of times.
During Phase 2, both signals were of a 10-sec. duration
and the intertrial interval length again varied from 30
sec. to 2^ min. These signals were invariably followed by
shock. A flexion of the right foreleg during the presenta-
tion of the CS was counted as a CR. The procedure for
each animal is summarized in Table 1.

Results
The learning score (i.e., the number of
reinforced presentations of the signal before
achieving the tenth CR) for the novel test
stimulus and the adapted test stimulus is
also presented in Table 1. Except for two small
FIG. 1. Experimental room with goat in harness. deviations, all the animals conditioned more
Stimuli, rotor, and light, to right of 5. Shock electrodes slowly to the pre-exposed stimulus. The Wil-
and lever for recording leg flexion on left foreleg. This coxon matched-pairs signed-rank test (Siegel,
was the procedure employed in Experiment II. In 1956) yields a T of 2 (p < .01, two tails).
Experiment I the electrodes were attached to the right
The results indicate that nonreinforced
pre-exposure to a stimulus results in a rate
of conditioning which is slower than to a novel
stimuli for the experiment, the bulb came on for \ sec.
each second; the rotor made one revolution each second. stimulus. It should also be noted that there
Both visual stimuli had auditory components. were two factors present that would tend to
Two electrodes were attached to the right foreleg of countereffect and minimize these differences
the 6\ The US was 11-v. a.c. Leg flexion was recorded by in rate of conditioning. Sensory generalization
means of a lever tied to the lower part of the right fore-
leg. The lever activated an overhead microswitch. When
may have occurred, since both stimuli were
5 lifted its leg, the switch closed. The closed micro- in the same modality. But, even more im-
switch moved a signal magnet on a polygraph in the portant, the signals were alternated, and 5
next room. During the experiment each animal was was required to make the same response to
alone in the room. Type of signal, intertrial intervals,
and shock presentations were programed in advance both signals. This situation maximizes positive
and presented automatically. All recording and observa- transfer. That the significant differences in
tion through a "one-way" glass was done from an ad- rate of learning did occur in spite of these
jacent room.

Procedure TABLE 1
Procedural Outline and Number of Trials to
The animals were brought into the testing room Conditioning Criterion, Experiment 1
individually and placed in the stock. The electrodes and
lever for recording leg flexions were attached to the Number of Trials
Subject to Criterion
right foreleg. The S was then allowed to stand for 15
Pre-exposed
min. before the first phase of the experiment was begun. Nonrein- Order of
Phase 1, -nonreinforced stimulus presentation. Each forced Alternation Pre-
Stimulus ex- Nov-
animal received 10 nonreinforced presentations of one Species Sex posed el S Diff.
of the signals which would later be used in the second or S
reinforced phase. Half the animals received the flashing
light, and half received the moving rotor. In each case Goat F (L)ight LRL--.R 57 25 32
the signal duration was 10 sec. The length of the interval Goat F (R)otor RLR • • • L 12 13 -1
between signals varied randomly from 30 sec. to 1\ min. Goat M L RLR. . - L 16 14 2
Phase 2, reinforced stimulus presentation. This phase, Goat M R LRL- - - R 24 14 10
in which all signals were followed by shock, began 2 min. Sheep F L RLR.--L 25 22 Q

after the presentation of the last nonreinforced signal Sheep F R LRL. . . R 28 24 4

of Phase 1. Half the animals were presented with the old Sheep M TJ LRL---R 31 29 ! 2
signal and half were presented with the new signal for Sheep j M R RLR — L 13 |14 -1
the first conditioning trial. On the second conditioning
trial each animal received that signal which was not Mean 25.819.4; 6.4
presented on the first trial. Thereafter, both signals con-
 LATENT INHIBITION 417

unfavorable conditions indicates that the TABLE 2

phenomenon of pre-extinction is indeed very Procedural Outline and Number of Trials lo
potent. Conditioning Criterion, Experiment 2

Subject Number of Trials

EXPERIMENT II to Criterion
Pre-exposed
The inhibitory effects of nonreinforced pre- Nonrein- of
forced Alterna Lion Pre-
exposure to a stimulus can, it is believed, be Spe- Stimulus ex- Nov-
cies Se\ posed el S Ditf
accounted for in terms of an S-R contiguity S
paradigm. Since only one relevant stimulus
was reliably pre-exposed, negative transfer Goat F (L)ight LRL- •R 85 45 40
•on the basis of a competing S-S association is Goat F (R)otor RLR- • • L |2S 20 5
Goat M L RLR. • • L |26 27 -1
difficult to assume. It is more plausible to infer Goat M R LRL- •R |16 18 -2
that some competing response has become Sheep F L RLR- ••L 22 17 5
associated with the nonreinforced signal. An Sheep F R LRL- •R |30 32 -2
analysis of the responses in the pre-exposed, Sheep M L LRL- •R 37 12 15
Sheep M R RLR • ••L 28 23 .1
nonreinforced phase and the reinforcement
phase is suggestive. The characteristic re- Mean I 33. 6 24.3 8.1
sponse to the pre-exposed stimulus is a re-
flexive turning of the head in the direction of
the stimulus (cf. Pavlov's investigatory reflex, This, then, would lead to an increase in flexion
1927). In the experimental situation reported tonus in that limb for which flexion will later,
in this paper the stimuli were located directly during the shock-reinforced phase, become
to the right of S, and right head turns were, in the "correct" response.
fact, frequently observed. Sherrington's (1906)
analysis of the various reflex figures indicates Method
that when the head of an animal is turned, The subjects were again four sheep and four goats.
there is an increase in extensor tonus in the These animals were not the same as those used in Ex-
limb on the side toward which the head is periment I. They received the same procedure as those
in the previous experiment except that the US of Phase
pointed. There is thus some inferential evidence 1 was delivered to the left leg of the 5 (and of course
for the pairing of the pre-exposed nonrein- remained on the left during the reinforced phase), while
forced stimulus with a reliable response. CS was again presented to the right of the S.
Furthermore, this extensor response is clearly
in competition with the response that is to be Results
learned during the reinforcement phase, The number of trials to reach the learning
flexion- of the right foreleg. criterion to the adapted and novel stimulus
The hypothesis that conditioning to the is presented in Table 2. It is quite obvious
nonreinforced stimulus was inhibited as a that the direction of the differences is opposite
result of the association of an incompatible that predicted from the hypothesis, and similar
extension of the right foreleg with the unre- to Experiment I, although the differences
inforced stimulus during the adaptation period are not significant (.10 > p > .05, Wilcoxon
has a readily deducible test. If, instead of matched-pairs signed-rank test).
having both the nonreinforced stimulus of
Phase 1 and the unconditioned stimulus of DISCUSSION AND CONCLUSIONS
Phase 2 delivered to the same side of the These results, then, do not enable one to
animal, as was done in Experiment I, these identify a competing mediating response in
stimuli are presented on opposite sides from terms of the previously described Sherring-
each other, it would be predicted that facilita- tonian analysis. That type of analysis would
tion of subsequent learning, rather than lead to the prediction of a slower rate of learn-
inhibition, would occur. In placing the CS ing for the pre-exposed CS when the CS and
and US on opposite sides, the orienting re- the US are on the same side of animal. This
sponse results in the S's turning away from was verified in Experiment I. However, in
the side on which the US will later be presented. Experiment II an equally valid deduction of
418 R. E. LUBCHV AXD A. U. MOORE

facilitated learning when the pre-exposed CS inhibition." It would seem that current learn-
and the US arc on opposite sides was not veri- ing theories might have some difficulty in-
fied. corporating these results. That "latent in-
That the position of the pre-exposed CS in hibition" may not be limited to a strictly
relation to the US has no effect is further classical conditioning situation is indicated
indicated by comparing the number of trials by two other studies which employ avoidance
to criterion for the pre-exposed stimulus conditioning (Bahrick, 1953; Lyons, 1954).
when the shock is given to the right leg These results have some implication for the
(Experiment I) as compared with the left leg design and interpretation of experiments
(Experiment II). The mean values are 25.8 employing either a sensory preconditioning or
and 33.6, respectively. This is in the opposite latent-learning paradigm. Both types of ex-
direction predicted from the hypothesis. periments use some form of nonreinforced
(Combining the number of trials to criterion for pre-exposure to a stimulus which will later
pre-exposed and novel CS, and comparing the become paired with a relevant response and
right leg with the left leg, yields no significant reinforcement. These studies emphasize facili-
difference: p > .20, Mann-Whitney U test, tation as the hallmark for the occurrence of
two tails.) learning. However, it should now be apparent
The effective conditions for inhibition of that the latent learning may result in a
learning can be refined further by combining decrement in the association of the relevant
the results of both experiments. A comparison response and stimulus. A study of the variables
of each stimulus type, rotor or light, inde- which determine whether nonreinforced pre-
pendently, in terms of whether it was a pre- exposure to a stimulus results in inhibition or
exposed or novel stimulus indicates an inter- facilitation would be enlightening. Muenzinger
action of type of stimulus and pre-exposure. and Conrad (1953) have already provided one
The mean number of trials to the conditioning clue, the amount of pre-exposure.
criteria for the pre-exposed light is 37.38,
and to the novel light, 19.75. Using the Mann- SUMMARY
Whitney U test, this difference is significant By using a classical conditioning paradigm
(p < .02). However, the mean number of with shock to the foreleg as the US, leg flexion
trials for conditioning to the adapted rotor is as the UR, and either a flashing light or a
22.00, to the novel rotor, 23.88. This difference turning rotor as the CS, two experiments were
is not significant. Pre-exposure to the flashing performed to evaluate the effects of non-
light significantly inhibits later learning, while reinforced pre-exposure of the CS on sub-
pre-exposure to the rotor has no effect. It is sequent conditioning. The first experiment,
the large differences between the adapted and in which the CS and the US were presented to
the novel light stimuli which determine the the same side of the animal, resulted in a
significance levels. partial inhibition of learning the association
The inability to identify a mediating re- between the pre-exposed CS and the US as
sponse mechanism in this quasi sensory- compared with a novel CS and the US. It
preconditioning, latent-learning experiment was suggested that this inhibition was produced
docs not, of course, nullify an S-R interpreta- by the pairing of the pre-exposed CS with a
tion (nor any interpretation) of the results, response that was incompatible with the
but rather eliminates a specific S-R hypothesis.
response which was later to be learned. An
It might be more fruitful to search for an
attempt was made to identify the competing
autonomic mediating response, particularly in
a situation such as this one which employed response in terms of an increase in extensor
shock as reinforcement. However, in the to nus in that leg for which flexion subse-
absence of data it would be ill-advised to quently becomes the correct response. On this
elaborate possible mechanisms. basis it was predicted that placing the CS and
Theory aside, one should not lose sight of US on opposite sides of the animal would
the fact that we are left with a phenomenon result in facilitation of learning to the pre-
that can be descriptively labeled "latent exposed CS when it is later paired with the
 LATENT INHIBITION 419

US. However, again, there was an inhibition LYONS, J. D. Sensory pre-conditioning and stimulus
of the CR to the pre-exposed CS. habituation. Library of Congress No. MICA
54-576.
The implications of latent inhibition for MUENZINGER, K. F., & CONRAD, D. G. Latent Learning
experiments employing latent-learning and observed through negative transfer, /. comp.
sensory-preconditioning designs were discussed. physiol, Psychol., 1953, 46, 1-8.
OSGOOD, C. E. Method and theory in experimental psy-
REFERENCES chology. New York: Oxford Univer, Press, 1953.
BAHRICK, H. P. Sensor)' pre-conditioning under two PAVLOV, I. P. Conditioned reflexes. London: Oxford
degrees of deprivation. J. com p. physiol, PsychoL, Univer. Press, 1927.
1953, 46, 39-42. SIIERRINGTON, C. S. The iiilegra/ive action of the nervous
BLODGETI, H. C. The effect of the introduction of system. London: Constable, 1906.
reward upon the maze performance of rats. Univer. SIEGEL, S. Nonparumetric statistics for the behavioral
Calif. Pi/bl. Psychol,, 1929, 4, 113-134. sciences. New York: McGraw-Hill, 1956.
BROGDEK, W. J. Sensory pre-conditioning. J. cxp. TOLMAN, E. C. Purposive behavior in animals and men.
PsychoL, 1939, 25, 323-332. New York: Appleton-Century, 1932.
HULL, C. L. Principles of behavior. New York: Applcton TOLMAN, E. C., & HONZIK, H. C. Introduction and re-
Century-Crofts, 1943. moval of reward, and maze performance of rats.
LTJMSDAINE, A. A. Conditioned eye-lid responses as Unher. Calif. Pull. Psychol., 1930, 4, 257-275.
mediating generalized finger reactions. Psychol.
Bull., 1939, 36, 650. (Abstract) Received Septeml>er[ 4} 1958.