Content uploaded by Xuan Zhao
Author content
All content in this area was uploaded by Xuan Zhao on Dec 04, 2020
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=psai20
Self and Identity
ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/psai20
Kind words do not become tired words:
Undervaluing the positive impact of frequent
compliments
Xuan Zhao & Nicholas Epley
To cite this article: Xuan Zhao & Nicholas Epley (2020): Kind words do not become tired
words: Undervaluing the positive impact of frequent compliments, Self and Identity, DOI:
10.1080/15298868.2020.1761438
To link to this article: https://doi.org/10.1080/15298868.2020.1761438
View supplementary material
Published online: 31 May 2020.
Submit your article to this journal
Article views: 71
View related articles
View Crossmark data
ARTICLE
Kind words do not become tired words: Undervaluing the
positive impact of frequent compliments
Xuan Zhao and Nicholas Epley
University of Chicago, Booth School of Business, Chicago, IL, USA
ABSTRACT
Compliments can satisfy others’need to belong, but expressers may
underestimate their positive impact on recipients, creating a barrier to
giving them more often. We assess how people expect recipients will
react to multiple compliments over time, compared to recipients’
actual experiences. Participants expected that recipients would adopt
to multiple compliments, with each feeling less positive and sincere
(Experiment 1). An experiment found no evidence of adaptation in
recipients’actual experience, while expressers underestimated recipi-
ents’positive reactions (Experiment 2). Participants expected less adap-
tation among recipients when they saw the actual compliments
received, suggesting that mistaken adaptation beliefs stem from over-
estimating similarity between multiple compliments (Experiment
3). Underestimating compliments’consistently positive impact may
restrain people from expressing them more routinely.
ARTICLE HISTORY
Received 22 December 2019
Accepted 22 April 2020
KEYWORDS
Prosocial behavior; social
cognition; compliments;
need to belong; kindness
Thinking back during an interview on the most important moments in his life, the actor
Brad Pitt recalled a time from his youth when he was called out of a large crowd at
a Harlem Globetrotter’s show to perform a trick with one of their legendary players,
Meadowlark Lemon (Marchese, 2019). This brief brush with celebrity left Pitt feeling so
great that he recognized the power of his own celebrity to make others feel great. “I’m
trying to say that I have the opportunity to brighten someone’s day. That’s a rare thing.”
Pitt’s celebrity is indeed a rare thing, but empirical research suggests that his ability to
brighten someone’s day is not nearly as rare as he –or as most people –seem to expect.
Human beings are the most social of all primates, whose long-term survival and repro-
duction depends on maintaining strong social bonds (Tomasello, 2014). Maintaining
positive interpersonal relationships is a fundamental need that operates in ways that
are similar to biological needs, such as eating or drinking (Baumeister & Leary, 1995;
Bowlby, 1969). Signs of others’positive regard, appreciation, or warmth that satisfy
a person’s need to maintain positive interpersonal relationships therefore create strong
positive emotions, just as satisfying other biological needs is pleasurable. Indeed, the
extent to which people feel valued and accepted by others is a powerful determinant of
a person’s own sense of self-esteem (Leary et al., 1995). Brightening another person’s day
may be as straightforward as showing warmth, kindness, or appreciation, such as by
CONTACT Nicholas Epley epley@chicagobooth.edu University of Chicago, Booth School of Business, 5807 South
Woodlawn Avenue, Chicago, IL 60637
Supplemental data for this article can be accessed here
SELF AND IDENTITY
https://doi.org/10.1080/15298868.2020.1761438
© 2020 Informa UK Limited, trading as Taylor & Francis Group
performing a random act of kindness (Curry et al., 2018), expressing gratitude (Algoe et al.,
2010), providing social support (Gleason & Iida, 2015), giving someone a compliment
(Zhao & Epley, 2020a), or even just acknowledging a passerby’s presence by making eye
contact (Wesselmann et al., 2012). None of these day-brightening activities requires
worldwide celebrity to accomplish.
More important, emerging research suggests that these prosocial activities brighten
recipients’moods more than those performing the activities may expect. In one series of
experiments, people asked to write agratitude letter underestimated how positive their
recipient would feel after reading the letter (Kumar & Epley, 2018). In another series of
experiments, people visiting a public garden were instructed to ask another visitor to help
by taking a picture of them in front of a nearby attraction. The helpers felt more positive
after taking the requested picture than the requesters expected (Zhao & Epley, 2020b). In
a third series of experiments, people asked to perform a random act of kindness for another
person –such as giving him or her a cup of hot chocolate on a cold winter day –under-
estimated how positive the recipient would feel (Kumar & Epley, 2020). In a final series of
experiments most closely related to our current hypotheses (Zhao & Epley, 2020a), people
asked to write three compliments to a friend or romantic partner underestimated how
positively their recipients would feel after reading the compliments, an effect that did not
emerge in a control condition where people simply predicted how the other person in their
pair felt at the moment. Undervaluing the positive consequences of prosocial actions
matters because people tend to behave in ways that are consistent with their expectations,
thereby creating a misplaced barrier to engaging in positive interpersonal behaviors more
often in daily life.
Miscalibrated expectations about the consequences of prosocial actions are not ran-
dom but seem to stem from two systematic mechanisms. First, being the recipient of
a kind act creates an empathy gap between those performing the act and those receiving
it (Loewenstein, 2005; Nordgren et al., 2011; Van Boven et al., 2013). A kind act is targeted
directly at a recipient, producing uniquely positive feelings for the recipient that are not
shared and hence not anticipated by the expresser (or by other observers; Zhao & Epley,
2020a). Second, those performing prosocial acts appear to evaluate their actions on
somewhat different dimensions than recipients, focusing more on their competency
(how well they perform the action) while recipients are focused more on the warmth
conveyed by their action (the positive intent and meaning of the action; Kumar & Epley,
2018; Zhao & Epley, 2020a). This reflects a more general tendency for actors to evaluate
their own behavior in terms of competency, whileobservers tend to evaluate the same
behavior more in terms of its warthm (Abele & Wojciszke, 2007; Bruk et al., 2018; Fiske
et al., 2007; Wojciszke et al., 1998). If expressers are attending to their competence while
recipients are attending to their warmth, then expressers are likely to underestimate how
positive a prosocial action will make a recipient feel (Kumar & Epley, 2018; Zhao & Epley,
2020a).
Of course, it is easy to imagine limits on the positive impact of prosocial acts on
others. As lovely as receiving one act of kindness might feel, receiving subsequent
acts may feel progressively less positive. Getting one compliment from a friend may
make one’s day, but getting a compliment each day from the same friend 5 days in
a row might become more mundane by the end of a week. As with almost any
positive stimulus (Kahneman & Snell, 1992; Rolls, Rolls, Towe, & Sweeney, 1981),
2X. ZHAO AND N. EPLEY
people may expect that repeatedly receiving compliments will become less and less
positive over time. However, rates of adaptation vary across stimuli (Campbell et al.,
2014; Frederick & Loewenstein, 1999; Galak & Redden, 2018;Holtetal.,1995;O’Brien
& Kassirer, 2019), and we expect adaptation to prosocial acts such as multiple
compliments may be relatively slow for three reasons. First, several experiments
suggest that people may adapt more slowly to the positive experience of performing
prosocial acts than to performing selfish acts (O’Brien & Kassirer, 2019). Similarly slow
rates of adaptation may characterize the positive experience of receiving prosocial
actsaswell.Second,although“compliments”describe a specific category of state-
ments, the precise details of those compliments are likely to vary a great deal,
making each compliment uniquely meaningful to recipients (O’Brien, 2019). Finally,
most needs are not satisfied by a single event but instead are recurring desires that
can be repeatedly satisfied. Just as hunger is not satisfied for long after a good meal,
so too one’s sense of belonging is not likely to be satisfied by a single expression of
warmth. If people fail to recognize that needs require repeated satisfaction over time
(Read & Loewenstein, 1995), then they may fail to appreciate how consistently the
recipient will enjoy multiple prosocial acts that occur over time.
Here we present experiments investigating people’s experience of receiving
repeated acts of kindness compared to expectations about recipients’experiences.
Specifically, we designed an experimental procedure in which we asked one member
of an acquainted pair to write 5 compliments to his or her partner, with one
compliment being shown each day to the recipient, and then to predict how the
recipient would feel each day after he or she read the compliment. We measured
people’s expectations of how this experience would unfold for recipients in
Experiment 1 by describing eight different variations of this procedure and asking
participants to predict the recipients’experiences. Based on existing theory and past
research (Kahneman & Snell, 1992), we predicted that people would generally expect
recipients to “tire”of receiving these compliments, feeling a little less positive after
reading each one. In contrast, we expected that actual recipients would feel similarly
positive after reading each compliment. We also predicted, consistent with prior
research (Zhao & Epley, 2020a), that compliment expressers would generally under-
estimate how positive their recipients would feel. We tested these predictions in
Experiment 2 by conducting the procedure described at the beginning of this
paragraph. We further explored people’s expectations about adaptation in
Experiment 3 by asking participants to report how they expected compliment
recipients to feel either after reading concrete compliments, or without reading
them. If failing to appreciate the uniqueness of specific compliments leads people
to expect more adaptation than is actually experienced, then seeing the exact
compliments should diminish this tendency.
All experiments were pre-registered. We reported all measures, manipulations, and
exclusions in the paper. All materials, protocols, data, analyses, and preregistration files
can be found online (https://osf.io/wfmqb/).
SELF AND IDENTITY 3
Experiment 1: Expecting adaptation?
People’s behavior is guided at least partly by the expected value of their choices, making
people’s expectations of how others might respond to repeated prosocial acts potentially
important for understanding behavior. We asked participants in Experiment 1 to predict
how a person would respond each day after receiving a compliment for 5 days in a row.
Participants did so by imagining themselves in the role of a compliment expresser or
recipient, who communicated the compliment via text or in person, with compliments
prepared in advance or generated each day. We included these latter factors to test
whether changing the context in Experiment 2, where compliments were delivered over
text and generated at the beginning of a 5-day period, would affect participants’expecta-
tions about adaptation. We predicted, as with most stimuli (Kahneman & Snell, 1992), that
people would generally expect that people would adapt to receiving multiple compli-
ments, with each producing a little less positive response than the one preceding it.
Method
Participants
We targeted a sample size of 400 participants in order to obtain 50 in each of our eight
experimental conditions. A total of 410 Amazon Mechanical Turk (MTurk) workers com-
pleted our study (M
age
= 38.20, SD
age
= 11.22, range
age
=18–73; 48% female) in exchange
for 0.50 USD.
Procedure
We first instructed participants in the survey to think of someone they were close to and
interacted with often. We then randomly assigned participants to one of eight conditions in
a 2 (perspective: expresser vs. recipient) × 2 (modality: text delivery vs. in-person delivery) ×
2 (preparation: prepared in advance vs. generated each day) between-participants design.
Participants who imagined expressing compliments first predicted how their reci-
pient would react on Monday after reading their compliment: how positive their
recipient would feel (“How positive/negative do you predict the other person would
feel after receiving this compliment?”), how awkward their recipient would feel (“How
awkward do you predict the other person would feel after receiving this compli-
ment?”), and how sincere their recipient would rate the compliment to be (“How
sincere do you think the other person would perceive this compliment to be?”).
Those who imagined receiving a compliment reported how they thought they would
feel on the same constructs. All items were presented with a scale of 0 (notatall)to
10 (extremely), except for the positive/negative item, which had a scale ranging from
−5(much more negative than normal)to5(much more positive than normal), with 0
(no different than normal) as the midpoint. Participants completed the same proce-
dure for the remaining four days from Tuesday through Friday.
Finally, participants completed additional questions on exploratory measures and mem-
ory checks (see the Supplemental Material for full details), provided demographic informa-
tion (gender, age, race, education), and were debriefed and then compensated.
4X. ZHAO AND N. EPLEY
Results
To detect changes in participants’expectations across five days, we analyzed the
data using growth curve modeling. To account for interdependence among
responses from the same participant across multiple days, we constructed a linear
mixed model
1
for each of the three primary DVs (mood, awkwardness, sincerity) by
entering the linear effect of time, perspective, delivery modality, preparation, and
their interaction terms as fixed effects and participant-specificinterceptsasrandom
effects.
2
Results indicated that people generally expected recipients to adapt to the experi-
ence of receiving multiple compliments, regardless of what perspective they ima-
gined taking, in what modality their compliments were exchanged, or whether the
compliments were prepared in advance. As shown in Figure 1(a), participants
expected recipients to feel a little less positive each day after receiving a new
compliment, b=−.22, SE =.02, 95% confidence interval (CI) = [−.25, −.19], t
(1640) = −14.48, p< .001. Participants also expected the perceived sincerity of
each compliment to steadily decline over the course of the week (see Figure 1(c)),
b=−.30, SE = .02, 95% CI = [−.33, −.26], t(1640) = −16.30, p< .001. Expectations
about awkwardness were somewhat more complicated (see Figure 1(b)), and did not
show the same pattern of steady decline observed on the other measures (see the
Supplemental Material for full results and graphs by each condition). Participants
expected repeated compliments to grow somewhat “tired”over time, with each one
leaving the recipient feeling a little less positive than the one before it. Experiment 2
tests the extent to which these expectations match recipients’actual experiences.
Figure 1. Mean expected positive mood (a), awkwardness (b), and sincerity ratings (c) for recipients who
read one compliment each day over a period of five days, among those who imagined either expressing
or receiving the compliments. Error bars indicate ±1 SE.
Ratings on positive mood ranged on a scale of −5 (much more negative than normal) to 5 (much more positive than
normal). Ratings on awkwardness and sincerity ranged on a scale of 0 (not at all) to 10 (extremely).
SELF AND IDENTITY 5
Experiment 2: Compliment week
Method
Participants
Pairs of participants were recruited to an experiment advertised as the “Interpersonal
Relationship Study”in one of three different locations: in a privately-owned coffee shop in
Chicago (n= 45 pairs), in a public park located in Chicago (n= 12 pairs), or in a research
laboratory on the University of Chicago campus (n= 9 pairs). Each participant received either
a small novelty gift (in the coffee shop or park) or a 1 USD bonus (in the laboratory) for signing up
and a 5 USD gift card upon completion. We targeted a total sample size of 50 pairs. Anticipating
an attrition rate of at least 20%, we planned to recruit 63 pairs and finished collecting data
through the end of our last scheduled shift as we approached that target. We ended that shift
recruiting a total of 66 pairs who completed the initial session. Sadly, we excluded one pair from
data analysis because the expresser died in a car crash mid-week, dramatically altering the
recipient’s experience compared to other recipients in the experiment. This yielded 65 pairs of
participants (M
age
= 23.47, SD
age
=7.48;range
age
=18–62; 67% female, 31% male, 2% other) that
were composed of friends (n= 40), romantic couples (n=13),spouses(n=7),family(n=3),and
colleagues (n= 1) and had known each other for an average of 5.64 years (range = 2 months to
39 years; SD =7.51years).
Procedure
We recreated one scenario from Experiment 1 in which one participant wrote 5 compli-
ments all at once that were then delivered one at a time over a period of 5 days to the
recipient over text. Specifically, we first recruited participants in acquainted pairs to
complete an initial session where one person wrote compliments and reported how
they expected their recipient to respond. In the week following this initial session,
recipients received one compliment each day (from Monday through Friday) and reported
their reactions. Finally, both expressers and recipients completed an exit survey on the
subsequent Saturday. We describe the details of each of these time periods below.
Initial session. After both participants agreed to participate, one was randomly assigned
to be the compliment expresser and the other to be the compliment recipient.
Participants were then separated and given brief verbal descriptions of the experiment’s
timeline before receiving a tablet to start their respective surveys.
The first survey block was identical for expressers and recipients and asked basic
relationship information questions. Participants first reported their current relationship
quality on two bipolar scales, the first measuring how close they felt to their partner on
a scale ranging from −5(feels like we’re miles apart)to5(feels like we’re really close), and
the second measuring how satisfied they were with their relationship on a scale ranging
from −5(extremely dissatisfied)to5(extremely satisfied). Participants then reported how
frequently they communicated with each other (ranging from a few times per day to once
or twice per month). Participants then reported how often they gave compliments to and
received compliments from the other person on two 7-point scales ranging from −3(a lot
less often than I think I should)to3(a lot more often than I think I should), with 0 (exactly as
often as I think I should) as the midpoint. Finally, participants provided numeric estimates
6X. ZHAO AND N. EPLEY
of how many times they gave compliments to and received compliments from the other
person over the past week.
Expressers then proceeded to the second part of their survey, where they received the
following compliment-writing instructions followed by five text boxes on the same page
where they were to write their compliments, each labeled with a specific day from
Monday to Friday:
In this study, we want you to write down five separate compliments you could give to your
study partner. Please think about positive characteristics or actions you see in your study
partner that would be worth complimenting him or her on.
These should be positive things you have noticed but have not, for whatever reason, had
a chance to compliment your study partner on yet.
Below, please write down five separate compliments you could give to your study partner.
Over the next week from Monday to Friday, we’re going to send your compliments to your
study partner in the order below.
After writing the five compliments, expressers were presented with the first compliment
they wrote and predicted how their study partner would report feeling after reading
that compliment on Monday. Expressers first predicted the recipient’smoodontwo
separate items: “How positive/negative do you predict your partner will report feeling?”,
and “How pleasant do you predict your partner will report feeling?”Expressers then
predicted how awkward . . . their study partner would report feeling. Finally, participants
reported how sincere their compliment was and then predicted how sincere their study
partner would perceive the compliment to be. Participants answered all items on scales
ranging from 0 (not at all)to10(extremely), except for the first item, which ranged from
−5(much more negative than normal)to5(much more positive than normal), with 0 (no
different than normal) as the midpoint. Expressers completed the same procedure for
the remaining four compliments matched to each day from Tuesday through Friday.
Finally, expressers predicted how their recipient’s week-long experience of receiving
compliments would change over the course of the week on a scale of −5(get worse over
time)to5(get better over time), with 0 (stay the same over time) as the midpoint, and also
predicted the extent to which their partner would get tired of receiving one compliment
each day by the end of the week on a scale from 0 (not at all)to10(extremely).
Expressers finished the second block of their survey by providing an e-mail address to
receive the exit survey on the following Saturday.
The recipient’s second survey block was much shorter than the expressers’second
block. Recipients first reported their relationship type and duration. They then learned
that they would receive one survey each day over the following week and provided an
e-mail address to receive those surveys.
The final survey block for both expressers and recipients asked them to report basic
demographic information including gender, age, race/ethnicity, and education level and
also provided a text box for writing any additional comments or feedback about the
experiment.
Receiving compliments (Monday through Friday). We sent recipients an e-mail con-
taining a unique survey link at noon each day of the week following the initial session.
SELF AND IDENTITY 7
Upon opening the survey, recipients saw their compliment for that given day of the week
and a short survey asking them to report how positive/negative, pleasant, and awkward
they felt after receiving this compliment, and also how sincere they perceived the
compliment to be. The next survey page presented recipients with an open-ended
question asking them to describe how they felt. All surveys automatically expired after
midnight.
Final exit survey. At noon on Saturday following the week of compliments, we sent both
expressers and recipients e-mails directing them to an exit survey measuring their overall
perception of the experience. In this survey, participants first reported their relationship
satisfaction on two bipolar items identical to those in the first survey. Expressers then
reported how they thought their recipient’s experience had changed over the course of
the week and also how tired the recipients were of receiving compliments by the end of
week, using the same items included in the initial experimental session. Expressers then
reported how interested they thought their recipient would be in receiving a new
compliment the following Monday on a 11-point scale ranging from 0 (not at all)to10
(extremely). Recipients responded to the same items phrased from their perspective to
measure their actual experience and beliefs.
Next, to understand whether the recipients attempted to reciprocate compliments
back to the expresser over the course of the week, we asked both expressers and
recipients to report how often the recipient complimented the expressers by providing
a numeric estimate as well as responding to a 11-point scale ranging from −5(a lot less
often than usual)to5(a lot more often than usual), with 0 (as often usual) as the
midpoint.
Finally, we gave participants an opportunity to give additional feedback (in a text box),
a full debriefing, and a compensation voucher.
Results
Response rates
All expressers reported their expectations of the recipient’s reactions at the initial screen-
ing session, but not all recipients responded to each daily survey. Of the 65 recipients, we
received 59, 51, 57, 55, 54 recipients on each day of the week, respectively. All but 1
participant filled out at least one daily survey. We received completed exit surveys from 52
expressers and 53 recipients. Although less than perfect, these response rates (from 78%
to 91%) do not suggest that participants were slowly dropping out over the course of the
week because they were losing interest in receiving compliments, as we observed the
lowest response rates on Tuesday. We also note that recipients needed to open the survey
URL in order to read their compliment. Our records indicate that all participants who
opened a survey completed the survey, meaning that any attrition came from participants
not opening the survey to see the compliment in the first place. Therefore, we know that
recipients were not first reading the compliments and then selectively responding to
favorable compliments while ignoring less favorable ones. These results suggest that
attrition is not creating systematic selection biases in our analyses.
8X. ZHAO AND N. EPLEY
Reactions to multiple compliments
We were primarily interested in how expressers’expectations and recipients’experiences
of mood, awkwardness, and sincerity might change over the course of the compliment
week. Given that time is nested within individual participants and that expressers and
recipients are nested within pairs, we again analyzed our data using growth curve
modeling. We accounted for interdependence by constructing a linear mixed model for
each of the three primary DVs (mood, awkwardness, sincerity), with the linear effect of
time (zero-centered on Wednesday), perspective (−1 = expresser, 1 = recipient), and their
interactions as the fixed effects and participant-specific intercepts and pair-specific inter-
cepts as random effects.
To enable easy comparison between expectations and experiences over the course of
the week, we also preregistered our intention to compare participants’Monday versus
Friday responses using a 2 (day: Monday vs. Friday) × 2 (perspective: expresser vs.
recipient) repeated measures ANOVA. However, we later realized that repeated measures
ANOVA could not properly account for the interdependence between expressers and
recipients. We therefore constructed the same linear mixed model as specified above,
except that we treated time as a categorical variable (−1 = Monday, 1 = Friday) instead of
a continuous variable as in the case of growth curve modeling.
Positive mood. We first created a composite score of expected and actual positive mood
by adding 5 to the negative/positive mood item so that it was on a 0–10 scale and then
averaged it together with the pleasantness item (r
expresser
= .78; r
recipient
= .86; ps < .001).
A linear mixed model revealed a significant main effect of time, F(1, 475.49) = 10.39,
p= .001, and a marginally significant main effect of perspective, F(1, 64.31) = 2.77, p= .10,
qualified by a significant interaction between time and perspective, F(1, 475.49) = 5.55,
p= .019. As shown in Figure 2(a), recipients remained in a very positive mood consistently
over the course of the week, b= .03, SE = .05, 95% CI = [−.06, .12], t(215.77) = .59, p= .56.
Figure 2. Expressers’and recipients’ratings regarding positive mood (a), awkwardness (b), and
sincerity (c) on each compliment across 5 days (Experiment 2). Error bars indicate ±1 SE.
“Positive Mood”is a composite of positive/negative mood and pleasantness (with 5 added to the positive/negative mood
item to make the scale range from 0–10). Awkwardness and sincerity ranged from 0 (not at all) to 10 (extremely).
SELF AND IDENTITY 9
Interestingly, expressers expected recipients’mood to become increasingly positive over
the course of the week, b= .18, SE = .04, 95% CI = [.09, .27], t(259) = 4.08, p< .001. We did
not predict this result, which is the opposite pattern observed in Experiment 1. We will
return to this issue in the Discussion section of this experiment.
An additional analysis comparing participants’responses on Monday and Friday sup-
ported the above findings. The linear mixed model revealed a significant main effect of
time, F(1, 119.56) = 15.03, p< .001, and a marginally significant main effect of perspective,
F(1, 63.08) = 2.96, p= .09, qualified by a significant interaction, F(1, 119.56) = 8.50, p= .004.
Specifically, compliment expressers significantly underestimated how positive their reci-
pient would feel on Monday (Ms = 7.91 vs. 8.69, SDs = 1.50 vs. 1.29, respectively), F(1,
60.81) = 10.91, p= .001, supporting prior research suggesting that people tend to
underestimate the positive impact of their compliments on the recipients (Zhao &
Epley, 2020a). By contrast, compliment expressers did not significantly underestimate
how positive their recipient would feel on Friday (Ms = 8.85 vs. 8.80, SDs = 1.36 vs. 1.72,
respectively), F(1, 63.21) = 0.03, p= .86. As mentioned earlier, expressers expected their
recipient’s positive mood to improve over the course of the week, such that they no
longer significantly underestimated recipients’positive mood by the end of the week.
Awkwardness. A linear mixed model on awkwardness revealed only a significant main effect
of perspective, F(1, 62.85) = 8.78, p= .004. Consistent with prior research (Zhao & Epley, 2020a),
expressers overestimated how awkward recipients would feel (see Figure 2(b)).
Comparing participants’responses on Monday vs. Friday mirrored these results,
although the main effect of perspective was only marginally significant, F(1,
62.32) = 3.22, p= .077, likely due to a smaller number of observations on two days as
opposed to five days.
Compliment sincerity. The same linear mixed model on expressers’expectation versus
recipients’experienced sincerity revealed a significant main effect of time, F(1,
475.58) = 6.63, p= .010, a marginally significant main effect of perspective, F(1,
62.46) = 3.41, p= .069, and a non-significant interaction between perspective and time,
F(1, 475.58) = 1.15, p= .28. The main effect of time indicates that both expected and
experienced sincerity increased over the course of the week. The marginally significant
main effect of perspective indicates that expressers somewhat underestimated how
sincere their compliments would be perceived to be by recipients (see also Zhao &
Epley, 2020a).
Comparing participants’responses on Monday vs. Friday again mirrored these results,
yielding a significant effect of time, F(1, 121.60) = 4.46, p= .037, a marginally significant
main effect of perspective, F(1, 60.81) = 3.87, p= .054, and a nonsignificant interaction
between perspective and time, F(1, 121.60) = 1.41, p= .24.
Because expressers also reported how sincere each of their compliments actually was,
we explored the extent to which recipients’perceptions of sincerity were calibrated, as
well as the extent to which expressers expected their recipients’perceptions to be
calibrated. To assess the calibration of recipients’perceptions, we replaced the perspec-
tive factor in the previous linear mixed model –that is, expressers’expectations versus
recipients’actual experience –with a comparison between expressers’self-reported
sincerity against recipients’actual experience. This analysis revealed only a significant
10 X. ZHAO AND N. EPLEY
main effect of perspective, F(1, 61.32) = 9.74, p= .003, suggesting that recipients per-
ceived the compliments to be less sincere than expressers reported them to be (Ms = 8.67
and 9.30, respectively SDs = 1.80 and 1.39). To assess the extent to which expressers
expected their recipients to be calibrated, we replaced the perspective factor in the linear
mixed model with a comparison between expressers’self-reported sincerity and their
expectations on how the recipients would think. This analysis revealed a significant main
effect of perspective, F(1, 582) = 98.04, p< .001, and a significant main effect of time, F(1,
582) = 6.84, p= .009, with a non-significant interaction between perspective and time, F(1,
582) = 1.71, p= .19. The significant main effect of perspective indicates that expressers
expected their recipient to perceive the compliments as considerably less sincere than
they actually were (Ms = 8.27 and 9.30, respectively, SDs = 1.83 and 1.39) –and, as already
described, as even less sincere than they were perceived by the recipients to be.
Exit survey: Perceived change
In addition to inferring changes from participants’ratings of positive mood, awkwardness,
and sincerity across 5 days, we also directly asked expressers how they expected their
recipient’s experience to change over the course of the week (both in the initial session
and at the final survey), and we also asked recipients to report how they perceived their
experience to have actually changed over the week. Interestingly, expressers’expecta-
tions (both in the initial survey and the exit survey) as well as recipients’reported
experience all indicated similar levels of improvement over the course of the week
(Ms = 2.14, 2.08, and 1.79, respectively, SDs = 2.25, 1.87, and 1.86); F(2, 167) = .47,
p= .63. Although expressers did not expect their recipients to grow especially tired of
their complements (on a 0–10 scale), expressers did expect their recipients to grow more
tired of receiving one compliment each day than their recipient actually did (Ms = 2.14
and 1.17, respectively, SDs = .30 and .25), t(52) = 2.65, p= .011. Finally, recipients reported
being moderately interested in receiving another new compliment on Monday (M= 7.34,
SDs = 2.81), a level of interest that did not differ significantly from what the expressers
predicted (M= 6.73, SD = 3.10), t(46) = −1.08, p= .29.
Exploratory analyses
We compared participants’self-reported relationship quality before and after the week-
long experience and found that a weeklong series of compliments did not significantly
influence the overall quality of people’s long-term relationships. We also analyzed peo-
ple’s self-reported compliment frequency before and during the compliment week and
found that people overall thought they gave fewer compliments than they should, and
that recipients reciprocated the expressers’compliments over the course of the week by
giving more compliments than they normally did. We describe these exploratory analyses
fully in the Supplemental Material.
Discussion
Contrary to the expectations for adaptation we observed from survey respondents in
Experiment 1, an actual week-long experience of receiving compliments did not turn kind
words into tired words in the recipients’minds. Instead, recipients responded very positively
to each new compliment they received, with no decrease in reported positive mood over
SELF AND IDENTITY 11
the course of the week, and also rated each compliment as being similarly sincere over the
course of the week. This discrepancy between imagined recipients’expectations in
Experiment 1 and actual recipients’responses in Experiment 2 suggests that people’s
expectations about adapting to receiving multiple compliments may be miscalibrated. We
suggest that this could come either from an overgeneralization of the typical experience of
affective adaptation that people have with most stimuli (Kahneman & Snell, 1992), or from
a failure to recognize that each new compliment is distinct and so repeatedly receiving
compliments is more unique than people abstractly imagine (O’Brien, 2019).
In line with prior research (Zhao & Epley, 2020a), expressers underestimated how
positively recipients would react to the initial compliments early in the week. Expressers
did not, however, significantly underestimate recipients’positive response to the last
compliment, nor did they expect recipients’adaptation to receiving multiple compli-
ments. In fact, expressers expected the opposite, believing that their recipient’s positive
mood would actually increase as they received more compliments. This pattern seems
unlikely to have resulted from expressers’belief that recipients might get happier as they
moved closer to the weekend, because respondents in Experiment 1 also made predic-
tions from Monday to Friday without exhibiting a similar pattern. Instead, this pattern
could either reflect a systematic tendency among people anticipating the outcome of
actual compliments expressed over time in daily life, or it could reflect an artifact of our
experimental design.
If it reflects a systematic tendency, then it may stem from the difference between
knowing the concrete details of a compliment (as expressers did in Experiment 2) versus
only considering compliments in the abstract (as participants did in Experiment 1).
Knowing a compliment’s concrete details may mitigate the application of a more
general heuristic about affective adaptation. If so, then we would expect that observers
who read the same compliments actually written by expressers would also expect little
adaptation, if any, in the recipients’reactions over time. We test this directly in
Experiment 3.
If it reflects an artifact of our experimental design, then it could come from allowing
expressers to write their compliments in whatever order they wanted, meaning that the
expressers may have assigned what they thought was their best compliment to be
delivered at the end of the week rather than at the beginning, perhaps hoping to create
an improving sequence over the course of the week that ends “on a high note”
(Loewenstein & Prelec, 1993;O’Brien & Ellsworth, 2012). A closer look at expressers’
predictions in Figure 1(a,c) is consistent with this potential artifact. Note that expressers’
expectations did not follow a linear pattern, but instead showed a notable increase on the
last day of the week (Friday). In fact, excluding Friday from the growth curve models
reveals a much more gradual change in the expressers’expectations of their recipients’
positive mood (b= .12, SE = .06, 95% CI = [0, .24], t(194) = 1.97, p= .051), no change in
expected awkwardness (b=−.05, SE =.11, 95% CI = [−.26, .16], t(194) = −.44, p= .66), and
no change in expected ratings of sincerity (b= .11, SE = .08, 95% CI = [−.05, .26], t
(194) = 1.36, p= .18). Furthermore, the compliments themselves also differed over the
course of the week, with those written for the last day being significantly longer com-
pared to those written for the first day, t(256) = −2.86, p= .004 (see Table 1).
Although we can only speculate about what these extra words in the Friday compli-
ments mean, they could be viewed as a proxy for how much effort expressers put into
12 X. ZHAO AND N. EPLEY
creating a positive experience for their recipient on the last day (even though the number
of words in a compliment did not significantly affect the recipients’reactions in reality,
p= .60). Given that expressers were free to specify the order in which their compliments
were delivered, we cannot conclude that expressers’expectations over the course of the
week were primarily driven by the repeated nature of the compliments rather than by the
presumed quality of the compliments themselves. Experiment 3 again addresses this issue
in more detail by examining expectations about adaptation among a separate sample of
participants who either saw the actual compliments that were delivered, or did not see
the actual compliments and instead predicted reactions to unspecified compliments.
Experiment 3: Expectations of concrete vs. abstract compliments
People imagining how others would respond to a series of compliments anticipated that
the recipients would grow tired of them over time (Experiment 1), while those who were
randomly assigned to actually write compliments to a recipient did not (Experiment 2).
Experiment 3 was designed to clarify the nature of people’s expectations and to better
explain these seemingly inconsistent results. Specifically, we tested whether having
specific compliments in mind would prevent people from applying the concept of
adaptation when predicting recipients’reactions to multiple compliments. We did so in
the concrete compliment condition by yoking one participant to each pair in Experiment 2
and asked them to read the actual compliments expressers delivered to the recipients and
then make the same predictions that the actual expressers did. In comparison, we asked
participants in two abstract compliment conditions to predict, with no specific compli-
ments in mind, how a recipient would react in the procedure we used in Experiment 2. We
predicted that people expect adaptation to experiencing the same event repeatedly,
meaning that participants in the two abstract compliment conditions would expect more
adaptation to receiving multiple compliments than would participants in the concrete
compliment condition.
Method
Participants
Amazon MTurk workers agreed to participate in a three-minute “survey on social interac-
tion”and received .50 USD in exchange. We targeted 67 participants in each of our 3
experimental conditions, so that one participant in the concrete compliment condition
could be yoked to every one of the 67 pairs in Experiment 2.
3
A total of 226 participants
completed our experiment. We excluded 6 participants who failed our bot-screening
check and 21 in the concrete compliment condition who were the second participants
randomly assigned to the same expressers in Experiment 2,
4
yielding 199 participants for
Table 1. Means (and standard deviations) of word count for each day’s compliments in Experiment 2.
Day 1 Day 2 Day 3 Day 4 Day 5
18.91 (16.70) 15.80 (14.89) 18.55 (16.81) 17.45 (12.17) 23.42 (20.40)
–Count compared
to Day 1
–
–
t=−1.97
p=.049
t=−.22
p=.82
t=−.93
p=.35
t= 2.86
p=.004
SELF AND IDENTITY 13
data analysis (M
age
= 35.37, SD
age
= 10.38, range
age
=18–70; 41% female). Including these
21 duplicates does not, however, meaningfully alter any of the following analyses.
Procedure
Participants were assigned to one of three conditions. Participants in the concrete compli-
ment condition were each yoked to one pair in Experiment 2, read a brief summary of
Experiment 2’s procedure, and then saw the exact compliments delivered to their yoked
recipient on each day of the experiment. These participants were told that those compli-
ments were sent to the recipient from Monday to Friday and reported how they expected
the recipient to respond on the same three items used in Experiment 2 (positive/negative
mood, awkwardness, and compliment sincerity) across five days. Participants in the
abstract compliment (observer) condition followed the same procedure except that they
were not shown the actual compliments sent; instead, they were asked to imagine how
someone would respond to abstract, unspecified, “compliments.”Finally, participants in
the abstract compliment (expresser) condition were asked to identify one specific person
they were close in their life as their recipient (as in Experiment 1), imagined going through
Experiment 2’s procedure as the expresser, and predicted how that person would react to
their compliments without actually writing any.
Finally, all participants predicted how the recipient’s overall experience would change
over the week on two items identical to those in Experiment 2 (i.e., getting better/worse,
getting tired), provided their demographic information (i.e., gender, age, race, and educa-
tion), and were debriefed and compensated.
Results
As in Experiment 2, we constructed a linear mixed model for each of the three primary DVs
(positive mood, awkwardness, and perceived sincerity) with the linear effect of time, the
effects of experimental condition, and their interaction terms as fixed effects and partici-
pant-specific intercepts as random effects. Given that our experimental manipulation has
three conditions, we conducted two a priori Helmert contrasts to examine the effect of
compliment concreteness: the first comparing the two abstract compliment conditions
against each other, and the second comparing the two abstract compliment conditions
against the concrete compliment condition. Following reviewers’suggestions during the
publication process, we also compared participants’expectations in the current experi-
ment against the yoked expressers’expectations and the yoked recipients’experiences in
Experiment 2. Overall, we found that participants in Experiment 3 expected even less
positive reactions from the recipients than the actual expressers in Experiment 2 did, and
thereby underestimated even further how positive the recipients in Experiment 2 actually
felt. We report these additional analyses in the Supplemental Material.
Positive mood
Participants generally expected recipients’positive mood to decline over the course of the
week, b=−.18, SE = .03, 95% CI = [−.24, −.13], t(796) = −6.67, p< .001. As shown in Figure 3
(panel A), this expected decline was similar in the two abstract compliment conditions, as
indicated by the lack of an interaction between time and the contrast between the two
abstract conditions, b=0,SE = .03, p= .95. By comparison, this expected decline was more
14 X. ZHAO AND N. EPLEY
gradual in the concrete compliment condition, as indicated by a significant interaction
between time and the contrast between the concrete condition and two abstract condi-
tions, b= .04, SE = .02, 95% CI = [0, .08], t(796) = 2.18, p= .029. Specifically, participants
expected significant adaptation to repeatedly receiving compliments over the course of
the week in both the abstract compliment (observer) condition, b=−.23, SE = .04, 95%
CI = [−.31, −.15], t(256) = −5.54, p< .001, and in the abstract compliment (expresser)
condition, b=−.23, SE = .04, 95% CI = [−.31, −.15], t(272) = −5.39, p< .001, but they
expected a more gradual decline in the recipients’positive mood in the concrete compli-
ment condition, b=−.10, SE = .06, 95% CI = [−.22, .02], t(268) = −1.71, p= .088.
Awkwardness
Similar to Experiment 1, participants’expectations about awkwardness were somewhat
more complicated than expectations about positive mood. Overall, participants expected
little change in awkwardness over the course of the week, b= .04, SE = .04, p= .20, but this
varied by experimental conditions. As shown in Figure 3(b), expected awkwardness of
receiving multiple compliments varied somewhat between the two abstract compliment
conditions, as indicated by a marginally significant interaction between time and the
contrast between two abstract conditions, b=−.09, SE = .05, 95% CI = [−.19, 0], t
(796) = −1.94, p= .052. In addition, the pattern was also somewhat different in the
concrete compliment condition compared to the two abstract conditions, as indicated
by a marginal interaction between time and the contrast between the concrete condition
and the two abstract conditions, b=−.05, SE = .03, 95% CI = [−.11, 0], t(796) = −1.94,
p= .053. Specifically, participants in the abstract compliment (expresser) condition
imagined that someone in their own life would feel increasingly awkward over the course
of the week, b= .19, SE = .05, 95% CI = [.09, .30], t(272) = 3.68, p< .001, while participants
expected little change in awkwardness over the course of the week in the abstract
Figure 3. Expected positive mood (a), awkwardness (b), and perceived sincerity (c) among participants
in the concrete compliment, abstract compliment (observer), and abstract compliment (expresser)
conditions in Experiment 3. Error bars indicate ±1 SE.
Ratings on positive mood ranged from a scale of −5 (much more negative than normal) to 5 (much more positive than
normal). Ratings on awkwardness and compliment sincerity ranged from a scale of 0 (not at all) to 10 (extremely).
SELF AND IDENTITY 15
compliment (observer) condition, b=0,SE = .06, p= .98, or in the concrete compliment
condition, b=−.07, SE = .09, p= .44.
Compliment sincerity
Participants generally expected recipients to perceive multiple compliments to be
increasingly less sincere over the course of a week, b=−.22, SE = .03, 95% CI = [−.28,
−.16], t(796) = −6.79, p< .001. As shown in Figure 3(c), this expectation of decline was
similar in the two abstract compliment conditions, as indicated by the lack of an interac-
tion between time and the contrast between the two abstract conditions, b= .04, SE = .04,
p= .34. By comparison, this decline was more gradual in the concrete compliment
condition, as indicated by a significant interaction between time and the contrast
between the concrete condition and the two abstract conditions, b= .05, SE = .02, 95%
CI = [.01, .10], t(796) = 2.30, p= .022. Specifically, participants expected other people to
perceive each compliment as a bit less sincere over the course of the week in both the
abstract compliment (observer) condition, b=−.23, SE = .05, 95% CI = [−.33, −.13], t
(256) = −4.62, p< .001, and in the abstract compliment (expresser) condition, b=−.31,
SE = .04, 95% CI = [−.40, −.22], t(272) = −6.78, p< .001. Participants in the concrete
compliment condition, however, expected perceived sincerity to decline more gradually,
b=−.11, SE = .07, 95% CI = [−.25, .02], t(268) = −1.67, p= .096.
Predicted change
Finally, we analyzed the two items that directly asked participants to report their expecta-
tions about changes over the course of the week (i.e., get better or worse/get tired by the
end of the week) and found no main effect of condition in either measure, Fs = .15 and .25,
ps = .86 and .78, respectively. Overall, participants expected the recipients’experience to
get slightly better (M= .45, SD = 2.44; t(198) = 2.645, p= .009) and that the recipients
would feel moderately tired of the repeated experience at the end of the week (M= 4.88,
SD = 3.15; t(198) = 21.90, p< .001).
Discussion
These results from Experiment 3 indicated that participants in the abstract conditions –
regardless of whether imagining being an expresser or an observer –generally expected
the recipients to adapt to multiple compliments, consistent with the survey results
reported in Experiment 1. In contrast, those who read the actual compliments in
Experiment 2 expected somewhat weaker adaptation over the course of the week. This
suggests that one difficulty in appreciating the value of giving compliments repeatedly
may come from failing to recognize the diversity of compliments when considering the
act abstractly. Expressers may only come to recognize their consistently positive impact
on the recipients once they have actually generated the compliments, yet anticipated
adaptation could serve as a barrier to routinely generating compliments for a recipient in
daily life. We observed no systematic tendency for participants in any condition to expect
recipients’experience to improve over the course of the week, suggesting that those
results among expressers in Experiment 2 may stem from an artifact of the procedure
used in that experiment. In hindsight, that artifact could have been removed by informing
expressers in Experiment 2 that their compliments would be delivered to their recipients
16 X. ZHAO AND N. EPLEY
in a random order. Further research will be necessary to understand whether expressers
still expect increasing positivity in the recipients' mood when the quality of their compli-
ments is rendered independent of the order in which they are expressed.
General discussion
“My child, I could live on a good compliment two weeks with nothing more to eat.”
-Mark Twain, 1906
Twain was obviously exaggerating when noting that kind words could satisfy his
hunger needs for two weeks, but it is no exaggeration to say that satisfying a person’s
need for belonging may be just as fundamental as satisfying a person’s hunger
(Baumeister & Leary, 1995). Behavior that signals strong relationships is deeply satisfying,
both to one’s happiness and sense of self-worth. However, just as one good meal does not
satisfy hunger for weeks, signs of positive regard –like a good compliment –are also
unlikely to satisfy belonging needs for weeks. Indeed, recipients in Experiment 2 did not
grow tired of receiving one compliment every day for a week, but instead found each one
to make them feel just as positive (on average) as the one before it. A compliment, like
a good meal, may leave people feeling satisfied for a while, but the hunger for a sense of
belonging likely will not go away for long.
More important, our research also suggests that people can misunderstand how others
react to compliments, at times overestimating the extent to which others will adapt and
“grow tired”of receiving multiple compliments. These results provide an important
extension beyond existing research demonstrating that people underestimate how posi-
tively others respond to receiving compliments at a single point in time (Zhao & Epley,
2020a). This extension is important because relationships require maintenance in the form
of repeated affirmations, assurances, and relational bonding in order to succeed over
time. A person who assumes that another person will be satisfied with a single compli-
ment for two weeks may do less of this maintenance, at least in the form of positive
affirmations and compliments, than would be optimal for both their own and others’
wellbeing.
Several findings from the current experiments indicate that people can expect recipi-
ents to adapt to multiple compliments more than recipients actually do. In Experiment 1,
those who imagined delivering or receiving one new compliment each day for 5 days
expected that each compliment would leave the recipient feeling a little less positive than
the one before, and would also seem a little less sincere. These results emerged regardless
of how the compliment was expressed (in person or through text) or generated (all at
once or one at a time each day). Describing similar results for a host of different stimuli
(i.e., food and music), Kahneman and Snell (1992) argued that people have a “stereotyped
prediction of declining liking for an initially liked stimulus,”suggesting that our partici-
pants’expectations about compliments are not especially unique to prosocial behaviors.
Experiment 3 further suggests that people may expect others to adapt to compliments
primarily when they are considered abstractly, rather than concretely, likely because this
makes each new compliment seem less novel or unique than a series of real compliments
actually is. People’s emotional experience adapts to experiencing the same stimulus over
time, but adapts more slowly (if at all) to experiencing different stimuli (Galak & Redden,
2018;O’Brien, 2019). Mistaken expectations of adaptation could come from
SELF AND IDENTITY 17
underestimating how unique repeated experiences will actually seem to those who are
experiencing them (O’Brien, 2019).
It is worth noting, however, that we do not know from this research whether people
would adapt as expected to receiving the very same compliment repeatedly. Although it is
possible that people adapt to hearing the very same compliment repeated over time, we
would still hypothesize that recipients adapt less than the expressers would anticipate. In
fact, the gap between the expected and actual impact of a compliment could even
increase if expressers assume that recipients adapt even more extremely to the very
same compliment while recipients continue deriving positive value from it. After all, even
the same words expressed over different days can take on new meaning for a recipient,
and hence renewed emotional impact, because the context in which those words are
expressed are likely to have changed over the course of different days. In addition, we
presume that the same words can repeatedly satisfy a person’s need for belonging simply
by reaffirming another person’s positive regard. Hearing a spouse say, “you are beautiful,”
on multiple days is unlikely to grow tired over time, in the same way that a favorite meal is
unlikely to grow tired even when eaten repeatedly (Read & Loewenstein, 1995). Future
research is needed to test these hypotheses.
Interestingly, those who were actually expressing compliments to a recipient in
Experiment 2 did not expect their compliments to grow tired, but instead expected
their recipient to feel more positive in response to the last compliment than they did in
response to the first. This result was unexpected and requires more research to under-
stand fully. At this time, we believe it is most likely that this result emerged because
expressers in this experiment were able to specify the order in which their compliments
would be delivered to a recipient, and so may have arranged them in a way that would
yield an improving sequence over the course of the week. Future research that removes
this confound is necessary to test this hypothesis directly.
Although our experiments suggest that compliments are a surprisingly good way to
make others feel positive, a week of receiving compliments did not significantly increase
the recipients’self-reported relationship satisfaction in Experiment 2. One possible expla-
nation is that we were studying relationships with very high levels of satisfaction even
before recipients entered the week-long experience, which then remained high through
the end of the experiment (Ms = 4.13 and 4.08; SDs = 1.24 and 1.26, on a scale ranging
from −5 to 5). Furthermore, recipients reported that their study partners already gave
them 6.02 compliments per week, on average, prior to entering the week-long experi-
ence. The impact of compliments on a recipient’s mood may be independent of relation-
ship quality, but its impact on relationship satisfaction may be significantly larger in
weaker relationships where satisfaction is lower. Indeed, a large literature on relationship
maintenance strategies has shown that communication of positivity (such as giving
compliments) and assurance (such as telling the relationship partner how much they
mean) often strongly predicts relationship satisfaction among romantic partners (Stafford
& Canary, 1991; Stafford et al., 2000), family members (Myers et al., 2001), and coworkers
(Madlock & Booth-Butterfield, 2012). Considering that participants in a satisfying relation-
ship with plenty of compliments exchanged still underestimated the positive impact of
their compliments in Experiment 2, we believe it would be important to examine how
dyads who rarely exchange compliments expect them to to influence their recipient, and
18 X. ZHAO AND N. EPLEY
whether it meaningfully impacts relationship satisfaction even over the course of a single
week.
Although the experiments we report here have focused primarily on expected versus
actual adaptation to prosocial acts, in this case of receiving compliments, it is worth
noting that our experiments also indicated that recipients generally felt more positive
than others expected. This result emerged when comparing those who wrote and
expressed their compliments to a friend or a romantic partner in Experiment 2, and was
especially strong when examining more generalized expectations about how others
respond to repeated compliments in Experiment 3 (see analyses in the Supplemental
Material). These results add to a growing body of new research reporting similar findings.
People underestimate how positively others respond tocomplete honesty (Levine &
Cohen, 2018), expressions of gratitude (Kumar & Epley, 2018), random acts of kindness
(Kumar & Epley, 2020), requests for help (Zhao & Epley, 2020b), and constructive con-
frontations (Dungan & Epley, 2020). Human beings are deeply social, valuing positive
signals of warmth and liking from others presumably because they serve to satisfying
belonging and relational needs (Leary et al., 1995). These emerging results, however,
suggest that people widely underestimate just how important those social signals are to
others.
Finally, misunderstanding how others respond to prosocial actions matters because
miscalibrated expectations can serve as a barrier to engaging in prosocial actions more
often in daily life. Although we did not test this hypothesis in the experiments we report
here, people’s behavior is often guided by the expected value of possible actions. If
a person believes that others will not react favorably to a repeated series of compliments,
then they refrain from giving as many as they could. As just one anecdotal example of this
possibility, one of the authors (Epley) suggested a new family tradition for the Christmas
holiday to his wife while in the midst of writing this very paper. The idea was to have each
family member write 5 compliments to each other member of the family, one each on
a strip of paper, which could then be stapled into a series of rings and hung up on the
Christmas tree. Each family member would then tear offand read one of the rings
each day. The idea was almost immediately rejected because, “It would get old.”Failing
to fully appreciate just how positive repeated signs of warmth and affirmation could feel
may even keep one of the author's f amily–at least for one holiday season –from feeling
as positive and connected as they could otherwise be.
Notes
1. All analyses were conducted using the R statistical software (R Core Team, 2013) with
functions from the lme4 package and the lmerTest package (Bates et al., 2015; Kuznetsova
et al., 2017).
2. We did not include random slopes in this analysis because the random slope models did not
provide superior model fit compared to the random intercept models based on our data; in
addition, it has been suggested that random slope models may be unreliable for estimating
growth when there are fewer than six time points (Wright, 2017), and we only have five time
points in our design.
3. Two pairs in Experiment 2 each had one participant below the age of 18 according to their
self-reported demographic information at the end of the survey and so had to be removed
from all analyses. Because all participants indicated that they were at least 18 years old during
SELF AND IDENTITY 19
the consent procedure prior to study enrollment, we only noticed those two underage
participants after we completed data collection for all experiments. We therefore based our
sample size targets on the number of participants we believed were included in Experiment 2
at the time we conducted Experiment 3.
4. Due to the technological limitations of our Qualtrics survey design, during the initial data
collection period, some pairs from Experiment 2 got yoked to two participants, while some
did not get any participants. We therefore recruited an additional 21 participants and
randomly assigned them to one of the pairs that did not get a yoked observer in our initial
data collection period.
Disclosure statement
No potential conflict of interest was reported by the authors.
Funding
This work was supported by the Neubauer Family Faculty Fellowship, University of Chicago Booth
School of Business.
ORCID
Xuan Zhao http://orcid.org/0000-0002-0893-8920
References
Abele, A. E., & Wojciszke, B. (2007). Agency and communion from the perspective of self versus
others. Journal of Personality and Social Psychology,93(5), 751–763. https://doi.org/10.1037/0022-
3514.93.5.751
Algoe, S. B., Gable, S. L., & Maisel, N. C. (2010). It’s the little things: Everyday gratitude as a booster
shot for romantic relationships. Personal Relationships,17(2), 217–233. https://doi.org/10.1111/j.
1475-6811.2010.01273.x
Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects models using lme4.
Journal of Statistical Software,67(1), 1–48. https://doi.org/10.18637/jss.v067.i01
Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for interpersonal attachments as
a fundamental human motivation. Psychological Bulletin,117(3), 497–529. https://doi.org/10.
1037/0033-2909.117.3.497
Bowlby, J. (1969). Attachment and Loss, Volume I: Attachment. Basic Books. https://doi.org/10.2307/
588279
Bruk, A., Scholl, S. G., & Bless, H. (2018). Beautiful mess effect: Self-other differences in evaluation of
showing vulnerability. Journal of Personality and Social Psychology,115(2), 192–205. https://doi.
org/10.1037/pspa0000120
Campbell, T., O’Brien, E., Van Boven, L., Schwarz, N., & Ubel, P. (2014). Too much experience:
A desensitization bias in emotional perspective taking. Journal of Personality and Social
Psychology,106(2), 272–285. https://doi.org/10.1037/a0035148
Curry, O. S., Rowland, L. A., Van Lissa, C. J., Zlotowitz, S., McAlaney, J., & Whitehouse, H. (2018). Happy
to help? A systematic review and meta-analysis of the effects of performing acts of kindness on
the well-being of the actor. Journal of Experimental Social Psychology,76, 320–329. https://doi.
org/10.1016/j.jesp.2018.02.014
Dungan, J., & Epley, N. (2020). We need to talk: Misunderstanding others creates a barrier to
constructive confrontation. Manuscript in Revision.
20 X. ZHAO AND N. EPLEY
Fiske, S. T., Cuddy, A. J. C., & Glick, P. (2007). Universal dimensions of social cognition: Warmth and
competence. Trends in Cognitive Sciences,11(2), 77–83. https://doi.org/10.1016/j.tics.2006.11.005
Frederick, S., & Loewenstein, G. (1999). Hedonic adaptation. In D. Kahneman, E. Diener, & N. Schwarz,
(Eds.), Well-being: The foundations of hedonic psychology (p. 302–329). Russell Sage Foundation.
Galak, J., & Redden, J. P. (2018). The properties and antecedents of hedonic decline. Annual Review of
Psychology,69(1), 1–25. https://doi.org/10.1146/annurev-psych-122216-011542
Gleason, M. E. J., & Iida, M. (2015). Social support. In M. Mikulincer, P. R. Shaver, J. A. Simpson, & J. F.
Dovidio, (Eds.), APA handbook of personality and social psychology, volume 3: Interpersonal rela-
tions (p. 351–370). American Psychological Association. https://doi.org/10.1037/14344-013
Holt, S. H., Miller, J. C., Petocz, P., & Farmakalidis, E. (1995). A satiety index of common foods.
European Journal of Clinical Nutrition,49(9), 675–690. https://www.ncbi.nlm.nih.gov/pubmed/
7498104
Kahneman, D., & Snell, J. (1992). Predicting a changing taste: Do people know what they will like?
Journal of Behavioral Decision Making,5(3), 187–200. https://doi.org/10.1002/bdm.3960050304
Kumar, A., & Epley, N. (2018). Undervaluing gratitude : Expressers misunderstand the consequences
of showing appreciation. Psychological Science,29(9), 1423–1435. https://doi.org/10.1177/
0956797618772506
Kumar, A., & Epley, N. (2020). A little good goes an unexpectedly long way: Underestimating the
positive impact of kindness on recipients [unpublished manuscript]. University of Chicago.
Kuznetsova, A., Brockhoff, P. B., & Christensen, R. H. B. (2017). lmerTest package: Tests in linear mixed
effects models. Journal of Statistical Software,82(13), 1–26. https://doi.org/10.18637/jss.v082.i13
Leary, M. R., Tambor, E. S., Terdal, S. K., & Downs, D. L. (1995). Self-esteem as an interpersonal
monitor: The sociometer hypothesis. Journal of Personality and Social Psychology,68(3), 518–530.
https://doi.org/10.1037/0022-3514.68.3.518
Levine, E. E., & Cohen, T. R. (2018). You can handle the truth: Mispredicting the consequences of
honest communication. Journal of Experimental Psychology. General,147(9), 1400–1429. https://
doi.org/10.1037/xge0000488
Loewenstein, G. (2005). Health psychology. Hot-cold Empathy Gaps and Medical Decision Making,24
(4 Suppl), S49–S56. https://doi.org/10.1037/0278-6133.24.4.S49
Loewenstein, G. F., & Prelec, D. (1993). Preferences for sequences of outcomes. Psychological Review,
100(1), 91–108. https://doi.org/10.1037/0033-295X.100.1.91
Madlock, P. E., & Booth-Butterfield, M. (2012). The influence of relational maintenance strategies
among coworkers. Journal of Business Communication,49(1), 21–47. https://doi.org/10.1177/
0021943611425237
Marchese, D. (2019). Brad pitt on the kind of leading man he doesn’t want to be. The New York Times.
https://www.nytimes.com/interactive/2019/12/09/magazine/brad-pitt-interview.html
Myers, S. A., Black, J., Bukaty, A., Callin, A., Davis, L. A., Fairbanks, S. L., Gieron, M. E., Ferry, M. F.,
Kappenman, K., King, M., McGuire, M., Nix, F. D., Saniuk, J., Tracy, M. M., Triolo, T. N., & Valentino, T.
(2001). Relational maintenance behaviors in the sibling relationship. Communication Quarterly,49
(1), 19–34. https://doi.org/10.1080/01463370109385612
Nordgren, L. F., McDonnell, M. H. M., & Loewenstein, G. (2011). What constitutes torture?:
Psychological impediments to an objective evaluation of enhanced interrogation tactics.
Psychological Science,22(5), 689–694. https://doi.org/10.1177/0956797611405679
O’Brien, E. (2019). Enjoy it again: Repeat experiences are less repetitive than people think. Journal of
Personality and Social Psychology,116(4), 519–540. https://doi.org/10.1037/pspa0000147
O’Brien, E., & Ellsworth, P. C. (2012). Saving the last for best: A positivity bias for end experiences.
Psychological Science,23(2), 163–165. https://doi.org/10.1177/0956797611427408
O’Brien, E., & Kassirer, S. (2019). People are slow to adapt to the warm glow of giving. Psychological
Science,30(2), 193–204. https://doi.org/10.1177/0956797618814145
Read, D., & Loewenstein, G. (1995). Diversification bias: explaining the discrepancy in variety seeking
between combined and separated choices.Journal of Experimental Psychology: Applied, 1(1), 34–
49. https://doi.org/10.1037/1076-898X.1.1.34
Rolls, B. J., Rolls, E. T., Rowe, E. A., & Sweeney, K. (1981). Sensory specific satiety in man. Physiology &
Behavior,27(1), 137–142. https://doi.org/10.1016/0031-9384(81)90310-3
SELF AND IDENTITY 21
Stafford, L., & Canary, D. J. (1991). Maintenance strategies and romantic relationship type, gender
and relational characteristics. Journal of Social and Personal Relationships,8(2), 217–242. https://
doi.org/10.1177/0265407591082004
Stafford, L., Dainton, M., & Haas, S. (2000). Measuring routine and strategic relational maintenance:
Scale revision, sex versus gender roles, and the prediction of relational characteristics.
Communication Monographs,67(3), 306–323. https://doi.org/10.1080/03637750009376512
Tomasello, M. (2014). The ultra-social animal. European Journal of Social Psychology,44(3), 187–194.
https://doi.org/10.1002/ejsp.2015
Van Boven, L., Loewenstein, G., Dunning, D., & Nordgren, L. F. (2013). Changing places: A dual
judgment model of empathy gaps in emotional perspective taking. In J. M. Olson & M. P. Zanna
(Eds.), Advances in Experimental Social Psychology: Vol. 48. Advances in experimental social psy-
chology,(pp.117–171). Elsevier Academic Press. https://doi.org/10.1016/B978-0-12-407188-9.
00003-X
Wesselmann, E. D., Cardoso, F. D., Slater, S., & Williams, K. D. (2012). To be looked at as though air:
Civil attention matters. Psychological Science,23(2), 166–168. https://doi.org/10.1177/
0956797611427921
Wojciszke, B., Bazinska, R., & Jaworski, M. (1998). On the dominance of moral categories in impres-
sion formation. Personality & Social Psychology Bulletin,24(12), 1251–1263. https://doi.org/10.
1177/01461672982412001
Wright, D. B. (2017). Some limits using random slope models to measure academic growth. Frontiers
in Education,2(November), 58. https://doi.org/10.3389/feduc.2017.00058
Zhao, X., & Epley, N. (2020a). Insufficiently complimentary?: Underestimating the positive impact of
compliments creates a barrier to expressing them [unpublished manuscript]. University of
Chicago.
Zhao, X., & Epley, N. (2020b). Under-asking: Underestimating others’happiness to help as a barrier to
help-seeking [unpublished manuscript]. University of Chicago.
22 X. ZHAO AND N. EPLEY