Abstract and Figures

Connecting with others increases happiness, but strangers in close proximity routinely ignore each other. Why? Two reasons seem likely: Either solitude is a more positive experience than interacting with strangers, or people misunderstand the consequences of distant social connections. To examine the experience of connecting to strangers, we instructed commuters on trains and buses to connect with a stranger near them, to remain disconnected, or to commute as normal (Experiments 1a and 2a). In both contexts, participants reported a more positive (and no less productive) experience when they connected than when they did not. Separate participants in each context, however, expected precisely the opposite outcome, predicting a more positive experience in solitude (Experiments 1b and 2b). This mistaken preference for solitude stems partly from underestimating others' interest in connecting (Experiments 3a and 3b), which in turn keeps people from learning the actual consequences of social interaction (Experiments 4a and 4b). The pleasure of connection seems contagious: In a laboratory waiting room, participants who were talked to had equally positive experiences as those instructed to talk (Experiment 5). Human beings are social animals. Those who misunderstand the consequences of social interactions may not, in at least some contexts, be social enough for their own well-being. (PsycINFO Database Record (c) 2014 APA, all rights reserved).
Content may be subject to copyright.
Journal of Experimental Psychology: General
Mistakenly Seeking Solitude
Nicholas Epley and Juliana Schroeder
Online First Publication, July 14, 2014. http://dx.doi.org/10.1037/a0037323
CITATION
Epley, N., & Schroeder, J. (2014, July 14). Mistakenly Seeking Solitude. Journal of
Experimental Psychology: General. Advance online publication.
http://dx.doi.org/10.1037/a0037323
Mistakenly Seeking Solitude
Nicholas Epley and Juliana Schroeder
University of Chicago
Connecting with others increases happiness, but strangers in close proximity routinely ignore each other.
Why? Two reasons seem likely: Either solitude is a more positive experience than interacting with
strangers, or people misunderstand the consequences of distant social connections. To examine the
experience of connecting to strangers, we instructed commuters on trains and buses to connect with a
stranger near them, to remain disconnected, or to commute as normal (Experiments 1a and 2a). In both
contexts, participants reported a more positive (and no less productive) experience when they connected
than when they did not. Separate participants in each context, however, expected precisely the opposite
outcome, predicting a more positive experience in solitude (Experiments 1b and 2b). This mistaken
preference for solitude stems partly from underestimating others’ interest in connecting (Experiments 3a
and 3b), which in turn keeps people from learning the actual consequences of social interaction
(Experiments 4a and 4b). The pleasure of connection seems contagious: In a laboratory waiting room,
participants who were talked to had equally positive experiences as those instructed to talk (Experiment
5). Human beings are social animals. Those who misunderstand the consequences of social interactions
may not, in at least some contexts, be social enough for their own well-being.
Keywords: social cognition, social connection, mind perception, affective forecasting, wellbeing
Humans are among the most social species on the planet, with
brains uniquely adapted for living in large groups (Dunbar, 1998;
Herrmann, Call, Herna
`ndez-Lloreda, Hare, & Tomasello, 2007;
Sallet et al., 2011). Feeling socially connected increases happiness
and health, whereas feeling disconnected is depressing and un-
healthy (Holt-Lunstad, Smith, & Layton, 2010; House, Landis, &
Umberson, 1988; Kahneman & Deaton, 2010; Myers & Diener,
1995). Nevertheless, modern life affords many opportunities for
connecting with others that are routinely foregone. From trains to
cabs to airplanes to waiting rooms, strangers may sit millimeters
apart while completely ignoring each other, treating one another as
objects rather than as sources of well-being. As Milgram and
Sabini (1978) noted in their seminal study of modern urban life,
The requirements of appropriate social behavior on the subway are, on
the face of it, simple. [One] implicit rule...discourages passengers
from talking to each other. Even though riders are often squeezed into
very close proximity, they are rarely observed to converse. (pp.
32–33)
For a species that seems to benefit so much from connecting to
others, why would people in close proximity so routinely seem to
prefer isolation instead? Why are such highly social animals, at
times, so distinctly unsocial?
There are two plausible answers to this apparent social paradox.
One is that connecting with a stranger in conversation is truly less
pleasant than remaining isolated for a variety of possible reasons.
Preferring isolation in the company of random strangers may
therefore maximize one’s well-being. The other is that people
systematically misunderstand the consequences of social connec-
tion, mistakenly thinking that isolation is more pleasant than
connecting with a stranger, when the benefits of social connection
actually extend to distant strangers as well. We designed a series
of field and laboratory experiments to test between these two
hypotheses, to identify underlying mechanisms and moderators for
the behavior we observe, and to examine the broader consequences
of distant social connections.
The Pleasure of Disconnection?
Modern life provides overwhelming opportunities for social
engagement, and so, social connections have to be regulated. Just
like other drive states such as hunger, where people consume food
that appears satisfying and avoid food that appears nauseating,
people regulate their social drive by connecting with people who
seem satisfying (e.g., close others) but avoiding those who seem
unsatisfying (e.g., distant others). For instance, a person’s overall
well-being appears to be driven by the quality of connections with
close others rather than the quantity of connections with more
distant others (Coan, Schaefer, & Davison, 2006; Cohen, 2004;
Davis, Morris, & Kraus, 1998; King & Reis, 2012; Pinquart &
Sorensen, 2003; Williams & Solano, 1983). People often evaluate
distant strangers and outgroup members as relatively poor sources
of social support—even lacking humanlike mental capacities of
rational thought and secondary emotions (Cortes, Demoulin, Ro-
driguez, Rodriguez, & Leyens, 2005; Haslam, 2006; Haslam, Bain,
Nicholas Epley and Juliana Schroeder, Booth School of Business, Uni-
versity of Chicago.
We thank Julia Arozena, Rachel Choi, Jesus Diaz, Janet Flores, Jasmine
Kwong, Justin Liang, Rachel Meng, Michael Pang, Maimouna Thioune,
Roisleen Todd, Lester Tong, and Emily Wolodiger for assistance conduct-
ing experiments, Thomas Gilovich for helpful comments on a draft of the
manuscript, and the Booth School of Business for financial support.
Correspondence concerning this article should be addressed to Nicholas
Epley, 5807 South Woodlawn Avenue, Booth School of Business, Univer-
sity of Chicago, Chicago, IL 60637. E-mail: epley@chicagobooth.edu
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Journal of Experimental Psychology: General © 2014 American Psychological Association
2014, Vol. 143, No. 5, 000 0096-3445/14/$12.00 http://dx.doi.org/10.1037/a0037323
1
Douge, Lee, & Bastian, 2005). When faced with a seemingly
unsatisfying stranger as a potential conversation partner, it could
be sensible to avoid conversation and remain disconnected instead.
Indeed, at least some people seem to think that remaining
disconnected from strangers is quite sensible. In a survey of 203
Amazon.com Mechanical Turk participants, we asked one group
of people to report the likelihood that they would talk (yes or no)
to a friend and a stranger (in counterbalanced order) in one of four
locations (waiting room, train, airplane, and cab). Virtually none of
these participants predicted that they would avoid talking to a
friend (7%, 0%, 0%, and 0% responded “no” in each context,
respectively), but a majority in each context thought they would
avoid talking to a stranger (93%, 76%, 68%, and 51%, responded
“no,” respectively).
1
Few would forgo the potential pleasure of
connecting with a friend, but most seem readily inclined to ignore
a stranger. This makes sense if talking to a stranger really is less
pleasant than sitting in solitude.
Mistakenly Seeking Solitude?
Connecting with strangers may not bring the same long-term
benefits as connecting with friends, but our interest is in whether
connecting with a stranger is less beneficial than remaining iso-
lated altogether. It is possible that people misunderstand the con-
sequences of distant social interactions such that people avoid
talking to strangers because they expect it will be less pleasant than
remaining isolated, when the opposite may actually be true.
Several existing findings suggest some misunderstanding about
the consequences of social connection. In one series of experi-
ments (Mallett, Wilson, & Gilbert, 2008), White participants ex-
pected to have more pleasant interactions with other White partic-
ipants than Black participants. In reality, they had equally pleasant
conversations with both groups. In another series (Zelenski et al.,
2013), extraverts believed they would have a more pleasant expe-
rience interacting with another person in an extraverted fashion
than introverts predicted. In fact, both introverts and extraverts
reported a more positive experience while acting extraverted in a
social interaction than while acting introverted.
In a third series of experiments (Dunn, Biesanz, Human, & Finn,
2007), undergraduates in one experiment expected to feel worse
while waiting to interact with an opposite-sex stranger than while
waiting to interact with their dating partner, but there were no
mood differences among people actually waiting to interact with
an opposite-sex stranger versus their romantic partner. A second
experiment involving an actual 4-min interaction with either a
romantic partner or an opposite-sex stranger found similar results.
Although it is unclear whether these two results tell us something
general about the hedonic benefits of interacting with strangers or
something more specific about the pleasure of opposite-sex flirta-
tion, they do suggest that people might misunderstand the pleasure
of interacting with strangers and therefore mistakenly prefer iso-
lation.
This misunderstanding of the actual consequences of social
connection could take at least two different forms. Research on the
impact bias (Gilbert, Driver-Linn, & Wilson, 2002) predicts a
relatively mild misunderstanding in which people might expect
that connecting with a stranger in conversation will be more
negative than remaining isolated but that it will not be quite as
negative in reality as they anticipate (e.g., Dunn et al., 2007;
Mallett et al., 2008). Talking to a stranger may be no worse (or
only slightly worse) than remaining isolated.
A second pattern could reflect a more extreme misunderstand-
ing, not just a mistake in the magnitude or duration of an effect but
a mistake in the actual valence of an effect. In particular, in
contexts that actually require interactions with others, people are
happier when told to act extraverted—to be more assertive, ad-
venturous, energetic, and talkative—than when told to act intro-
verted (Fleeson, Malanos, & Achille, 2002; McNiel & Fleeson,
2006; Sandstrom & Dunn, 2014; Zelenski, Santoro, & Whelan,
2012). Critically, these methods either utilize diary or experience
sampling studies in which the targets and contexts of the interac-
tions are unclear (people are most likely interacting with friends in
social contexts) or experiments that require engaging with a
stranger (or a group of strangers) to complete a task. Nevertheless,
they suggest it is possible, even in situations where social inter-
action is neither required nor the norm (such as on trains, on buses,
or in waiting rooms), that engaging a stranger in conversation may
actually be more pleasant than remaining isolated. This suggests a
more profound misunderstanding of social interactions: Members
of a highly social species may ignore other people because they
expect that connecting with a stranger will be more negative than
remaining isolated when in fact the exact opposite pattern is true.
Overview of Experiments
We conducted nine experiments, in both field and laboratory
settings, to at least partly explain an apparent social paradox: why
people who benefit greatly from social connection nevertheless
prefer isolation amongst strangers. We began our research in two
contexts where strangers come in very close proximity but almost
never connect: on commuter trains and public buses. In each
context, we first asked participants in one experiment (Experi-
ments 1a and 2a) to talk to a stranger, sit in solitude, or do
whatever they would normally do, to measure the actual conse-
quences of distant social engagement versus isolation. This
between-participants design allows us to measure whether inter-
acting with a distant stranger is indeed less pleasant than remaining
isolated in ecologically valid contexts (Keren & Raaijmakers,
1988).
In another experiment in each context (Experiments 1b and 2b),
we then asked a separate group of participants to predict their
experiences in the same conditions (talking to a stranger, sitting in
solitude, or doing whatever they would normally do) to measure
people’s expectations of the outcomes of these conditions as
precisely as possible. This within-participant comparison allows us
1
This survey included another between-participants condition that asked
participants to remember their past behavior in each of the four contexts
(waiting room, train, airplane, cab) to see whether their memory for their
past interactions matched their predictions. Overall, they did. Participants
in the memory condition considered the last time they were in each of the
contexts with a friend and a stranger (counterbalanced order) and indicated
whether they engaged in conversation (“yes,” “no,” “do not remember,” or
“not possible in this situation”). Simply looking at those who said “no,”
few participants again avoided talking to a friend (6%, 6%, 11%, and 12%,
in each context, respectively), but many avoided talking to a stranger (86%,
86%, 59%, and 43%, respectively). The increased frequency of talking to
a stranger in a cab—that is, to the driver—in both anticipated interaction
and recalled interactions is a result we turn to in Experiments 4a and 4b.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2EPLEY AND SCHROEDER
to examine people’s explicit theories about the consequences of
these actions in direct comparison against each other (Hsee, 2000).
These field experiments tested whether isolation is truly
preferable to connecting with a stranger and, if not, whether
people’s seeming preference for isolation among strangers
comes from a mistaken belief that isolation will be a more
positive experience than connecting with a stranger. Experi-
ments 3a–4b tested between different possible mechanisms for
the results observed in Experiments 1a–2b. Finally, Experiment
5 was a laboratory experiment that addressed two concerns with
the field experiments: whether the effects of social connection
affect only those we instructed or if they are contagious and
extend to people who are talked to or ignored as well, and
whether free choice affects the consequences of social connec-
tion. Overall, these experiments tested whether people are ap-
propriately social in their everyday lives or perhaps not social
enough for their own well-being.
Experiments 1a and 1b: Trains
Method
Experiment 1a procedure. Two research assistants recruited
commuters just before boarding inbound morning commuter
trains, surreptitiously recording their gender and ethnicity. The
research assistants recruited people walking alone to the train
platform in order to make it easier for participants in the connec-
tion condition to talk with a stranger (rather than with friends,
family, or acquaintances they happened to be traveling with that
day). One hundred eighteen commuters at the Homewood, Illinois,
Metra station agreed to participate. We chose this particular station
partly because it is one of the first on the rail line for the express
trains, meaning that passengers are boarding a nearly empty train
during the times when we conducted our experiments and would
therefore be able to begin the experiment sitting alone (as is the
strong norm) rather than choosing a stranger to sit next to.
Research assistants randomly assigned commuters to one of
three conditions: connection, solitude, or control. Participants in
the connection condition were told, “Please have a conversation
with a new person on the train today. Try to make a connection.
Find out something interesting about him or her and tell them
something about you. The longer the conversation, the better. Your
goal is to try to get to know your community neighbor this
morning.” Participants in the solitude condition were told, “Please
keep to yourself and enjoy your solitude on the train today. Take
this time to sit alone with your thoughts. Your goal is to focus on
yourself and the day ahead of you.” Finally, participants in the
control condition were told, “Please do not make any changes to
your normal commute. Your goal is to do as you would normally
do.” Although this condition is not an ideal control because par-
ticipants are free in this condition to talk to others (almost certainly
with friends) or sit alone in solitude, we nevertheless felt it useful
to include for the sake of completeness.
Commuters then received the experimental survey and the Ten
Item Personality Inventory (TIPI; Gosling, Rentfrow, & Swann,
2003) in a stamped and addressed envelope along with a $5 gift
card as compensation. They were asked to follow the instructions
(to talk, sit in solitude, or have their normal commute) during the
train ride and then open the envelope and complete the survey at
the end of their train ride and mail it back.
The first page of the survey asked participants to report the time
that they were completing the survey and to mark whether or not
they did the following during their morning commute: talk to
someone (friend or stranger not specified), talk on the phone, read,
sleep, think, work, or other. Participants then completed three
measures to assess the overall positivity of their commute: how
happy and how sad they felt after their commute on response
scales ranging from 0 (Not at all happy/sad)to6(Very happy/sad)
and how pleasant their commute was, compared to their usual
commute, on response scales ranging from 3(Much less pleas-
ant)to3(Much more pleasant). Because engaging a stranger in
conversation may come at a cost to other activities one might do on
a commuter train (such as work), we then asked participants to
report how productive their commute was, compared to their usual
commute, on a scale ranging from 3(Much less productive)to3
(Much more productive).
On the second page of the survey, we asked participants in the
connection condition to write “as much as you can remember
about the person with whom you spoke (name, ethnicity, age,
occupation, etc.),” to estimate the length of their conversation (in
minutes), and then to report how pleasant their conversation was
on a scale ranging from 0 (Not pleasant at all)to6(Very pleasant)
and their overall impression of their conversation partner on a
scale ranging from 3(Very negative)to3(Very positive). Fi-
nally, participants in all conditions marked the activities they
normally do on their commute from a list of the same seven
activities from the first page of the survey. The TIPI was always
included after the experimental survey in the envelope.
Experiment 1b procedure. Research assistants recruited
commuters in the same manner as Experiment 1a. One hundred
five commuters from the Homewood, Illinois, Metra station
agreed to participate. Those who agreed to participate received
a stamped and addressed envelope containing the TIPI and the
experimental survey in randomly determined order, along with
a $5 gift card as compensation.
The experimental survey described the procedure of the actual
experience experiment (Experiment 1a) as closely as possible.
Participants imagined following the instructions from the control
condition and, subsequently, the connection and solitude condi-
tions in counterbalanced order. Therefore, the design included the
same three conditions as Experiment 1a but manipulated within
participants.
In the control condition, participants read,
We would like you to imagine that you are participating in a study
about commuting on the train. Imagine that, as you walk to the
platform to catch your train in the morning, you see a student standing
near the platform who asks you to participate in a study. Imagine that,
just like you did today, you agree to participate and sign a consent
form. Imagine that the student gives you these instructions for the
study: “Please do not make any changes to your normal commute.
Your goal is to do as you would normally do.” Imagine that you
follow these instructions and then complete a questionnaire at the end
of your commute.
The connection and solitude conditions contained the same
information except that the instructions were changed to match
the actual instructions given in Experiment 1a. Following the
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
3
SOLITUDE
instructions for each condition, participants predicted their
mood (how happy and how sad they would feel after a commute
in that condition) and how pleasant and productive their com-
mute would be compared to usual on the same rating scales
described in Experiment 1a. At the end of the experimental
survey, participants marked the activities they normally do on
their commute from the same list of seven activities in the
Experiment 1a survey.
Results
To obtain an overall measure of positivity, we first calculated
positive mood (happy minus sad), then standardized positive mood
and pleasantness, and then averaged those two measures into a
single index. Positive mood and pleasantness were significantly
correlated in all experiments reported in this manuscript (rs.36,
.55, .48, .55, .42, .59, .49, .52, .62, Experiments 1a–5, respectively,
ps.01).
Experiment 1a. Eighty-nine percent of participants who took
a survey returned it in the mail, with no differences between
experimental conditions (
2
2.36, p.31). Of those, eight
reported being unable to follow the instructions (one in the solitude
condition, seven in the connection condition). Of these seven in the
connection condition, all reported being unable to talk because
nobody sat next to them on their train ride (a distinct possibility
later in the mornings when the trains are not full). This left a final
sample of 97 commuters in the following analyses (M
age
49
years, SD
age
13 years, 61% female).
Experiences. A one-way analysis of variance (ANOVA)
showed that reported positivity of the commute experience varied
across the three experimental conditions, F(2, 96) 3.10, p.05,
2
0.06.
Talking to a stranger on the train was not systematically un-
pleasant. In fact, participants in the connection condition reported
having the most positive experience out of all three of our exper-
imental conditions (see Figure 1). Most important, participants in
the connection condition reported having a significantly more
positive experience than participants in the solitude condition,
t(94) 2.49, p.02, d0.63.
2
Because our control condition allows a wide range of activities,
including talking to friends as well as sitting in solitude, the data
from the control condition are somewhat difficult to interpret.
Indeed, six participants in the control condition talked with a friend
during the commute (based on the content of their description of
the conversation), and these six participants reported a mean
positivity comparable to participants in the connection condition
(M.20, SD .77). Overall, ratings of positivity in the control
condition fell almost perfectly in between the solitude,
t(94) ⫽⫺1.22, p.22, d⫽⫺0.25, and connection conditions,
t(94) 1.39, p.17, d0.29. Studies 2a and 4a allow for
additional comparisons with a similar control condition to provide
a more informed assessment of how connection versus isolation
compares to whatever else participants might normally do in these
contexts.
The positive experience that participants had in the connection
condition did not appear to come at a significant cost to their
reported productivity as there were no significant differences on
this measure between the three experimental conditions, F(2,
96) 1.22, p.30. In particular, reported productivity in the
solitude and connection conditions was nearly identical (Ms.24
and .04, SDs.97 and 1.02, respectively), t(94) 1. Whether this
null effect on productivity comes from people not being very
productive on the train to begin with or from redefining what it
means to be productive depending on their experimental condition
is unclear. What is clear is that participants who talked to a
stranger did not leave the train feeling that a potentially productive
commute was wasted.
Conversation characteristics. Additional survey results showed
that participants in the connection condition talked for an average of
14.2 min (SD 11.3 min), had quite pleasant conversations
overall (M5.1, SD 0.9), and reported a positive impression
of their partner (M2.2, SD 1.1). Positivity of the commute
did not significantly correlate with either the reported pleasant-
ness of conversation or the impression of the conversation
partner (ps.10). Of course, these null effects do not mean
that the quality of a conversation is not related to a person’s
evaluation of his or her experience in that conversation. Instead,
none of our participants reported having a truly negative con-
2
The effect of talking compared to sitting in solitude on positivity of
commute experience remained significant after including the eight com-
muters who did not follow instructions, t(102) 2.11, p.04. It also
remained significant after controlling for normal train activities in a linear
regression (␤⫽0.24, p.05).
Figure 1. Actual and predicted positivity (top panel) and productivity
(bottom panel) from Experiments 1a, 1b, 2a, and 2b. Error bars represent
the standard error around the mean of each condition.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
4EPLEY AND SCHROEDER
versation with a truly unpleasant person, and hence, there was
little variability in participants’ evaluations to detect a relation-
ship with their resulting evaluations of the experience. We
suspect these generally positive conversations are not an acci-
dent but rather reflect a general feature of short-term interac-
tions that we discuss further in the General Discussion. Length
of the conversation, however, did correlate with positivity (r
.43, p.03). People liked their conversation partners, had
pleasant conversations, and had more positive commutes the
longer their conversations lasted.
Personality. The Appendix shows the correlations between
participants’ personality as measured by the TIPI and positivity
of commute in each of the experimental conditions. There are
few significant correlations with our experimental conditions.
Perhaps more important, the difference in positivity between
the connection and solitude conditions remained significant
even after controlling for the Big Five personality factors in a
linear regression (␤⫽0.31, p.01), and the difference
between these experimental conditions did not interact with any
of the personality factors (ps.10). As is often the case, strong
situational manipulations tend to influence people similarly,
regardless of their personality type (Bem & Allen, 1974; Mis-
chel, 1973).
Experiment 1b. Sixty-six commuters (M
age
44 years,
SD
age
13 years, 66% female) completed the survey during their
train ride and mailed it back (a 63% response rate).
Predicted experiences. If commuters in Experiment 1a had a
more positive experience connecting with a stranger, then why do
people in such circumstances so rarely do so? Experiment 1b
suggests an answer. Commuters predicted precisely the opposite
pattern of the actual experiences observed in Experiment 1a, F(1,
64) 4.69, p.03,
2
0.07, not simply mispredicting the
magnitude of these experiences but mispredicting their relative
valence. In particular, Figure 1 shows that commuters predicted
that they would have a significantly less positive commute in the
connection condition than in both the solitude condition, t(64)
2.46, p.02, d0.41, and the control condition, t(64) 2.18,
p.03, d0.45. Compared to the actual experiences of partic-
ipants in these conditions, connecting with strangers is surprisingly
positive.
Not only did participants expect to have the least positive
commute in the connection condition, they also expected to
have the least productive commute F(1, 64) 20.69, p.01,
2
0.24. Participants predicted that they would have a sig-
nificantly less productive commute in the connection condition
than in either the solitude condition, t(64) 5.04, p.01, d
0.74, or the control condition, t(64) 2.72, p.01, d0.83.
Overall, connecting with a stranger was seen as being an
unpleasant and unproductive use of time, neither of which
appeared to be true among those who actually did so in Exper-
iment 1a.
Personality. The Appendix again shows the correlations be-
tween participants’ personality as measured by the TIPI and pos-
itivity of commute in each of the experimental conditions. Some
significant and intuitive correlations emerged among participants
in the connection condition. The predicted difference in positivity
between the connection and solitude conditions remained signifi-
cant even after controlling for the Big Five personality factors in a
repeated measures ANOVA, F(1, 55) 4.77, p.03, p
20.08.
The difference between these experimental conditions did not
interact with any of the personality factors (ps.05). We continue
testing for potential personality moderators in the following ex-
periments.
Discussion
Commuters on a train into downtown Chicago reported a sig-
nificantly more positive commute when they connected with a
stranger than when they sat in solitude, and yet they predicted
precisely the opposite pattern of experiences. This pattern of
results demonstrates a severe misunderstanding of the psycholog-
ical consequences of social engagement. This mistake is particu-
larly unfortunate for a person’s well-being given that commuting is
consistently reported to be one of the least pleasant experiences in
the average person’s day (e.g., Kahneman, Krueger, Schkade,
Schwarz, & Stone, 2004). This experiment suggests that a surpris-
ing antidote for an otherwise unpleasant experience could be
sitting very close by.
Experiments 2a and 2b: Buses
Experiments 1a and 1b provide an ecologically valid setting for
our predictions (particularly compared to a university laboratory
setting with undergraduate participants; Henrich, Heine, & Noren-
zayan, 2010), but any field setting readily calls to mind idiosyn-
cratic features that limit its generalizability. In addition, partici-
pants in Experiments 1a and 1b completed the questionnaires at the
end of their commute, at a time when we could not control the
participants’ context or ensure 100% response rates. We therefore
sought convergent evidence in Experiments 2a and 2b in a differ-
ent field experiment (public buses) with final surveys completed in
a laboratory.
Method
Experiment 2a procedure. Participants recruited from a lab-
oratory participant pool in downtown Chicago enrolled in the
experiment by completing an online survey. The survey explained
that they could only participate if they took public transportation to
the laboratory in the morning, that they would receive a phone call
on the morning of their commute to the laboratory, and that they
would be compensated $5 for their time. Eligible and interested
participants then signed an electronic consent form to enroll;
reported their age, gender, and ethnicity; and indicated the time
they preferred to be called and the phone number to call. Eighty-
seven participants answered the telephone when called on the
morning of the experiment.
When the experimenter called participants, she asked them to
follow the same control, connection, or solitude condition instruc-
tions from Experiment 1a (randomly assigned), modified slightly
to accommodate differences in the change of context. She subse-
quently asked participants to brainstorm how they could follow the
instructions, in an attempt to increase the odds that they would
actually do so. Once in the laboratory after their commute, partic-
ipants completed a computerized survey with the same questions
as Experiment 1a. Participants also indicated what mode of public
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
5
SOLITUDE
transportation they had used to get to the laboratory. They received
$5 for their participation.
3
Experiment 2b procedure. Sixty-one people who had previ-
ously completed studies at the same laboratory as Experiment 2a
participated in exchange for entry into a $30 Amazon.com gift card
raffle. Demographics collected from a prior survey revealed our
sample had a median age between 21 and 25 years and was 71%
female. The experimental survey, completed online by partici-
pants, described the procedure of the actual experience experiment
(Experiment 2a) as closely as possible. Participants imagined
following the instructions from the control condition and, subse-
quently, the connection and solitude conditions in counterbalanced
order. The design therefore included the same three conditions as
Experiment 2a but was within participants.
In the control condition, participants read,
You are participating in a study about commuting using public trans-
portation. Imagine that you will take public transportation to the
downtown research laboratory. Imagine further that a research assis-
tant will call you in the morning before your commute to give you the
instructions for the study. The research assistant gives you these
instructions: “Please commute to the laboratory and then complete a
short survey after your commute.” Imagine that you follow these
instructions and then complete a questionnaire at the end of your
commute.
The connection and solitude conditions contained the same infor-
mation except that instructions were added to match the actual
instructions given in Experiment 2a. Following the instructions for
each condition, participants predicted their mood (how happy and
how sad they would feel after a commute in that condition) and
how pleasant and productive their commute would be compared to
usual on the same rating scales described in Experiment 2a.
Unlike Experiment 2a, participants did not complete the TIPI,
and we did not ask about their normal commuting activities.
Results
Experiment 2a. Of 87 participants who answered their
phones, 75 commuted to the laboratory and completed the exper-
imental survey (an 86.2% response rate that did not vary by
experimental condition,
2
2.89, p.10). All participants
reported following instructions, yielding a final sample of 75
participants (M
age
27 years, SD
age
7 years, 49% female).
Experiences. As in Experiment 1a, connecting with a stranger
was not unpleasant. In fact, Figure 1 shows that participants in the
connection condition again reported the most positive experience
of our experimental conditions, F(2, 74) 4.09, p.02,
2
0.10. Participants in the connection condition reported a signifi-
cantly more positive experience than participants in the solitude
condition, t(72) 2.14, p.03, d0.56,
4
and also a more
positive experience than participants in the control condition,
t(72) 2.69, p.01, d0.91.
Again replicating the results of Experiment 1a, we found no
significant difference in the reported productivity of the commute,
F(2, 73) 0.47, p.63. Reported productivity in the solitude and
connection conditions was nearly identical (Ms.05 and .15,
SDs.88 and 1.39, respectively), t(72) 1.
Conversation characteristics. Additional survey results showed
that participants in the connection condition talked for an average of
9.8 min (SD 6.1 min), had a relatively pleasant conversation
(M4.2, SD 1.4), and reported a positive impression of their
partner (M1.9, SD 1.2). As in Experiment 1a, positivity of the
commute was not significantly correlated with the pleasantness of
the conversation or the impression of the partner but was positively
correlated with the length of the commute (r.44, p.03). The
longer participants connected with a stranger, the more positive
their commuting experience was.
Personality. The Appendix again shows the correlations be-
tween participants’ personality as measured by the TIPI and pos-
itivity of the commute in each of the experimental conditions. The
difference in positivity between the connection and solitude con-
ditions remained significant even after controlling for the Big Five
personality factors in a linear regression (␤⫽0.38, p.01). The
difference between these experimental conditions did not interact
with any of the personality factors (ps.10).
Experiment 2b.
Predictions. As in Experiment 1b, participants again predicted
having the most negative experience connecting with a stranger
and most positive experience sitting in solitude, precisely the
opposite pattern of experiences than we actually observed in
Experiment 2a, F(1, 59) 3.28, p.04,
2
0.05. As shown in
Figure 1, participants again expected to have a significantly less
positive experience in the connection condition than in the solitude
condition, t(60) 2.03, p.05, d0.37. Predictions in the
control condition fell roughly in between, not differing from the
connection condition, t(59) 0.89, p.38, but differing from
the solitude condition, t(59) 2.29, p.03.
Predictions for productivity followed a similar pattern as Ex-
periment 1b, F(1, 59) 9.68, p.01,
2
0.14, such that
participants again anticipated a less productive commute in the
connection condition than in both the solitude condition, t(60)
3.74, p.01, d0.66, and the control condition, t(60) 2.72,
p.01, d0.47.
3
To explore whether any effects from the morning commute might
affect participants’ evaluations of their entire day, we e-mailed them a link
to a final survey at approximately 7:00 p.m., promising entry into a raffle
to win an iPod Shuffle if they completed the survey. The overall response
rate was 92%, with no differences between experimental conditions (
2
1.11, p.10). This second survey first asked commuters to enter their
participant number and then asked if they had talked to someone (or
planned to talk to someone) on their commute home that night, with “yes”
or “no” response options. Commuters rated how their day was overall on
a scale ranging from 3(Very bad)to3(Very good) and how happy they
felt during the day, from 0 (Not at all happy)to6(Very happy). Finally,
they entered comments into a textbox if they had any. This survey allowed
us to examine whether any effects from the morning commute carried
through to color their entire day. Experimental condition affected the extent
to which participants talked to someone (or planned to talk to someone) on
their evening commute, F(2, 66) 3.67, p.03. Interestingly, partici-
pants in the solitude condition reported being directionally more likely to
talk to someone during their evening commute than those in the control
condition, t(66) 1.68, p.10, and more than those in the connection
condition, t(66) 2.68, p.01. Whether this is simply seeking a variety
of experiences or learning that sitting in solitude was not very pleasant is
unclear. There were no significant differences in participants’ evaluations
of their overall day by experimental condition (p.10). Having a more
pleasant commute in the morning did not have an equally large effect on
the rest of participants’ days.
4
The difference on the positivity of the commute between the connec-
tion and control conditions of commute experience remained significant
after controlling for normal commuting activities in a linear regression
(␤⫽0.29, p.03).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
6EPLEY AND SCHROEDER
Discussion
People riding on a public bus had a more positive experience
when they talked to a nearby stranger than when they sat in
isolation and yet predicted precisely the opposite. This again
suggests that people may routinely choose to sit in isolation
because they mistakenly think it will be more pleasant than talking
to a nearby stranger.
One possible alternative interpretation is that the pleasure of
connecting with a stranger in Experiments 1a and 1b comes not
from actual social connection but rather from simply violating
social norms. That is, people on both the commuter trains and
public buses almost never engage strangers in conversation, and
violating this norm may be what people enjoy rather than actually
connecting with a stranger.
We think two points argue strongly against this alternative.
First, many behaviors would violate social norms and would also
be unquestionably miserable experiences. For instance, one Inter-
net reader who commented on a news report of Experiment 1a
suggested that the experiment’s authors might want to tie them-
selves to the front of the train engine “if they like noise on the train
so much.” We did not feel compelled to test this norm violation
empirically. Riders could also travel naked, lie in the aisle, sing in
falsetto, or perform any number of other creative norm violations
that need no experiment to confirm that they would be more
miserable than sitting in quiet solitude. Violating norms is clearly
insufficient for a positive experience. Second, the large empirical
literature on conformity actually demonstrates that violating social
norms is a systematically negative experience that people try to
avoid by going along with whatever others are doing (Miller,
2006). If anything, the counternormative nature of talking to
strangers in the contexts we studied would seem to work against a
positive experience—perhaps leading to social censure from oth-
ers—rather than in favor of it.
Experiments 3a and 3b: Barriers to Engagement
or Bad Experiences?
Experiments 1 and 2 provide an answer to our first research
question: People may avoid connecting with strangers and choose
isolation instead because they misunderstand the consequences of
social interaction, not because connecting with strangers is actually
more negative than remaining isolated. Indeed, participants’ ex-
pectations about the consequences of social interaction were not
just mistaken about the magnitude of an emotional experience;
they were mistaken about the valence of the experience. Commut-
ers expected that isolation would be more pleasant than connec-
tion, when precisely the opposite was true.
These results, however, raise a second major question. If con-
necting with strangers is truly more pleasurable than sitting in
isolation, then why do people not learn this and then behave
differently? There are at least two plausible mechanisms for this
significant misunderstanding.
First, there may be some barrier that keeps people from con-
necting with strangers and thereby learning that their expectations
are mistaken. The most obvious barrier could come from the
existing social norms that discourage connecting with strangers
(see also Milgram & Sabini, 1978). These norms could evolve out
of a complete disinterest in connecting with strangers, but they
could also evolve even among highly social people who are
otherwise very interested in connecting with strangers. People may
feel like they are being polite by not intruding on another person,
fear being rejected when attempting to start a conversation, or feel
that they have little or nothing in common with a stranger. This
could create the perfect context for pluralistic ignorance (e.g.,
Prentice & Miller, 1993), whereby people believe that others are
less interested in connecting than they are themselves. If other
people’s silence around strangers is interpreted as disinterest rather
than as politeness, then attempting conversation would seem more
unpleasant than it would actually be (Miller & McFarland, 1991;
Vorauer & Ratner, 1996). Highly social animals could sit in the
company of strangers, all be interested in connecting with each
other, and yet misread others’ silence as disinterest and therefore
prefer solitude. This pluralistic ignorance, whereby people consis-
tently think others are less interested in connecting than they are
themselves, not only could make an attempted conversation seem
unpleasant but could also create a barrier to learning that one’s
expectations are mistaken.
Second, expectations can be based on memories of past expe-
riences (Wirtz, Kruger, Scollon, & Diener, 2003). Past negative
experiences of talking to strangers may be more memorable than
positive or even typical experiences (Hastie & Kumar, 1979). If a
prediction is biased by one’s memory of unusually negative inter-
actions (Morewedge, Gilbert, & Wilson, 2005) or biased by the
relative ease of imagining negative outcomes (Baumeister, Brat-
slavsky, Finkenauer, & Vohs, 2001), then the average actual
experience would be better than expected. By this account, it is not
a lack of experience connecting with strangers that creates mis-
taken expectations but rather learning the wrong lesson from past
experiences because of biases in imagination or memory.
We tested between these two possibilities—one of misreading
others’ behavior that creates a barrier to engagement and the other
of negativity bias in memory or imagination—in two surveys of
train and bus commuters (Experiments 3a and 3b) recruited from
the same populations as Experiments 1 and 2, respectively. To test
for a barrier to engagement, participants imagined trying to have a
conversation and then predicted others’ interest and willingness to
talk. If people misread others’ behavior as disinterest, then partic-
ipants will systematically think they are more interested in con-
necting than others are and also underestimate others’ willingness
to connect. To test whether negativity bias in memory leads to
mistaken expectations, participants were randomly assigned to
imagine having a positive conversation, a negative conversation,
or simply a conversation (the control condition). If expectations
are biased by memories for negative past experiences or the ease
of imagining negative experiences, then those who imagine simply
having a conversation will make predictions more similar to those
in the negative conversation condition than in the positive conver-
sation condition.
Method
Experiment 3a procedure. Research assistants recruited
commuters in the same manner as Experiment 1a. Eighty-six
commuters from the Homewood, Illinois, Metra station agreed to
participate. Those who agreed to participate received a stamped
and addressed envelope containing the experimental survey, along
with a banana as compensation.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
7
SOLITUDE
In the experimental survey completed during their commute,
participants first imagined following the connection condition in-
structions from Experiment 1a. They reported how interested they
thought the other person would be to talk to them and how
interested they would be to talk to the other person on a scale
ranging from 0 (Not at all interested)to6(Very interested), how
difficult it would be to start the conversation on a scale ranging
from 0 (Not at all difficult)to6(Very difficult), and then circled the
percentage of people they thought would be willing to talk to them
(from 0% to 100% in 10% increments).
On the second page, all participants were asked to imagine
having a conversation. This conversation varied by random
assignment to one of three between-participant conditions. Par-
ticipants randomly assigned to the control condition imagined
simply having a conversation for the duration of the ride.
Participants randomly assigned to the positive condition imag-
ined a positive interaction in which the conversation partner
was extremely interesting, and those assigned to the negative
condition imagined a conversation partner who was extremely
uninteresting. Participants then predicted how much they would
have in common with the other person on a scale ranging from
0(Very little in common)to6(A lot in common) and then
predicted how stimulating the conversation would be on a scale
ranging from 0 (Not at all stimulating)to6(Very stimulating).
Last, they completed the same questions from prior experiments
about their predicted happiness, sadness, pleasantness, and pro-
ductiveness of commute.
Experiment 3b procedure. Sixty-five people recruited from
the same subject pool as Experiment 2b participated in exchange
for entry into a $30 Amazon.com gift card raffle. Demographics
collected from a prior survey revealed our sample had a median
age between 21 and 25 years and was 77% female. The comput-
erized survey described the procedure of the actual experience
experiment (Experiment 2a) as closely as possible, following the
same procedure as Experiment 3a, modified where necessary to
match the change in context from trains to buses.
Results
Experiment 3a. Sixty-four commuters (M
age
38 years,
SD
age
13 years, 74% female, 74% response rate) completed the
survey during their train ride and mailed it back.
As shown in Table 1, participants reported being significantly
more interested in talking to others than they thought others were
in talking to them, t(63) 4.77, p.01, d0.44. Participants
also predicted that fewer than 47% of commuters would be willing
to engage in conversation and that it would be relatively difficult
to start the conversation (M4.38, SD 1.69). In reality, all who
returned our surveys reported talking with the person they at-
tempted to connect with, and not a single person reported being
rebuffed. The seven who reported being unable to talk in Experi-
ment 1a reported that nobody sat next to them. Commuters ap-
peared to think that talking to a stranger posed a meaningful risk
of social rejection. As far as we can tell, it posed no risk at all.
If predicted interactions are biased by memories of especially
negative interactions, then predictions in the control condition
should mimic those in the negative interaction condition. That
is, when people spontaneously think of having a conversation
with a stranger, the context that comes to mind should be closer
to an explicitly negative interaction than an explicitly positive
interaction. As shown in Table 1, they are not. Participants in
the negative condition predicted having less in common,
t(61) ⫽⫺5.08, p.01, d1.53; a less stimulating conver-
sation, t(61) ⫽⫺3.99, p.01, d1.18; and a less positive
experience, t(60) ⫽⫺2.86, p.01, d0.93, than participants
in the control condition. In contrast, participants in the positive
condition predicted having more in common, t(61) 2.41, p
.02, d0.81; a marginally more stimulating conversation,
t(61) 1.85, p.07, d0.57; a less productive commute,
t(60) 2.35, p.02, d0.82; but not a significantly more
positive experience, t(60) 0.64, p.10, than did participants
in the control condition. There was no systematic evidence that
control condition predictions were negatively biased.
Table 1
Estimates of Own and Others’ Sociality and Predicted Positivity and Productivity Following an
Imagined Positive, Negative, or Control Conversation (Experiments 3a and 3b)
Measure
Experiment
3a: Train 3b: Bus
Perceived sociality
Your interest in talking 3.11 (1.70) p.01 3.18 (1.65) p.01
Others’ interest in talking 2.44 (1.34) 2.55 (1.48)
Percentage willing to talk 46.4 44.6
Predicted conversation
Positivity
Positive conversation 0.35 (0.87) p.52 0.55 (0.64) p.12
Control conversation p.01 0.20 (0.58) p.01 0.19 (0.54)
Negative conversation 0.47 (0.58) 0.67 (0.94)
Productivity
Positive conversation 1.26 (1.33) p.02 0.14 (1.96) p.78
Control conversation p.47 0.33 (0.91) p.17 0.00 (1.27)
Negative conversation 0.61 (1.44) 0.70 (1.64)
Note. Values are means, with standard deviations in parentheses. The pvalues indicate the statistical
significance of pairwise comparisons between conditions. Evaluations of interest in talking (both yours and
others’) are shown on the same scale ranging from 0 to 6, with more positive numbers indicating more interest
in talking.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
8EPLEY AND SCHROEDER
Experiment 3b. Once again, participants reported being sig-
nificantly more interested in talking to others than they thought
others were in talking to them, t(64) 3.21, p.01, d0.41;
predicted fewer than 45% of commuters would be willing to talk
with them; and predicted it would be difficult to start a conversa-
tion (M4.03, SD 1.8). Again, all participants who arrived to
the laboratory in Experiment 2a reported talking to the person they
attempted to connect with. Participants predicted that whether a
stranger would agree to talk or not was roughly the same as a coin
flip. Again, as far as we can tell, it is closer to a sure thing.
As shown in Table 1, the control condition predictions again
showed no evidence of being negatively biased. Participants in
the negative condition imagined having less in common,
t(62) ⫽⫺5.35, p.01, d1.69; a less stimulating conver-
sation, t(62) ⫽⫺5.33, p.01, d1.55; and a less positive
experience, t(62) ⫽⫺3.89, p.01, d1.13, than participants
in the control condition. No significant differences emerged in
evaluation between the positive and control conditions.
Discussion
These results suggest that misunderstanding the consequences
of social connection comes, at least in part, from barriers to
engaging in conversation rather than from biased memory for past
conversations or biased imagination for conversations once en-
gaged. Solitude seems preferable to connecting with a stranger, it
appears, because people interpret others’ actions as signs of dis-
interest and therefore do not engage in the very conversations that
would correct their expectations.
This mechanism suggests that misunderstanding the conse-
quences of social interaction comes from inexperience connecting
with strangers rather than from negative experiences doing so. This
further implies an important moderator of both the misunderstand-
ing of social connection and the preference for isolation observed
in Experiments 1 and 2. In particular, it implies that those who are
more likely to connect with strangers in conversation, or those in
contexts that more readily enable such connections, should obtain
the experience necessary to calibrate one’s expectations. The mis-
taken preference for solitude we observed earlier should therefore
emerge among those who rarely engage strangers in conversation
or in contexts (like those of Experiments 1 and 2) that create
barriers to engaging strangers in conversation. We tested this idea
in Experiments 4a and 4b by studying a context that more naturally
enables conversations between strangers: taxicabs.
Experiments 4a and 4b: Calibrating
Through Experience
A person may sit on a train, bus, or plane or in a waiting room
without engaging a stranger in conversation because there is no
explicit need to do so. This is not true in a taxicab, where a person
must engage in at least some minimal conversation with the driver
at the beginning to state one’s destination. In addition, the private
nature of the cab, especially when traveling alone, is more likely to
encourage conversation with a driver than a more public setting
like a train, plane, bus, or large waiting room. Indeed, in the survey
we described in the introduction, 49% of respondents said they
would be likely to talk to a stranger (presumably the driver) in a
cab. This stands in stark contrast to only 23% of people in
Experiment 1a who said they normally talk to strangers on the train
and 0% of people in the control condition who actually reported
talking to a stranger. If experience calibrates expectations, then
those who routinely talk to their drivers should make more cali-
brated predictions than those who rarely talk. Those who rarely
talk should make the same mispredictions about the experience of
connecting with a stranger as in Experiments 1b and 2b.
Method
Experiment 4a procedure. Two research assistants recruited
travelers waiting in the taxicab line to leave the Chicago Midway
International Airport. Only people traveling alone were asked to
participate. One hundred fifty travelers agreed to participate in
exchange for a candy bar. When travelers signed the consent form,
they also reported how tired they felt on a scale from 0 (Not at all
tired)to6(Very tired).
Research assistants randomly assigned participants to the con-
trol, connection, or solitude conditions. The corresponding instruc-
tions were identical to Experiment 1a except that participants in
the connection condition were asked to talk to their cab driver
instead of talk to someone else on the train. Travelers received the
experimental survey and the TIPI in a stamped and addressed
envelope. They were asked to follow the instructions (to talk, sit in
solitude, or have a normal ride) during the cab ride and then open
the envelope and complete the survey at the end of their cab ride
and mail it back. Travelers were also told that they would be
entered into a $30 gift card lottery if they returned their survey.
The experimental survey was identical to Experiment 1a except
for four initial questions asking participants about the ride (its
length and purpose), where they live (in Chicago or visiting, and
zip code), and a final question asking how tired they felt (on the
same scale as on the consent form). Additionally, the list of normal
cab ride activities was slightly modified from our previous exper-
iments: Instead of asking whether or not participants normally talk
to “someone,” we asked whether they talk to their driver as well as
whether they talk to someone else in the cab. The TIPI was always
included after the experimental survey in the envelope.
Experiment 4b procedure. Research assistants recruited
travelers in the same manner as Experiment 4a. Seventy travelers
agreed to participate and received a stamped and addressed enve-
lope containing the TIPI and the experimental survey in randomly
determined order, along with a candy bar as compensation.
The experimental survey described the procedure of the actual
experience experiment (Experiment 4a) as closely as possible.
Participants imagined following the instructions from the control
condition and, subsequently, the connection and solitude condi-
tions in counterbalanced order. Therefore, the design included the
same three conditions as Experiment 4a but manipulated within
participants.
In the control condition, participants read,
We would like you to imagine that you are participating in a study
about commuting in cabs. Imagine that, as you leave the airport to
catch your cab, you see a student standing near the cab stand who asks
you to participate in a study. Imagine that, just like you did today, you
agree to participate and sign a consent form. Imagine that the student
gives you these instructions for the study: “Please do not make any
changes to what you would normally do during your cab ride today.
Your goal is to do as you normally would when you are in a cab.”
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
9
SOLITUDE
Imagine that you follow these instructions and then complete a ques-
tionnaire at the end of your commute.
The connection and solitude conditions contained the same infor-
mation except that the instructions were changed to match the
actual instructions given in Experiment 4a. Following the instruc-
tions for each condition, participants predicted their mood (how
happy and how sad they would feel after a ride in that condition)
and how pleasant and productive their ride would be compared to
usual on the same rating scales described in Experiment 4a. At the
end of the experimental survey, participants marked the activities
they normally do on their commute from the same list of eight
activities in the Experiment 4a survey.
Results
Experiment 4a. One hundred four travelers returned a com-
pleted survey (a 69.3% response rate, with no difference by ex-
perimental condition,
2
3.64, p.16). Of those, 11 reported
being unable to follow instructions, all of whom were in the
solitude condition. This meant that the rate of following instruc-
tions varied by experimental condition (
2
21.36, p.01). Our
final sample of people who returned a survey and followed in-
structions was therefore 93 travelers (M
age
39 years, SD
age
13
years, 49% female).
We expected that participants would be more likely to report
normally talking to strangers—namely, the driver—in cabs than
we observed in Experiments 1a–3b. Indeed, unlike the prior ex-
periments where talking to strangers was unusual, a majority of
participants (65%) checked the box indicating that they normally
talked to their driver.
Experiences. As before, a one-way ANOVA showed that re-
ported positivity of the commute experience varied across the three
experimental conditions, F(2, 92) 6.84, p.01,
2
0.13. As
shown in Figure 2, commuters has a significantly more positive
experience in the connection condition than in the solitude condi-
tion, t(90) 3.58, p.01, d0.93.
5
In the control condition,
participants had a significantly more positive commute than in the
solitude condition, t(90) 2.75, p.01, d0.58, but did not
differ significantly from the connection condition, t(90) 1.01,
p.32. There was no interaction with whether participants
indicated routinely talking to their driver or not, F(2, 86) 1.23,
p.29. If anything, the result was directionally larger for loners
than for talkers.
There were no differences in reported productivity between any
of the three experimental conditions, F(2, 91) 1.14, p.32.
Reported productivity in the solitude and connection conditions
was nearly identical (Ms⫽⫺.11 and .03, SDs.97 and .57,
respectively), t(90) 1, with no interaction by whether partici-
pants indicated routinely talking to their driver or not.
Conversation characteristics. Additional survey results showed
that participants in the connection condition talked for an average
of 19 min (SD 10.3 min), had relatively pleasant conversations
(M4.8, SD 1.0), and had a relatively positive impression of
their driver (M1.9, SD 1.2). Unlike Experiments 1a and 2a,
positivity of the commute was correlated with both the pleasant-
ness of the conversation (r.60, p.01) and their impression of
the driver (r.55, p.01). Positivity of the commute was also
positively correlated with the length of the conversation, but only
marginally so (r.31, p.10).
Personality. The Appendix again shows the correlations be-
tween participants’ personality as measured by the TIPI and re-
ported positivity in each of the experimental conditions. The
difference in positivity between the connection and solitude con-
ditions remained significant even after controlling for the Big Five
personality factors in a linear regression (␤⫽0.42, p.01). The
difference between these experimental conditions did not interact
with any of the personality factors (ps.10).
Additional analysis. To assess another potential cost of con-
versation compared to isolation, we asked participants to report
how tired they were. In particular, conversation may be relatively
exhausting and isolation relatively energizing. We found no sup-
port for this possibility, as there were no significant differences in
reported tiredness between the connection and solitude conditions
either before the commute (Ms3.13 and 2.70, SDs1.38 and
1.35, respectively), after the commute (Ms3.10 and 3.07, SDs
5
The effect of talking compared to sitting in solitude on positivity of
commute experience remained significant after controlling for normal cab
activities in a linear regression (␤⫽0.40, p.01).
Figure 2. Actual and predicted positivity (top panel) and productivity
(bottom panel) from Experiments 4a and 4b. Error bars represent the
standard error around the mean of each condition.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
10 EPLEY AND SCHROEDER
1.69 and 1.49, respectively), or in the difference between them,
t(55) 1.22, p.10. Connecting with others does not appear to
be especially tiring, nor does solitude in this context appear par-
ticularly energizing.
Experiment 4b. Forty three travelers (M
age
38 years,
SD
age
13 years, 52% female) completed the survey during their
cab ride and mailed it back (a 61% response rate).
Predicted experiences. Unlike prior experiments, cab riders
predicted no overall difference in positivity between the three
experimental conditions, F(2, 82) 1.42, p.25. This is con-
sistent with our hypothesis that misunderstanding the conse-
quences of social action would be reduced in contexts where
engaging with random strangers is more common. Indeed, 56% of
riders checked the box indicating they normally talked to their
drivers (talkers), whereas the remaining reported not normally
talking (loners). As shown in Figure 2, participants’ predictions
were significantly moderated by whether they were talkers or
loners, F(2, 82) 18.31, p.01,
2
0.30. Consistent with
learning from experience, talkers predicted a significantly more
positive experience in the connection condition than in the solitude
condition, t(23) 5.32, p.01, d1.52. Consistent with failing
to learn because of inexperience, loners showed the same pattern
of predictions observed among train (Experiment 1b) and bus
(Experiment 2b) commuters. That is, loners predicted a less pos-
itive experience in the connection than in the solitude condition,
t(18) ⫽⫺2.35, p.03, d0.93. Predictions in the control
conditions matched what talkers and loners said they normally do
(i.e., connection and control conditions were similar for talkers,
whereas solitude and control conditions were similar for loners).
This stands in contrast to what we observed among the actual
experiences of participants in Experiment 4a, in which prior ex-
perience as a talker or loner did not affect the actual positivity of
the commute.
Cab riders also predicted no difference in productivity among
the three conditions, F(2, 82) 1.92, p.15, an effect that did
not differ between talkers or loners, F(2, 82) 0.80, p.45.
Personality. The Appendix again shows the correlations be-
tween participants’ personality as measured by the TIPI and pos-
itivity of ride in each of the experimental conditions. The predicted
difference in positivity between the connection and solitude con-
ditions remained nonsignificant after controlling for the Big Five
personality factors in a repeated measures ANOVA, F(1, 34)
0.01, p.10. More important, the interaction between experimen-
tal condition and whether participants normally talked to their
drivers remained significant after controlling for personality fac-
tors, F(1, 34) 22.07, p.01, p
20.39. The difference between
the experimental conditions did not interact with any of the per-
sonality factors (ps.10).
Discussion
When expectations and reality diverge, the gap can be explained
either by a failure of learning or a failure of imagination. Exper-
iments 4a and 4b clearly suggest that failing to understand the
consequences of social connection stem from a failure of learning.
The talkers who reported routinely engaging a stranger in conver-
sation—in this case, a cab driver—had the most calibrated pre-
dictions. They expected that connecting with the driver would
make the cab ride more pleasant than sitting quietly in solitude.
Talkers’ expectations were right: In Experiment 4a, those who
talked to their driver reported a more positive experience than
those who sat in solitude. In contrast, the loners who did not
normally talk with their driver showed the same pattern of misun-
derstanding observed in Experiments 1 and 2. The loners predicted
that they would have a more pleasant experience in isolation than
in conversation. Loners’ expectations were wrong: in fact, pre-
cisely the opposite of what these participants actually experienced
in Experiment 4a.
Cab riders’ predictions were guided by the amount of informa-
tion they had from past behavior. Their actual experience, how-
ever, was not. Cab riders enjoyed their commute more when they
talked to the drivers than when they sat in solitude regardless of
whether they normally talked to their driver or not.
Experiment 5: Is the Pleasure of
Connection Contagious?
Our experiments suggest that people fail to maximize their own
well-being because they mistakenly prefer isolation over connect-
ing with strangers. However, the broader implications of correct-
ing this misunderstanding are unclear for three reasons. First, we
could only sample participants who agreed to be in our experi-
ments, not their conversation partners. Talking to a stranger may
be surprisingly pleasant, but being talked to by a stranger could be
predictably miserable, meaning that the overall positivity of a
train, plane, cab, or waiting room of talkers could be no higher than
one full of loners. Second, all participants in our experiments were,
by scientific necessity, told to behave one way or another. For a
variety of reasons, participants may only get pleasure from a
conversation they were instructed to have—for instance, because
they achieved their (imposed) goal to talk, because they find it
easier to initiate a required conversation, or because they would
perceive themselves to be lonely when voluntarily initiating a
conversation where the norm is not to talk. Finally, all participants
in the preceding solitude conditions sat in isolation, but in close
proximity, to another participant. It is possible that this experience
of being “alone together” made the solitude conditions particularly
unpleasant. To test this possibility, Experiment 5 included a con-
trol condition in which participants simply were left in solitude,
alone.
To measure evaluations from both members of the conversation
dyad and also to make people feel that they had free choice to talk
(instead of feeling required), we created a situation resembling a
waiting room at, say, the dentist. Two participants had a 10-min
break in the middle of unrelated tasks in a laboratory experiment.
During this waiting period, we instructed one participant to either
talk to the other person in the room or to remain isolated and avoid
conversation. We gave the other participant no instructions.
Among those given instructions, some were told they did not have
to follow the instructions if they did not want to (although 79 of 80
pairs did so), thereby giving them some sense of free choice over
their behavior (e.g., Cooper, 1980). For other instructed partici-
pants, their free choice was not emphasized (as in the preceding
experiments). We then measured both participants’ evaluations of
the waiting period.
If being talked to is just as pleasant as initiating a conversation,
and if initiating a conversation is just as pleasant under a sense of
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
11
SOLITUDE
free choice as no choice, then explicit instruction or forced con-
versation is not what makes connecting with a stranger pleasant.
Method
Participants. One hundred seventy-nine people (20 pairs in
each of the four randomly assigned dyad conditions and 19 people
in the control condition) recruited to a laboratory in downtown
Chicago participated in exchange for $6 (M
age
32 years,
SD
age
13 years, 39.1% female).
Procedure. We recruited unacquainted participants in dyads
and randomly assigned each participant into one of nine possible
conditions in a 2 (participant: instructed vs. not instructed) 2
(condition: connection vs. solitude) 2 (instruction type: free
choice vs. no choice) experimental design, plus a single control
condition in which participants sat alone. The participant factor
was manipulated within dyad, whereas condition and instruction
type were manipulated between dyads.
The experimenter asked participants to leave their belongings in
the laboratory waiting room and then instructed them to complete
a consent form and a 44-item personality measure (the Big Five
Inventory; John, Donahue, & Kentle, 1991) in separate rooms. The
experimenter then randomly assigned one participant in each dyad
(or the sole control participant) to be in the uninstructed condition
and told this participant that there would be a short delay while
preparing materials for the second part of the experiment.
The experimenter told the other participant, in the instructed
condition, that there would be a short delay. The instructed con-
dition participants were randomly assigned to one of four possible
conditions: 2 (instruction type: free choice vs. no choice) 2
(condition: connection vs. solitude). In the free-choice solitude
condition, the experimenter told instructed participants,
Would you mind not talking at all to that other participant? Just leave
him or her alone for the entire time, and keep to yourself. Of course,
you don’t have to do this. If you really want to talk to the person, we
can’t stop you, but it would be great if you could choose to just keep
to yourself and not talk to that person. Again, not talking to the other
participant is optional for the study. It is your choice.
In the no-choice solitude condition, the experimenter told in-
structed participants,
Please do not talk to that other participant. Just leave him or her alone
for the entire time, and keep to yourself. It is important for you to keep
to yourself and not talk to that person. Not talking to the other
participant is a mandatory requirement for this study.
In the free-choice connection condition, the experimenter told
instructed participants,
Would you mind chatting for a while with that other participant? Just
make small talk with him or her while you wait. Of course, you don’t
have to do this. If you really want to sit in silence, we can’t stop you,
but it would be great if you could choose to engage the other person
in conversation for as long as I’m gone. Again, talking to the other
participant is optional for this study. It is your choice.
In the no-choice connection condition, the experimenter told in-
structed participants,
Please chat for a while with that other participant. Just make small talk
with him or her while you wait. It is important for you to engage the
other person in conversation for as long as I’m gone. Talking to the
other participant is a mandatory requirement for this study.
The experimenter then led both participants into the waiting
room and left them in the room along with the uninstructed
participants. After 10 min, participants were separated into indi-
vidual cubicles to complete the experimental survey about their
waiting experience.
In order to ensure compliance for each step of this relatively
long (30-min) experiment, participants completed and signed a
Decision Sheet for each component of the study: before the per-
sonality survey, before their wait experience, and before their final
experimental survey. Decision Sheets also clarified whether the
instructions were mandatory or optional, consistent with partici-
pants’ experimental condition, to reinforce the free-choice manip-
ulation.
Materials. The experimental survey first asked participants to
mark all of the things they did during their break from the follow-
ing list: talked to someone, talked on the phone, read a book or
newspaper, slept, thought (by yourself), worked on a computer,
worked on a phone, or other (with a space to write the activity).
Participants then rated how pleasant and productive their break
was on a scale ranging from 0 (Not pleasant/productive at all)to
6(Very pleasant/productive) and how happy and sad they felt after
the break from 0 (Not at all happy/sad)to6(Very happy/sad).
The second page of the survey included secondary measures to
be completed only by participants who spoke to the other partic-
ipant during the break. We first asked participants to “write as
much as you can remember about the person with whom you spoke
(name, ethnicity, age, occupation, etc.) and the conversation you
had.” Then we asked participants to estimate how long they talked
in minutes, whether or not they knew the other person before the
experiment, how pleasant their conversation with the person was
on a scale ranging from 0 (Not at all pleasant)to6(Very pleasant),
and what their overall impression of the person was on a scale
ranging from 3(Very negative)to3(Very positive).
The final page of the survey included a manipulation check that
asked participants if the experimenter gave them instructions on
how to behave prior to their wait. If participants reported they had
received instructions, it then asked them to rate how much control
they had in following the instructions from 1 (I had no choice but
to follow instructions—they seemed entirely mandatory)to4(It
was completely my choice to follow instructions—they seemed
entirely optional). As an exploratory measure, all participants then
estimated how long the break was (in minutes) and how long it felt
on a scale ranging from 1 (Time dragged)to7(Time flew). Finally,
participants reported their demographics (gender, age, ethnicity,
education level, and whether English was their first language).
Results and Discussion
One pair of participants in the solitude free-choice condition did
not follow experimental instructions and both reported talking to
each other during the session. The following analyses are not
meaningfully affected by either including or excluding this one
group. We have included them here to be conservative, but note
that the hypothesized effects get slightly stronger when this pair is
removed. Degrees of freedom vary slightly because some partici-
pants skipped items on the final survey.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
12 EPLEY AND SCHROEDER
Manipulation check. Even though 79 of 80 pairs of partici-
pants in the following analyses followed the experimenter’s in-
structions, participants in the free-choice condition reported feel-
ing that they had more free choice in whether to talk to their
partner or not during the waiting period (M3.34, SD 0.90)
than participants in the no-choice condition (M1.85, SD
1.05), F(1, 62) 37.16, p.01, p
20.38.
Positivity. As shown in Figure 3, participants had a more
positive experience when they connected with each other than
when they avoided each other, regardless of the other experimental
conditions. A 2 (participant: instructed vs. not instructed) 2
(condition: connection vs. solitude) 2 (instruction type: free
choice vs. no choice) mixed-model ANOVA on reported positivity
revealed only a significant main effect for condition, F(1, 78)
23.63, p.01, p
20.24. None of the other main effects or
interactions approached significance, Fs(1,76) 1.05. Apparently,
being talked to by a stranger is every bit as positive as talking to
one, regardless of whether the initiator has a sense of free choice
or not.
If anything, the difference between the connection and solitude
conditions was somewhat stronger among instructed participants
in the free-choice condition than in the no-choice condition, F(1,
76) 3.16, p.08. Having no choice to either connect or sit in
solitude does not seem to be a likely explanation for the results of
our preceding experiments.
6
Participants in the control condition, who sat in the room alone,
reported an experience (M⫽⫺0.18, SD 0.74) that was statis-
tically indistinguishable from uninstructed participants in the sol-
itude condition (M⫽⫺0.25, SD 0.81), t(96) ⫽⫺0.31, p.76,
but that was significantly less positive than uninstructed partici-
pants in the connection condition (M0.33, SD 0.65), t(96)
2.51, p.01, d0.51. Sitting in solitude for 10 min is signifi-
cantly less pleasant than engaging a stranger in conversation,
regardless of whether one is sitting alone or in close proximity to
a stranger. Being alone together does not seem to explain the
results of the solitude condition in the preceding experiments.
Productivity. Unlike the preceding experiments, participants
in Experiment 5 also reported having a more productive experi-
ence when they connected than when they sat in isolation. A 2
(participant: instructed vs. not instructed) 2 (condition: connec-
tion vs. solitude) 2 (instruction type: free choice vs. no choice)
mixed-model ANOVA on reported productivity revealed a main
effect of condition, F(1, 76) 8.61, p.01, p
20.10. There was
no main effect of participant and no interactions, Fs(1, 78) 1.56,
except for a marginal interaction of experimental condition and
instruction type, F(1, 76) 2.79, p.099, p
20.04. Decom-
posing the interaction, participants in the no-choice condition
showed a significantly smaller effect of condition (M
connection
3.05, SD 2.91, vs. M
solitude
2.68, SD 2.91) than participants
in the free-choice condition (M
connection
4.05, SD 2.88, vs.
M
solitude
2.45, SD 2.88), Fs(1, 38) 2.55 and 6.63, ps.12
and .01, p
2s0.06 and 0.15. There was also a marginal main
effect of instruction type, F(1, 76) 2.79, p.099, p
20.04,
such that participants in the no-choice condition felt more produc-
tive than those in the free-choice condition (Ms3.34 vs. 2.78,
SDs2.13 vs. 2.13). Finally, participants in the control condition,
who sat in the room alone, reported the same level of productivity
as participants in the connection and solitude conditions, F(2,
96) 0.58, p.56.
Participants felt somewhat more productive when talking than
when sitting in solitude, but this was strongest for participants
instructed to talk. This suggests that the sense of productivity in
this experiment comes partly from following the experimenter’s
instructions. Consistent with the previous experiments, we do not
conclude that connecting with a stranger is systemically perceived
as productive but rather that, for whatever reason, it is not sys-
tematically perceived as an unproductive waste of time.
Personality. The Appendix shows the correlations between
participants’ personalities as measured by the Big Five Inventory
and positivity of their waiting experiences in each of the experi-
mental conditions. As in the preceding experiments, there is no
consistent pattern of moderation. We report them for the sake of
completeness. The difference in positivity between the connection
and solitude conditions remained significant even after controlling
for the Big Five personality factors in linear regressions for both
the instructed (␤⫽0.47, p.01) and uninstructed participants
(␤⫽0.32, p.01).
7
The results affirm the positive consequences of distant social
connection observed in the preceding experiments (see also Sand-
strom & Dunn, 2014). Connecting with a stranger in a laboratory
context was more pleasant than sitting in solitude for those who
initiated the conversation as well as for those who were initiated
into it, both for those who initiated conversation with a sense of
free choice and for those given no sense of free choice. Solitude,
6
We note that Experiment 5 is a more elaborate version of another
experiment we conducted that follows a very similar procedure but in-
cludes only the no-choice condition. This simpler experiment replicates the
main effects in the no-choice condition of Experiment 5, bolstering support
for the reliability of this result. In this experiment, participants (N88)
again experienced a 10-min wait in a laboratory. One participant was
explicitly told either to connect with the other participant or to sit in
solitude. The other (uninstructed) participant was told nothing. All depen-
dent measures were identical to Experiment 5. Results again demonstrated
that participants had a more positive experience when they connected with
each other (M0.47, SD 0.68) than when they avoided each other
(M⫽⫺0.47, SD 0.90), regardless of whether they were the instructed
or uninstructed participant. A 2 (condition: connection vs. solitude) 2
(participant: instructed vs. uninstructed) ANOVA on reported positivity
revealed only a significant main effect for condition, F(1, 38) 28.09, p
.01,
2
0.43. The main effect of participant, F(1, 38) 1.42, p.24,
and the overall interaction, F(1, 38) 0.32, p.58, was nonsignificant.
Again, being talked to by a stranger is every bit as positive as talking to
one. Participants in this simpler experiment also reported having a more
productive experience when they connected (M2.98, SD 1.70) than
when they sat in isolation (M2.11, SD 1.77). A 2 (condition:
connection vs. solitude) 2 (participant: instructed vs. uninstructed)
ANOVA on reported productivity revealed only a significant main effect of
condition, F(1, 38) 6.80, p.01,
2
0.15. The main effect of
participant, F(1, 38) 0.00, p.95, and the overall interaction, F(1,
38) 0.86, p.36, were nonsignificant.
7
Estimated wait time: It is often said that time flies when you are having
fun. Consistent with this possibility, participants in the connection condi-
tion reported time flew more (M4.80, SD 1.92) than those in the
solitude condition (M3.95, SD 1.90), F(1, 73) 7.61, p.01, p
2
0.09. Whether participants were instructed or uninstructed and whether
their instructions involved free choice or no choice had no significant main
effect or interaction on their perception of time, Fs(1, 73) 0.57. Partic-
ipants’ estimations of their actual waiting time, however, was not affected
by any of the experimental conditions, Fs(1, 70) 1.43, except for the
instruction condition, F(1, 70) 8.57, p.01, p
20.11. Participants
who were uninstructed more accurately guessed the 10-min break time
(M9.69 min, SD 4.79), whereas instructed participants believed the
break was nonsignificantly shorter (M7.56 min, SD 4.23).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
13
SOLITUDE
in contrast, was relatively unpleasant in this context whether
experienced in the company of another person or not. The benefit
of social connection, or the cost of isolation, seems both conta-
gious and robust.
General Discussion
Aristotle famously argued that man is by nature a social animal,
but people in the company of strangers often look to be anything
but social. Instead of treating each other as possible sources of
well-being, strangers in close proximity often ignore each other
completely, treating each other more like objects than like fellow
social beings. For one of the most highly social species on the
planet, whose members benefit significantly from forming connec-
tions with other people, this seems paradoxical. Why would highly
social animals in the company of strangers so routinely ignore each
other?
A series of nine experiments conducted on trains, buses, and
taxicabs and in a laboratory suggests an answer: People misunder-
stand the consequences of social connection. On trains and buses,
participants in both Experiments 1b and 2b predicted that they
would have a more positive commute sitting in solitude than
connecting with a random stranger. These predictions are consis-
tent with the common behavior in these contexts, where conver-
sations between strangers are rare. Yet, consistent with the broad
benefits of social connection and the pain of social isolation,
participants in Experiments 1a and 2a who actually experienced
these two situations had precisely the opposite experience: They
had a more positive commute when they connected with a stranger
than when they sat in solitude. People seem to ignore strangers
because they mistakenly think that forming a connection with them
would be systematically unpleasant, whereas isolation would be
pleasurable. Humans may indeed be social animals but may not
always be social enough for their own well-being.
Of course, life is not always lived to maximize well-being.
People may therefore put off positive interactions because it comes
at a cost for some other goal. When commuting, a person may
Figure 3. Positivity (top panel) and productivity (bottom panel) from Experiment 5. Error bars represent the
standard error around the mean of each condition.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
14 EPLEY AND SCHROEDER
want to get work done. When riding in a cab after a long plane
flight, a little sleep might be rejuvenating. When sitting in a
waiting room, homework may need to get done. Indeed, in both
Experiments 1b and 2b, participants predicted that their socializing
would come at a cost to productivity, expecting they would have
a less productive commute if they connected with a stranger than
if they sat in solitude. Neither experiment, however, found that
participants who actually connected with a stranger reported hav-
ing a less productive commute when they connected with a
stranger than when they sat in solitude. Either people do not get as
much done on the train sitting alone as they expect to (Buehler,
Griffin, & Ross, 1994), or forming a new connection comes to be
defined as a reasonably productive use of time after having done it.
Likewise, cab riders in Experiment 4a did not report being any
more tired after connecting with the driver than when sitting in
solitude, and those sitting in a waiting room in Experiment 5
actually reported having a more productive time when they talked
than when they sat in isolation. These results do not mean that
social interaction is never distracting or draining, but they do mean
that the benefits our participants received from connecting with a
stranger in our experiments did not come with easily imagined
costs.
Such a dramatic disconnect between the actual and anticipated
impacts of connecting with a stranger is puzzling. If connecting
with a stranger is so much more pleasant than sitting in solitude,
then why do people expect precisely the opposite? At least part of
the answer comes from Experiments 3a and 3b. In order to have
beliefs that accurately reflect reality, a person must learn about
reality either directly or indirectly. Any barrier that keeps a person
from learning from reality could create mistaken beliefs.
Experiments 3a and 3b both suggest that one possible barrier is
a widely shared belief that other people are not interested in
connecting, a belief that could come directly from existing behav-
ioral norms. Just as in emergencies, where a concerned onlooker
who hesitates to intervene can conclude that other equally con-
cerned onlookers doing the very same thing are not actually
concerned at all (Miller & McFarland, 1987), so too do strangers
who follow the existing social norms of isolation seem to exhibit
pluralistic ignorance, which leaves the majority of people feeling
that others are not as interested in connecting as they are them-
selves. This belief serves as an obvious barrier to connecting with
a distant other, one that could keep people from gaining the very
experience they would need to accurately understand the hedonic
consequences of social interaction. Indeed, Experiments 4a and 4b
make this point directly. Cab riders had a more positive commute
when they talked to their driver than when they sat in isolation, an
effect completely anticipated by people who reported routinely
talking to their drivers but completely unanticipated by people who
reported rarely talking to their drivers. A preference for solitude in
the company of strangers seems to come at least partly from failing
to learn from experience.
This research broadly suggests that people could improve their
own momentary well-being—and that of others—by simply being
more social with strangers, trying to create connections where one
might otherwise choose isolation. Our experiments, however, con-
tained some necessary artifacts that may create concerns about
accepting this broader implication. In particular, participants in our
experiments were asked to engage others in conversation as part of
an experiment. Although Experiment 5 clearly showed that being
the recipient of connection is just as positive as being the instigator
of it, perhaps this feature served as an icebreaker that allowed
people to have a more pleasant conversation than they might have
had otherwise. Perhaps being able to start the conversation by
saying, “I’m in an experiment and was asked to talk,” enabled our
results?
Regrettably, we did not ask how participants began their con-
versations in the experiments, but this alternative did occur to us
afterward. We therefore e-mailed all of the participants from the
connection conditions of Experiments 1a, 2a, and 4a who gave us
valid e-mail addresses (n62) to ask if they could recall how they
started their conversation. This distant retrospection is obviously
less valid than asking people immediately after the experiment, but
we assumed that participating in the experiment would be mem-
orable enough for at least some people to give informative evi-
dence. The overall response rate of the participants we e-mailed
was 50% (n31). Of these, 16% reported being unable to
remember how they started their conversation. Of the remaining
who did remember how they started their conversation, 29% asked
about current events (e.g., “What do you think of the weather?”,
discussing an upcoming election), and 39% asked a personal
question (e.g., “How are you?”, “Where are you from?”, or com-
mented on the other person’s hair/clothes/shoes, etc.) The final
16% began in other idiosyncratic ways, such as talking about
themselves or saying “bless you” after a sneeze. No participant
reported starting the conversation by mentioning the experiment.
Instead, they seemed to do what most do to start a conversation—
they tried to say something worth responding to.
However, even if all of these participants are misremembering
their conversation or if all of the participants who did not respond
to our survey actually began their conversation by mentioning the
experiment (a possibility we find very implausible), this would not
invalidate our results. It would only show that what intuitively
seems to be a weak opening line might be a surprisingly good one.
Moderators, Likely and Unlikely
Our experiments demonstrate a consistent disconnect between
the anticipated and actual consequences of connecting with a
stranger. However, there are surely important moderators of either
the anticipated or actual experience of social connection that
would affect the disconnect we have documented. In fact, our
experiments suggest both a likely moderator and an unlikely
moderator.
A likely moderator, suggested by Experiments 3 and 4, is the
frequency with which people actually connect to strangers. Actual
contact with outgroup members tends to lead to more positive
attitudes about them (Pettigrew & Tropp, 2006), and actual expe-
rience connecting to distant others should lead to more calibrated
expectations about future interactions (as we observed among cab
riders who reported routinely talking to their drivers in Experiment
4a). Indeed, we suspect that the variance in existing cultural norms
about engaging strangers in conversation (Forgas & Bond, 1985;
Triandis, 1972) would moderate the magnitude of the disconnect
we observed. In cultures or contexts where connecting with strang-
ers is more routine and encouraged, we would expect calibration
between the expected and actual outcomes of distant social inter-
actions because people in these cultures and contexts learn about
the consequences of social interaction through their own experi-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
15
SOLITUDE
ence. But in cultures or contexts where connecting with strangers
is more actively discouraged, such as in the contexts we studied,
we would expect a mistaken preference for solitude simply be-
cause people have failed to learn that their expectations are wrong.
An unlikely moderator suggested by our experiments is a per-
son’s existing personality. We found no evidence that global
personality types, measured (in all but Experiment 2b) by a short
or long measure of the Big Five Personality Inventory, influenced
either predicted or actual experiences with social interaction in any
consistent way across our experiments. Instead, the actual conse-
quences of social interaction in Experiments 1a, 2a, 4a, and 5 did
not differ consistently by personality type. Personality did not
significantly moderate the actual effects of connecting with a
stranger, of sitting in isolation, or of the difference between the
two. This is consistent with a general finding that strong situational
contexts tend to affect people with differing personality types
relatively similarly (Mischel, 1973) and also with the more directly
related finding that people who are asked to behave extravertedly
in a social interaction report having a more pleasant experience
than those asked to behave introvertedly regardless of their trait
level of extraversion (Fleeson et al., 2002; McNiel & Fleeson,
2006; Zelenski et al., 2012). Pleasant social interactions seem
pleasant for most people, as isolation seems unpleasant for most
people.
Interestingly, two experiments (Experiments 1b and 4b) also
failed to find consistent personality effects on the predicted con-
sequences of connecting with strangers. These results may appear
somewhat discrepant with one published series of experiments in
which trait extraversion predicted how positively people expected
to feel in social situations where they were instructed to behave
either extravertedly (i.e., to act bold, talkative, energetic, active,
assertive, and adventurous) or introvertedly (Zelenski et al., 2013).
For instance, extraverts predicted that they would have a less
pleasant experience if they were instructed to act introverted than
if instructed to act extraverted while completing a jigsaw puzzle
with another person, getting to know another person, or planning
a day together in a group. Although Experiment 1b found direc-
tionally consistent results, with extraversion positively correlated
with predicted positivity of the connection experience, we did not
replicate the direction of this result in Experiment 4b, and the
relationship between extraversion and predicted positivity did not
vary between experimental conditions. Why did our extraverts not
think that sitting in isolation on the train would be unpleasant?
One possible answer is that our experiments were underpowered
to detect any stable personality moderators. However, because our
prediction experiments were all within participants involving rel-
atively large samples (ns66 and 43 in Experiments 1b and 4b,
respectively), these experiments would at least theoretically be
able to detect modest effects. Another possible answer is that we
used only a 10-item measure of the Big Five in our prediction
experiments (Gosling et al., 2003) rather than a more specific
measure of extraversion in Zelenski et al. (2013; Saucier, 1994).
Although the shorter version of the Big Five appears to be an
empirically valid measure of the longer and more elaborate version
(Gosling et al., 2003), it is certainly possible that two different
measures of the same thing yield different results because they are
not, in fact, measuring exactly the same thing.
However, we think our findings suggest a third, more interesting
possibility. Notice that extraversion is a much broader concept
than connecting with a stranger, one that operationally involves
talking but also involves behaving in a bold, adventurous, ener-
getic, active, and assertive fashion. These latter behaviors are
relatively self-focused and need not involve any interaction with
others. Predicting the consequences of extraversion may then focus
more one’s own actions than others’ reactions. In contrast, our
participants predicted the consequences of connecting with a
stranger, a prediction that Experiments 3a and 3b suggest is guided
by a judgment of how others are likely to respond (Dunn et al.,
2007; Mallett et al., 2008; Vorauer & Miller, 1997). Note also that
few would predict that connecting with a friend would be an
unpleasant experience, presumably because a friend would react
favorably compared to the uncertain (and potentially negative)
reactions of seemingly disinterested strangers. Extraversion, and
perhaps all of the individual difference measures, failed to predict
the anticipated consequences of social connection in any consistent
fashion, we believe, because the latter is guided by social judg-
ments as much as, or perhaps more so than, by self-assessments.
Experiment 4b is consistent with this possibility. Here, partici-
pants who learned about others’ reactions through their own per-
sonal experience made different and more calibrated predictions
than those who failed to learn from such experience, regardless of
their own trait extraversion (or any of the other Big Five traits).
Indeed, even measures of trait extraversion in that experiment did
not differ between those who reported normally talking to their
drivers versus those who reported normally remaining silent, either
in riders’ actual experiences, Experiment 4a, t(89) 0.60, ns, or
in riders’ predicted experiences, Experiment 4b, paired t(41)
0.47, ns. People may fail to connect with strangers because others
seem disinterested, rather than because it is inconsistent with one’s
own personality.
This suggests that people do not so much have a preference for
solitude in the presence of strangers as they do a fear of the
negative consequences that might come from attempted interac-
tions. We suspect this is why there are specific social functions and
customs intended to alleviate this fear, whether it is a cocktail
party, a social networking event, or a welcome reception—all
places where strangers come in contact but are known to be
interested in connecting. We also suspect this is why concepts like
icebreakers and pickup lines exist, customs intended to make it
easier to start a conversation, whereas ice-makers and drop-dead
lines that would end a conversation are relatively uncommon.
Removing the barrier to starting a conversation, rather than
trying to increase a person’s own trait extraversion, may there-
fore be the most effective way to encourage interactions with
distant strangers.
Extraversion, Social Connection, and Well-Being
Prior research suggests that acting extravertedly—that is, acting
bold, assertive, energetic, active, adventurous, and talkative (the
exact list has varied by study)—in laboratory experiments involv-
ing group tasks like solving jigsaw puzzles and planning a day
together generally leads to greater positive affect than acting
introvertedly—lethargic, passive, and quiet—in those same situa-
tions (Fleeson et al., 2002; Zelenski et al., 2013). Our studies
extend these findings in at least two major ways. First, our exper-
iments manipulated the goal of connecting with a stranger directly,
whereas acting extravertedly affects a wide variety of behaviors,
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
16 EPLEY AND SCHROEDER
not all of which require engagement with other people. Indeed, in
one experiment, the subcomponents of state extraversion signifi-
cantly correlated with 17 different goals, ranging from trying to
convey information (being talkative) to trying to have fun (being
spontaneous) to trying to strive hard for something (being bold) to
trying to be attractive or interesting (being energetic; McCabe &
Fleeson, 2012). Exactly why trait and state extraversion increases
well-being is unclear. By restricting our manipulation to one goal,
our experiments suggest that social connection is at least one
potentially important component of this more general result.
Second, asking participants to act introvertedly in a context in
which they must necessarily interact with other people (e.g., a
group decision task) creates a mismatch between their behavior
and the situation. Our studies do the opposite: The behavioral
norms in Experiments 1, 2, and 3 strongly pushed for remaining
disengaged from strangers, whereas we asked participants in some
conditions to talk to a stranger. This means that the behavior–
situation mismatch occurred in the connection conditions rather
than in the solitude conditions. This provides very strong evidence
that participants’ positive experience connecting with a stranger
does not come simply from an alignment between participants
behavior and the surrounding situation but that connecting with a
stranger is positive even when it is inconsistent with the prevailing
social norm.
Qualifications
Of course, our results do not demonstrate that all interactions
with strangers will be pleasant, any more than studies showing the
hedonic benefits of close family and friends demonstrate that all
family relationships are pleasant. Moreover, we suspect that there
are some important boundary conditions on the pleasures of con-
necting with strangers that our experiments were unable to test.
Our experiments tested interactions that lasted anywhere from a
few minutes to as long as 40 min (considerably longer than
existing laboratory experiments), but they did not require repeated
interactions or particularly long interactions with the same random
stranger. Nobody in the connection condition, for instance, spent
the weekend with a stranger on a train. Indeed, some research
suggests that liking for a stranger may peak at a relatively short
interaction and then decline over time as more is learned about
another person (Norton, Frost, & Ariely, 2007).
If, however, the amount of time spent in conversation with a
distant stranger is inversely related to its pleasantness at some
point along the time spectrum, then this only makes the results of
our experiments even more surprising. On trains and buses and in
waiting rooms, the duration of the conversation is relatively lim-
ited. These could be the kinds of brief social snacks (Gardner,
Pickett, & Knowles, 2005) with distant others that are maximally
pleasant and yet people still routinely avoid them. Understanding
how the anticipated length of an interaction affects the likelihood
of engaging someone in an interaction, compared to its actual
effects on the quality of an experience, is a potentially rich domain
for future research.
Likewise, our experiments examined the consequences of only
a single social connection. We did not test the consequences of
talking repeatedly to strangers over a longer period of time. It
could be people habituate to the pleasures of social engagement
quite quickly. Although possible, existing research on adaptation
suggests that the pleasures of connecting with strangers could be a
repeated source of well-being. In particular, people habituate more
quickly to identical repeated experiences than to subtly different
repeated experiences (Epstein, Caggiula, Rodefer, Wisneiwski, &
Mitchell, 1993; Redden, 2008; Rolls, Van Duijvenvoorde, &
Rolls, 1984; Schumann, Petty, & Clemons, 1990). The 10th iden-
tical chocolate will not be as delightful as the first, but the 10th
subtly different dessert could be every bit as delightful as the first.
Notice that conversations are more like subtly different desserts
than they are completely identical chocolates because each con-
versation with a new stranger is unique. We would therefore
predict that the 10th conversation with a stranger on a morning
commute would still be a significantly more pleasant experience
than the 10th foregone conversation. We think testing this hypoth-
esis in particular, and the consequences of repeated social engage-
ment more generally, is a very promising avenue for future re-
search.
Concluding Thought
Being civil toward distant others or random strangers is typi-
cally believed to benefit others—society at large or those who are
befriended. The results of our experiments, however, join a grow-
ing body of research suggesting positive consequences of proso-
ciality for oneself. Whether it is spending money on others versus
oneself (Dunn, Aknin, & Norton, 2008), behaving equitably rather
than selfishly (Condon & DeSteno, 2011), or expressing gratitude
versus disdain (Algoe, Gable, & Maisel, 2010), prosociality seems
not only to benefit others but also to benefit oneself. On an
increasingly crowded planet, misunderstanding the benefits of
social engagement could be increasingly problematic. At least in
this respect, the hedonist who seeks happiness and the idealist who
seeks civility should choose the same path.
References
Algoe, S. B., Gable, S. L., & Maisel, N. (2010). It’s the little things:
Everyday gratitude as a booster shot for romantic relationships. Personal
Relationships, 17, 217–233. doi:10.1111/j.1475-6811.2010.01273.x
Baumeister, R. F., Bratslavsky, E., Finkenauer, C., & Vohs, K. D. (2001).
Bad is stronger than good. Review of General Psychology, 5, 323–370.
doi:10.1037/1089-2680.5.4.323
Bem, D. J., & Allen, A. (1974). On predicting some of the people some of
the time: The search for cross-situational consistencies in behavior.
Psychological Review, 81, 506–520. doi:10.1037/h0037130
Buehler, R., Griffin, D., & Ross, M. (1994). Exploring the “planning
fallacy”: Why people underestimate their task completion times. Journal
of Personality and Social Psychology, 67, 366–381. doi:10.1037/0022-
3514.67.3.366
Coan, J. A., Schaefer, H. S., & Davidson, R. J. (2006). Lending a hand:
Social regulation of the neural response to threat. Psychological Science,
17, 1032–1039. doi:10.1111/j.1467-9280.2006.01832.x
Cohen, S. (2004). Social relationships and health. American Psychologist,
59, 676684. doi:10.1037/0003-066X.59.8.676
Condon, P., & DeSteno, D. (2011). Compassion for one reduces punish-
ment for another. Journal of Experimental Social Psychology, 47, 698
701. doi:10.1016/j.jesp.2010.11.016
Cooper, J. (1980). Reducing fears and increasing assertiveness: The role of
dissonance reduction. Journal of Experimental Social Psychology, 16,
199–213.
Cortes, B. P., Demoulin, S., Rodriguez, R. T., Rodriguez, A. P., & Leyens,
J. P. (2005). Infrahumanization or familiarity? Attribution of uniquely
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
17
SOLITUDE
human emotions to the self, the ingroup, and the outgroup. Personality
and Social Psychology Bulletin, 31, 243–253. doi:10.1177/
0146167204271421
Davis, M. H., Morris, M., & Kraus, L. A. (1998). Relationship-specific and
global perceptions of social support: Associations with well-being and
attachment. Journal of Personality and Social Psychology, 74, 468481.
doi:10.1037/0022-3514.74.2.468
Dunbar, R. I. M. (1998). The social brain hypothesis. Evolutionary An-
thropology, 6, 178–190. doi:10.1002/(SICI)1520-6505(1998)6:5178::
AID-EVAN53.0.CO;2-8
Dunn, E. W., Aknin, L. B., & Norton, M. I. (2008, March 21). Spending
money on others promotes happiness. Science, 319, 1687–1688. doi:
10.1126/science.1150952
Dunn, E. W., Biesanz, J. C., Human, L. J., & Finn, S. (2007). Misunder-
standing the affective consequences of everyday social interactions: The
hidden benefits of putting one’s best face forward. Journal of Person-
ality and Social Psychology, 92, 990–1005. doi:10.1037/0022-3514.92
.6.990
Epstein, L. H., Caggiula, A. R., Rodefer, J. S., Wisneiwski, L., & Mitchell,
S. L. (1993). The effects of calories and taste on habituation of the
human salivary response. Addictive Behaviors, 18, 179–185. doi:
10.1016/0306-4603(93)90048-E
Fleeson, W., Malanos, A. B., & Achille, N. M. (2002). An intraindividual
process approach to the relationship between extraversion and positive
affect: Is acting extraverted as “good” as being extraverted? Journal of
Personality and Social Psychology, 83, 1409–1422. doi:10.1037/0022-
3514.83.6.1409
Forgas, J. P., & Bond, M. H. (1985). Cultural influences on the perception
of interaction episodes. Personality and Social Psychology Bulletin, 11,
75–88. doi:10.1177/0146167285111007
Gardner, L., Pickett, C., & Knowles, M. (2005). Social snacking and
shielding: Using social symbols, selves, and surrogates in the service of
belonging needs. In K. D. Williams, J. P. Forgas, & W. von Hippel
(Eds.), The social outcast: Ostracism, social exclusion, rejection, and
bullying (pp. 227–242). New York, NY: Psychology Press.
Gilbert, D. T., Driver-Linn, E., & Wilson, T. D. (2002). The trouble with
Vronsky: Impact bias in the forecasting of future affective states. In L.
Feldman Barrett & P. Salovey (Eds.), The wisdom in feeling: Psycho-
logical processes in emotional intelligence (pp. 114–143). New York,
NY: Guilford Press.
Gosling, S. D., Rentfrow, P. J., & Swann, W. B. (2003). A very brief
measure of the Big-Five personality domains. Journal of Research in
Personality, 37, 504–528. doi:10.1016/S0092-6566(03)00046-1
Haslam, N. (2006). Dehumanization: An integrative review. Personality and
Social Psychology Review, 10, 252–264. doi:10.1207/s15327957pspr1003_4
Haslam, N., Bain, P., Douge, L., Lee, M., & Bastian, B. (2005). More
human than you: Attributing humanness to self and others. Journal of
Personality and Social Psychology, 89, 937–950. doi:10.1037/0022-
3514.89.6.937
Hastie, R., & Kumar, P. A. (1979). Person memory: The processing of
consistent and inconsistent person information. Journal of Personality
and Social Psychology, 37, 25–38. doi:10.1037/0022-3514.37.1.25
Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in
the world? Behavioral and Brain Sciences, 33, 61–83. doi:10.1017/
S0140525X0999152X
Herrmann, E., Call, J., Herna
`ndez-Lloreda, M. V., Hare, B., & Tomasello,
M. (2007, September 7). Humans have evolved specialized skills of
social cognition: The cultural intelligence hypothesis. Science, 317,
1360–1366. doi:10.1126/science.1146282
Holt-Lunstad, J., Smith, T. B., & Layton, J. B. (2010). Social relationships
and mortality risk: A meta-analytic review. PLoS Medicine, 7, Article
e1000316. doi:10.1371/journal.pmed.1000316
House, J. S., Landis, K. R., & Umberson, D. (1988, July 29). Social
relationships and health. Science, 241, 540–545. doi:10.1126/science
.3399889
Hsee, C. (2000). Attribute evaluability and its implications for joint-
separate evaluation reversals and beyond. In D. Kahneman & A. Tversky
(Eds.), Choices, values and frames (pp. 543–563). New York, NY:
Cambridge University Press.
John, O. P., Donahue, E. M., & Kentle, R. L. (1991). The Big Five
Inventory—Versions 4a and 54. Berkeley: University of California,
Berkeley, Institute of Personality and Social Research.
Kahneman, D., & Deaton, A. (2010). High income improves evaluation of
life but not emotional well-being. PNAS: Proceedings of the National
Academy of Sciences, USA, 107, 16489–16493.
Kahneman, D., Krueger, A. B., Schkade, D. A., Schwarz, N., & Stone,
A. A. (2004, December 3). A survey method for characterizing daily life
experience: The day reconstruction method. Science, 306, 1776–1780.
doi:10.1126/science.1103572
Keren, G. B., & Raaijmakers, J. G. W. (1988). On between-subjects versus
within-subjects comparisons in the study of choice behavior. Organiza-
tional Behavior and Human Decision Processes, 41, 233–247. doi:
10.1016/0749-5978(88)90028-3
King, K. B., & Reis, H. T. (2012). Marriage and long-term survival after
coronary artery bypass grafting. Health Psychology, 31, 55–62. doi:
10.1037/a0025061
Mallett, R. K., Wilson, T. D., & Gilbert, D. T. (2008). Expect the unex-
pected: Failure to anticipate similarities when predicting the quality of
an intergroup interaction. Journal of Personality and Social Psychology,
94, 265–277. doi:10.1037/0022-3514.94.2.94.2.265
McCabe, K. O., & Fleeson, W. (2012). What is extraversion for? Integrat-
ing trait and motivational perspectives and identifying the purpose of
extraversion. Psychological Science, 23, 1498–1505. doi:10.1177/
0956797612444904
McNiel, J. M., & Fleeson, W. (2006). The causal effects of extraversion on
positive affect and neuroticism on negative affect: Manipulating state
extraversion and state neuroticism in an experimental approach. Journal
of Research in Personality, 40, 529–550. doi:10.1016/j.jrp.2005.05.003
Milgram, S., & Sabini, J. (1978). On maintaining social norms: A field
experiment in the subway. In A. Baum, J. E. Singer, & S. Valins (Eds.),
Advances in environmental psychology: Vol. 1.The urban environment
(pp. 31–40). Hillsdale, NJ: Erlbaum.
Miller, D. T. (2006). An invitation to social psychology: Expressing and
censoring the self. Belmont, CA: Wadsworth-Thomson.
Miller, D. T., & McFarland, C. (1987). Pluralistic ignorance: When simi-
larity is interpreted as dissimilarity. Journal of Personality and Social
Psychology, 53, 298–305. doi:10.1037/0022-3514.53.2.298
Miller, D. T., & McFarland, C. (1991). When social comparison goes awry:
The case of pluralistic ignorance. In J. Suls & T. Wills (Eds.), Social
comparison: Contemporary theory and research (pp. 287–313). Hills-
dale, NJ: Erlbaum.
Mischel, W. (1973). Toward a cognitive social learning reconceptualiza-
tion of personality. Psychological Review, 80, 252–283. doi:10.1037/
h0035002
Morewedge, C. K., Gilbert, D. T., & Wilson, T. D. (2005). The least likely
of times: How remembering the past biases forecasts of the future.
Psychological Science, 16, 626630. doi:10.1111/j.1467-9280.2005
.01585.x
Myers, D. G., & Diener, E. (1995). Who is happy? Psychological Science,
6, 10–19. doi:10.1111/j.1467-9280.1995.tb00298.x
Norton, M. I., Frost, J. H., & Ariely, D. (2007). Less is more: The lure of
ambiguity, or why familiarity breeds contempt. Journal of Personality
and Social Psychology, 92, 97–105. doi:10.1037/0022-3514.92.1.97
Pettigrew, T. F., & Tropp, L. R. (2006). A meta-analytic test of intergroup
contact theory. Journal of Personality and Social Psychology, 90, 751–
783. doi:10.1037/0022-3514.90.5.751
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
18 EPLEY AND SCHROEDER
Pinquart, M., & Sorensen, S. (2003). Risk factors for loneliness in adult-
hood and old age: A meta-analysis. In S. P. Shohov (Ed.), Advances in
psychology research (Vol. 19, pp. 111–143). Hauppauge, NY: NOVA
Science Publishers.
Prentice, D. A., & Miller, D. T. (1993). Pluralistic ignorance and alcohol
use on campus: Some consequences of misperceiving the social norm.
Journal of Personality and Social Psychology, 64, 243–256. doi:
10.1037/0022-3514.64.2.243
Redden, J. P. (2008). Reducing satiation: The role of categorization level.
Journal of Consumer Research, 34, 624634, doi:10.1086/521898
Rolls, B. J., Van Duijvenvoorde, P. M., & Rolls, E. T. (1984). Pleasantness
changes and food intake in a varied four-course meal. Appetite, 5,
337–348. doi:10.1016/S0195-6663(84)80006-9
Sallet, J., Mars, R. B., Noonan, M. P., Andersson, J. L., O’Reilly, J. X.,
Jbabdi, S.,...Rushworth, M. F. S. (2011, November 4). Social network
size affects neural circuits in Macaques. Science, 334, 697–700. doi:
10.1126/science.1210027
Sandstrom, G., & Dunn, E. W. (2014). Is efficiency overrated? Minimal social
interactions lead to belonging and positive affect. Social Psychological &
Personality Science, 5, 437–442. doi:10.1177/1948550613502990
Saucier, G. (1994). Mini-Markers: A brief version of Goldberg’s unipolar
Big Five markers. Journal of Personality Assessment, 63, 506–516.
doi:10.1207/s15327752jpa6303_8
Schumann, D. W., Petty, R. E., & Clemons, D. S. (1990). Predicting the
effectiveness of different strategies of advertising variation: A test of the
repetition-variation hypothesis. Journal of Consumer Research, 17,
192–202. doi:10.1086/208549
Triandis, H. C. (1972). The analysis of subjective culture. New York, NY:
Wiley.
Vorauer, J. D., & Miller, D. T. (1997). Failure to detect the effect of
implicit social influence on the presentation of self. Journal of Person-
ality and Social Psychology, 73, 281–295. doi:10.1037/0022-3514.73.2
.281
Vorauer, J. D., & Ratner, R. K. (1996). Who’s going to make the first
move? Pluralistic ignorance as an impediment to relationship formation.
Journal of Social and Personal Relationships, 13, 483–506. doi:
10.1177/0265407596134001
Williams, J. G., & Solano, C. H. (1983). The social reality of feeling
lonely: Friendship and reciprocation. Personality and Social Psychology
Bulletin, 9, 237–242. doi:10.1177/0146167283092007
Wirtz, D., Kruger, J., Scollon, C. N., & Diener, E. (2003). What to do on
spring break? The role of predicted, on-line, and remembered experience
in future choice. Psychological Science, 14, 520–524. doi:10.1111/
1467-9280.03455
Zelenski, J. M., Santoro, M. S., & Whelan, D. C. (2012). Would introverts
be better off if they acted more like extraverts? Exploring emotional and
cognitive consequences of counterdispositional behavior. Emotion, 12,
290–303. doi:10.1037/a0025169
Zelenski, J. M., Whelan, D. C., Nealis, L. J., Besner, C. M., Santoro, M. S.,
& Wynn, J. E. (2013). Personality and affective forecasting: Trait
introverts underpredict the hedonic benefits of acting extraverted. Jour-
nal of Personality and Social Psychology, 104, 1092–1108. doi:10.1037/
a0032281
(Appendix follows)
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
19
SOLITUDE
Appendix
Correlations Between Reported Personality Traits (as Measured by the Ten Item Personality Inventory
or Big Five) and Positivity of Commute or Wait
Experimental conditions Extraversion Agreeableness Openness to
Experience Conscientiousness Emotional
Stability
Experiments 1a & 1b: Trains
Actual (Experiment 1a)
Control (n37) 0.208 0.063 0.200 0.211 0.278
Connection (n27) 0.080 0.116 0.278 0.299 0.227
Solitude (n32) 0.086 0.339 0.432
ⴱⴱ
0.017 0.428
ⴱⴱ
Predicted (Experiment 1b, n66)
Control 0.168 0.089 0.031 0.008 0.306
ⴱⴱ
Connection 0.259
0.141 0.425
ⴱⴱ
0.277
0.330
ⴱⴱ
Solitude 0.187 0.108 0.150 0.205 0.099
Experiment 2a: Buses
Actual (Experiment 2a)
Control (n24) 0.111 0.407
0.419
0.333 0.197
Connection (n26) 0.287 0.113 0.051 0.379 0.269
Solitude (n25) 0.323 0.140 0.306 0.532
ⴱⴱ
0.053
Experiments 4a & 4b: Cabs
Actual (Experiment 4a)
Control (n36) 0.066 0.166 0.381
0.180 0.019
Connection (n27) 0.127 0.176 0.030 0.089 0.241
Solitude (n30) 0.028 0.283 0.217 0.153 0.039
Predicted (Experiment 4b, n43)
Control 0.148 0.142 0.154 0.044 0.032
Connection 0.169 0.242 0.164 0.039 0.191
Solitude 0.027 0.206 0.220 0.116 0.004
Experiment 5: Is the pleasure of connection contagious?
Instructed
Connection–no choice (n20) 0.396 0.275 0.088 0.122 0.081
Connection–free choice (n20) 0.222 0.730
ⴱⴱ
0.213 0.374 0.525
Solitude–no choice (n20) 0.254 0.356 0.160 0.105 0.289
Solitude–free choice (n20) 0.396 0.275 0.088 0.122 0.081
Not instructed
Connection–no choice (n20) 0.217 0.406 0.142 0.047 0.337
Connection–free choice (n20) 0.295 0.781
ⴱⴱ
0.319 0.249 0.063
Solitude–no choice (n20) 0.120 0.256 0.006 0.583
ⴱⴱ
0.306
Solitude–free choice (n20) 0.628
ⴱⴱ
0.525
ⴱⴱ
0.082 0.309 0.625
ⴱⴱ
Control (n19) 0.030 0.219 0.470
0.290 0.254
p.05.
ⴱⴱ
p.01.
Received December 11, 2013
Revision received June 2, 2014
Accepted June 5, 2014
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
20 EPLEY AND SCHROEDER
... When it comes to strangers, people may be reluctant to have any sort of interaction, let alone a CIC. Many people report feeling uncomfortable or awkward at the thought of engaging in any kind of conversation with a stranger [21]. Despite this, when people are induced to actually have these dreaded conversations, at least in non-ideological domains, they find them to be more enjoyable than imagined [21,22]. ...
... Many people report feeling uncomfortable or awkward at the thought of engaging in any kind of conversation with a stranger [21]. Despite this, when people are induced to actually have these dreaded conversations, at least in non-ideological domains, they find them to be more enjoyable than imagined [21,22]. It is likely that similar, perhaps even greater, fears exist towards having a CIC with a stranger. ...
... These inaccurate forecasts demonstrate the reality that many people have misconceptions about their ideological opponents, highlighting the so-called 'perception gap'. It may be that these initially negative perceptions contribute to avoidance, creating barriers to engaging in CIC [20][21][22][23], and ultimately preventing these misperceptions from ever being corrected. Although future studies are needed to investigate the role these misperceptions play in discouraging CIC, one participant's comment after opting out of the study prior to the conversation lends support for this possibility: ...
Article
Full-text available
The rise of ideological polarization in the U.S. over the past few decades has come with an increase in hostility on both sides of the political aisle. Although communication and compromise are hallmarks of a functioning society, research has shown that people overestimate the negative affect they will experience when viewing oppositional media, and it is likely that negative forecasts lead many to avoid cross-ideological communication (CIC) altogether. Additionally, a growing ideological geographic divide and online extremism fueled by social media audiences make engaging in CIC more difficult than ever. Here, we demonstrate that online video-chat platforms (i.e., Zoom) can be used to promote effective CIC among ideologically polarized individuals, as well as to better study CIC in a controlled setting. Participants ( n = 122) had a face-to-face CIC over Zoom, either privately or publicly with a silent ingroup audience present. Participant forecasts about the interaction were largely inaccurate, with the actual conversation experience found to be more positive than anticipated. Additionally, the presence of an ingroup audience was associated with increased conflict. In both conditions, participants showed preliminary signs of attitude moderation, felt more favorable toward the outgroup, and felt more informed about the issue after the CIC. These results suggest that face-to-face CIC’s are generally positive and beneficial for polarized individuals, and that greater effects may be achieved through private conversations, as opposed to more public social media-like interactions. Future researchers studying ideological conflict may find success using similar Zoom paradigms to bring together ideologically diverse individuals in controlled lab settings.
... Engaging with others in conversation has well-documented hedonic consequences of increasing a person's mood (5-7), well-being (8,9), and liking for one's conversation partner (10). These hedonic consequences can be unexpectedly large: people tend to underestimate how much they'll enjoy their conversation (5), feel connected to their conversation partner (11), and be liked by their conversation partner (10). ...
Article
A meaningful amount of people's knowledge comes from their conversations with others. The amount people expect to learn predicts their interest in having a conversation (pretests 1 and 2), suggesting that the presumed information value of conversations guides decisions of whom to talk with. The results of seven experiments, however, suggest that people may systematically underestimate the informational benefit of conversation, creating a barrier to talking with-and hence learning from-others in daily life. Participants who were asked to talk with another person expected to learn significantly less from the conversation than they actually reported learning afterward, regardless of whether they had conversation prompts and whether they had the goal to learn (experiments 1 and 2). Undervaluing conversation does not stem from having systematically poor opinions of how much others know (experiment 3) but is instead related to the inherent uncertainty involved in conversation itself. Consequently, people underestimate learning to a lesser extent when uncertainty is reduced, as in a nonsocial context (surfing the web, experiment 4); when talking to an acquainted conversation partner (experiment 5); and after knowing the content of the conversation (experiment 6). Underestimating learning in conversation is distinct from underestimating other positive qualities in conversation, such as enjoyment (experiment 7). Misunderstanding how much can be learned in conversation could keep people from learning from others in daily life.
... Engaging with others in conversation has well-documented hedonic consequences of increasing a person's mood (5-7), well-being (8,9), and liking for one's conversation partner (10). These hedonic consequences can be unexpectedly large: people tend to underestimate how much they'll enjoy their conversation (5), feel connected to their conversation partner (11), and be liked by their conversation partner (10). ...
Preprint
A meaningful amount of people’s knowledge comes from their conversations with others. The amount people expect to learn predicts their interest in having a conversation (Pretests 1-2), suggesting that the presumed information value of conversations guides decisions of whom to talk with. The results of seven experiments, however, suggest that people may systematically underestimate the informational benefit of conversation, creating a barrier to talking with—and hence learning from—others more often in daily life. Participants who were asked to talk with another person expected to learn significantly less from the conversation than they actually reported learning afterwards, regardless of whether they had conversation prompts or not and whether they had the goal to learn or not (Exp 1-2). Undervaluing conversation does not stem from having systematically poor opinions of how much others know (Exp 3), but is instead related to the inherent uncertainty involved in conversation itself. Consequently, people underestimate learning to a lesser extent when uncertainty is reduced, as in a nonsocial context (surfing the web, Exp 4), when talking to an acquainted conversation partner (Exp 5), and after knowing the content of the conversation (Exp 6). Underestimating learning in conversation is distinct from underestimating other positive qualities in conversation, such as enjoyment (Exp 7). Misunderstanding how much can be learned in conversation could keep people from learning more from others in daily life.
... Further, people compare how they did in a conversation to their ideal of how the conversation could have gone, or to past conversations that went better, failing to realize that their conversation partner does not have access to these standards (Chambers et al., 2008). For all of these reasons, people often underestimate how well a conversation with a stranger will go (Epley & Schroeder, 2014). We suggest that because they have such a dim view of themselves as conversationalists, people may believe that the best strategy to be liked is to lay low and let the other person do most of the talking. ...
Article
Full-text available
We hypothesized that people would exhibit a reticence bias, the incorrect belief that they will be more likable if they speak less than half the time in a conversation with a stranger, as well as halo ignorance, the belief that their speaking time should depend on their goal (e.g., to be liked vs. to be found interesting), when in fact, perceivers form global impressions of each other. In Studies 1 and 2, participants forecasted they should speak less than half the time when trying to be liked, but significantly more when trying to be interesting. In Study 3, we tested the accuracy of these forecasts by randomly assigning participants to speak for 30%, 40%, 50%, 60%, or 70% of the time in a dyadic conversation. Contrary to people’s forecasts, they were more likable the more they spoke, and their partners formed global rather than differentiated impressions.
Article
Subjective well-being is characterized by relatively frequent positive emotions, relatively infrequent negative emotions, and high life satisfaction. Although myriad research topics related to subjective well-being have been explored – from how it should be measured to how it affects physical health – a key finding is that social connections are crucial. Researchers are therefore increasingly exploring whether subjective well-being can be improved through interventions that encourage specific types of social behaviors, including prosociality, gratitude, extraversion, and brief social interactions. We review this recent work, highlighting potential behavioral and psychological mechanisms underlying the effectiveness of such interventions, along with their boundary conditions.
Article
Performing acts of kindness increases well-being, yet people can be reluctant to ask for help that would enable others’ kindness. We suggest that people may be overly reluctant because of miscalibrated expectations about others’ prosocial motivation, underestimating how positively others will feel when asked for help. A pretest identified that interest in asking for help was correlated with expectations of how helpers would think and feel, but a series of scenarios, recalled experiences, and live interactions among adult participants in the United States (total N = 2,118) indicated that those needing help consistently underestimated others’ willingness to help, underestimated how positively helpers would feel, and overestimated how inconvenienced helpers would feel. These miscalibrated expectations stemmed from underestimating helpers’ prosocial motivation while overestimating compliance motivation. This research highlights a limitation of construing help-seeking through a lens of compliance by scholars and laypeople alike. Undervaluing prosociality could create a misplaced barrier to asking for help when needed.
Article
In everyday life, people often have opportunities to improve others’ lives, whether offering well-intentioned advice or complimenting someone on a job well done. These are opportunities to provide “prosocial input” (information intended to benefit others), including feedback, advice, compliments, and expressions of gratitude. Despite widespread evidence that giving prosocial input can improve the well-being of both givers and recipients, people sometimes hesitate to offer their input. The current paper documents when and why people fail to give prosocial input, noting that potential givers overestimate the costs of doing so (e.g., making recipients uncomfortable) and underestimate the benefits (e.g., being helpful) for at least four psychological reasons. Unfortunately, the reluctance to give prosocial input results in a short supply of kindness.
Article
Converging evidence shows that mindfulness is associated with various indicators of interpersonal behavior and well-being. Although promising, the effects of mindfulness should ultimately be expressed during interpersonal interactions and observed by interaction partners. The current study assessed the associations between trait mindfulness, interpersonal stress, and interpersonal perceptions during stressful interpersonal tasks between strangers. Sixty-seven same sex stranger dyads (134 individuals; all females) participated in a laboratory study. Trait mindfulness was measured via an online questionnaire. In the lab, participants were asked to engage in two tasks with a stranger: (1) a stressful interaction task (they were asked to introduce themselves standing only 27 cm apart) and (2) a joint coordination task. Afterwards, both partners’ levels of interpersonal stress and interpersonal perceptions (i.e. liking of the interaction, perceived attentiveness, and perceived coping) were assessed. Results of Actor Partner Interdependence Models (APIM) showed a negative association between trait mindfulness and experienced interpersonal distress. Trait mindfulness was positively associated with liking of the interaction, perceived attentiveness and perceived coping. Actors’ trait mindfulness was positively associated with the partners’ liking of the interaction (marginally significant), but no other partner effects were found. There was no association between trait mindfulness and performance on the joint coordination task. The current findings underscore the importance of studying trait mindfulness dyadically. In actual interpersonal interactions, trait mindfulness positively affects interaction experiences of actors, but we found little support for a transfer to experiences of interaction partners. We discuss the implications of these findings in light of several theoretical models.
Article
Decades of research show that people's social lives are linked to their well-being. Yet, research on the relationship between social interactions and well-being has been largely inconclusive with regard to the effects of person-situation interactions, such as the interplay between contextual factors (e.g., interactions occurring in physical vs. digital contexts, different interaction partners) and dispositional tendencies (e.g., Big Five personality traits). Here, we report on exploratory and confirmatory findings from three large studies of college students (Study 1: N = 1,360; Study 2: N = 851; Study 3: N = 864) who completed a total of 139,363 experience sampling surveys (reporting on 87,976 social interactions). We focus on the effects of different modes of communication (face-to-face [FtF] interactions, computer-mediated communication [CMC], and mixed episodes [FtF + CMC]), and types of interaction partners (close peers, family members, and weak ties). Using multilevel structural equation modeling, we found that FtF interactions and mixed episodes were associated with highest well-being on the within-person level, and that these effects were particularly pronounced for individuals with high levels of neuroticism. CMC was related to lower well-being than FtF interactions, but higher well-being than not socializing at all. Regarding the type of interaction partner, individuals reported higher well-being after interactions with close peers than after interactions with family members and weak ties, and the difference between close peers and weak ties was larger for FtF interactions than for CMC. We discuss these findings with regard to theories of person-situation interactions and research on well-being and social interactions. (PsycInfo Database Record (c) 2022 APA, all rights reserved).
Article
Receiving social support is critical for well-being, but concerns about a recipient’s reaction could make people reluctant to express such support. Our studies indicate that people’s expectations about how their support will be received predict their likelihood of expressing it (Study 1, N = 100 online adults), but these expectations are systematically miscalibrated. Participants who sent messages of support to others they knew (Study 2, N = 120 students) or who expressed support to a new acquaintance in person (Study 3, N = 50 adult pairs) consistently underestimated how positively their recipients would respond. A systematic perspective gap between expressers and recipients may explain miscalibrated expectations: Expressers may focus on how competent their support seems, whereas recipients may focus on the warmth it conveys (Study 4, N = 300 adults). Miscalibrated concerns about how to express support most competently may make people overly reluctant to reach out to someone in need.
Article
Full-text available
When we buy our daily cup of coffee, sometimes we engage in a social interaction with the barista, and sometimes we are in a rush. Every day we have opportunities to transform potentially impersonal, instrumental exchanges into genuine social interactions, and the happiness literature suggests that we may reap benefits by doing so; in other words, treating a service provider like we would an acquaintance (i.e., weak tie) might make us happier. In the current study, people who had a social interaction with a barista (i.e., smiled, made eye contact, and had a brief conversation) experienced more positive affect than people who were as efficient as possible. Further, we found initial evidence that these effects were mediated by feelings of belonging. These results suggest that, although people are often reluctant to have a genuine social interaction with a stranger, they are happier when they treat a stranger like a weak tie.
Article
Two studies demonstrated that individuals can fail to detect changes in their actions that are induced by implicit social influence. In both studies, observers' impressions indicated that actors matched the positivity of their remarks about themselves to the positivity of another person's self-description. However, actors' own judgments of the types of impressions they conveyed revealed that they did not perceive the effect of the other's self-description on their self-presentation. Study I suggested that actors' relatively poor access to their own nonverbal behavior could not fully account for their failure to perceive how they were influenced. Study 2 indicated that actors' metaperceptions were connected to actors' general beliefs about themselves, whereas observers' impressions were not. The ''blindness'' effect was driven primarily by actors low in self-esteem. Implications for self-presentation and other social phenomena are discussed.
Article
Conventional wisdom over the past 160 years in the cognitive and neurosciences has assumed that brains evolved to process factual information about the world. Most attention has therefore been focused on such features as pattern recognition, color vision, and speech perception. By extension, it was assumed that brains evolved to deal with essentially ecological problem-solving tasks. 1.
Article
Gratitude and indebtedness are differently valenced emotional responses to benefits provided, which have implications for interpersonal processes. Drawing on a social functional model of emotions, we tested the roles of gratitude and indebtedness in romantic relationships with a daily-experience sampling of both members of cohabiting couples. As hypothesized, the receipt of thoughtful benefits predicted both gratitude and indebtedness. Men had more mixed emotional responses to benefit receipt than women. However, for both men and women, gratitude from interactions predicted increases in relationship connection and satisfaction the following day, for both recipient and benefactor. Although indebtedness may maintain external signals of relationship engagement, gratitude had uniquely predictive power in relationship promotion, perhaps acting as a booster shot for the relationship.
Article
Six studies were conducted to test the hypothesis that individuals contemplating `making the first move' in initiating a romantic relationship experience pluralistic ignorance (N = 544). Studies 1 and 2 established the preconditions for the phenomenon: participants indicated that they were more likely than a potential partner to be inhibited from making an initiative by a fear of rejection. Studies 3 and 4 demonstrated pluralistic ignorance: although participants explained their own inaction in terms of their fear of being rejected, they attributed a potential partner's inaction to a lack of interest in developing a relationship with them. Study 5 utilized an open-ended thought-listing measure to demonstrate that individuals spontaneously perceive a potential partner's inaction as reflective of disinterest more so than they perceive their own inaction in these terms. Finally, Study 6 provided more definitive evidence that participants' divergent perceptions of their own vs their potential partner's underlying feelings stemmed from the biased interpretation of inaction. Implications for the formation of social bonds are discussed.