tained similar results. That is, when References and Notes animals in these two groups.

In five of
ball-pulling produced programed conse- 1. B. F. Skinner, Amer. Psychologist 8, 69 the six pairs, the observer cat (solid
quences, rocking movements did not (1953). circles) learned the conditioned avoid-
2. C. B. Ferster, Psychol. Bull. 5, 263 (1953).
occur. Second, chlorpromazine blocked 3. J. H. Hollis, Working Paper No. 168 (Par- ance response much more rapidly than
the stereotyped rocking movements dur- sons Research Center, Parsons, Kansas, 1967). the student cat (operant, open circles).
4. , Percept. Mot. Skills 24, 465 (1967).
ing the periods when the alternative 5. , ibid., p. 156. One observer cat did not acquire the
activity produced no programed conse- 6. F. L. Fitz-Gerald, unpublished thesis, McGill response faster than his group 2 con-
quences. University (1964). trol, although his initial performance
7. J. J. Boren, in Operant Behavior, W. K.
Finally, the data clearly show that re- Honig, Ed. (Appleton-Century-Crofts, New was better. This cat became ill after 3
inforcement had stimulus properties. York, 1966). days of training and his schedule was
8. 0. R. Lindsley, in Psychosomatic Medicine:
That is, when a free reinforcer was de- First Hahnemann Symposium, J. H. Nodine interrupted for several days. It is par-
livered during extinction of the ball- and J. H., Moyer, Eds. (Lea & Febiger, ticularly striking that two cats in group
New York, 1962).
pulling response, immediate resumption 9. C. B. Ferster and B. F. Skinner, Schedules 1 immediately performed at high levels,
of ball-pulling occurred (10). This effect of Reinforcement (Appleton-Century-Crofts, requiring a total of only one and two
New York, 1957).
was not changed as a function of the 10. J. E. Spradlin, F. L. Girardeau, G. L. Hom, shocks each to reach criterion. A two-
drugs used. J. Exp. Child Psychol. 4, 369 (1966). tailed t-test comparing the total number
JOHN H. HOLLIS 11. Supported by grant No. 00870-04 from the of failures to perform on the part of the
National Institute of Child Health and Hu-
Parsons Research Center, man Development. student and observer cats throughout
Parsons, Kansas 67357 5 February 1968 the 6-day period was significant (P
< 0.001).
In our second experiment, 22 young
adult cats were used. Group 1 consisted
Observation Learning in Cats of six naive observer cats. Two trained
cats served as "lever press teachers."
Abstract. In two experiments cats acquired a stimulus-controlled approach or In order to establish social compatibil-
avoidance response by observational or conventional shaping procedures. Observer ity in the training cage, each naive cat
cats acquired the avoidance response (hurdle jumping in response to a buzzer shared a home cage with its teacher
stimulus) significantly faster and made fewer errors than cats that were conven- throughout the training period. An ob-
tionally trained. Observer cats acquired the approach response (lever pressing for server cat, together with its teacher,
food in response to a light stimulus) with significantly fewer errors than cats that both deprived of food for 24 hours,
were conventionally trained. In some cases, observer cats committed one or no were placed in a standard, operant con-
errors while reaching criterion. ditioning cage. No barrier separated the
two cats, and neither was restrained.
Electrophysiological studies in our Each naive observer cat from group During three observation sessions, the
laboratories have made it necessary to 1 was first placed in the training cage teacher cat performed approximately 30
train many cats to perform conditioned alone. A buzzer was presented for 15 approach responses daily, pressing a
approach and avoidance responses. Ex- seconds, and one foot shock was ad- lever within 15 seconds after onset of
perimental schedules have sometimes ministered unless a hurdle jump had a 5-cycle/sec flickering light. The lever
been disrupted because of the slow ac- been performed. After one such "em- was mounted in one wall of the cage,
quisition of such responses or by the pathy" trial, the observer cat was placed 7 inches above the floor and 3 inches
failure of some cats to reach criterion in a small cage with a mesh front, di- to the left of a dipper, which delivered
after extensive training. In the course rectly behind the training cage. Al- food. The exact number of trials pre-
of seeking more effective methods of though no attempt was made to coerce sented to the teacher cat was limited
rapidly and reliably training our ani- -observation, the observer cat was in a by the observer cat's attentiveness or
mals to make discriminative responses, position to watch a matched naive stu- apparent readiness to perform. The
we designed two experiments to inves- dent cat from group 2 receive 20 con- number of teacher responses apparently
tigate the acquisition of stimulus-con- ventional training trials of the condi- observed was recorded, and an attempt
trolled approach and avoidance re- tioned avoidance response daily, fol- was made to achieve daily observation
sponses via an observational procedure. lowed by performance of the same of 30 trials. After the observation ses-
Each experiment required the observing number of trials by one of the fully sion, the teacher was removed from the
animal to acquire and perform a re- trained teacher cats. This procedure was training cage and approximately 30 test
sponse not previously in his repertoire, repeated daily until the student cat trials were presented to each naive ob-
without overtly performing it during the reached a criterion of 90 percent per- server cat. The number of test trials
observation period. formance for 3 days in a row. The presented to observer cats during the
In the first experiment, 14 young observer cat was then subjected to 20 first 3 days varied considerably, but
adult cats were used. Six naive "ob- training trials daily until he reached they all received 30 test trials for the
server" cats composed group 1, six the same criterion. In three cases, ob- last 3 post-observation days.
naive "student" cats composed group 2, server cats began avoidance training To insure that a stimulus-controlled
and two fully trained cats served as before their matched controls reached bar press is not somehow facilitated by
"teachers." The apparatus consisted of criterion, because they had performed the presence of two cats in the training
a standard operant conditioning cage. an avoidance response on their daily cage or by any effects of exploration
A lucite hurdle, 6 inches (15 cm) high, empathy trial. or familiarization per se, a second
bisected the 24- by 24-inch shock-grid Figure 1 compares the acquisition of group of six cats (group 2) similarly
floor of the cage. the conditioned avoidance response by, observed a cage-mate receive food by
29 MARCH 1968 1489
 O 100-
Z 1007 100-
o >80- 80-
eoo- V914
.*-.*
0
do
60- 60- 60-
o 20-
0w
/N- 40-
_°/
* o A
'40-
Z. 40- 00
.s O' 0ooo 20-
0~
20
. a I A 5 a I. I a I I I5 I v- ;5
n
lw 5 I I I ].
a 4 5a 5a 5 . I 5 U ' 0 a A A -
A a A 0 A a a a a

2 4 6 8 10 12 14 16 2 4 6 8 10 12 14 16 2 4 6 8 10 1214 16

*-. OBSERVER °-@° OPERANT

100- es.* 0OO-
C)
lo0, O
I,80 80- oo 80-
w 0AI
tr 60 60- 60-
0 40- 40-
C) C 40 i / ' /\/ /
< 20 o/ _ 20- / 0 20-
_ _ _,-o
> 0 a
1
-T II 0 I12 14 1 _ -L 00-. . . . -. . . . . . . . . 0- a I 5I I I l
4c 24 6 811012 1416 2 4 6 8 10 12 14 16 2 4 6 8 10 12 14 16
SESSIONS
Fig. 1. Observation learning of an avoidance response (hurdle jump) to an auditory stimulus.
merely moving to the food dipper with- a third group of six cats (group 3) was Figure 2 shows that the observer cats
in 15 seconds after stimulus onset. The operantly "shaped" to press a lever for in group 1 (solid circles) acquired a
naive observers in this group watched food in response to the flickering light. discriminated lever-pressing response
30 teacher responses each day for 3 They received 30 trials daily for a pe- quickly, with four of the six cats reach-
days, and received 30 test trials daily riod of 6 days. For this group, any ef- ing 90 percent correct performance by
for a total 6-day period. During these fective depression of the lever, whether the third day. One of these cats per-
test trials, the observers were rewarded directly by paw, by elbow, or by oral formed at the 100 percent level on the
for lever pressing as well as for dipper manipulation, was considered a correct first day, after observing only three
approach. response. It should be noted that this teacher trials. A second cat reached
Finally, in order to compare the is a broader definition of an acceptable 100 percent performance on the second
standard operant acquisition of a stim- response than was used with the ob- day, after observing only 16 teacher
ulus-controlled lever press with that server cats, who were required to press trials. The mean number of trials ob-
acquired by an observational procedure, the lever with their paws. served by group 1 cats throughout the

LLI
(a
100 - 100-
m
°80~ 80-
a 60- 60- X~~~~_~__-_
x JII
" 40- 40-
= 20- 20-
0O
O. 0-
1 2i3 4 56
0- x--X--X--ax-x-x
1 2 3 4 5 6

*- OBSERVER I x---x DIPPER CONTROL E1 °- o° OPERANT CONTROL m

100 100-
z 'retr
S100~~~~ I
° 80- , . *80- 80-
cn 0'~

cr 60 60- 60-
'-
C)O~
40-
ixI 20-
t40-
0
40-
20-
I
.,6- I
1 . m
40-
20-
010-*

'0-
0
1 2 3 4 5 6 1 23 45 6
0- --
ml =J
6-5
SESS IONS
Fig. 2. Observation learning of an approach response (lever press) to a visual stimulus.
1490 SCIENCE, VOL. 159
 experiment was 32. These results are observational learning which have led 2,5-Dimethoxy-4-Methyl-
apparently attributable to observation of to a variety of experimental designs Amphetamine: New
a lever-pressing response, rather than using different species.
to any facilitation offered by the pres- Undoubtedly, conditioning techniques Hallucinogenic Drug
ence of two cats in the cage, since cats have been of great utility in the quan- The report (1) on the effects of a
in group 2 (crosses) failed to press the titative study of learning. Yet the im- hallucinogen, 2,5-dimethoxy-4-methyl-
lever even once during the entire ex- pressive speed and efficiency of obser- amphetamine (STP, or DOM), due to a
periment. The group 2 situation actu- vational learning, contrasted with the number of methodological inadequacies
ally seems to inhibit lever-pressing, but potentially catastrophic slowness and can be accepted only as an interesting
as a function of observation learning. need for repetition which often charac- collection of observations and not as a
Conventionally shaped animals (group terize conventional conditioning, suggest definitive assessment.
3) acquired the lever-pressing response that the latter may well be a phenome- That suggestion or expectation in a
more slowly than the observers, with non of limited relevance, utilizing rela- subject affects the outcome of an exper-
two animals failing to make a single tively unnatural mechanisms. Observa- iment is common knowledge, particu-
lever press throughout the 6-day train- tional learning may be the primary larly where hallucinations are induced
ing period (open circles). A two-tailed method of acquiring language, ideas, (for example, in sensory deprivation
t-test comparing group 1 and group 3 and social habits in man, and such experiments). Snyder, Faillace, and
for total number of failures to perform learning may also play an important Hollister state "subjects were told that
during the 6-day period was significant role in the adaptation and survival of they would receive a drug ... presum-
(P < 0.01). Moreover, the level of lower organisms (7). Thus, the be- ably a hallucinogen ...." These in-
stimulus discrimination was significantly havioral and physiological study of structions are somewhat more than a
higher in the observer group, as meas- more natural learning situations may suggestion that hallucinations were to be
ured by the number of spontaneous be essential for adequate understand- expected. Hallucinogens have achieved
lever presses performed by criterion ing of learning mechanisms. great notoriety, and their psychological
animals in both groups. Specifically, on E. RoY JOHN, PHYLLIS CHESLER effects are widely known. It is probable
the sixth day, the operant group had a FRANK BARTLETT, IRA VICTOR that suggestion influenced the results
mean number of 89 intertrial lever Brain Research Laboratories, ascribed to DOM.
presses, as compared to an observer Department of Psychiatry, New York The authors quote one subject who
mean of 28 intertrial presses. Medical College, New York 10029 compared the supposed effects of DOM
It should be stressed that several of References and Notes to a "halfway decent pot experience."
our observer cats, in both the approach 1. R. C. Gonzalez and B. Shepp, Amer. J. Even minimal prior drug experience
and avoidance situations, performed Psychol. 78, 441 (1965); E. R. Strain, J.
Exp. Psychol. 46, 391 (1953); D. Thistle- might have been an additional uncon-
correctly at the first opportunity and thwaite, Psychol. Bull. 48, 97-129 (1951). trolled factor. The report points out
committed few or no errors while 2. H. A. Adler, J. Genetic Psychol. 86, 159
(1955); A. L. Bandura, in Nebraska that sensitization to the effects of DOM
reaching criterion. Some of these ani- Symposium on Motivation, M. R. Jones, may be a result of previous drug experi-
mals behaved as though they "knew" Ed. (Univ. of Nebraska Press, Lincoln,
1962), pp. 211-274; R. M. Church, J. Comp. ences. Without an adequate control the
what they were doing. Thus, learning Physiol. Psychol. 50, 315 (1957); J. A. Cor- effects of suggestion and of past drug
son, Psychon. Sci. 7, 197 (1967); C. L.
mechanisms seem to exist which are Darby and A. J. Riopelle, J. Comp. Physiol. use in producing DOM hallucinations
capable of integrating diverse perceived Psychol. 52, 94 (1959); B. V. Dawson and cannot be determined.
B. M. Foss, Animal Behav. 13, 470 (1965);
stimuli into a meaningful whole, with- V. V. Gerasimov, Dokl. Akad. Nauk SSSR The method of obtaining reports of
out direct reinforcement and without 146, 1456 (1962); K. R. C. Hall, Brit. J.
Psychol. 54 (3), 201-226 (1963); M. J. symptoms is somewhat suspect. Self-
overt performance of the response. Herbert and C. M. Harsh, J. Comp. Psychol. reports and questionnaires in general
Many physiological theories of learn- 37, 81 (1944); P. Klopfer, Amer. Naturalist
91, 61 (1957); N. E. Miller and J. Dollard, are of dubious validity and reliability,
ing assume that learning is a gradual Social Learning and Imitation (Yale Univ. and answers may be easily faked. The
Press, New Haven, 1941); K. W. Spence,
phenomenon, requiring repeated re- Psychol. Bull. 34, 806-850 (1937); V. E. physiological signs reported are appar-
inforced performance of a response, Stimbert. R. W. Schaeffer, D. L. Grimsley,
Psychon. Scd. 5, 339 (1966); C. J. Warden ently objective enough, but cannot be
and consisting of the establishment of and T. A. Jackson, J. Genet. Phychol. 46, presumed to be due solely to the effect
neural pathways that connect brain re- 103-125 (1935); J. C. Welty, Physiol. Zool.
7, 85-128 (1934); R. M. Yerkes, J. Soc. of the drug and not to fatigue, stress, or
gions receptive to the sensory stimulus Psychol. 5, 271 (1934). the attention given to the subjects by
to areas which mediate the behavioral 3. 163 R. M. Church, J. Abnormal Soc. Psychol. 54,
(1957); W. H. Thorpe, Ibis 93, 1-52, the experimenters. The authors did not
response. Such learning theories have 252-296 (1951). indicate whether reported deviations
been derived from the facts of classical 4.5. P. Klopfer, Behavior 14, 282 (1959).
R. Gilbert and M. Beaton, Psychon. Sci. 8, differed significantly from baseline
or instrumental conditioning. Yet, nu- 1 (1967); P. Klopfer, Science 128, 903
(1958); 0. J. Sexton and J. Finch, Psychon. measurements of healthy, normal per-
merous experiments in latent learning Scd. 7, 5 (1967); E. L. Thomdike, Psychol. sons. Quantitative estimates of the rela-
(1) and observational or vicarious Rev. Monogr. 2, No. 4 (whole No. 8), 1-109
(1898). tive potencies of DOM, lysergic acid
process learning (2-5) demonstrate 6. W. MacDougall, An Introduction to Social diethylamide (LSD), and mescaline are
that learning can take place without Psychology (Methuen, London, 1908); G.
Humphrey, Pedagog. Sem. 28, 1 (1921); B. therefore unwarranted.
reinforcement and with little or no per- F. Skinner, Verbal Behavior (Appleton, New
York, 1957); 0. H. Mowrer, Learning PATRICK A. CABE
formance of the response which is re- Theory and the Symbolic Processes (Wiley, Human Factors Laboratory,
quired. Perhaps because of its theoret- 7. New York, 1960).
A. L. Bandura, in Advances In Experimental Goodyear Aerospace Corporation,
ical significance, controversy has existed Social Psychology, S. Berkowitz, Ed. (Aca- Akron, Ohio 44315
regarding the mediating mechanisms 8. demic Press, New York, 1965), vol. 2.
Supported under NIMH grant MH 08579.
(3, 4, 6, 7) and reproducibility (4, 5) We thank Dr. J. A. Corson for making avail- Reference
able information about his comparable ex- 1. S. H. Snyder, L. Faillace, L. Hollister, Science
of observational learning. In part, this periments with rats prior to their publication. 158, 669 (1967).
may be due to different definitions of 7 February 1968 22 January 1968
29 MARCH 1968
1491