P SY CH OL OG I C AL S CIE N CE
Research Report
The Ability to Judge the
Romantic Interest of Others
Skyler S. Place,1 Peter M. Todd,1 Lars Penke,2 and Jens B. Asendorpf3
1
Indiana University, 2University of Edinburgh, and 3Humboldt University of Berlin
ABSTRACT—The
ability to judge another individual’s romantic interest level—both toward oneself and toward
others—is an adaptively important skill when choosing a
suitable mate to pursue. We tested this ability using videos
of individuals on speed dates as stimuli. Male and female
observers were equally good at predicting interest levels,
but they were more accurate when predicting male interest: Predictions of female interest were just above chance.
Observers predicted interest successfully using stimuli as
short as 10 s, and they performed best when watching clips
of the middle or end of the speed date. There was considerable variability between daters, with some being very
easy to read and others apparently masking their true
intentions. Variability between observers was also found.
The results suggest that the ability to read nonverbal behavior quickly in mate choice is present not only for individuals
in the interaction, but also for third-party observers.
It is adaptively important for an individual to be able to evaluate
the interest level of a potential mate. Choosing a mate is a key
component of gene promotion, and it is one of the most central
decisions concerning reproduction across species (Andersson,
1994). Accurately appraising interest minimizes wasted time
and resources and allows for a greater chance of success in a
competitive mating market (Wiegmann & Angeloni, 2007). In
terms of evolutionary life-history theory, it is thus fundamental
for an efficient allocation of mating effort (Kaplan & Gangestad,
2005). Correctly perceiving interest is useful not only for
choosing a mate but also for determining one’s own mate value
(Simão & Todd, 2002), which is important for future mating
decisions (Penke, Todd, Lenton, & Fasolo, 2007; Penke &
Denissen, 2008). Thus, it is beneficial for humans to be able to
pick up on cues that allow them to excel at such appraisals.
These cues could include information available through lan-
Address correspondence to Skyler Place, Psychological and Brain
Sciences, Indiana University, 1101 E. 10th St., Bloomington, IN
47405, e-mail: ssplace@indiana.edu.
22
guage content and tone of voice, as well as nonverbal behaviors
such as body language, social signaling, and eye contact (Ambady & Rosenthal, 1992; Penke & Asendorpf, 2008).
In addition to evaluating a potential mate’s level of interest in
oneself, it is advantageous to be able to evaluate levels of interest between others via observed interactions. This is important for building knowledge of the surrounding social network
(Pentland, 2007), including the availability and desirability of
future potential mates (Simão & Todd, 2002; see also the literature on mate copying in animals—e.g., Dugatkin, 1992,
2000—and in humans—e.g., Jones, DeBruine, Little, Buriss, &
Feinberg, 2007). Observer perception in general has been a
fruitful field for social psychologists: Kenny and colleagues
(Kenny, 1994; Kenny & Albright, 1987; Kenny, Bond, Mohr, &
Horn, 1996) studied ‘‘third-party metaperceptions,’’ with participants observing interactions between pairs of individuals,
and found that people performed above chance at predicting
who feels friendly toward whom. This and other social perceptions can be made accurately with limited information (see
Ambady & Rosenthal, 1992, for a review).
Given the results on accurate observer predictions regarding
friendship, along with the adaptive need for an efficient mechanism to predict interest in mate choice, we hypothesized that
individuals will be able to accurately predict others’ interest in
themselves and in third parties. Here we focus on the latter,
third-party metaperceptions of how romantically interested
other people are in each other. To be adaptive in everyday
situations, the ability to determine this should require only a
limited amount of information, suggesting that performance
should not be hindered by shortened stimuli-presentation times.
Furthermore, because women face greater risks during mate
choice due to their inevitably higher minimal parental investment in potentially resulting offspring (Trivers, 1972), we
predicted that they would behave more cautiously, covertly, and
ambiguously during initial interactions, making their intentions
more difficult to read than those of men (Grammer, Kruck,
Juette, & Fink, 2000; Haselton & Buss, 2000). Finally, we also
investigated the observers’ relationship status as a potentially
confounding factor.
Copyright r 2009 Association for Psychological Science
Volume 20—Number 1
S.S. Place et al.
To test these ideas, we needed a set of mate-choice-relevant
interactions that observers could watch and judge, and for which
there was information on actual romantic interest so we could
assess the observer’s accuracy. Videos of speed dating interactions fulfilled these requirements and also allowed us to limit the
information available to our judges by presenting them with
clips of various durations.
METHOD
Participants
The study included 54 participants—28 women (mean age 5
19.8 years, SD 5 3.8 years; 14 in relationships) and 26 men
(mean age 5 19.5, SD 5 1.1 years; 9 in relationships). Participants were recruited from the Indiana University psychology
participant pool and were compensated with research credits
required for undergraduate coursework. Participants were
screened to be over 18 years old, to be heterosexual, and to have
no knowledge of the German language (because the stimuli were
in German; see next section).
Participants watched shortened video clips that were either
10 s or 30 s long and came from the beginning, middle, or end of
the date (three temporal locations). For each of the 24 interactions we used, each participant saw four clips (in randomized
order, both within and across interactions): 10-s clips from
all three locations and one 30-s clip from a location that was
randomized across interactions. The experimental design was
therefore a 2 (observer sex: male, female) 2 (relationship
status: single, in relationship) 3 (clip location: beginning,
middle, end) 2 (clip length: 10 s, 30 s) mixed factorial design.
Procedure
Participants first provided their age, sex, ethnicity, and relationship status. Our dependent measure was the observing
participant’s perception of the interest within each speed dating
interaction they watched. Observers answered two questions
after each video clip: ‘‘Do you think the man was interested in
the woman?’’ and ‘‘Do you think the woman was interested in
the man?’’ Their binary ‘‘yes’’ or ‘‘no’’ answers were then compared to the binary decisions of the actual speed daters.
RESULTS
Stimuli
The videos of mate-choice situations were gathered during a
series of laboratory-based speed dating sessions run at Humboldt University in Berlin, Germany. Speed dating is a paradigm
designed to allow singles to meet a large number of possible
mates in a short period of time (Finkel & Eastwick, 2008). The
individuals who participated in the Berlin Speed Dating Study
(BSDS) were recruited using advertising and publicity in media
outlets; in exchange for free speed dating, they agreed to have
their interactions videotaped and to provide additional data on
themselves. Seventeen sessions of speed dating were run as part
of the study, for a total of 382 participants.
The ‘‘dates’’ took place in separated booths, and each lasted
for 3 min, at the end of which each individual wrote down
whether he or she was interested in seeing that date again (an
‘‘offer’’). Pairs making mutual offers were given each other’s
contact information after the session so they could meet again.
The videos of these interactions were the stimuli used in our
experiment. Each of the two individuals in a speed date was
filmed with a separate over-the-shoulder camera, and these two
videos (with audio in German) were shown in a synchronized
side-by-side combination to our participants. These combined
video presentations, which we refer to as a video clip, allowed a
naturalistic view of the date. Videos of 24 interactions were used
in this experiment, randomly selected from two different sessions comprising speed daters in their 20s; each person appeared in only one video. This sample matched the entire
population of interactions from the BSDS sessions with regard to
offer rates from men (41%) and women (33%), as well as rates of
mutual interest between individuals (15%).
Volume 20—Number 1
The first question posed was whether observers could predict
romantic or dating interest between others accurately. Figure 1
presents the results for prediction of male interest and for
prediction of female interest separately, collapsing across all
within-subjects conditions. A paired-sample t test showed a
significant difference between the two measures, t(53) 5 3.64,
prep 5 .986, d 5 1.00. It is important to note that observers could
achieve a chance accuracy above 50% in this task if they had
knowledge of the fact that daters make offers less than half
the time (see above), and they could have had such knowledge
through past dating experience (participants were not explicitly
informed of the interest rates prior to the start of the experiment).
If they took account of the actual offer prevalence rates, the best
that observers could do at chance would be 52% (.41 .41 1
[1 .41] [1 .41]) for predicting male interest and 56% (.33
.33 1 [1 .33] [1 .33]) for predicting female interest,
calculated using the base rates of interest present in the stimuli
set.
Observer performance on both of the dependent measures
was significantly better than these adjusted chance levels for
predicting male interest, t(53) 5 10.76, prep 5 .986, d 5 2.94,
but was just above chance for predicting female interest, t(53) 5
2.24, prep 5 .908, d 5 0.62. (See Ambady & Rosenthal, 1992, for
similar accuracy ranges for other thin-slice social perceptions.)
Each dependent measure was analyzed for contributing factors using a mixed-factor analysis of variance (ANOVA). For
predicting male interest in females, there was no effect of sex.
There was an effect of relationship status, such that individuals
in relationships outperformed individuals who were single,
23
Judging Romantic Interest
Accuracy (proportion correct)
.70
.68
Accuracy (proportion correct)
.66
.64
.62
Male Interest in Females
Female Interest in Males
.65
.60
.55
.50
.45
Beginning
.60
Middle
End
Video-Clip Location
.58
Fig. 2. Accuracy in predicting romantic interest of speed-dating participants by video-clip location. Error bars show standard errors of the
means.
.56
.54
.52
.50
Male Interest
in Females
Female Interest
in Males
Fig. 1. Overall accuracy in predicting romantic interest of videotaped
speed-dating participants. Dashed lines indicate chance performance
levels for predicting interest for daters of each sex. Error bars show
standard errors of the means.
F(1, 50) 5 6.18, prep 5 .935, Zp 2 ¼ :11. The length of the video
clip presented had no effect on accuracy. There was, however,
an effect of video-clip location, F(2, 100) 5 16.86, prep 5 .986,
Zp 2 ¼ :25. None of the possible interactions reached significance. For predicting female interest in males, sex, relationship
status, and video presentation length were all not significant. As
in the male-interest data, there was a significant effect of videoclip location, F(2, 100) 5 16.18, prep 5 .986, Zp 2 ¼ :41.
We further analyzed the within-subjects factor of location
using a single-factor ANOVA (with levels beginning, middle,
and end), revealing a significant difference of location for predicting both male interest in females, F(2, 106) 5 29.35, prep 5
.986, Zp 2 ¼ :36, and female interest in males, F(2, 106) 5
36.52, prep 5 .986, Zp 2 ¼ :41. Figure 2 shows that the best
performance at judging interest came from viewing clips from
the middle and end of the interaction. Post hoc comparisons
using the Bonferroni correction showed significant differences
between the beginning clip and both the middle and end clips for
both male interest and female interest.
To look at how ‘‘readable’’ individual speed daters were, the
data were further analyzed at the per-dater level (see Fig. 3).
Daters were sorted from most-accurately predicted (mean ac-
24
.70
curacy across all observers) to least. This was done by individual, not by interaction—a single video could include a woman
who was very easy to read (yielding high accuracy) and a difficult-to-read man (yielding low accuracy), and we assessed
readability of these two individuals separately. In fact, being
accurate at predicting the dating interest of one sex does not
help in predicting the interest level of the other: The correlation
between accuracy in predicting male interest and accuracy in
predicting female interest in the same video was zero, r 5 .00,
prep 5 .083.
The solid lines in Figure 3 show participants’ mean accuracy
at predicting all males and all females. The steep downward
slope of both lines indicates the wide range in observers’ ability
to predict the interest level of different individuals. To find out
if these results are different from those expected by chance, we
ran a Monte Carlo simulation designed to determine chancelevel performance based on guessing. The simulation generated
a set of responses for each observer, for each video, for each of
the four partial clips. Predictions of interest in each case were
chosen randomly according to the interest-judgment rates
(60%) of observers. The simulation was run 1,000 times, and
the responses were averaged within each observer for each dater
in the videos and then averaged across observers. The results
rank-ordered across daters (Fig. 3, dashed lines) have a slope
that is less steep than that of the experimental data. Both the
male and the female human-observer data fall clearly above the
95% confidence intervals of the Monte Carlo simulation in the
11 daters who were easiest to predict. In addition, the five women daters who were the hardest for observers to read fall below
the 95% confidence intervals, showing that observers were
systematically fooled in these cases.
DISCUSSION
The data supported our two main hypotheses: Observers were
able to assess the dating interest of others at above-chance
Volume 20—Number 1
S.S. Place et al.
1.0
.9
Accuracy (proportion correct)
.8
.7
.6
.5
.4
.3
Male Interest in Females
Female Interest in Males
.2
Monte Carlo Simulated Male Interest in Females
.1
Monte Carlo Simulated Female Interest in Males
.0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Daters, Rank-Ordered by Observer Accuracy
Fig. 3. Accuracy in predicting each dater’s interest, comparing human judges with Monte Carlo simulations. Error bars for observed data show
standard errors of the means. Error bars for simulated expected data show 95% confidence intervals. Open circles indicate daters who were not
interested; closed circles indicate daters who were interested.
levels, and the length of time required to do so was brief. For
both sexes, accurately perceiving romantic interest both of and
toward potential mates holds evolutionary benefits through the
efficient allocation of mating effort. Our results suggest that men
and women possess this adaptive ability. Whether it is the result
of a domain-specific adaptation or a more general ability for
social perception remains to be determined. Furthermore, as
predicted, it was on average easier for observers to gauge men’s
intentions than it was to gauge women’s intentions (though there
was high variance in observers’ performance levels across individual daters of both sexes). The lower overall accuracy concerning women’s intentions was not due to observers guessing or
performing at chance but to a systematic overperception of female daters’ interest (Fig. 3)—surpassing 80% erroneous interest predictions for the five hardest-to-read women.
This dramatic rate of incorrect perception supports our hypothesis that women are harder to read, presumably because
they mask their true intentions: As Grammer et al. (2000) argued, the biologically deep-rooted sex inequality in parental
investment (Trivers, 1972) puts greater risks on the females of
a species during mate choice. As a result, females, including
women in speed dating (Todd, Penke, Fasolo, & Lenton, 2007),
are much more critical and picky when making mate-choice
decisions. And, in order to evaluate potential mates longer
Volume 20—Number 1
without signaling their true intentions, women behave more
covertly and ambiguously during initial interactions with the
opposite sex. Men, in contrast, face lower risks and consequently should be less likely to hide their intentions.
In our study, observers only saw an individual interacting on
one date, but perhaps if multiple dates with the same individual were presented, observers would be better able to
differentiate instances of deceptive and true interest from that
individual.
Whereas the degree of observer accuracy seems to depend
heavily on the individual dater being watched, the length of
time spent watching has almost no effect. However, a systematic
difference in observer performance appears when comparing
across video-clip locations: In our study, the best observer
judgment performance came for video clips taken from the
middle and end of the dates. This may arise because daters are
using the information they gather throughout their brief encounter to make their ultimate decisions, so that their decisions
are not fully determined, and therefore not fully readable by
others, until later in the encounter. If true, this would counter a
major critique of speed dating as a method of finding a long-term
partner: that people are using only physical attractiveness to
make their dating decisions because they do not have the time to
assess much else (Eastwick & Finkel, 2008; Kurzban & Weeden,
25
Judging Romantic Interest
2005; Todd et al., 2007). Other data are needed to determine
whether daters are using multiple cues over time or just taking
time to register an attractiveness-driven decision.
Some observers also appear to be better at using the available
information for making some judgments. Whereas we did not
aim to identify the individual differences underlying good observers (see Funder, 2001), we did find that observers who
indicated they were currently in a relationship did better at
predicting male interest than did those who were currently
single. This suggestive finding could stem in part from learning
through relationship experiences. Alternatively, it is possible
that the social skills necessary to succeed in finding and
maintaining a relationship also support the ability to correctly
perceive romantic interest. Studying younger observers before
they have much relationship experience could help to disentangle these (and potentially other) hypotheses.
The results of this study add to the body of findings on the
abilities of naive observers to make quick and accurate judgments, demonstrating that this ability extends to assessments of
romantic interest in the mate-choice domain as well. We have
shown this through a novel method that provides a strong criterion against which the observer judgments were evaluated:
unambiguously stated, consequential mate-choice decisions of
actual partner-seeking singles meeting available potential
mates while speed dating. With limited information, observers
can make accurate judgments of mate-choice decisions, though
their abilities may be hampered by the desire of some daters to
mask their true intentions.
Acknowledgments—This research was supported by Grant As
59/15 of the German Research Foundation (DFG), awarded
to Jens Asendorpf. We would like to thank Marie-Luise Haupt,
Karsten Krauskopf, Harald Schneider, and Sebastian Teubner
for their help with the Berlin Speed Dating Study. Lars Penke is
supported by the United Kingdom Medical Research Council
(Grant No. 82800) and is part of the University of Edinburgh
Centre for Cognitive Ageing and Cognitive Epidemiology.
Funding from the Biotechnology and Biological Sciences Research Council, Engineering and Physical Sciences Research
Council, Economic and Social Research Council, and Medical
Research Council is gratefully acknowledged.
REFERENCES
Ambady, N., & Rosenthal, R. (1992). Thin slices of expressive
behavior as predictors of interpersonal consequences: A metaanalysis. Psychological Bulletin, 111, 256–274.
Andersson, M.B. (1994). Sexual selection. Princeton, NJ: Princeton
University Press.
Dugatkin, L.A. (1992). Sexual selection and imitation: Females copy
the mate choice of others. American Naturalist, 139, 1384–1389.
Dugatkin, L.A. (2000). The imitation factor: Evolution beyond the gene.
New York: Free Press.
26
Eastwick, P.W., & Finkel, E.J. (2008). Sex differences in mate preferences revisited: Do people know what they initially desire in a
romantic partner? Journal of Personality and Social Psychology,
94, 245–264.
Finkel, E.J., & Eastwick, P.W. (2008). Speed-dating. Current Directions in Psychological Science, 17, 193–197.
Funder, D.C. (2001). Accuracy in personality judgment: Some research and theory concerning an obvious question. In B. Roberts
& R. Hogan (Eds.), Personality psychology in the workplace (pp.
121–140). Washington, DC: American Psychological Association.
Grammer, K., Kruck, K., Juette, A., & Fink, B. (2000). Non-verbal
behavior as courtship signals: The role of control and choice in
selecting partners. Evolution and Human Behavior, 21, 371–390.
Haselton, M.G., & Buss, D.M. (2000). Error management theory:
A new perspective on biases in cross-sex mind reading. Journal
of Personality and Social Psychology, 78, 81–91.
Jones, B., DeBruine, L., Little, A., Burriss, R., & Feinberg, D. (2007).
Social transmission of face preferences among humans. Proceedings of the Royal Society B: Biological Sciences, 274, 899–903.
Kaplan, H.S., & Gangestad, S.W. (2005). Life history theory and
evolutionary psychology. In D.M. Buss (Ed.), The handbook of
evolutionary psychology (pp. 68–95). New York: Wiley.
Kenny, D.A. (1994). Interpersonal perception: A social relations analysis. New York: Guilford Press.
Kenny, D.A., & Albright, L. (1987). Accuracy in interpersonal perception: A social relations analysis. Psychological Bulletin, 102,
390–402.
Kenny, D.A., Bond, C.F., Jr., Mohr, C.D., & Horn, E.M. (1996). Do we
know how much people like one another? Journal of Personality
and Social Psychology, 71, 928–936.
Kurzban, R., & Weeden, J. (2005). HurryDate: Mate preferences in
action. Evolution and Human Behavior, 26, 227–244.
Penke, L., & Asendorpf, J.B. (2008). Beyond global sociosexual
orientations: A more differentiated look at sociosexuality and
its effects on courtship and romantic relationships. Journal of
Personality and Social Psychology, 95, 1113–1135.
Penke, L., & Denissen, J.J.A. (2008). Sex differences and lifestyledependent shifts in the attunement of self-esteem to selfperceived mate value: Hints to an adaptive mechanism? Journal
of Research in Personality, 42, 1123–1129.
Penke, L., Todd, P.M., Lenton, A.P., & Fasolo, B. (2007). How selfassessments can guide human mating decisions. In G. Geher &
G. Miller (Eds.), Mating intelligence: Sex, relationships, and the
mind’s reproductive system (pp. 37–75). Mahwah, NJ: Erlbaum.
Pentland, A. (2007). On the collective nature of human intelligence.
Adaptive Behavior, 15, 189–198.
Simão, J., & Todd, P.M. (2002). Modeling mate choice in monogamous
mating systems with courtship. Adaptive Behavior, 10, 113–136.
Todd, P.M., Penke, L., Fasolo, B., & Lenton, A.P. (2007). Different
cognitive processes underlie human mate choices and mate
preferences. Proceedings of the National Academy of Sciences,
USA, 104, 15011–15016.
Trivers, R.L. (1972). Parental investment and sexual selection. In B.
Campbell (Ed.), Sexual selection and the descent of man (pp. 136–
179). Chicago: Aldine-Atherton.
Wiegmann, D.D., & Angeloni, L.M. (2007). Mate choice and uncertainty in the decision process. Journal of Theoretical Biology,
249, 654–666.
(RECEIVED 4/2/08; REVISION ACCEPTED 7/9/08)
Volume 20—Number 1
Lars Penke