You are on page 1of 10


PSSXXX10.1177/0956797616687364Lee, AndersonReading Mind From Eyes

Research Article

Psychological Science

Reading What the Mind Thinks From 1­–10
© The Author(s) 2017
Reprints and permissions:
How the Eye Sees
DOI: 10.1177/0956797616687364

Daniel H. Lee1,2 and Adam K. Anderson3,4
Department of Psychology and Neuroscience, University of Colorado Boulder; 2Institute of
Cognitive Science, University of Colorado Boulder; 3Department of Human Development,
Cornell University; and 4Human Neuroscience Institute, Cornell University

Human eyes convey a remarkable variety of complex social and emotional information. However, it is unknown
which physical eye features convey mental states and how that came about. In the current experiments, we tested
the hypothesis that the receiver’s perception of mental states is grounded in expressive eye appearance that serves
an optical function for the sender. Specifically, opposing features of eye widening versus eye narrowing that regulate
sensitivity versus discrimination not only conveyed their associated basic emotions (e.g., fear vs. disgust, respectively)
but also conveyed opposing clusters of complex mental states that communicate sensitivity versus discrimination
(e.g., awe vs. suspicion). This sensitivity-discrimination dimension accounted for the majority of variance in perceived
mental states (61.7%). Further, these eye features remained diagnostic of these complex mental states even in the
context of competing information from the lower face. These results demonstrate that how humans read complex
mental states may be derived from a basic optical principle of how people see.

facial expressions, emotions, social perception

Received 3/22/15; Revision accepted 12/12/16

Humans can read highly complex mental states from the Expressive emotional features that reconfigure the eyes’
eyes of other humans (Baron-Cohen, Wheelwright, Hill, appearance may have arisen as sensory adaptations for
Raste, & Plumb, 2001). Eye reading is a critical ability the expresser that were organized according to Darwin’s
associated with people’s capacity for mentalizing and principle of antithesis, which posits opposing form and
appreciating that other individuals have minds, inten- function (Darwin, 1872/1998; Susskind et al., 2008). We
tions, and feelings (Baron-Cohen, 1995). The additional have shown recently that expressive eye widening (e.g.,
contrast afforded by the white sclera in human eyes, in fear) and eye narrowing (e.g., in disgust) are associ-
unique among primates (Kobayashi & Kohshima, 1997), ated with opposing optical consequences and serve
highlights how people’s eyes have physically evolved to opposing perceptual functions (Lee, Mirza, Flanagan, &
support their salient role in human social and emotional Anderson, 2014). Operating by the same physical princi-
communication (Lee, Susskind, & Anderson, 2013). Fur- ples as the lens aperture of a camera, eye widening
ther, the superior temporal sulcus and gyrus, regions that enhances sensitivity, gathering more light information for
are responsive to the eyes (Allison, Puce, & McCarthy, vigilance (e.g., during fear; Öhman & Mineka, 2001;
2000; Calder et al., 2007), are also neighbors of regions Whalen, 1998), whereas eye narrowing enhances acuity,
that support theory of mind (Saxe & Powell, 2006). exerting sharper focus of light information for discrimina-
Despite the prominence of the eyes and surrounding tis- tion (e.g., during disgust; Chapman & Anderson, 2012;
sue in mental-state signaling, the physical features that
convey such complex states, and how they came to do
Corresponding Author:
so, remains unknown. Daniel H. Lee, Department of Psychology and Neuroscience,
Although the eyes may be the proverbial windows to University of Colorado Boulder, 345 UCB, Boulder, CO 80309
the soul, they originated as conduits of light to the retina. E-mail:

fear is similar thus hypothesized that expressive features associated to disgust in their negative valence. For these analyses. Susskind et al. ination. & Clore. and eyebrow curva- tures. eyebrow distance. This is important for two reasons. Our thesis is that the opposing effects of physical properties (not conceptual properties) are para- eye widening and narrowing on the expresser’s visual mount because they constitute the initial signals sent to be perception have been socially co-opted to denote oppos. We thus created averaged statistical models of the eyes’ tively. 1969) because similar facial features see the Supplemental Material available online). The notion that the form of facial expressions origi- nates in adaptive sensory function for the sender (Lee General Method: Expression and et al. 1980) may with this eye-opening dimension that dynamically regu. respectively (Ungerleider & Mishkin. 1987). including six basic emotions. ing of expressions in sensory function implies that the 2007). Thus. Barlett. diverge in their expressive form and function (e. decoded by the receiver. another. eyebrow slope.. curvature. creating an exemplar for each extracted seven eye features from each exemplar: vertical expression by averaging across affective faces from two eye aperture. Although have been shown to communicate not just basic emotions participants rated these exemplar expressions for per- but also complex social information such as personality traits ceived mental states. discrimination would group along eye-narrowing features.. respec. Euclidean coordinates from the different mental states.g. (slope from start to end of eyebrow). 2008) for the six basic expres- may originate from this simple perceptual opposition. because how people have come to read a variety of mental states we are examining how mental states are transmitted. 2008) to understand. our analysis examined the expressive (Gill. fear lates the quality of how humans see may help explain opposes disgust. which mental states. sions (Ekman et al. 2014. We Eye-narrowing disgust expressions are sometimes con. 2008). & Friesen. 2014. the ground- 2008. Although eyes were presented in isolation in Exper- out the visual system. ity and discrimination functions dovetail with a division and that these mental-state groups would oppose one of information-processing needs that is present through. We then appearance models were used to compute eye aperture took a dimensional approach (Oosterhof & Todorov. 2008. 1). 2008) supports the idea that there would be similarity in signaling across cultures Feature Modeling (Ekman. Aligned with our thesis that mental states are rooted ture (angular change from start to end of eyebrow). 1976. Haidt. features (rather than the emotion categories) underlying 2009). states).. we examined their influence in the context of full crude but fast magnocellular system and slow but sharp faces in Experiment 2. information. We thus modeled the eyes and surrounding features these mental-state perceptions. eyebrow databases (Ekman & Friesen. which suggests a they do because they are adaptive interfaces with the envi- fundamental dimension of appearance and meaning that ronment (Darwin. Matsumoto & Ekman. for variations across Examining how mental states are read from the eyes began cultures. 1872/1998). eyebrow slope ing how people read mental states from physical eye fea. If the eyes are windows into inner parvocellular system (Livingstone & Hubel. . some mental states is not restricted to just categories of fear and disgust. 1988. These emotionally expressive effects occur along appearance of expressions is not arbitrary. 1976.. Susskind et al. based our examination on the physical similarity of fea- fused with physically similar anger expressions but are less tures (rather than the conceptual similarity of mental often confused with eye-widening fear (Aviezer et al. temporal wrinkles.. Littlewort.. both alone and in the context of competing facial 1982). eyebrow distance Russell & Barrett. Susskind.. we of six basic expressions. Anderson Sherman. Sebe.g. 2012). 2013) and contexts: with a simple but important methodological approach. tures should convey mental states of sensitivity and discrim- tral “what” streams. in  features shaped for sensory function. see also Nelson & Russell.. & Friesen. Sorenson. they look as a physical continuum (Lee et al.2 Lee. Movellan. the from the eyes. 2013. Russell. 2014). & Todorov. 1999. our hypothesis Wrinkle features were computed as the amount of high- was  that mental states conveying sensitivity would group spatial-frequency information extracted from correspond- along eye-widening features. Second. then eye-widening and eye-narrowing fea- channel information to dorsal “where” streams and ven. sions using cross-ethnic facial-identity databases (Ekman We grounded our investigation in basic emotional expres. These opposing sensitiv. 1969. & Anderson. for details. whereas those conveying ing areas (see the Supplemental Material for details). Said. (distance from top to bottom of eye). Matsumoto & Ekman.. First. from retinal rods and cones to the iment 1. such that even opposing complex mental states appearance (Susskind et al. we collected participant ratings of 50 below the eyes (Fig. Jack. Susskind et al. We that are judged as conceptually similar (e. Garrod. (distance from top of eye to eyebrow). nasal wrinkles. & Schyns. and wrinkles 1988).. In Experiment 1. ing mental states of sensitivity and discrimination.

1. 2006) to chart a navigable map of percentage of times the fear rating was higher than all of mental states based on features (Fig. The tar.6% (95% CI = [75.6%. PA). mean overall accuracy of 88. not at all.6. We computed the basic. followed eye-feature mental-state map. 93.7° from a viewing distance pendent correlations across participants for each of the of 40 cm). to 9. we com- Twenty-eight undergraduates at the University of pared each eye stimulus’s matching basic-emotion rating Toronto participated for course credit (approximately 30 with those for all remaining 49 mental states in pairwise participants were planned. tures reveals their conceptual similarity.000 ms or until response. exemplar eyes for 50 different mental-state terms (six one-sample t test. r2 = . Psychology Software Tools. states × 2 repetition blocks.6%.0001. The eye-feature space. region’s ability to convey basic emotions (i. mation about discrete mental states. Below that was a mental-state term and a rat.Reading Mind From Eyes 3 Disgust Disgust Fear Temporal Wrinkles. how all 50 mental states were perceived. p < .2. From left to right. very strongly).696. t(27) = 20. Experiment 1: Mental-State Map counting as chance. p < . ing scale (from 1. p < .0%. t(27) = 7. the eye region reliably conveyed diagnostic infor- Participants rated the eyes in a within-subjects. the images (extracted from expressions of disgust and fear) illustrate differences in eye aperture and eyebrow slope. 100%]) for sadness. or mental states. Hubert.4% (95% confidence interval. p < .9. terms in this space to capture the relationships between Pittsburgh. in a seven-dimension break of up to 1 min was provided halfway through. and this test produced similar results. plemental Material). and the regions from which wrinkle information was extracted.0. Two mental the remaining five basic-emotion ratings. & Meulman. differences in eyebrow distance and eyebrow curve.0. perceived mental states in terms of features (by contrast.5%]). We conducted an extended test of the eye variate relationship to the seven eye features.  Illustration of the eyes and eye features used in the experiments. 94. for definitions of mental for anger. 1980). CI = [91.e. 50%). Thus. on the basis of eye features. r2 = .0001. across all basic-expression eyes. 350 combinations of seven features and 50 mental states. We then analyzed the ratings’ multi.5° × 4. accuracy for fear eyes was the Arabie. t(27) = 24. wise comparison (e. ran. we aimed to map.2%.. see Table S1 in the Sup.0001. 1980.. The resulting 7 × 50 matrix of r values was treated as 50 get screen remained on for 8. repetitions were averaged). Nasal Wrinkles. Eyebrow Slope and Eye Aperture Eyebrow Curve and Eyebrow Distance and Wrinkles Under the Eyes Fig. ple (for demographics. to 96. r2 = . t(27) = 19. A coordinates. we collected participants’ ratings of the six CI = [86.8%. domized. 2b). We then computed the dissimilarities experiment was run on an IBM-compatible computer (euclidean distance) among the ratings of all mental-state running E-Prime (Version 1. matching previous behavioral comparison). scaling (distance variance accounted for 54. To create this Each trial first showed a fixation cross (500 ms).0001. We first examined whether the eye region alone was examining the relationships between ratings without fea- diagnostic of basic emotions.934.4% (95% states used in this experiment. with studies).937.7%]) expanded on Plutchik. Averaging the percentage of correct recognition across participants revealed a high To examine the relationship between eye features and mean accuracy of 90. we first computed inde- by eyes (subtending 7. This was an ethnically diverse population sam.g. full factorial design (6 exemplars × 50 mental Next. with ties states that were similarly correlated across features were . Mean basic emotions and 44 complex mental states that accuracy ranged from 84. see the Supplemental Material). emotion recognition accuracy for each eye stimulus We created a map of the eyes’ capacity to communicate according to whether the matching basic-emotion rating various mental states by using circular unidimensional or some other basic-emotion rating was higher in pair. one for each mental state. Russell.

Eye features associated with joy mental states (indicated by distinct colors in Fig. 2a. On the opposite eye narrowing or by the dimension of valence. eye-widening features were aligned ducted a principal components analysis of the (7 × 50) with a cluster of mental states that convey information eye-feature mental-state dissimilarity relationships. The diagram in (b) summarizes the eye-feature patterns for each cluster. sus. brow slope and nasal wrinkles for disgust and brow distance and eye aperture for fear). had inverse correlation patterns (Fig.25 . and similarly correlated. color-coded as in (a).e. The k-means cluster analysis identified four clusters of mental states.  Results from Experiment 1: relationship between mental states and eye features. which are denoted by different colors. 2b). cowardice. such as awe. 2. sensitivity. The brighter the square. the labels for the basic emotions matching the exemplar eye stimuli are in larger type than the labels for complex mental states. S1 in the Supplemental Mate.4 Lee. see Fig. interest. states. For reference. the greater the magnitude of the Pearson r value (|r|). anticipation. These clusters were characterized by distinct To calculate the placement of mental-state groups and eye-feature combinations that opposed each other and their boundary transitions along the mental-state circle. To examine whether mental-state distance was states that convey social discrimination. Largely orthog- we submitted the same dissimilarity data to a k-means onal to this opposition was a dimension anchored by the cluster analysis. This revealed four distinct clusters of two remaining clusters. and contempt. The opposition of disgust and anger (in which eye narrowing enhances discrimination) to fear and surprise (in which eye widening enhances sensitivity) is illustrated by the fact that the former were maximally distant from the latter on the circle. In the map of mental states shown in (a). aggressiveness. we con- side of the circle. Correlation of mental states to each eye feature was averaged across all mental states within each cluster.. The illustration of eyes of disgust and fear (c) highlight key features of those expressions (i. The . such as hate. closer together relative to two mental states that were dis. whereas distinctly opposing eye features of sad- rial).50 Contempt Defiance c Bravery Disgust Pride Boastfulness Calm Anticipation Serenity Joy Interest Harmony Acceptance Trust Love Fear Vitality Optimism Sincerity Admiration Desire Gratitude Fig. for aligned with a cluster of positively valenced mental further information. The mental-state map revealed that eye-narrowing ness aligned with a cluster of negatively valenced mental features of disgust were aligned with a cluster of mental states. better captured by the dimension of eye widening versus picion. Anderson a b Wrinkles Below Eyes Apprehension Uneasiness Temporal Wrinkles Distraction Remorse Boredom Nasal Wrinkles Brow Distance Submission Eye Aperture Passiveness Brow Curve Brow Slope Tension Neglect Fatigue Sadness Insult Conflict Pensiveness Disapproval Puzzlement Pessimism Suspicion Cowardice Envy Annoyance Fear Awe Insincerity Disgust Surprise Aggressiveness Anger Hate |r | Deceptiveness 0 . the perceived similarity of mental states across eye features is indicated by the mental states’ proximity to one another. the plus and minus signs indicate the direction of correlation of each eye feature in each cluster.

The experimental ence in cluster isolation.. We first tested ity cluster had a smaller set of mental states and was more this for congruent expressions (wide-eyed fear and sur- isolated from other mental-state clusters. By contrast. eyes of disgust and anger with all nonmatching to read their meaning. If eye features associated with wid- eigenvalue threshold of 1. comparing observer to diagnose a particular mental state.e. r2 = . and narrow eyes alone in Experiment 1. see Fig. pared ratings between wide-eyed and narrow-eyed ceived mental states. S2 in the Supplemental Material). mouths of fear all mental-state distances across all participants revealed and surprise with all nonmatching eyes vs. coward- ter had the fewest number of mental states and was more ice. surprise.7. The widening-narrowing ening versus narrowing are used as diagnostic signals of dimension was the primary component. states in the sensitivity cluster were the most isolated We hypothesized that wide-eyed and narrow-eyed full (ps < ..720. 95% CI = [0. expressions would convey opposing mental states of whereas the remaining clusters were similarly distant from sensitivity and discrimination.0001. the opposing discrimi.e.489. nation cluster had more mental states that were closer eyes of fear and surprise with all nonmatching lower together. covered the mental-state map in Figure 2a: six basic states The distribution along the mental-state map and the (disgust. ratings between wide-mouthed and narrow-mouthed Supporting this relative distance. respectively.5° × 7. We then selected a sub- ponents captured 88. disgust and anger with all nonmatching eyes).0001. hypothesizing that the faces would convey conveyed by expressions (sensitivity or discrimination) the mental states associated with the eyes despite the were mirrored by the perceptual needs of the observer— incongruence. which revealed a significant differ. six com- k-means cluster analysis revealed that the sensitivity clus. to determine the importance of narrower eyes required more discrimination for the the eyes. narrow-eyed disgust and anger). For mental states of These results reflect judgments of eyes in isolation. 46) = 14. awe. and apprehension). explaining 61. we measured the distance from that mental Toronto participated for course credit (approximately 30 state to each of the 49 other states and calculated the participants were planned).44].5°. These results show that the sensitiv. We then com- greater diagnosticity (i. We subtracted ratings for narrow-eyed expressions from those for wide-eyed expressions to calculate difference Experiment 2: Reading Eyes in Full scores. boredom. picion) were highly significantly greater for expressions . a complete analysis of expressions with incongruent eyes (i. To measure cluster isola. plars modeled in Experiment 1. cantly different from the mean distance between mental states. tion. and temporal wrinkles.. Eye is visible. expressions (Fig.28).868. suspicion. η2 = setup and trial structure were identical to the setup and . t(27) = 8. pride.32. seam- narrowing component. 1. states (interest. We discrimination. Together. Therefore. Participants rated all 36 faces mean distance). the valence dimension was the second. 1. S1a in the Supplemental Material). sizing that they would not as strongly convey the same The distance between hate and awe.. then of the variance. We first compared congruent wide-eyed. the mental contents lower faces). and remorse). anger. and eyebrow curvature. fear.8% of the total variance. expressions with incongruent lower-face features (i. which suggests prise vs. less ambiguity) of their per. which suggests greater discrimination required faces vs.15. we created a set of 36 mixed aperture and eyebrow distance loaded on the widening. For detailed set of appropriate mental-state terms that uniformly results. hypothe- tal states communicated through physical eye features.0001. full- Face Expressions face expressions (disgust and anger). full-factorial design (for details.7% mental states of sensitivity versus discrimination. p < . found in lessly combining the six upper regions (eyes) and six joy eyes. gust and anger) and complex mental states (hate and sus- tions toward evaluating the mental contents in complex. and four in-between isolated than the other clusters. p < . found in sad eyes. Faces subtended 7. mental states of sensitivity and discrimination. see the Supplemental Material. we first computed an isolation index for each mental Twenty-nine new undergraduates at the University of state (i. joy. F(3. these com. full-face expressions (fear and surprise) with narrow-eyed. tal Material). To test this. we examined the converse situation.e.Reading Mind From Eyes 5 analysis indicated that two components met the critical full-face expressions. observers should rely on them even when the lower face ary component.e. lower regions (nose and mouth) of the expression exem- loaded on the valence component. Finally.8% of the variance. was signifi. plementary complex states (hate. and sadness). averaged ratings of basic emotions (dis- next investigated the importance of the eyes’ contribu. random- by cluster and submitted the four groups to a one-way ized. after distance scores across all 946 mental-state pairs Congruent expressions (44 mental states × 43/2) were mean-centered at 0. We then grouped these isolation indices on all 16 mental-state terms in a within-subjects. see the Supplemen- analysis of variance. mouths of that hate and awe were the most opposed complex men. A Bonferroni post hoc analysis showed that mental structure used in the previous experiment. admiration. explaining 26. as did wide each other (ps > .

19]. 3).725. see Fig. SD = ences in eyes alone.55]. difference lar basic and complex mental states whether the eyes are score = −2. p < . we averaged our data set across basic-emotion ratings. Difference scores for eyes only and for congruent faces were calculated by subtracting participants’ ratings of mental states when viewing expressions with narrow eyes of disgust and anger from their ratings of mental states when viewing expressions with wide eyes of fear and surprise.58. the lower face dimension (excluding congruent expres- 95% CI = [3. Eyes. Error bars represent ±1 SEM.36. p < . −8. Narrow Mouths With Incongruent ** Eyes * Disgust and Anger Fear and Surprise Hate and Suspicion Awe and Cowardice Ratings Ratings Ratings Ratings Fig.  Effect of the eyes on mental-state perception alone and in full-face expressions. Paired-samples t tests revealed significant differ.833.99. The graphs show difference scores for averaged ratings of (a) basic emotions and (b) complex mental states when participants viewed eyes only (Experiment 1) and congruent expressions and incon- gruent expressions (Experiment 2). p < . SD = 1.59. r2 = . 3. 95% CI = [2. 95% CI = [−4. see the Supplemental Material).05.56. for statistics of differ- for basic-emotion ratings. and for complex-mental-state ratings.838.8. t(28) = 8. ing could maintain signal value in the context of compet- Paired-samples t tests revealed significant differences ing information from the lower face. which were . r2 = which suggests that features of eye opening convey simi- . and sions.96].714. r2 = . 1. Asterisks indicate significant differences between expressions (*p < . 5.  We next addressed whether features of eye open- row eyes than for expressions with wide eyes (Fig. −1.6 Lee. t(28) = shown alone or in the context of the full face.0001.19. Narrow Eyes ** With Incongruent ** Mouths ** Wide vs.62. t(28) = −11. and by subtracting ratings of expressions with narrow mouths from ratings of expressions with wide mouths. −2.0001.85. (Fig. SD = 1. difference score = 4.001).0001.84. 3. r2 = . t(28) = 12.0. Using for complex mental-state ratings. **p < . To examine between wide-eyed and narrow-eyed expressions for incongruent expressions. Conversely. with narrow eyes than for expressions with wide eyes SD = 1.62. 95% CI = [−3. 1) ** ** ** Congruent ** Expressions ** ** ** Wide vs. p < .43. for mental states of sensitivity. S2 in the Supplemental Material).15. Difference scores for incongruent faces were calculated in two ways: by subtracting participants’ ratings of expres- sions with narrow eyes from ratings of expressions with wide eyes. 3). averaged ratings of basic emotions (fear and Incongruent expressions surprise) and complex mental states (awe and cowardice) were highly significantly greater for expressions with nar. difference score = −3. these ratings of incongruent expressions. Anderson a b Basic Emotions Complex Mental States Difference Scores Difference Scores (Wide Minus Narrow Expressions) (Wide Minus Narrow Expressions) –5 –4 –3 –2 –1 0 1 2 3 4 5 –5 –4 –3 –2 –1 0 1 2 3 4 5 ** Eyes Only ** (Exp. difference score = 2.0001.68. These results were similar to those found for the ences between wide-eyed and narrow-eyed expressions eyes alone in Experiment 1 (Fig.94].18. 3.

21. 2014).66].67. continuous feature dimensions. results followed the same overall pattern as found in p < .Reading Mind From Eyes 7 defined according to incongruence with the eyes.83].. t(28) = −11. r2 = .37. 2011) and sis.22.0001.0001. which suggests firmed this hypothesis (Fig. 95% CI = [0. 3). along which ference score = 0. we subtracted rat- conducted the same analyses as above. apertures associated with fear and disgust were not dis- t(28) = 2. vious findings that the opposing optical effects of eye difference score = 0. feelings. 1969) but may undergo exaptation for more about sensitivity or discrimination. which corroborated previous and suspicion ratings.55.92. p < . & Jolliffe. 2008) and now serve as social signals.22. This dimensional perspective was supported by our pre- tal states of awe and cowardice. t(28) = −0.0001. p < .6. eye widening and eye narrowing. from a dimension of opening versus closing of the sen- sory apertures that have egocentric perceptual functional Mouth. 95% CI = [0. 1997.32. consequences (Darwin. Hence.76.84. In the context of conflicting information from may remain relatively invariant across cultures (Ekman the eyes. The results con. tests revealed that. albeit to a lesser degree. S2 in the Supplemental Material). We then showed that samples t tests revealed no significant differences between the dimensional variance of these eye features communi- wide. difference face.. The results con. t(28) = −1. paired-samples t tests revealed Baron-Cohen et al. t(28) = 5.11. lower faces from the same kind et al. SD = 0. t(28) = 4. mental states of sensitivity.77. 95% CI = [−3. are not arbitrary but appear to reflect a basic structural t(28) = 10. 2014) the wide-versus-narrow dimension of eye features con. p < . 95% CI = [1. Experiment 1 and in the congruent full-face expressions.01]. Lee et al. expressions (e. 1.0001.7.83. averaging our data set across the eyes 2013).10. Opposing features around the eyes that harness ice ratings. but we focused on may be further shaped by specific cultures (Elfenbein. the results from Experiment ences between wide-eyed and narrow-eyed expressions 2 show that opposing mental states were attributed to for disgust and anger ratings.464. 3).g.01. SD = 0. and for awe and coward.89].14]. we related to sensitivity.656. 2014. This result follows dimension of mental states despite the incongruent fea. 2.27. p = .1. t(28) = 5.25. dif.  If the eyes are indeed most important for con. logic. Wheelwright. r2 = .811.05. 2001). Paired-samples t tests revealed Discussion highly significant differences between wide-eyed and narrow-eyed expressions for fear and surprise ratings. p < .17]. the eyes Although there was a small reduction in endorsement of had a greater influence on fear and surprise ratings. r2 = . The eye features observers use in attributing mental states difference score = 2. 2. dif- incongruent faces compared with congruent faces. p < . r2 = .0001.0001..87.63. ings of expressions that were incongruent with the mouth firmed our hypothesis that wide eyes convey sensitivity from ratings of expressions that were incongruent with and narrow eyes convey discrimination.137. 1872/1998. basic expressions’ features were used as points of anchor. even in the context of SD = 1. t(28) = 7. −2.. r2 = . p < . as well as complex men. we conducted another analysis. from the theory that emotional expressions may originate tures from the lower face.552.69.502.0. a coupling logic between the expressive form of the eyes Examining mental states of discrimination. 95% CI = [0. difference score = 1.06.42. Paired-samples t tests revealed highly significant differ. multiple. directly comparing results for incongruent et al.. were highly diagnostic of basic emotions. r2 = . and on awe and cowardice ratings. score = −2. p < .. the eyes to calculate difference scores. dimension (again excluding congruent expressions.. Our multidimensional analyses revealed that expressions revealed a stronger influence of the eyes eye-widening and eye-narrowing features associated than of the lower face on perception of mental states with fear and disgust likewise supported how the eyes .25. paired. compared with the mouth. difference score = 1.17.0001. crete but exerted along an eye-opening continuum (Lee However.31. Paired-samples t ence of incongruent lower-face information (Fig. the ference score = 1. see We first showed that specific features of the eye region Fig. −1. fear grimace) should not have the same the functions and thus the appearance of these features influence. difference score = −2. and their conveyed social signals. 95% CI = invariant properties of light to bias sensory acquisition [1. SD = 1.49. SD = 1. 95% CI = [−2.12. and for hate and suspicion ratings. 2. To gain further understanding that comparing wide-versus-narrow-mouth expressions of how the eyes convey these mental states. 95% CI = [0. and perception in opposing ways (Lee et al. appear to be socially co-opted to convey our inner tinued to convey the sensitivity-versus-discrimination thoughts. p = .825. Therefore. Suss- veying these mental states. potentially incongruent information from the rest of the r2 = .54]. even in the pres. In sum. lower-face information should convey less et  al. complex social meaning (Shariff & Tracy. and intentions.and narrow-mouth expressions for disgust and cated not only basic emotions but also a wide variety of anger ratings. For this analysis. r2 = .786. we used did distinguish basic emotions of disgust and anger. Examining findings (Baron-Cohen. SD = 1. the lower face. 0. or for complex hate complex mental states. To test this hypothe. t(28) = −11.14].

1995). direction of gaze can ral wrinkles (e. as Darwin (1872/1998) sug.g.. gests that eye features have been co-opted in the service Russell & Barrett. tion of what. submission and remorse were between expressions. Wide eyes are strong physical signals accuracy of these attributions to the sender’s mental states. 2008). 2014). embodies human mental states in basic perception and (Among our exemplars. Lanzilotto. may reflect the physical signal bias involving humans’ such as suspicion toward and discrimination of poten. were fewer and more isolated than those related to dis- sive antagonism may be engaged in contexts far removed crimination (see Fig. Another notable aspect of the mental-state map (Fig. In the current study.. 2005) and is an important contributor to mental- Reinforcing the dimensional perspective of how people state decoding (Baron-Cohen. by shifting the salience away wide not only enhances the sensitivity of the visual sys. whereas narrow state map (Fig. which indicates that a requiring greater scrutiny. It has tially unfair social transactions. direction of eye gaze constant. Squinting when suspicious of another stronger signal. & Anderson. between disgust and joy. elicit asymmetric approach and avoidance . Lee et al. Kleck. Kim. such as fear and anger sion and remorse. where) or to the parvocellular visual system (discrimina. 1872/1998. 2007). & Kleck. It is possible that the relative sensitivity-discrimination organizational principle was the abundance of mental states on one side reflects a sam- primary dimension accounting for the ability to read pling bias in the mental states we selected for the experi- mental states from the eyes. sion anchored by joy and sadness. lessening the need for discrimination. are not the only variables that influence how humans These opposing clusters of positive and negative mental read mental states. Although wide Although we have shown that such expressive actions eyes may represent a relatively unambiguous signal and have direct perceptual consequences for the expresser unambiguous associated mental states. 2008. such as frequency information (e. For (Marsh. For example. For example. A variety of nuanced mental states arises expresser. 2005. which may enhance signal detection.1 SD in action states (Niedenthal. there were increases of 3.g. which evoke sociomoral been shown that revealing more sclera in eye widening disgust (Chapman. the opposing expressions & Ghazanfar. 2014). 2008). associated with complex states. and are more diagnostic of mental states of sensitivity. Orthogonal to the primary dimension in our mental. similarly negative expressions. 1997). unique eye whites (Kobayashi & Kohshima. in which their (eye widening versus eye narrowing) that enhance visual social communicative role is paramount.8 Lee. such as fear and sadness. 2a) was a positive-negative valence dimen. Likewise. Whalen et al. from eye whites to other features surrounding the eyes. and this finding underscores ment. that mental states related to sensitivity (Darwin. 1999. widen- exploration of the mental-state map integrating the pri. require greater discrimination to differentiate among gested. This increase in socially exapted for purposes beyond its role in biasing low-level contrast and luminance sends a physically visual encoding.) Conversely. wrinkles) across a wider interest or awe. Opposing facial actions may bias perception from a combination of these two basic dimensions of eye toward the magnocellular visual system (sensitivity to features. This expres. enhances processing of the eyes (Adolphs et al. 2004).. Evidence also sug- perceive facial information (Oosterhof & Todorov.7 SD in luminance from lowest disgust to enhanced perceptual discrimination. influence emotional expression perception (Adams & sad eyes. a broader of more complex social pressures. a true asymmetry suggested by our their link to the sensory origins of their expressiveness cluster analysis. our sensitivity versus discrimination in the sender (Lee et al. Anderson convey other opposing complex mental states. In this sense. 2013).. boredom (between The reading of emotions and mental states can be sadness and disgust) was opposite interest (between joy understood in part from the perceptual needs of the and surprise). Susskind et al. Susskind.. 2005). joy eyes) and eyebrow curvature (e. pride and boastfulness were opposite submis. eyes are weaker signals and are much less diagnostic. Alternatively. underlying mental states.. narrow eyes (Lee et al. 2b. and pride and boastfulness were awe and hate. tem but also embodies complex mental states related to eye narrowing requires finer discrimination of higher- the need for sensitivity to incoming information. sensitivity-discrimination opposition is not the only It is worth noting that the eye features we used here dimension by which the eyes communicate internal states. even if it does not confer contrast and 2.. Adams. largely vestigial or ritualized (Sheperd. results are limited to the mental states the receiver reads 2014) may resonate in how the receiver decodes them from these physical features and cannot speak to the true (Lee et al. see the Supplemental Material online). which may explain why instance. 2012) in their modern usage. our examination held states were expressed by key eye features such as tempo. permutation of possible configurations. from its sensory origins. this 2b) is its salient asymmetry. they may be. S1a in the Supplemental Material)... Thus.g. Opening the eyes highest fear. Fig.. This highlights how expressive eye aperture has been Lee et al. Susskind et al.. 2009). ing the eyes makes the face appear more juvenile and mary and secondary dimensions revealed interesting narrowing the eyes makes the face appear more mature potential accounts of other oppositional mental states.. 2013.

Separate coding of different gaze directions in the superior temporal sul- cus and inferior parietal lobule. (2001). Lee). 164. (2014). MA: MIT Press/Bradford Books. 2008. A.. Wheelwright. Kim.. 241–251. & Jolliffe. K. 4. & Kleck. Tranel. D.2006..1038/nature03086 Allison.. R. the eyes receive differential weighting in conveying doi:10. Nonverbal dialects and accents in This work was supported by the Canada Research Chairs pro. Hassin. facial expressions of emotion. Y.1017/ of awe or the narrow-mindedness of hate. How expres. D. 19. Lee and A. H. Journal of explain how people convey the eye-widening experience Child Psychology and Psychiatry. doi:10. disgusted. C. Relative to the eyes. S. P. S. H. doi:10. J.. and by an NSERC post. Wheelwright. 323. E. & Plumb. see. S. V.. 9280. (2008). Susskind et al. graduate fellowship (to D. Raste. (1997).1165565 experiments and J. (1969). & Anderson. (2006). S. Acknowledgments (2009). recognition after amygdala damage. 68–72. Pan-cultural respect to their authorship or the publication of this article. & Anderson. (2014). H.. Alice J. Chapman. B. Unique morphology of References the human eye. & McCarthy. Susskind. 745–752. T. Mahmood and Z. (2000). O’Toole served as action editor for this article. 20–25..Reading Mind From Eyes 9 responses. Jenkins. (1998). & Kleck. New York. 5. The authors declared that they had no conflicts of interest with Ekman.. .1016/S1364-6613(00)01501-1 cally associated with sensory aperture narrowing (e. H. Grady.06369.1177/0956797613514451 . A.. F. Effects of direct and Lee.86 Funding Elfenbein. M. or or widening (e. Flanagan. the manuscript and approved the final version for submission.. In bad taste: Evidence for the oral origins of moral dis- We thank R. windows into inner mental states. Susskind for providing feedback on the Darwin. averted gaze on the perception of facially communicated Optical origins of opposing facial expression actions.1111/j. & Schyns. gram. 2005). D. R.. Anderson developed the experiments. Adolphs. Winston. S. The “reading the mind in the eyes” test revised version: A study with normal adults.1037/1528-3542. 767–768. A.. Ryan.2008.. I.. R.3 logical Science. Nature.. P. 1079–1086.. (1995). Psycho- emotion. & Kohshima. (1997). Is there nizing principle underlying people’s ability to read the a “language of the eyes”? Evidence from normal adults. K. The expression of the emotions in man and facial-expression models. doi:10. Facial movements strategically camouflage involuntary social signals of face morphology.. G. from perceivers (Marsh.. 1222–1226. W. Anderson). and adults with tures that harness the physical principles of light may Asperger syndrome or high-functioning autism. (2007). S. Supplemental Material 25. & Friesen. Buchanan. doi:10. 25.. A.02148. J. G. 42. 86–88. The eyes are conduits of light to the retina as well as Baron-Cohen. A. J.1177/1754073912451332 (NSERC) grant (to A. Kobayashi. H. CA: Consulting Psychologists Press. J. 724–732. Jr. R. Katsiapis for conducting the gust. Beaver.. 4.. R. Palo Alto.. and complex states of another person’s mind. Pictures of facial affect. 1. Current Biology. R. J. W.1177/0956797616687364 PA: ASA-SIAM. Trends in Cognitive weight. H.1080/713756761 Action Editor Calder. doi:10. adults with autism or Asperger syndrome. A.10.1126/science. A. S. respectively. H. H. Adams. 267–278. J. A mechanism for impaired fear Our results suggest that even when the full face is vis. Anderson wrote gust.. K. Jack.. K. R.1016/j. Suss. A. M. O. E. A.. R.x This finding underscores how the origins of decoding Baron-Cohen. Bentin.. R. T. 3–11. doi:10.. Boston.g. Lee and A. Science. Gill. & Friesen.. kind & Anderson. & Damasio.. D.... D. Philadelphia. 311–331. nose wrinkle and lip raise in disgust. A. P. Ambady. Understanding dis- Lee conducted the analyses. J. doi:10. Eger. Baron-Cohen. Psychological Science. (2005). Puce. animals. lower face features canoni- Sciences. H. E. journals. Emotion Review.. T. ible. Susskind. L. 433. 90–96. NY: Oxford University Press.1177/0956797614522274 Hubert. The structural repre- Additional supporting information can be found at http:// sentation of proximity matrices with MATLAB. (1976).1111/j.sagepub. Sorenson. Aviezer. H. K. & Henson.. R. C. (2012). N. P. Conspicuous facial fea.164.2011..1467- significantly less sensitivity or discrimination. Dolan. elements in facial displays of emotion. Angry. B.5. G. (2013). (Original work published 1872) Declaration of Conflicting Interests Ekman.cub.. Hill. 2008) conveyed Psychological Science. Gosselin. open mouth in fear and surprise. K. & Anderson. Arabie.1749-6632. & Meulman. P..052 D. G. S.. Mirza. J. D. Nature. Garrod. R. A. 17. . by a Natural Sciences and Engineering Research Council doi:10.. Susskind et al. 1251. Science. 2008) Anderson. doi:10. doi:10. 387. D.3875. E.g.x Chapman. respectively.1126/science. A. J.. Annals of the New York Academy of Sciences. 62–76. (2005). doi:10. Social perception complex mental states.. S0021963001006643 sions alter the way the eyes see reveals an important orga. W. Visual Cognition. K.. Emotion... and the eye region has the most from visual cues: Role of the STS region. R. afraid? Studies on the malleability of emotion perception.1. Mindblindness: An essay on autism and mental states from the eyes relate in part to how the eyes theory of mind. .. Schyns. Author Contributions doi:10.

J. j. W. L. C. MA: research. (2011). Department of ical Science. 805–819.. Johnstone. doi:10.. for? Current Directions in Psychological Science. J. Specific brain regions for one component of theory of color. R. Psychophysical evi. The Journal of Neuroscience.1103617 . & Kellerman. . doi:10.76. Fear. R. L. D. San Fran­ of dirt: Disgust enhances the detection of impurity. E. Embodying emotion. doi:10 in Psychological Science.5. Why do Shariff. M.. The Journal avoidance-related behaviors. & Kleck. Psycholog- cisco. F. R.. & Ekman.. Susskind. E. A. & Barrett. 11. cal emotional episodes. 6105–6116. CA: San Francisco State University. Two cortical visual sys- .ep10836912 Russell. A.. Journal of Personality and Social sivity to masked fearful eye whites.1126/science. Cambridge. M. R.1016/j. doi:10.. M.). doi:10. Proceedings of the National Academy 152–162.3. H.1037/h0077714 8721. doi:10.neuropsychologia. N.. (2013). 177–188. J. doi:10. Intercultural and Emotion Research Laboratory.and facial expressions and ingestive movements.1037/0022-3514. S. 32. Mansfield (Eds.. J. 11087–11092... 957–965. doi:10. doi:10. J. Susskind.1177/0146167204271306 Sheperd. doi:10. Psychology.). J. Saxe. and experience: Vol. Emotion. mind.1038/ ness: Toward an evolved module of fear and fear learning.1037/0033.. F. G. B. Polis. blance to emotional expressions predicts evaluation of emo- ing. Psychological Science. form and function.. & 1002–1005. (2004). (2012). A.. Anderson Lee. The faintest speck facial expressions of emotion (JACFEE) [Slides]. N. Psychological Science. S. M. 17. P.. & Anderson. Lanzilotto. 20.1177/0963721411424739 Bulletin. (2008).. R. 31. Adams.1111/ 7. & Russell. J.. & Ghazanfar. & Hubel. Whalen. The effects Facial muscle coordination in monkeys during rhythmic of fear and anger facial expressions on approach.2006. Theories of emotion.1126/science. 1. H. & Powell.2012 Matsumoto. J. A circumplex model of affect. V.. L.1177/ tionally neutral faces. P.5.. & Mineka. Kagan.001 of  Sciences. doi:10.. K. 108. R. Haidt. J.805 doi:10. doi:10. P. 1161–1178. Japanese and Caucasian Sherman. . M. M. Social Said. and ambiguity: Initial neuro- Russell. acquisition. & Clore. D. 260–264. R. Ambady. Universality revisited. Core affect. and other things called emotion: S. 483–522. K.1177/1754073912457227 1. (2006). Marsh.108. & Todorov. (2005).1467-9280. Neuropsychologia.. Goodale. Human and computer recognition Oosterhof. K. A. A.05. Emotion. N. A. Fears. 395– tion in facial expressions. K. (2013). 1506–1514.. A. In D. A. York. 39. Current Directions Personality and Social Psychology.. and depth. & Kleck.. phobias. 73–86. Cook.1. 8–15.6136-11. J.6999 Niedenthal. 843–850. doi:10. & 295X. (2007). prototypi. The functional basis of facial expressions of emotion. A. Barlett. Jr. G. doi:10. movement. and prepared.01768. Movellan. doi:10. Susskind. Structural resem- transmission of the sensory benefits of fear eye widen.. N. A. 549–586). Science. 76. T. M. Cusi. (2008).1523/JNEUROSCI . A. C. What are emotion expressions fear and anger look the way they do? Form and social func. Emotion Review. 5.0805664105 tems. G. & Mishkin.1037/1528-3542. S. 3416–3468. Grabski. USA. 148–149. M. (1980). (2005). M. Nature Neuroscience.1. (1980). J. J. G. doi:10 of Neuroscience. 692–699. A. (1987)..10 Lee. 5. D.2. 7. A.4161/cib. Facial expression Nelson. NY: Academic Press. Kim. L... (1988). J. Davis..119 . 105. 2061. & Anderson. . Whalen.. H. & R. Journal of imaging studies of the human amygdala.483 Anderson... (1982). H. (2012). Analysis of visual behavior (pp. 316. (2009). of face evaluation. N. 306. (2008). It’s the thought that counts: dence for separate channels for the perception of form. D. Littlewort. A. L. A.1177/0956797612445318 Psychology. A. 9. doi:10. A. P. Susskind. J. Ingle. M. Expressing fear enhances sensory Öhman. P. Plutchik. Lee. vigilance.1037/ 0956797612464500 a0014681 Livingstone. (1999).2006. (2001).x Marsh. 119–124.. Sebe. Feiman. 24. Communicative & Integrative Biology.. Science.2138 Psychological Review. nn.. 23. D.. G. J. 45. & Todorov. Personality and Social Psychology 399. A. (2007).1073/pnas Ungerleider. (Eds. Human amygdala respon- Dissecting the elephant. J. New MIT Press. H. doi:10. Emotion: Theory. S. & Tracy. (1998).1136930 Anderson.... A.1111/1467- .