181
BRIEF ARTICLES
HARVEY, 0. J. Reactions to unfavorable evaluations
of self by others. Technical Report No. 8, 1958,
Vanderbilt University, Contract NONR 2149(02),
Office of Naval Research.
HARVEY, 0. J. Personality correlates of concept functioning and change across situations. Technical
Report No. 3, 1959, University of Colorado,
Contract NONR 1147(07), Office of Naval Research.
HARVEY, 0. J., & BEVERLY, G. D. Some personality
correlates of concept change through role playing.
Journal of Abnormal and Social Psychology, 1961,
63, 125-130.
HARVEY, 0. J., & CALDWELL, D. F. Assimilation and
contrast in response to environmental variation.
Journal of Personality, 1959, 27, 125-135.
HARVEY, 0. J., KEIXEY, H. H., & SHAPIRO, M. M.
Reactions to unfavorable evaluations of the self
made by other persons. Journal of Personality,
1957, 25, 393-411.
HOVLAND, C., HARVEY, 0. J., & SHERIT, M. Assimilation and contrast effects in relation to communication and attitude change. Journal of Abnormal
and Social Psychology, 1957, 55, 244-252.
OSGOOD, C. E., & TANNENBAUM, P. H. The principle
of congruity in the prediction of attitude change.
Psychological Review, 1955, 62, 42-55.
ROBERTS, A. H., & JESSOR, R. Authoritariansm, punitiveness, and perceived social status. Journal of
Abnormal and Social Psychology, 1958, 56, 311314.
THIBAUT, J. W., & RIECKEN, H. W. Some determinants and consequences of the perception of social
causality. Journal of Personality, 1955, 24, 113133.
WAGMAN, M. Attitude change and authoritarian personality. Journal of Psychology, 1955, 40, 3-24.
(Received June 10, 1963)
Journal of Personality and Social Psychology
1965, Vol. 1, No. 2, 181-186
FACILITATING EFFECTS OF "EATING-WHILE-READING" ON
RESPONSIVENESS TO PERSUASIVE COMMUNICATIONS1
IRVING L. JANIS, DONALD KAYE, AND PAUL KIRSCHNER
Yale University
This experiment was designed to test the hypothesis that food, as an extraneous
gratification accompanying exposure to a persuasive communication, will increase acceptance, even though the donor of the food is not the source of the
communication and does not endorse it. 2 replicating experiments were carried
out with 216 male college students. In both experiments there were 3 groups
of Ss, assigned on a random basis to the following conditions, which involved
exposure to: (a) 4 persuasive communications while eating desirable food;
(b) the same 4 communications with no food present; (c) no relevant communications (control condition). Both experiments provide confirmatory evidence, indicating that more opinion change tends to be elicited under conditions
where the Ss are eating while reading the communications. The theoretical
implications are discussed with respect to psychological processes involved in
changing attitudes.
It is commonly assumed that people are more
likely to yield to persuasion at a time when they
are eating or drinking than at a time when they
are not engaged in any such gratifying activity.
Salesmen, business promoters, and lobbyists often
try to "soften up" their clients by inviting them
to talk things over at a restaurant or cafe. Representatives of opposing economic or political
groups, when unable to settle their disputes
while seated formally around a conference table,
1
This experimental investigation was conducted
under the auspices of the Yale Studies in Attitude
and Communication, which is supported by a grant
from the Rockefeller Foundation.
may find themselves much more amenable to
mutual influence, and hence more conciliatory,
while seated comfortably around a dinner table.
Little systematic research has been done, as
yet, to determine the conditions under which
pleasant stimulation will augment the acceptance
of persuasive communications. One might expect
that when the communicator is the perceived
source of the gratifying stimulation, a more
favorable attitude toward him will ensue, which
would tend to lower the recipient's resistance
to his persuasive efforts (see Hovland, Janis, &
Kelley, 1953, pp. 19-55). But a more complicated
situation often arises at educational symposia,
182
BRIEF ARTICLES
political conventions, cocktail parties, and informal dinners where: (a) the donor (that is, the
person who is perceived as being responsible for
the gratification) is not the communicator and
(b) the donor does not endorse the persuasive
communications that happen to be presented at
the particular time when the recipients are being
indulged. If a positive gain in effectiveness is
found to occur under these conditions, where the
gratifying activity is entirely extraneous to the
content, source, or endorsement of the communications, a number of important theoretical questions will arise—questions concerning some of
the basic processes of attitude change which will
require systematic experimental analysis. For
example, when eating has a facilitating effect on
acceptance of persuasive messages, does it always
depend entirely upon the heightened motivation
of the recipients to conform with the donor's
wishes? If so, a positive outcome under nonendorsement conditions will be paradoxical unless
it turns out that there is a general tendency for
people to assume, consciously or unconsciously,
that the donor would like them to be influenced
by whatever communications are presented (even
though he explicitly says that he does not endorse
the point of view being expressed). Or does the
extraneous gratification operate as a source of
reinforcement independently of the recipient's
attitude toward the donor? If this is the case,
we might be led to assume that the food corresponds to an "unconditioned stimulus," and its
facilitating effects might be accounted for in
terms of the laws of conditioning.
The latter theoretical possibility is suggested
by Razran's (1940) brief research note, published
25 years ago, in which he gave a summary
statement of the following two experimental observations : (a) an increase in ratings of "personal
approval" occurred when a series of sociopolitical
slogans were presented to experimental subjects
while they were enjoying a free lunch and (b) a
decrease in such ratings occurred when the
slogans were presented while the subjects were
being required to inhale a number of unpleasant,
putrid odors. In his report, however, Razran
does not mention certain important details, such
as whether the experimenter was the donor of
the free lunch and whether he said anything to
the subjects about his personal attitude toward
the slogans.
So far as the authors have been able to
ascertain, no subsequent experiments have been
published pertinent to checking Razran's observations. Nor has any published research been
found bearing on the related questions of whether
or not (and under what limiting conditions)
extraneous pleasant or unpleasant stimulation can
affect the degree to which a recipient will accept
a series of persuasive arguments that attempt
to induce him to change a personal belief or
preference.
As a preliminary step toward reopening experimental research on the above-mentioned set of
theoretical problems, the present study was
designed to investigate the alleged phenomenon
of enhanced communication effectiveness arising
from "eating-while-reading," The research was
designed primarily to answer the following
question: If an experimenter gives the subjects
desirable food and drink but states explicitly
that the persuasive messages to be presented are
ones with which he does not necessarily agree,
will there be a significant increase in acceptance
from the gratifying activity of eating that
accompanies exposure to the communications?
METHOD AND PROCEDURE
Experimental Design
The basic design involved randomly assigning the
subjects to two different experimental conditions. One
was a condition in which a substantial quantity of
food was offered to the subjects during the time
they were engaged in reading a series of four persuasive communications. Upon entering the experimental room, the subjects found the experimenter
imbibing some refreshments (peanuts and PepsiCola) and they were offered the same refreshments
with the simple explanation that there was plenty
on hand because "I brought some along for you too."
The contrasting "no-food" condition was identical
in every respect except that no refreshments were
in the room at any time during the session.
The same measures of opinion change were used
in the two experimental groups and also in a third
group of unexposed controls, who were included in
the study in order to obtain a base line for ascertaining the effectiveness of each communication per
se. The subjects randomly assigned to the control
condition were given the same pre- and postcommunication questionnaires, separated by the same
time interval as in the other two experimental conditions, but without being exposed to any relevant
communications.
The Communications and the Opinion Measures
On the basis of extensive pretesting, we prepared
four communications, each of which advocated an
unpopular point of view and had been found to
be capable of inducing a significant degree of
opinion change. These communications were attributed to fictitious authors who were described as
journalists or news commentators. The main conclusions, all of which involved quantitative predictions or preferences about future events, were as
follows:
183
BRIEF ARTICLES
1. It will be more than 25 years before satisfactory progress can be expected in the search
for a cure for cancer.
2. The United States Armed Forces do not need
additional men and can be reduced to less than
85% of their present strength.
3. A round-trip expedition to the moon will be
achieved within the next decade.2
4. Within the next 3 years, three-dimensional
films will replace two-dimensional films in practically all movie theaters.
In order to assess opinion changes, four key
questions were included in both the pre- and the
postcommunication questionnaires, each of which
asked the subject to express his opinion in the
form of a quantitative estimate (for example, "How
many years do you think it will be before an
extremely effective cure is found for cancer so that
cancer will no longer be a major cause of death?
About
years.")
Experiments I and II: Similarities and
ences
Differ-
The same experimental design, described above,
was used in two separate experiments, during successive semesters at the same college. In all essential
features the first (Experiment I) was identical with
the second (Experiment II) in that exactly the same
experimental variations were used along with the
same instructions, the same communications, and
the same pre- and postcommunication questionnaires.
But the two experiments differed in several minor
ways. The main difference was that in Experiment I
the time interval between the precommunication
questionnaire and exposure to the communications
was about 2 months; whereas in Experiment II
the precommunication questionnaire was given at the
beginning of the experimental session, immediately
preceding the communications.
In Experiment I, the initial questionnaire was
administered in regular undergraduate class sessions.
It was introduced as a "survey of student opinions"
and the key questions were embedded among numerous filler questions on a variety of other controversial issues. After a period of 2 months, the
subjects were contacted by telephone and asked to
be unpaid volunteers for a study on reading preferences. The vast majority volunteered and each subject was seen in a private interview session, at the
beginning of which he was randomly assigned to
the "food with communication" condition or the
"no food with communication" condition or the
unexposed control condition. After answering the
2
This study was carried out before the major
developments in space flights had occurred, at a
time when few people were optimistic about the
rate of technical progress in this field. In response
to the moon-flight question on the initial questionnaire, almost all the students gave estimates of 10
years or more before a successful round-trip flight
could be expected.
final set of postcommunication questions, each subject was briefly interviewed concerning his reactions
to the experimental situation.
In Experiment II, the same essential procedures
were used except for the fact that the precommunication questionnaire was given at the beginning
of the experimental session. Another minor difference was that the unexposed controls were given
some extracts from a popular magazine on irrelevant
topics, which took approximately the same reading
time as the four persuasive communications. Moreover, unlike the unexposed controls in Experiment I,
those in Experiment II were given the same food in
the same way as in the main experimental condition,
so that they too were eating while reading the
(irrelevant) articles.
In addition to the three conditions that were set
up to replicate the essential features of Experiment I,
a fourth experimental condition was introduced in
Experiment II in order to investigate a subsidiary
problem, namely, the effects of extraneous unpleasant
stimuli. The fourth experimental group, while reading
the four persuasive communications, was exposed to
an unpleasant odor (produced by a hidden bottle
of butyric acid), for which the experimenter disclaimed any responsibility.
In both experiments, the experimenter explained
that the purpose was to assess the students' reading
preferences. He asserted that he did not endorse the
communications and casually mentioned that he happened to agree with certain of the ideas expressed
and not with others (without specifying which).
He asked the subjects to read the articles as though
they were at home reading a popular magazine. In
line with the alleged purpose, the postcommunication
questionnaire in both Experiments I and II included
20 filler questions asking for interest ratings of the
articles (for example, ratings of how much interest
they would expect the average college student to
have in each topic).
Subjects
A total of 216 Yale undergraduate students were
used in the two experiments. In Experiment I, 35
men were in the unexposed control group, 32 in the
"no food with communication" condition, and 33 in
the "food with communication" condition. In Experiment II, the corresponding numbers were 23, 31,
and 31, respectively. There were also 31 subjects in
the fourth experimental group exposed to the
"unpleasant" condition.
RESULTS
In both experiments, observations of the subjects' eating behavior in the "food" condition
showed that every one of them ate at least one
handful of peanuts and drank at least one-half
glass of the soft drink. The main findings concerning the effects of eating desirable food on
the acceptance of the four persuasive communications are shown in Table 1. In general, the
184
BRIEF ARTICLES
TABLE 1
OPINION CHANGES INDUCED BY EXPOSURE TO FOUR PERSUASIVE COMMUNICATIONS
UNDER Two DIFFERENT CONDITIONS: "FOOD" VERSUS "No FOOD"
GIVEN BY THE EXPERIMENTERS
% opinion change
Communication
topic
1. Cure for cancer
Positive change
Negative change
No change
Total
Net change
P
2. Preferred size of
United States
Armed Forces
Positive change
Negative change
No change
Total
Net change
P
3. Round trip to
moon
Positive change
Negative change
No change
Total
Net change
P
4. Three dimensional movies
Positive change
Negative change
No change
Total
Net change
P
Experiment I
Experiment II
Combined data from
Experiments I and II
No food
(N = 32)
Food
(2V = 33)
No food
(AT =31)
Food
(N =31)
No food
(N = 63)
Food
(N = 64)
68.7
21.8
9.5
100.0
46.9
81.8
12.1
6.1
100.0
69.7
80.7
3.2
16.1
100.0
77.5
*
93.5
0.0
6.5
100.0
93.5
74.6
12.7
12.7
100.0
61.9
87.4
6.3
6.3
100.0
81.1
<
'll
81.8
0.0
18.2
100.0
81.8
65.6
9.4
25.0
100.0
56.2
< "05
48.4
19.4
32.2
100.0
29.0
< 01
58.0
12.9
29.1
100.0
45.1
75.9
12.1
12.0
100.0
63.8
= .20
results indicate that "eating-while-reading" has
a facilitating effect on the amount of opinion
change. In Experiment I, the differences between
the food and no-food conditions are consistently
in the predicted direction for all four communications, two of which are significant at the .05
level. (All p values are one-tailed and were
obtained on the basis of the formula for assessing the difference between two net percentage
changes, given by Hovland, Lumsdaine, & Sheffield, 1949.) The results for Experiment II show
differences in the same direction for three of the
four communications, two of which are significant
at the .10 level. There is a very small, nonsignificant difference in the reverse direction on
the fourth communication.
74.2
0.0
25.8
100.0
74.2
< 05
77.4
6.5
16.1
100.0
70.9
=140
i
67.2
12.5
20.3
100.0
54.7
50.8
20.6
28.6
100.0
30.2
20
05
67.2
0.0
32.8
100.0
67.2
47.6
4.8
47.6
100.0
42.8
< OS
75.9
12.1
12.0
100.0
63.8
68.7
21.8
9.5
100.0
46.9
51.6
0.0
48.4
100.0
51.6
29.0
0.0
71.0
100.0
29.0
< 05
53.2
21.9
24.9
100.0
31.3
loT"
76.6
9.4
14.0
100.0
67.2
71.5
11.1
17.4
100.0
60.4
<.20
The p values based on the combined data from
both experiments, shown in the last column of
Table 1, can be regarded as a satisfactory summary of the overall outcome inasmuch as: (a)
the numbers of cases in each experiment are
almost equal; and (6) the two experiments differed only in minor features that are irrelevant
to the main comparison under investigation. The
combined data show that all four communications
produced differences in the predicted direction
and for three of them the differences are large
enough to be statistically significant. Thus, the
results support the conclusion that, in general,
the extraneous gratification of eating while reading a series of persuasive communications tends
to increase their effectiveness.
185
BRIEF ARTICLES
That each communication was effective in
inducing a significant degree of opinion change,
whether presented under food or no-food conditions, is indicated by the comparative data from
the unexposed controls. In both experiments, the
control group showed very slight positive
changes, if any, on each of the four key questions
and the amount of change was always significantly less than the corresponding net change
shown by the food and no-food experimental
groups.3 There were no consistent differences
between the control group in Experiment I and
the one in Experiment II, which indicates that
the different time intervals between the before
and after measures and the other minor procedural differences between the two experiments
had no direct effect on the opinion measures.
The condition of unpleasant stimulation introduced into Experiment II had no observable effect on the amount of opinion change. The net
changes obtained from the group exposed to the
foul odor (N = 31) were as follows: cancer cure,
67.7%; size of armed forces, 25.8%; round trip
to moon, 38.7%; three-dimensional movies,
64.5%. These values differ only very slightly
from those obtained from the group exposed to
the no-food condition in Experiment II (see
Table 1); none of the differences are large
enough to approach statistical significance. As
expected, however, all the net changes for the
unpleasant odor condition are smaller than those
for the food condition and in two of the
four instances the differences are statistically
significant at beyond the .05 level.
3
In all but one instance, the net change shown
by the unexposed controls was not significantly different from zero. The one exception occurred in the
control group in Experiment I with respect to the
first issue (cancer cure), on which a significant net
change of —34% was found. This change, however,
was in the reverse direction from that advocated by
the communication (probably as a consequence of
optimistic publicity concerning new advances in
cancer research that appeared in the newspapers
during the months between the before and after
questionnaires). Thus, on this item, as well as on
the other three, the control group showed significantly less change in the expected direction than
the two experimental groups.
An analysis of responses to the precommunication
questionnaire from both experiments showed that
initially, on each of the four key opinion questions,
there were only very slight, nonsignificant differences
among the experimental and control groups. None of
the results in Table 1 and none of the other observed
differences in amount of opinion change are attributable to initial differences.
DISCUSSION
Our finding that the extraneous gratifying
activity of eating tended to increase the degree
to which the accompanying persuasive messages
were accepted may prove to have important
implications for the psychology of attitude
change, especially if subsequent research shows
that the gains tend to be persistent, giving rise
to sustained modifications of personal beliefs or
preferences. Since the control group in Experiment II (which received food along with irrelevant communications) showed net opinion
changes that were practically zero and were
significantly less than those shown by the main
experimental group, the food alone appears to
have had no direct effect on any of the opinion
measures. Hence the observed outcome seems to
implicate psychological processes involved in the
acceptance of persuasive influences.
Our results on the positive effects of food are
similar to Razran's (1940) findings on the increase in favorable ratings of sociopolitical
slogans induced by a free lunch. Razran has
indicated that he regards his observations as
evidence of Pavlovian conditioning, resulting
from the contiguity of the conditioned stimuli
(the slogans) and the unconditioned pleasant
stimuli (food). Before accepting any such interpretation, however, further investigations are
needed to check systematically on the possibility
that the change in acceptability is brought about
by creating a more favorable attitude toward
the donor. We attempted to minimize this possibility in both Experiments I and II by having
the experimenter give the subjects an introductory explanation in which he clearly stated
that he was not sponsoring the persuasive communications. Despite this attempt, however, the
subjects may have ignored or forgotten his remarks and assumed that he was sponsoring them.
We have no evidence bearing directly on this
matter, but we did note that in the informal
interviews conducted at the end of each experimental session, many more favorable comments
about the experimenter were made by the subjects who had been in the food condition than
by those who had been in the no-food condition.
Our failure to confirm Razran's findings on
the negative effects of unpleasant stimulation
might be accounted for in terms of attitude
toward the experimenter. In Razran's experiment,
the experimenter "required" the subjects to
sniff the putrid odors, and hence he might have
been directly blamed for the unpleasant stimulation; whereas in our Experiment II, the unpleasant odor was presented as an accidental
186
BRIEF ARTICLES
occurrence for which the experimenter was not
responsible. Further experimental analysis is
obviously needed to determine if the effects of
pleasant and unpleasant stimulation observed in
our experiment are dependent upon whether or
not the experimenter is perceived as the causal
agent.
The fact that the experimenter himself participated in eating the food might have influenced the subjects' perceptions of the general
atmosphere of the reading session and hence
needs to be investigated as a possible variable,
independently of the subjects' food consumption.
The limiting conditions for positive effects from
"eating-while-reading" also require systematic investigation, particularly in relation to unpleasant
interpersonal stimuli, such as those provoking
embarrassment, outbreaks of hostility, or other
forms of emotional tension that could counteract the positive atmosphere created by the
availability of desirable food.
It is also important to find out whether variations in the experimenter's endorsement of the
communications play a crucial role in determining
the facilitating effects of the preferred food. For
example, if subsequent research shows that the
experimenter's positive versus negative endorsements make a difference, then an explanation
in terms of increased motivation to please the
donor will be favored, rather than a simple
conditioning mechanism, and a more complicated
explanation will be required to account for the
positive effects obtained under conditions where
the experimenter explicitly detaches himself from
sponsorship of the communications.* These implications are mentioned to illustrate the new lines
of research suggested by comparing the results
from the present experiment with those from
Razran's earlier study.
4
The potential importance of positive versus negative endorsement by the experimenter as an interacting variable was suggested by some unexpected
results obtained in a pilot study by Dabbs and
Janis, which was carried out as a preliminary step
toward replicating the present experiment under
conditions where the experimenter indicates that he
personally disagrees with the persuasive communications. The pilot study results led us to carry out a
new experiment in which we compared the effects
of eating-while-reading under two different endorsement conditions (the experimenter agreeing or disagreeing with the communications). A report on
the effects of the interacting variables, as revealed
by the data from the Dabbs and Janis experiment,
is currently being prepared for publication.
REFERENCES
HOVLAND, C. I., JANIS, I. L., & KELLEY, H. H. Communication and persuasion. New Haven: Yale
Univer. Press, 1953.
HOVLAND, C. I., LUMSDAINE, A. A., & SHEFFIELD,
F. D. Experiments on mass communication. Princeton: Princeton Univer. Press, 1949.
RAZRAN, G. H. S. Conditioned response changes
in rating and appraising sociopolitical slogans.
Psychological Bulletin, 1940, 37, 481.
(Early publication received July 21, 1964)
ERRATUM
In Table 2 (page 311) of the article "Cooperation and Competition
in Means-Interdependent Triads," by Bertram H. Raven and H.
Todd Eachus (Journal of Abnormal and Social Psychology, 1963,
67, 307-316), the interaction effect was actually not significant. The
A X B row of the table should indicate that the mean sum of squares
is 14.40 and the F value is .40, with no indication of significance. This
change is in no way relevant to the substantive findings of the article,
since there had been no expectation of an interaction effect. However,
in the interests of accuracy, the last two sentences on page 310, which
refer to the significance of the interaction, should be omitted.