You are on page 1of 10

Journal of Comparative and Physiological Psychology

1973, Vol. 82, No. 3, 434-443

SIMULTANEOUS AND BACKWARD 1FEAR CONDITIONING


IN THE RAT
C. DONALD HETH2 AND ROBERT A. RESCORLA
Yale University
Simultaneous and backward Pavlovian conditioning paradigms were ex-
amined using a US event which was longer in duration than the CS. Three
experiments paired a 4-sec. electric shock with a 2-sec. tone-light stimulus
under conditions in which the onset of the stimulus occurred .0, .25, 1.0, 2.0,
4.0, or 4.5 sec. after the onset of the shock. Relative to nonpaired control
procedures, response-contingent presentations of the CSs in these paradigms
significantly suppressed a food-rewarded free operant, indicating that these
temporal relationships can produce excitatory associative conditioning. It
was suggested that the distinctions between "forward," "simultaneous,"
and "backward" procedures be modified to include a more molecular analysis
of the US event.

Historically, research in classical condi- relates the informational value of the CS to


tioning has shown that conditioning proce- its temporal priority within some stimulus
dures are effective only when the presenta- compound (Egger & Miller, 1962, 1963).
tion of the CS with a US occurs within cer- Although not limited to such an interpreta-
tain temporal constraints. A long line of tion, variations in the temporal relation of
studies from American laboratories has the onset of the CS to the US can be consid-
usually demonstrated no excitatory associa- ered a paradigmatic case of informational
tive conditioning when the CS occurs simul- manipulations through temporal priority.
taneously with or after the US (cf. Kimble, When the onset of the CS is simultaneous
1961). Thus, although investigators fre- with or follows the onset of the US, then it
quently describe the presentation of the CS no longer serves as a unique predictor of the
with the US simply as a "pairing," there is US. The CS in these Pavlovian procedures
general recognition that, in addition, some is made redundant by the simultaneous or
portion of the CS must occur prior to the prior occurrence of US onset. The failure to
US onset. This realization characterizes find excitatory associative conditioning
many of the conceptual schemes which the- using these paradigms is therefore consist-
orists have adopted to describe the Pav- ent with the informational position that a
lovian learning situation. redundant stimulus acquires little or no as-
For example, many theorists conceive of sociative strength when paired with the US
conditioning as establishing informational (Egger & Miller, 1962).
relationships between the CS and US. Res- However, it should be noted that investi-
corla (1972) has described several exper- gations of simultaneous or backward para-
imental operations as manipulations of the digms have typically employed a rather
informational value of the CS. One of these narrow range of conditioning procedures.
For instance, in most experiments US
1
This research was supported by National Sci- events are short while CSs are of equal or
ence Foundation Grants GB-6493 and GB-12897. greater duration (e.g., Bitterman, 1964;
The authors would like to thank Barbara Steinfeld Kamin, 1963). A study by Mowrer and
for her help in tabulating the data and preparing
the manuscript. Portions of the data from Experi- Aiken (1954) and its replication by Mat-
ments 1 and 2 were previously reported by Rescorla sumiya (1960) suggest that the failure to
(1972). C. Donald Heth is a National Science find simultaneous or backward conditioning
Foundation Predoctoral Fellow. may not generalize to a broader range of pro-
2
Requests for reprints should be addressed to cedures. The Mowrer and Aiken study, al-
C. Donald Heth, Department of Psychology, 333
Cedar Street, Yale University, New Haven, Con- though not addressed to the issue under dis-
necticut, 06510. cussion here, used a 10-sec. shock as the US
434
FEAR CONDITIONING IN THE RAT 435

and a 3-sec. light as the CS. Their data Method


show that initiating the CS 3 sec. before,
Subjects
simultaneously with, or 7 sec. after the
onset of the US were each sufficient to es- The subjects were 32 male Sprague-Dawley rats
tablish it as a conditioned punisher. Unfor- periment. They95
approximately days old at the start of the ex-
were maintained during the course
tunately, they used as a control condition a of the experiment at 80% of their free-feeding
procedure in which the CS and US were weights.
explicitly unpaired. This procedure has
since been found to endow the CS with con- Apparatus
ditioned inhibitory properties under some The subjects were run in eight experimental
circumstances (e.g., Rescorla & LoLordo, chambers, each housed within a box constructed
1965). This questions the claim that condi- of sound-attenuating material and equipped with
a ventilation fan. During bar-press training and
tioned excitation has been demonstrated in subsequent test sessions, the experimental chambers
the Mowrer and Aiken and the Matsumiya were 22.9 X 20.3 X 20.3 cm. Skinner boxes. The
studies. Nevertheless, the possibility of si- front and rear walls were made of aluminum; the
multaneous conditioning under conditions in side walls and top were Plexiglas. Th'3 floors con-
sisted of stainless-steel rods, 1.9 mm. in diameter
which the CS is temporally embedded within and spaced 19.0 mm. apart. The front wall of each
the US raises serious implications for even contained a recessed food trough located in the
the more traditional conceptions of Pavlo- center, a stainless-steel lever to the left of it, and
vian conditioning. This possibility was ex- a jeweled lamp mounted above the :'ood trough.
amined in the present investigation, using Pavlovian conditioning training was administered
in chambers identical to the Skinner boxes in di-
conditions similar to those of Mowrer and mensions and design with the exception that the
Aiken. Experiments 1 and 2 consider the front wall did not contain a lever, lamp, or food
question of simultaneous conditioning; Ex- trough.
periment 3 extends this analysis to the case Shocks were administered to the floor rods by
means of a high-voltage-high-resistance shock
of backward paradigms. source delivered through a relay-sequence scram-
bler of the type described by Hoffman and Fleshier
EXPERIMENT 1 (1962). The conditioning stimulus (C3) consisted
The conditioned punishment procedure of the onset of a 6'/i-w. light and an 1,800-Hz. tone
delivered simultaneously through a bulb and a
used by Mowrer and Aiken (1954) was loudspeaker mounted on the inside wall of the
adopted in the present investigation, since it sound-attenuation chambers. Programming and
allows the use of relatively short CS events recording equipment were located in an adjacent
and longer US events. The procedure, some- room.
what modified, consisted of three main Procedure
stages: (a) a pretraining stage, in which an
operant bar-pressing response was estab- Magazine and bar-press training. On Day 1 of
the experiment, all subjects were placed in the
lished and maintained by food reward; (b) Skinner boxes and given .045-gm. Noy>;s pellets on
Pavlovian conditioning in the form of pair- a random delivery schedule which averaged one
ing a tone-light compound (CS) with an pellet per minute. In addition, each bir press was
electric shock (US); (c) a testing stage, in rewarded with a pellet. The subjects remained in
which the Pavlovian-trained CS was used the Skinner boxes until they had made 50 responses
and were hand shaped if necessary. On Days 2-5,
as a conditioned punisher of the previously all subjects were placed in the Skinner boxes for
trained operant response. Presumably, the 1 hr. and allowed to bar-press for food on a 1-min.
relative efficacy of the CS in suppressing the variable-interval (VI-1) schedule.
response should provide a measure of the Conditioning. On Days 6 and 7 the subjects were
divided into four groups of eight animals each and
strengths of the conditioning procedures. In given Pavlovian fear training in the oonditioning
Experiment 1, two simultaneous paradigms chambers. All groups received a 2-sec. tone-light
and one forward paradigm were studied. compound CS, and a 4-sec. i/2-ma. electric shock
These were assessed against a random US. For the three experimental groi.ps, the CS
schedule of CS and US presentations—a pro- was paired with the US in the following ways: For
Group F-2 (forward—2 sec.), the onset of the CS
cedure hypothesized by Rescorla (1967) to occurred 2 sec. before the onset of the US. For
yield no associative strength to the CS. Group E-0 (embedded—0 sec.), the onset of the
436 C. DONALD HETH AND ROBERT A. RESCORLA

CS occurred simultaneously with the onset of the rate during punishment. The punishment phase of
US. For Group E-l, the onset of the OS occurred 1 each session was divided into three periods of ten
sec. after the onset of the US. Each session con- minutes, and separate ratios were computed for
tained 10 such conditioning trials, with the inter- each period.
trial intervals determined by a rectangular distri-
bution of 4, 5, 6, 7, and 8 min.; the subjects were Results and Discussion
removed 6 min. after the last trial. A fourth group,
Group R (random), received the same USs deliv- The four groups did not differ in number
nr
ered at the same intertrial intervals; however, the of responses during the first 1/2 - of either
10 CSs of each session were delivered randomly test session, as revealed by a Kruskal-Wal-
throughout the hour, independently of US occur- lis test (Hs = 2.9 and 1.2, pa > .10). Conse-
rences. In Experiments 1 and 2 three subjects es-
caped from the conditioning chambers. One, a quently, the suppression observed during
member of Group R, escaped on both conditioning the second half of each session can be eval-
days. Consequently, this animal was dropped from uated against base-line response rates which
the experiment. The other two subjects escaped were comparable for all groups. The sup-
on only one of the two conditioning days; their pression ratios for the six 10-min. periods of
data were included in the analyses.
Recovery and Testing. Days 8 and 9 consisted the two sessions are plotted in Figure 1. As
of recovery days in which the subjects were re- Figure 1 shows, response-contingent presen-
turned to the Skinner boxes and received 1 hr. tations of the CS produced considerable sup-
of VI bar-press training. On Days 10 and 11 each pression of responding in the forward group,
animal was given 1-hr, test sessions to assess the
degree to which the CS could serve as a condi- Group F-2, across most of the testing peri-
tioned punisher. For the first 30 min. of each ods; the control group, Group R, showed
test session, the subjects were allowed to bar-press appreciable suppression only during the
for food on the VI-1 schedule as before. Beginning first 10-min. period. Of the two simultane-
with Minute 31, punishment was initiated. Each ous groups, Group E-0 showed more sup-
bar press produced a 2-sec. presentation of the CS.
A bar press during the CS had no effect upon its pression than Group E-l, with both falling
duration. The VI-1 schedule remained in force between Group F-2 and Group R.
during the punishment phase of the test session. These differences were evaluated by pool-
Performance during the test session was assessed ing individual suppression ratios over the
by means of suppression ratios (Annau & Kamin, three periods of the first test session. A
1961; Church, 1969) in the form of A/(A + B),
where B refers to the response rate prior to the Kruskal-Wallis test found group differences
initiation of punishment and A to the response to be significant (H = 18.42, p < .001).

.50 r-

£ -30
UJ
tr
a.
a.
w .20

.10 -

J_ _L
t 2 3 1 2 3
DAY 10 DAY 11
1 0 - M I N U T E PERIODS
FIG. 1. Mean suppression ratios for each 10-min. period of the two testing sessions of Experiment 1.
FEAR CONDITIONING IN THE EAT 437

Mann-Whitney U tests revealed that Group US events (Rescorla, 1972). The use of a
F-2 differed from Group R and Group E-l tone-light compound as a CS in the present
(Us = 2 and 6.5, ps < .01). Comparisons experiment and the substantial length of
involving the two simultaneous groups the US could have allowed such preasymp-
showed that Group E-0 differed from Group totic excitation. In that case Group R pro-
R (U = 11, p < .05), but Group E-l did vides a highly conservative criterion
not (U = 20.5, p > .10). No other compari- against which to evaluate excitatory condi-
sons were significant. tioning.
This pattern of results is supported by An alternative interpretation of the
multiple U tests for each 10-min. period. On embedded data should also be considered. It
all six periods, Group F-2 differed from could be that the onset of CS and US were
Group R (ps < .05) and Group E-l (ps < not effectively simultaneous for the orga-
.01), Group E-0 differed from Group R on nism. Different factors can be alluded to,
Periods 2-5 (ps < .05). In addition, Group both mechanical (e.g., contact with the grid
E-0 showed more suppression than Group bars not initially electrified due to the
E-l on Periods 4-6 (ps < .05) and less scrambling sequence) and physiological
suppression than Group F-2 on Periods 1 (e.g., different neural transmission rates of
and 4 (ps < .05). CS and US) which might have resulted in
The significant response suppression ob- processing of the CS prior to the US. This
served to Group E-0 demonstrates the ca- interpretation would explain not only the
pacity of a CS to acquire conditioned aver- demonstration of conditioning in Group E-0
sive properties when its onset occurs simul- but the failure to find reliable conditioning
taneously with that of shock. The data of to Group E-l.
Experiment 1 indicate that this procedure is These questions might be easily answered
somewhat less effective than one in which by examining other temporally embedded
the CS is presented prior to the US, al- relationships of CS and US. Experiment 2
though the comparison was not so robust studies three such paradigms in which the
statistically as might be expected. Further- onset of the CS occurs .25, 1.0, or 2.0 sec.
more, the possibility exists that paradigms after the onset of the US. In addition, since
in which the onset of the CS occurs subse- there are both theoretical and empirical
quent to the onset of the US might also grounds for believing that under the experi-
prove sufficient for conditioning. Indeed, the mental conditions of this investigation the
data of Experiment 1 do not conclusively use of a random control is somewhat con-
establish the paradigm of Group E-l as the servative, a different control procedure was
limiting condition of such embedded para- introduced. The subjects in this procedure
digms. received the same number of CSs and USs
In this respect, it can be noted that the as the experimental groups, but all the CSs
random control, Group R, exhibited consid- were presented on Day 1 of conditioning
erable suppression during the first 10 min. and all the USs on Day 2. This procedure
of the testing period. There is some recent obviates the coincidental CS-US pairings
empirical evidence which indicates that a which may have produced conditioning in
random schedule of CSs and USs may yield Group R while not permitting the develop-
excitatory associative strength to the CS ment of conditioned inhibition expected
(Kremer & Kamin, 1971; Quinsey, 1971). A from an explicitly unpaired procedure
model of Pavlovian conditioning developed (Rescorla, 1967).
by Rescorla and Wagner (1972) provides
a theoretical basis for understanding this EXPERIMENT 2
outcome. According to that model, certain Method
conditioning parameters could result in pre-
asymptotic associative values of the CS Subjects and Apparatus
which are excitatory. Two of these relevant The subjects were 32 male Sprague-Dawley rats
to the present discussion are: (a) highly of the same age, and maintained in the same man-
salient stimuli as CSs, and (6) very potent ner, as those of Experiment 1. The apparatus and
438 C. DONALD HETH AND ROBERT A. RESCORLA

CS and US parameters were also identical to those sion (Hs — 3.17 and 5.13, ps > .05). Mean
used in Experiment 1. suppression ratios for the six 10-min. peri-
Procedure ods of testing are presented in Figure 2. As
this figure indicates, Group C-II showed lit-
Pretraining was administered in the same man- tle suppression throughout the testing stage,
ner as Experiment 1. Pavlovian conditioning was
likewise given on Days 6 and 7, with Groups E-!4, while Group E-i/4 suppressed substantially.
B-l, and E-2 receiving the CS onset .25, 1.0, and Groups E-l and E-2 demonstrated interme-
2.0 sec. after the onset of the US, respectively. diate levels of suppression with the latter
These groups were given 10 pairings each day on a showing some tendency to be less sup-
schedule like that of Experiment 1. The control pressed than the former. Although be-
group, Group C-II (Control—Experiment 2), re-
ceived 20 presentations of the CS alone on Day 6 tween-experiment comparisons cannot be
and 20 presentations of the US alone on Day 7. taken as definitive, Group C-II shows less
Group C-II received 2-hr, sessions during this overall suppression than Group R of the
stage, to allow the presentation of CSs and USs previous experiment. As discussed above, it
according to the same schedule and within-session
density as the experimental groups. The 2-day would seem reasonable to attribute the dif-
testing procedure was the same as that of Experi- ference to some amount of excitatory asso-
ment 1, with the exception that the VI-1 food ciative conditioning to Group R.
reinforcement schedule was discontinued for the These trends are supported by statistical
first 4 min. of the punishment stage. This feature analyses conducted on the sum of the indi-
was introduced to eliminate from the early part
of the punishment phase some of the between- vidual suppression ratios across the three
subject variance due to different self-administered periods of Test Session 1. A Kruskal-Wallis
reward experiences. test showed group differences to be highly
reliable (H = 17.72, p < .01). Multiple U
Results and Discussion tests revealed that Groups E-1/^, E-l, and
Statistical tests verified that the groups E-2 were each significantly more suppressed
did not differ in the number of bar presses than the control, Group C-II (Us = 0, 2,
during the base-rate period of each test ses- and 7; ps < .01). Groups E-l and E-2 did

.50

2 .40

z
o
a> .30
tO
Ul
tr
a. P"
a.
w .20

UJ
2
.10

D*
_L _L J_
2 3 1 2
DAY 10 DAY 11

10-MINUTE PERIODS

FIG. 2. Mean suppression ratios for each 10-min. period of the two testing sessions of Experiment 2.
FEAR CONDITIONING IN THE RAT 439

not differ from each other (U = 25.5, p > with, and .50 sec. after, the termination of the
.10); Group E-2 differed significantly from US, respectively. The control group, Group C-III,
received 20 CS presentations on Day 6 and 20
Group E-y4 (U = 13,p< .05), but Group USs on Day 7 in the same manner as Group C-II
E-l did not (U = 17, p > .05). More de- of the previous experiment. The testing procedure
tailed statistical comparisons for each 10- was the same as that of Experiment 2.
min. punishment period corroborated these
conclusions. Results and Discussion
The statistical analyses clearly support Appreciable differences among groups
the conclusion that excitatory associative were apparent only on Day 1 of testing;
conditioning can be established in other therefore, only these data are presented
embedded paradigms, specifically those in here. The groups did not differ in number of
which the onset of the 2-sec. CS occurs .25, bar presses during the first half of the test-
1.0, and 2.0 sec. after the onset of the 4-sec. ing period (H - .31, p > .10). Mean sup-
US. The data indicate that these relation- pression ratios for the punishment phase of
ships can be consistently ordered according the testing sessions are given in Figure 3. It
to their relative effectiveness in producing can be seen that Group C-III showed little
excitatory strengths, with a .25-sec. delay suppression of responding throughout the
being most effective and a 2.0-sec. delay testing stage. Group E-2 showed substantial
being the least. suppression during the first 10-min. period,
with decreasing amounts during subsequent
EXPEKIMENT 3 periods. Groups B-0 and B-i/fc showed inter-
The amount of suppression observed to mediate levels of suppression during the
Group E-2 suggests that longer US-CS in- first period, but their mean ratios did not
tervals might also prove sufficient to estab- appear to differ from those of the control
lish the CS as a conditioned punisher. With group on subsequent periods.
CS and US durations held constant, the As in the previous experiments, individ-
backward conditioning paradigm may rep-
resent the limit to which this interval can .50 r-
be increased and still produce conditioning.
Experiment 3 was run to provide the addi-
tional data points which might be relevant
to such an interpretation. In addition to the
E-2 condition of Experiment 2, two back-
ward paradigms were examined in which ir
the onset of the CS followed termination of
the US by delays of 0.00 or .50 sec. o .30
C/5
Method UJ
tc
o.
Subjects and Apparatus
.20 -
The subjects were 32 Sprague-Dawley rats,
similar in age to those of the previous experiments
and maintained on the same deprivation schedule.
The apparatus and stimulus parameters were those
used in Experiments 1 and 2. .10

Procedure
Magazine and pretraining were administered as
before. Days 6 and 7 consisted of Pavlovian condi- 1 1
tioning in which Groups E-2, B-0 (Backward—0 I 2 3
sec.), and B-Vi were given 10 CS-US pairings each 10-MINUTE PERIODS
day. Group E-2 received the onset of the CS 2.0
sec. after the onset of the US; for Groups B-0 and FIG. 3. Mean suppression ratios for each 10-min.
B-'/2, the onset of the CS occurred simultaneously period of the testing session of Experiment 3.
440 C. DONALD HETH AND ROBERT A. RESCORLA

ual ratios were summed across all three pe- GENERAL DISCUSSION
riods for each subject. A Kruskal-Wallis Taken together, the three experiments of
test of this measure showed that the groups this investigation have involved succes-
differed significantly (H = 9.55; p < .05). sively increasing the delay between onset of
Comparisons between groups snowed that the US and onset of the CS. Strong excita-
Group E-2 was more suppressed than tory conditioning occurred when the CS was
Groups C-III, B-0, and B-!/2 (Ua = 8, 14, presented in a forward relation with the
11; ps < .05). Neither Groups B-0 nor B-i/2 US; in addition, significant conditioning
differed reliably from Group C-III, al- could also be demonstrated when CS onset
though the former comparison approached occurred after the onset of the US—proce-
statistical significance (Us = 17 and 22; ps dures which have been commonly labeled as
= .065 and .164, respectively). simultaneous or backward. The pattern
Although neither backward group differed which emerges within each experiment
from the control group when suppression shows decreasing effectiveness of condition-
ratios were collapsed across the three test- ing with increases in the delay between US
ing periods, Figure 3 suggests that appreci- onset and CS onset. This trend can be illus-
able differences might be present on the first trated with data from the study as a whole.
10-min. period of testing. This impression Figure 4 presents mean suppression ratios
was confirmed by multiple comparisons for the first 10 min. of testing plotted
within this period. Both Groups B-0 and against delay between US and CS onsets.
B-i/2 differed significantly from the control Although between-experiment comparisons
group (Us = 13 and 11, respectively; pa < are complicated by the variation in overall
.05). The two backward groups did not dif- suppression between experiments, regular
fer between themselves (p > .10). Other decreases in conditioning strength as a
comparisons during the first 10-min. period function of the US-CS delay are quite ap-
showed that Group E-2 was significantly parent.
more suppressed than Groups C-III, B-0, What is remarkable about the data de-
and B-i/2 (ps < .05). Group E-2 also dif- picted in Figure 4 is this regularity even
fered from the other three groups on the across delays which would normally distin-
second period (ps < .05), but no other com- guish forward, simultaneous, and backward
parisons were significant for this period and paradigms. It strongly suggests that varia-
none were significant for the third. tions in US-CS delay can be interpreted as
.50r

2 .40
•—• Experiment 1
*—A Experiment 2
•—• Experiment 3
.30

.20

.10

-2 0 1 Z 3 CONTROLS
US-CS ONSET INTERVAL (ue)

FIG. 4. Mean suppression ratio during the first 10-min. period of testing as a function of the temporal
delay between onset of the US and onset of the CS.
FEAR CONDITIONING IN THE RAT 441

continuous manipulations of some underly- dicting these events. Thus, the findings of
ing variable sufficient for conditioning. the present investigation seem to necessitate
Specification of this variable may require revision of the conditions sufficient to estab-
revision of many common conceptions con- lish a stimulus as an informational event.
cerning the events which comprise the Pav- The present findings are at variance with
lovian learning situation. Many investiga- several lines of research from American
tors, for example, often treat the Pavlovian laboratories which have found no evidence
reinforcer as a single unitary event with no of excitatory associative conditioning when
extension in time and consider US duration CS onset occurs with or after the onset of
solely as an intensity parameter. The re- the US. This evidence has generally been of
sults of the present investigation require a two kinds: (a) Most studies employing re-
more molecular analysis. sponse measures sensitive to excitatory as-
The type of analysis suggested by tradi- sociations have found no conditioning with
tional conceptions of Pavlovian condition- simultaneous or backward paradigms; (b)
ing might consider the CS in an embedded alternatively, some recent studies specifi-
paradigm as being paired with some subset cally addressed to the question of inhibition
of the interval following US onset. The in backward conditioning have been able to
function depicted in Figure 4 suggests that demonstrate conditioned inhibition as the
this interval can be systematically related result of such pairings. Some mention
to decreasing effectiveness of the US event. should be made of factors which might ac-
It may be that the level of effectiveness count for these discrepancies.
begins to decline at some point after the on- With respect to the first class of findings,
set, but remains sufficient for conditioning there are several features of most simulta-
even past the US termination. For example, neous and backward conditioning studies
backward paradigms could pair the CS with which could be expected to attenuate the
some "trace" of the US (Cautela, 1965). magnitude of any associative conditioning.
Such a notion might retain the necessity of One which is characteristic of studies em-
forward conditioning relationships by relat- ploying a long latency response as a CR
ing amount of conditioning to the amount (e.g., Bitterman, 1964, Experiment 3) is the
of US or US trace which follows the CS. use of CSs of much longer duration than the
What have traditionally been designated as US. Consequently, there is a substantial pe-
forward paradigms may have simply op- riod of time in which the CS in these studies
timized this relationship. is paired with an interval free of shock,
More recent conceptions of the Pavlovian even in simultaneous paradigms. This inter-
learning situation emphasize the informa- val may serve to extinguish or otherwise
tional value of the CS in predicting the US. reduce any excitation which accrues to the
The existence of simultaneous conditioning CS.
is disturbing to many of the ways in which Another feature common to many studies
theorists have begun to speak of informa- of conditioning is the use of methodology
tional variables. In the embedded para- which assesses the amount of conditioning
digms of this investigation, the CS does not during acquisition. Typically, the CS is pre-
"predict" or "signal" the occurrence of the sented alone within a block of four or five
US. Indeed, in most of them the best pre- reinforced trials so that the CR can be sam-
dictor of the US is its own onset. Of course, pled independent of the disruptive effects of
it cannot be said that the CS provides no the US (e.g., Smith, Coleman, & Gorme-
information of other US variables which zano, 1969). There are two consequences of
would presumably yield excitatory strength such a procedure which would make it a
(such as the continuation of the US past CS conservative test of simultaneous and back-
onset); but according to most informational ward conditioning. Perlmuter, Kimble, and
accounts, the CS would be considered re- Leonard (1968) have argued that these test
dundant with respect to US onset in pre- trials may weaken the conditioning, since
442 C. DONALD HETH AND ROBERT A. RESCORLA

the CS is then only partially reinforced. In Campbell and R. M. Church (Eds.), Punish-
addition to this effect, the presentation of ment and aversive behavior. Appleton-Cen-
tury-Crofts: New York, 1969.
CS alone trials introduces a bias unique to EGQER, M. D., & MILLER, N. E. Secondary rein-
simultaneous and backward procedures. In forcement in rats as a function of information
these paradigms, the CS is presented on re- value and reliability of the stimulus. Journal
inforced trials in compound with stimulus of Experimental Psychology, 1962, 64, 97-104.
properties of the US itself; it is this com- EGQEB, M. D., & MILLER, N. E. When is a reward
reinforcing?: An experimental study of the in-
pound which is paired with the US, while formation hypothesis. Journal of Comparative
CS alone trials are always nonreinforced. It and Physiological Psychology, 1963, 56, 132-
is not unreasonable to expect the organism 137.
to make this discrimination and thus dem- HOFFMAN, H. S., & FLESHLEB, M. A relay sequenc-
ing device for scrambling grid shock. Journal
onstrate little or no conditioning to the CS of the Experimental Analysis oj Behavior, 1962,
alone. 5,329-330.
Two recent investigations have succeeded KAMIN, L. J. Backward conditioning and the con-
in demonstrating inhibition with ba'ckward ditioned emotional response. Journal of Com-
conditioning paradigms (Moscovitch & parative and Physiological Psychology, 1963,
56,517-519.
LoLordo, 1968; Siegel & Domjan, 1971). KIMBLE, G. A. Hilgard and Marquis' conditioning
These studies employed extinction-test pro- and learning. New York: Appleton-Century-
cedures and, unlike the present investiga- Crofts, 1961.
tion, gave extensive Pavlovian training. KREMER, E. F., & KAMIN, L. J. The truly random
control procedure: Associative or nonassocia-
Their results can be contrasted with those tive effects in rats. Journal of Comparative and
of Experiment 3, but need not be viewed as Physiological Psychology, 1971, 74, 203-210.
antithetical. The backward pairing of a CS MATSUMIYA, Y. The effects of US intensity and
with US is certainly overlaid with other CS-US pattern on conditioned emotional re-
critical events to which the organism may sponse. Japanese Psychological Research, 1960,
2, 35-42.
respond, such as the offset of the US and MOSCOVITCH, A., & LoLoRDO, V. M. Role of safety
the onset of a "safe" period following it. It in the Pavlovian backward fear conditioning
seems likely that the inhibitory strength is procedure. Journal of Comparative and Physi-
the asymptotic outcome after extensive ological Psychology, 1968, 66, 673-678.
training. Some of the points of Figure 4 MOWREB, 0. H., & AIKEN, E. G. Contiguity vs.
drive-reduction in conditioned fear: Temporal
might then represent preasymptotic values variations in conditioned and unconditioned
of CSs which would have been inhibitory stimulus. American Journal of Psychology,
had more training been given. Of course, 1954, 67, 26-38.
this qualification does not deny the exist- PERLMUTEB, L. C., KIMBLE, G. A., & LEONARD, T. B.
ence of excitatory conditioning with simul- Effect of a simultaneous conditioning proced-
ure upon subsequent extinction and acquisi-
taneous or backward paradigms at some tion. Journal of Experimental Psychology,
point in learning, nor does it deny the im- 1968, 77, 648-651.
plications for our understanding of the con- QTIINSEY, V. L. Conditioned suppression with no
ditioning process. CS-US contingency in the rat. Canadian Jour-
nal of Psychology, 1971, 25, 69-82.
REFERENCES RESCOHLA, R. A. Pavlovian conditioning and its
proper control procedures. Psychological Re-
ANNAU, Z., & KAMIN, L. J. The conditioned emo- view, 1967, 74, 71-80.
tional response as a function of intensity of the RESCORLA, R. A. Informational variables in Pav-
US. Journal of Comparative and Physiological lovian conditioning. In G. H. Bower and J. T.
Psychology, 1961, 54, 428-432. Spence (Eds.), The psychology of learning and
BITTEHMAN, M. E. Classical conditioning in the motivation. New York: Academic Press, 1972.
goldfish as a function of the CS-US interval. RESCORLA, R. A., & LoLoRDO, V. M. Inhibition of
Journal of Comparative and Physiological Psy- avoidance behavior. Journal of Comparative
chology, 1964, 58, 359-366. and Physiological Psychology, 1965, 59, 406-
CATJTELA, J. R. The problem of backward condi- 412.
tioning. Journal of Psychology, 1965, 60, 135- RESCORLA, R. A., & WAGNER, A. R. A theory of
144. Pavlovian conditioning: Variations in the
CHURCH, R. M. Response suppression. In B. A. effectiveness of reinforcement and nonrein-
FEAR CONDITIONING IN THE RAT 443

forcement. In A. Black and W. F. Prokasy Classical conditioning of the rabbit's nictitating


(Eds.), Classical conditioning II. New York: membrane response at backward, simultaneous,
Appleton-Century-Crofts, 1972. and forward CS-US intervals. Journal oj Com-
SIEQEL, S., & DOMJAN, M. Backward conditioning parative and Physiological Psychology, 1969,
as an inhibitory procedure. Learning and Moti- 59^ 226-231.
vation, 1971, 2, 1-11.
SMITH, M. C., COLEMAN, S. R., & GOEMEZANO, I. (Received January 14,1972)

You might also like