Emotion

© 2018 American Psychological Association 2019, Vol. 19, No. 5, 788 –798
1528-3542/19/$12.00 http://dx.doi.org/10.1037/emo0000484

You Are What I Feel: A Test of the Affective Realism Hypothesis

Jolie Baumann Wormwood Erika H. Siegel
Northeastern University University of California, San Francisco

Justin Kopec Karen S. Quigley

Syracuse University Edith Nourse Rogers Memorial Veterans Affairs Hospital,
Bedford, Massachusetts and Northeastern University

Lisa Feldman Barrett

This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Northeastern University, and Massachusetts General Hospital and Harvard Medical School
This document is copyrighted by the American Psychological Association or one of its allied publishers.

We present evidence for the affective realism hypothesis, that incidental affect is a key ingredient in an
individual’s experience of the world. In three studies, we used an interocular suppression technique
(continuous flash suppression [CFS]) to present smiling, scowling, or neutral faces suppressed from
conscious visual awareness while consciously perceived neutral faces were presented at three different
timing intervals: 150 ms before, 150 ms after, and concurrent with the suppressed affective faces (Studies
1 and 3) or at timing intervals of 100 ms (Study 2). Results for all three studies revealed that consciously
perceived neutral faces were experienced significantly more positively (e.g., as more trustworthy) when
concurrently paired with suppressed smiling faces than when concurrently paired with suppressed
scowling faces; there was no effect of suppressed affective faces on first impressions in the other timing
conditions. This pattern of results is consistent with the affective realism hypothesis but inconsistent with
both affective misattribution and affective priming interpretations. Incidental affect must be meaningfully
contiguous in time with the target stimulus to be experienced as a property of the target.

Keywords: affective realism, affective misattribution, priming, continuous flash suppression, incidental
affect

Supplemental materials: http://dx.doi.org/10.1037/emo0000484.supp

Affective feelings exert a powerful influence on behavior and eliciting source is presented outside of conscious awareness (e.g.,
decision making, even when the source of those feelings is unre- Niedenthal, 1990; Winkielman, Berridge, & Wilbarger, 2005;
lated (or incidental) to the decision at hand (Clore, Gasper, & Winkielman, Zajonc, & Schwarz, 1997). In fact, research suggests
Garvin, 2001; Johnson & Tversky, 1983; Loewenstein & Lerner, that incidental affect influences phenomena as diverse as food choice
2003), and the influence of incidental affect persists even when the and consumption (Garg, Inman, & Mittal, 2005; Garg, Wansink, &
Inman, 2007; Oliver, Wardle, & Gibson, 2000), detection of threat in
the environment (Baumann & DeSteno, 2010; Wormwood, Lynn,
This article was published Online First August 23, 2018. Feldman Barrett, & Quigley, 2016), prosocial behavior (Bartlett,
Jolie Baumann Wormwood, Department of Psychology, Northeastern Univer- Condon, Cruz, Baumann, & Desteno, 2012; Bartlett & DeSteno,
sity. Erika H. Siegel, Health Psychology, University of California, San Francisco. 2006; DeSteno, Bartlett, Baumann, Williams, & Dickens, 2010; Isen
Justin Kopec, Department of Psychology, Syracuse University. Karen S. Quigley, & Levin, 1972), risk taking (Baumann & DeSteno, 2012; Keltner &
Edith Nourse Rogers Memorial Veterans Affairs Hospital, Bedford, Massachu- Lerner, 2010; Lerner & Tiedens, 2006), and perceived life satisfaction
setts, and Department of Psychology, Northeastern University. Lisa Feldman (Schwarz & Clore, 1983). Incidental affect has also been shown to
Barrett, Department of Psychology, Northeastern University, and Department of
impact decisions with significant, real-life consequences like stock
Psychiatry, Massachusetts General Hospital and Harvard Medical School.
This research was supported by a contract from the United States Army market investments (Hirshleifer & Shumway, 2003) and medical
Research Institute for the Behavioral and Social Sciences (Contract W5J9CQ- school admission decisions (Redelmeier & Baxter, 2009).
12-C-0049 to Lisa Feldman Barrett) and by a grant to the National Science In the present paper, we propose a provocative explanation for
Foundation (Grant BCS-1422327 to Karen S. Quigley and Jolie Baumann some of these findings: affective feelings (incidental or not) nat-
Wormwood). The views, opinions, and/or findings contained in this paper are urally infuse our perceptions and give us a sense of confidence that
those of the authors and shall not be construed as an official Department of the they are valid windows onto the real world. We call this the
Army position, policy, or decision, unless so designated by other documents.
affective realism hypothesis (Anderson, Siegel, White, & Barrett,
Authors Jolie Baumann Wormwood and Erika H. Siegel contributed equally.
Correspondence concerning this article should be addressed to Jolie Bau-
2012; Barrett & Bar, 2009; Kring, Siegel, & Barrett, 2014). The
mann Wormwood, Department of Psychology, Northeastern University, 360 affective realism hypothesis builds on the writings of philosophers
Huntington Avenue, 125 NI, Boston, MA 02115. E-mail: jbwormwood@ who, for centuries, have argued that feelings of pleasure and
gmail.com distress are intrinsic elements in perceptions of the world similar to

788
 AFFECTIVE REALISM 789

the experiences of hue and brightness (for discussions, see Barrett realism may be thought of as a special case of affect-as-
& Bliss-Moreau, 2009; Lambie & Marcel, 2002). In much the information theory.
same way that color is treated as a property of an object itself In the present studies, we sought to empirically demonstrate that
instead of one’s own experience of the object—people experience affective realism is a special case of affective misattribution by
the sky as blue rather than experiencing sky color as their own manipulating the timing offset between the presentation of an
perception of light at 500 nm reflected off particles in the sky— affective stimulus and a to-be-judged (target) stimulus, allowing us
objects and people in the world are said to be “positive” or to carefully tease apart whether incidental affect can influence the
“negative” by virtue of their coincidence with a person’s affective experience of the target stimulus itself and not just the perceivers’
feelings. If the perception of a snake evokes extreme, unpleasant post hoc judgment of the target. We employed a paradigm used to
feelings, then the snake is said to be extremely negative. If a study visual consciousness called continuous flash suppression
defendant is viewed while a judge is unpleasantly hungry, the (CFS; Tsuchiya & Koch, 2005) in which flashing images are
defendant is experienced as untrustworthy and unreliable (Dan- presented to one eye (and seen) while static, low contrast images
ziger, Levav, & Avnaim-Pesso, 2011). Recent discoveries in neu- are presented to the other eye (and suppressed from awareness). In
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

roscience reveal that the human brain is creating a unified con- our studies, we utilized CFS to suppress affective faces from visual
This document is copyrighted by the American Psychological Association or one of its allied publishers.

scious experience, integrating all sources of sensation, both from consciousness and asked participants to provide their first impres-
inside the body and from without, with limbic circuitry as the sions of neutral target faces that remained in conscious awareness
driver (for a discussion, see Barrett, 2017; Barrett & Simmons, using a series of personality judgments (e.g., trustworthiness).
2015; Chanes & Barrett, 2016). As a consequence, a person’s Critically, we also directly manipulated the timing of the presen-
affect can serve as a source of realism in perception, even when tation of the seen target face across trials, presenting it either
incidental to the target of perception and when the person is before, concurrent with, or after the suppressed affective face,
unaware of it (Anderson et al., 2012; Kring et al., 2014; Siegel, while keeping the time between the suppressed affective face and
Wormwood, Quigley, & Barrett, 2018). the person perception judgments of the target stimulus the same
The affective realism hypothesis is consistent with a large across all conditions. Consistent with previous research (Anderson
literature demonstrating that affect shapes perception (for a review, et al., 2012; Kring et al., 2014; Siegel et al., 2018), we predicted
that neutral faces would be judged as more positive (e.g., more
see Zadra & Clore, 2011). Individuals experiencing negative affect
trustworthy) when paired with suppressed smiling faces than with
perceive tones as louder (Siegel & Stefanucci, 2011), are more
suppressed scowling faces, and we explored whether this effect
sensitive to visual contrast gradients (Phelps, Ling, & Carrasco,
held across timing conditions.
2006), and exhibit more local than global perceptual processing of
This approach provides a direct test of whether affective realism
images (Gasper & Clore, 2002). Moreover, affectively negative
is distinct from affective misattribution because the two hypothe-
stimuli tend to be perceived as larger compared with neutral or
ses make different predictions about whether the relative timing of
positive stimuli (Teachman, Stefanucci, Clerkin, Cody, & Proffitt,
the affective and target stimuli will influence the strength of the
2008; van Ulzen, Semin, Oudejans, & Beek, 2008), and perception
effect. The affective misattribution hypothesis would not predict
is influenced by individual differences in motivation and past
an effect of timing. The ordering of the stimuli should not alter the
experience (Balcetis & Dunning, 2006): people perceive a glass of
use of one’s feelings as information for evaluating the target face,
water as taller when they are feeling unpleasantly thirsty (Velt- particularly when the time between the presentation of the affec-
kamp, Aarts, & Custers, 2008), and spider-phobic individuals tive information (i.e., the suppressed affective face) and the eval-
perceive spiders but not wasps as larger, despite reporting that both uation (i.e., the personality rating) is constant across all trials. If
stimuli are unpleasant (Leibovich, Cohen, & Henik, 2016). the effect of the suppressed affective faces on personality ratings is
Traditionally, researchers have understood the effect of inciden- consistent across timing conditions, this would suggest that affec-
tal affect on perception in the context of affect-as-information tive realism and affective misattribution are not distinguishable in
theory (e.g., Murphy & Zajonc, 1993; Payne, Cheng, Govorun, & the present experiments. However, evidence consistent with affec-
Stewart, 2005; Winkielman et al., 2005). Typically, affect-as- tive realism (i.e., that affective feelings are integrated into the
information theory is explained as affective misattribution: inci- experience of the target face) would be found if the suppressed
dental affect is thought to exert influence when people are unaware affective faces influence personality ratings most strongly when
of the source of their feelings because, without an obvious source, affective stimuli are concurrent with neutral target stimuli. This
individuals are more likely to attribute their affective feelings as pattern of results would provide evidence that affective realism
related to or caused by the current situation (Clore et al., 2001; represents a unique class of affective misattribution effects.
Clore & Huntsinger, 2007; Schwarz & Clore, 1983). The affective
realism hypothesis is another variant of affect-as-information the-
ory that shares with affective misattribution the idea that affective Studies 1 and 2
feelings drive changes in judgment and behavior. The affective In Study 1, we utilized an offset of 150 ms: neutral target faces
misattribution hypothesis is relatively agnostic about whether feel- were presented 150 ms before, concurrent with, and 150 ms after
ings influence the experience of the object itself or just the judg- the suppressed affective faces, which were presented for 200 ms.
ment process, however. By comparison, the affective realism hy- In Study 2, we shortened the offset to 100 ms. These offsets were
pothesis explicitly hypothesizes that affect is a property of chosen in light of previous literature demonstrating that visual
consciousness and, as a result, is an integral part of the construc- stimuli presented for longer than 130 ms (as in the present studies)
tion of all conscious experiences (Barrett & Bar, 2009; Barrett & persist at multiple levels of the visual system (from activity in the
Bliss-Moreau, 2009; Barrett & Russell, 2015). Thus, affective peripheral retinal receptors to cortical activity) for a constant of
 790 WORMWOOD ET AL.

100 ms beyond their physical presentation (Bowling & Lovegrove, nondominant eye. Participants then completed a contrast adjust-
1980, 1981; Efron, 1970). Thus, we selected offsets that would put ment task during which the contrast level of images presented to
us outside the possible visual persistence window for our face the nondominant eye under CFS was adjusted to improve suppres-
stimuli, limiting the possibility that the affective and target stimuli sion on an individual basis. Each trial of the contrast adjustment
might be perceived or experienced as simultaneous in our offset task began with a 500 ms fixation cross and ended with a 500 ms
timing conditions. Across both studies, we predicted that seen backward mask. Following the fixation cross, high-contrast,
neutral faces would be evaluated more positively when paired with Mondrian-type images (similar to Tsuchiya & Koch, 2005) were
suppressed smiling faces than with suppressed scowling faces, and “flashed” in the participant’s dominant eye at 20 Hz for 1,200 ms.
that this effect would be strongest in the concurrent timing condi- During this 1,200 ms, a photograph of a house (either upside down
tion. or right-side up) was presented for 200 ms to the participant’s
nondominant eye, from 500 –700 ms after the fixation cross. Par-
Method ticipants reported the orientation of the suppressed house on each
trial by clicking one of two keys on their keyboard. They also rated
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

For conciseness and because methods and analyses were iden- their subjective awareness of the suppressed house using the
This document is copyrighted by the American Psychological Association or one of its allied publishers.

tical, we describe the samples, methods, and results jointly for 4-point Perceptual Awareness Scale (PAS; Ramsøy & Overgaard,
Studies 1 and 2 rather than separately by study. All procedures and 2004), from no experience to absolutely clear experience. Images
materials were approved by the Northeastern University Institu- of houses were presented at four discrete contrast levels, created by
tional Review Board. reducing the contrast and luminance levels of the original photo-
Participants. Participants in Studies 1 and 2 were students graphs to 75, 50, 25, and 12.5%. For the first 20 trials of this task,
and community members recruited from the greater Boston area all house images were presented at 75% contrast with half of the
through fliers on college campuses and advertisements on Craig- trials containing right-side up images and half containing upside-
slist.org and in the Boston-Metro newspaper. All participants down images. If participants correctly guessed the orientation of
reported normal or corrected-to-normal visual acuity and were the suppressed house on 70% of the trials or reported no experi-
naïve to the experimental hypotheses. Individuals wearing glasses ence on less than 75% of trials, the contrast level was reduced and
were excluded because glasses interfere with the proper function the participant completed another 20 trials of this task at the next
of the visual apparatus (mirror stereoscope). Target sample size for highest contrast level. This procedure was repeated until partici-
Study 1 was determined by conducting a power analysis in pants correctly guessed the orientation on 13 or fewer trials and
GⴱPower (Faul, Erdfelder, Lang, & Buchner, 2007) using the reported no experience on at least 15 trials, or until the 12.5%
average effect size from previous research in our lab that utilized
contrast level was reached. The contrast adjustment task deter-
a similar experimental task (Anderson et al., 2012; Study 4; Siegel
mined the individualized contrast level at which all suppressed
et al., 2018; Studies 1 and 2). This power analysis revealed that in
images would be presented for the remainder of the experimental
order for an effect size of ␩2 ⫽ 0.163 for affect condition to be
tasks.
detected within any of the timing conditions (80% chance) with
Person perception task. On each trial of the person perception
significance at the ␣ ⫽ .05 level, a sample of at least 28 partici-
task, participants were presented with flashing Mondrian-type
pants would be required. Thirty-three participants (18 males, 15
images interleaved with an image of an individual with a neutral
females; Age 18 – 60; Mage ⫽ 28.58, SDage ⫽ 14.23) completed
face in their dominant eye (neutral target face) while an image of
Study 1. Target sample size for Study 2 was determined by
an individual with an affective facial pose was presented to their
conducting a power analysis using the effect size of the critical
nondominant eye and was suppressed from awareness (suppressed
Affect Condition ⫻ Timing Condition interaction from Study 1.
affective face). Participants were asked to rate how likable, trust-
This power analysis revealed that in order for an interaction with
worthy, and reliable they found each seen, neutral target face on a
an effect size of ␩2 ⫽ 0.115 to be detected (80% chance) with
scale from 1 to 5. All face stimuli were pulled from a normed facial
significance at the ␣ ⫽ .05 level, a sample of at least 54 partici-
pants would be required. Sixty-one participants (18 males, 43 stimulus set, the Interdisciplinary Affective Science Laboratory
females; Age 18 –53, Mage ⫽ 23.82, SDage ⫽ 10.41) completed Face Set, which was developed by our laboratory with support
Study 2.1 Prior to data analysis, seven participants were removed from the National Institutes of Health Director’s Pioneer Award
because they either did not understand or follow the experimental (DP1OD003312) to LFB. Emotion expressions in this stimulus set
instructions (n ⫽ 3, Study 2), they experienced difficulty using the were posed. Models were given instructions concerning the facial
stereoscope (n ⫽ 2, Study 2), or there was a computer error muscles to move for each expression, and were shown several
during data collection (n ⫽ 2, Study 1). sample photos for each expression. For example, for neutral ex-
Experimental tasks. Participants viewed stimuli through a pressions, models were told “Let your face hang in a relaxed,
mirror stereoscope, a device that uses mirrors to simultaneously natural position. Look straight into the camera.” The specific
present a different image to each eye while the participant rests images utilized in the current studies were selected at random from
their chin and forehead on the rests of the device. All stimuli were this larger facial stimulus set. To control for potential differences
presented in grayscale on a 25-in. monitor (Study 1) or a 19-in. that might result from certain neutral faces being perceived as
monitor (Study 2). more attractive or trustworthy than other neutral faces, all seen
Contrast adjustment task. We first established eye dominance
for each participant using the hole-in-the-card test (Dolman meth- 1
Although participants vary widely in age, research suggests age-related
od; Dolman, 1919; Fink, 1938) since suppression of images under thinning of tissue in the visual cortices is not prominent until after age 60
CFS is more easily achieved when images are presented to the (McGinnis, Brickhouse, Pascual, & Dickerson, 2011).
 AFFECTIVE REALISM 791

neutral faces were presented in the various suppressed affective varied across three within-subject experimental conditions: the
conditions randomly across participants (i.e., Identity 1 was always affect before target condition, the concurrent condition, and the
paired with a smiling face for Participant 1, but always paired with affect after target condition (see Figure 1). In the concurrent
a scowling face for Participant 2, etc.). Consistent with the facial condition of Studies 1 and 2, the neutral target face was displayed
stimuli in Anderson et al. (2012), Gruber et al. (2016), and Kring from 500 –700 ms following fixation, concurrent with the presen-
et al. (2014), we used faces with no visible teeth that were cropped tation of the suppressed affective face. In the affect before target
to 150 (width) ⫻ 169 (height) pixels at 100 dpi. condition, the neutral target face was displayed with 150 ms
All trials in the person perception task began with a fixation between the offset of the suppressed affective face and the onset of
cross that was presented to both eyes for 500 ms and ended with the neutral target face in Study 1; this offset was reduced to 100 ms
a backward mask that was presented to both eyes for 500 ms. The in Study 2. In the affect after target condition, the neutral target
stimuli presented to each eye differed for the 1,200 ms between face was displayed with 150 ms between the offset of the neutral
fixation and the backward mask. Figure 1 displays visual repre-
target face and the onset of the suppressed affective face in Study
sentation of the trial structure. In the nondominant eye, a low-
1; this offset was reduced to 100 ms in Study 2. In all three timing
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

contrast, low-luminance face was displayed for 200 ms (from

conditions in both studies, Mondrian-type images were flashed at
This document is copyrighted by the American Psychological Association or one of its allied publishers.

500 –700 ms after fixation). The suppressed face displayed one of

three affective poses (smile, scowl, or neutral) and the contrast 20 Hz in the dominant eye before and after the presentation of the
level of the suppressed faces was individually determined for each neutral target face.
participant based on performance in the preceding contrast adjust- At the conclusion of each trial, participants completed four
ment task. The presentation of the suppressed affective face was ratings on a standard keyboard. First, they indicated the gender of
the same in all three timing conditions. In the dominant eye on the face they saw by choosing “male,” “female,” or “don’t know.”
each trial, a series of flashing, high-contrast Mondrian-type images They were instructed to choose don’t know if they had trouble
were presented and were consciously seen by participants. These determining the gender, saw more than one gender, or saw a blend
were interleaved with a 200-ms display of a full-contrast image of of two genders/faces. Because the suppressed face was a different
a neutral target face, which was also consciously seen. The timing gender than the seen neutral target face, this gender question was
of the presentation of the neutral target face in the dominant eye used as a trial-by-trial measure of conscious visual awareness of

Figure 1. Diagram of the trial structure for each of the timing conditions in Study 1. Trials each began with
a 500 ms fixation cross and ended with a 500 ms backward mask. All other images were presented in 50 ms bins.
(A) Shows the trial structure for the affect before target condition, (B) shows the trial structure for the concurrent
condition, and (C) shows the trial structure for the affect after target condition. The top row in each panel (labeled
D) represents the images shown to the dominant eye, which are consciously perceived, including the neutral
target face. Neutral target faces that were replaced with Mondrian images in Study 3 are marked with an X. The
bottom row in each panel (labeled S) represents the images shown to the nondominant eye, including the
affective faces; these images are typically suppressed from conscious awareness.
 792 WORMWOOD ET AL.

the suppressed face. All trials in which perceivers selected the sessed the participant’s eye dominance and led the participant into
gender of the suppressed face or the “don’t know” option were an individual testing room with a computer and a mirror stereo-
excluded from analyses (17.9% of all trials in Study 1; 17.0% of all scope. The researcher calibrated each participant to the stereo-
trials in Study 2). This trial-by-trial measure of conscious visual scope, adjusting the mirrors and rests so that the stimuli being
awareness, or a very similar one, has been used in five published presented were aligned properly. Before each of the experimental
experiments examining the impact of suppressed affective faces on tasks, the researcher read instructions and watched while the
judgments about seen neutral faces (Anderson et al., 2012; Siegel participant completed five practice trials. Participants completed
et al., 2018). This approach utilizes a fairly conservative check for each task alone in the testing room with the lights off. At the end
visual conscious awareness because it successfully captures trials of the experimental session, researchers administered a debriefing
on which perceptual blending or breakthrough of the suppressed questionnaire assessing participants’ awareness of the suppressed
image may have occurred, but also likely captures a large number stimuli and the purpose of the experiment, as well as their com-
of trials on which the gender of the seen neutral face was not prehension of task instructions. They were then debriefed about the
observed for reasons unrelated to suppression of the image in the nature of the study and compensated for their participation.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

nondominant eye (i.e., distraction, blinking, fatigue). Participants

This document is copyrighted by the American Psychological Association or one of its allied publishers.

then rated how likable, trustworthy, and reliable they found each
Results
neutral target face on three 5-point scales: unlikeable to likable,
untrustworthy to trustworthy, and unreliable to reliable. Each participant’s three personality ratings of the neutral target
Participants completed a total of 540 trials of the person per- face (likable, trustworthy, reliable) were averaged together to
ception task: 10 neutral target faces (5 male, 5 female) ⫻ 3 create a single personality rating measure for each trial of the
suppressed facial pose conditions ⫻ 3 timing conditions ⫻ 6 person perception task. This was done to simplify the reporting and
repetitions. While a given neutral target face was always paired interpretation of results and because predictions did not differ for
with the same suppressed affective face throughout the experiment the dependent variables. We found very strong and consistent
for a given participant, these pairings were counterbalanced across internal reliability across the three ratings within each timing
participants, such that approximately one third of participants saw condition in Study 1 (affect after target condition: ␣ ⫽ .98;
any given neutral target face matched with each of the possible concurrent condition: ␣ ⫽ .98; affect before target condition: ␣ ⫽
suppressed affective poses (i.e., smile, scowl, neutral). The task .98) and in Study 2 (affect after target condition: ␣ ⫽ .96; con-
was broken into 6 blocks of 90 trials each, and participants were current condition: ␣ ⫽ .96; affect before target condition: ␣ ⫽
given a 2-min break to rest their eyes after each block. .95). Separate analyses on the individual personality ratings re-
Objective awareness task. The final task served as a measure vealed the same general pattern of results and can be found in the
of a participants’ objective awareness of the suppressed images. online supplemental materials (Tables S1–S4).
These trials were nearly identical to the experimental trials in the Study 1 results
concurrent condition of the person perception task except that (a) Objective awareness. Two participants in Study 1 did not
suppressed affective faces were presented upside down on half of complete the objective awareness task because the experimental
the trials, and (b) a scrambled image of a face (designed to be session ran over the allotted time (and participants opted not to
unidentifiable as a face) was presented to the dominant eye (in- stay to complete it). Of the participants who did complete the
stead of a neutral target face). Participants completed 60 trials of objective awareness task, five participants were able to correctly
this task: each of the 30 unique suppressed affective faces they guess the orientation of the suppressed face on more than 61.67%
were shown during the person perception task were presented once of the trials (better than chance, p ⬍ .05, two-tailed), demonstrat-
right-side up and once upside down (rotated 180°) at the same ing that the majority of participants did not demonstrate evidence
contrast level as used in the person perception task for each of objective awareness of the suppressed affective faces. These
participant. At the conclusion of each trial, participants were asked participants were included in all analyses, but the general pattern
to guess the orientation of the face, and then to rate the quality of of results for Study 1 did not change whether we included (or
their visual experience on the same 4-point PAS used during the excluded) these participants.
contrast adjustment task. If images presented to the nondominant Conscious visual awareness. All trials of the person percep-
eye were successfully suppressed throughout the experiment, par- tion task in which perceivers selected the gender of the sup-
ticipants should have no conscious awareness of the faces and pressed face or the “don’t know” option were excluded from
should report the correct orientation of the suppressed affective analyses (17.9% of all trials). To examine whether conscious
faces at chance level during this task. Performance on the objective visual awareness varied across affect and timing conditions, we
awareness task is useful for describing group-level performance, conducted a 3 (affect condition: suppressed smile, suppressed
demonstrating whether the majority of participants exhibit evi- neutral, suppressed scowl) ⫻ 3 (timing condition: affect after
dence of objective awareness in a task with nearly identical pa- target, concurrent, affect before target) repeated-measures analysis
rameters to the main experimental task (the person perception of variance (ANOVA) with number of included trials as the
task). We use exploratory analyses to confirm that any pattern of dependent variable. This analysis revealed that the number of
results does not depend on the inclusion/exclusion of individuals included/excluded trials did not differ significantly across affect
who showed better-than-chance performance on this separate task. conditions, F(2, 60) ⫽ 0.50, p ⫽ .61, ␩p2 ⫽ .016, and there was no
Procedure. A research assistant greeted participants and pro- significant interactions between timing and affect condition on the
vided a verbal description of the experiment. Participants then number of included trials, F(4, 120) ⫽ 0.93, p ⫽ .45, ␩p2 ⫽ .030.
provided informed consent followed by demographic information, There was, however, a significant difference in the number of
including gender, race, age, and handedness. The researcher as- included trials across timing conditions, F(2, 60) ⫽ 5.19, p ⫽ .008,
 AFFECTIVE REALISM 793

␩p2 ⫽ .147. A post hoc Fischer’s least significant difference (LSD) suppressed neutral faces (M ⫽ 2.93, SE ⫽ .12) were intermediate;
test revealed that there were significantly more included trials, on however, comparisons to personality ratings of seen neutral faces
average, within the affect before target conditions (M ⫽ 50.24, paired with suppressed smiling (p ⫽ .10) or suppressed scowling
SE ⫽ 2.44) compared with both the affect after target conditions faces (p ⫽ .24) did not reach conventional levels of significance.
(M ⫽ 45.80, SE ⫽ 2.70), p ⫽ .03, and the concurrent conditions The omnibus repeated-measures ANOVA also revealed a main
(M ⫽ 44.95, SE ⫽ 2.72), p ⫽ .03, which did not differ, p ⫽ .16. effect for timing condition on personality ratings, F(2, 60) ⫽ 5.50,
Person perception task. A 3 (affect condition: suppressed p ⫽ .006, ␩p2 ⫽ .155. A post hoc Fischer’s LSD test revealed that,
smile, suppressed neutral, suppressed scowl) ⫻ 3 (timing condi- across all affect conditions, seen neutral faces were rated more
tion: affect after target, concurrent, affect before target) repeated- positively when shown concurrently with the suppressed face
measures ANOVA with personality ratings as the dependent vari- (M ⫽ 2.98, SE ⫽ .12) than when shown before the suppressed face
able revealed that the critical interaction between affect condition (M ⫽ 2.92, SE ⫽ .12), p ⫽ .03, or after the suppressed face (M ⫽
and timing condition was significant, F(4, 120) ⫽ 3.91, p ⫽ .005, 2.90, SE ⫽ .11), p ⫽ .01. Personality ratings did not differ
␩p2 ⫽ .115 (Figure 2). To examine this interaction, we conducted significantly between the affect after target and the affect before
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

separate repeated-measures ANOVAs assessing the impact of af- target conditions, p ⫽ .31.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

fect condition on personality ratings within each of the timing Study 2 results
conditions separately. These analyses revealed a significant effect Objective awareness task. Twelve participants in Study 2 did
of affect condition on personality ratings in the concurrent condi- not complete the objective awareness task because the experimen-
tion, F(2, 60) ⫽ 5.49, p ⫽ .006, ␩p2 ⫽ .155, but no significant effect tal session ran over the allotted time (and participants opted not to
in the affect after target condition, F(2, 60) ⫽ 2.16, p ⫽ .12, ␩p2 ⫽ stay to complete it). Of the participants who did complete the
.067, or in the affect before target condition, F(2, 60) ⫽ 0.14, p ⫽ objective awareness task, nine participants from Study 2 were able
.87, ␩p2 ⫽ .005. As predicted, a post hoc Fischer’s LSD test to correctly guess the orientation of the suppressed face on more
revealed that, within the concurrent condition, seen neutral faces than 61.67% of the trials (better than chance, p ⬍ .05, two-tailed),
were rated significantly more positively when paired with sup- demonstrating that the majority of participants did not demonstrate
pressed smiling faces (M ⫽ 3.08, SE ⫽ .12) than with suppressed evidence of objective awareness of the suppressed affective faces.
neutral faces (M ⫽ 2.97, SE ⫽ .12), p ⫽ .009, or with suppressed These participants were included in all analyses, but the general
scowling faces (M ⫽ 2.89, SE ⫽ .12), p ⫽ .009. Differences in pattern of results for Study 2 did not change whether we included
personality ratings of neutral faces paired with suppressed scowl- (or excluded) these participants.
ing faces and suppressed neutral faces were in the predicted Conscious visual awareness. All trials of the person percep-
direction, but did not reach conventional levels of significance, tion task in which perceivers selected the gender of the suppressed
p ⫽ .20. face or the “don’t know” option were excluded from analyses
Our omnibus repeated-measures ANOVA also revealed a sig- (17.0% of all trials). To examine whether conscious visual aware-
nificant main effect for affect condition on personality ratings, F(2, ness varied across affect and timing conditions, we conducted a 3
60) ⫽ 3.61, p ⫽ .03, ␩p2 ⫽ .107. As predicted, a post hoc Fischer’s (affect condition: suppressed smile, suppressed neutral, suppressed
LSD test revealed that, across all timing conditions, seen neutral scowl) ⫻ 3 (timing condition: affect after target, concurrent, affect
faces were rated significantly more positively when paired with before target) repeated-measures ANOVA with number of in-
suppressed smiling faces (M ⫽ 2.98, SE ⫽ .11) than with sup- cluded trials as the dependent variable. This analysis revealed that
pressed scowling faces (M ⫽ 2.89, SE ⫽ .12), p ⫽ .02. Also as the number of included/excluded trials did not differ significantly
predicted, personality ratings of seen neutral faces paired with across timing conditions, F(2, 110) ⫽ 2.20, p ⫽ .12, ␩p2 ⫽ .038, or
across affect conditions, F(2, 110) ⫽ 0.40, p ⫽ .67, ␩p2 ⫽ .007, and
there was no significant interaction between timing and affect
3.3 condition on the number of included trials, F(4, 220) ⫽ 1.14, p ⫽
* .34, ␩p2 ⫽ .020.
3.2
Person perception task. In Study 2, the influence of the sup-
3.1
pressed affective faces on personality ratings once again differed
significantly across timing conditions, F(4, 220) ⫽ 4.17, p ⫽ .003;
Average Personality Rating

3 ␩p2 ⫽ .070 (Figure 3). As in Study 1, the effect of affect condition

Smile on personality ratings was only significant within the concurrent
2.9 Neutral condition, F(2, 110) ⫽ 7.42, p ⫽ .001, ␩p2 ⫽ .119; we found no
Scowl
significant effect in the affect after target condition, F(2, 110) ⫽
2.8
1.48, p ⫽ .23, ␩p2 ⫽ .026, or the affect before target condition, F(2,
2.7 110) ⫽ 2.21, p ⫽ .11, ␩p2 ⫽ .039. Within the concurrent condition,
as predicted, a post hoc Fischer’s LSD test revealed that seen
2.6 neutral faces were rated significantly less positively when paired
with suppressed scowling faces (M ⫽ 2.90, SE ⫽ .06) than with
2.5
Affect Before Target Concurrent Affect After Target suppressed smiling faces (M ⫽ 3.02, SE ⫽ .06), p ⫽ .002, or with
Timing Condition suppressed neutral faces (M ⫽ 3.00, SE ⫽ .06), p ⬍ .001. Differ-
ences in the personality ratings of neutral faces paired with sup-
Figure 2. Average personality rating by affect and timing conditions for pressed smiling faces and suppressed neutral faces were in the
Study 1. Error bars represent ⫾SE (between-subjects). ⴱ p ⬍ .05. predicted direction, but did not reach significance, p ⫽ .74.
 794 WORMWOOD ET AL.

3.1
* select “don’t know” if they saw a blend of faces/genders or were
unsure of the gender of the face they saw for any reason. We then
3.05
excluded from all analyses any trials on which they reported seeing
3
the gender of the suppressed face or a perceptual blend of faces
Average Personality Rating

(i.e., trials on which they selected “don’t know”). Importantly,

2.95 across both studies, we saw no evidence of higher subjective
Smile
Neutral
awareness in the concurrent condition relative to the other two
2.9
Scowl timing conditions. Thus, it seems highly unlikely that differences
2.85 in conscious visual awareness or the amount of perceptual blen-
ding across conditions could account for any of the reported
2.8 findings. Nevertheless, in Studies 1 and 2, there were no flashing
images in the dominant eye during the presentation of the sup-
2.75
pressed affective stimulus within the concurrent timing condition.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

2.7 We designed Study 3 to explicitly address this concern and rule out
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Affect Before Target Concurrent Affect After Target

alternative explanations for our findings based on differences in
Timing Condition awareness across timing conditions.
Figure 3. Average personality rating by affect and timing conditions for
Study 2. Error bars represent ⫾SE (between-subjects). ⴱ p ⬍ .05. Study 3
In Study 3, we adjusted the task structure such that dynamically
In addition, the omnibus repeated-measures ANOVA again re- changing, high contrast images were always presented to the
vealed a significant main effect of affect condition on personality dominant eye while low contrast, unchanging images were pre-
ratings, F(2, 110) ⫽ 3.62, p ⫽ .03, ␩p2 ⫽ .062. As predicted, a post sented to the nondominant eye across all trials. Specifically, the
hoc Fischer’s LSD test revealed that, across all timing conditions, neutral target face presented to the dominant eye was flashed
seen neutral faces were rated significantly less positively when continuously across all conditions (vs. being presented statically
paired with suppressed scowling faces (M ⫽ 2.90, SE ⫽ .06) than for 200 ms in Studies 1 and 2; See Figure 1). Critically, this
with suppressed smiling faces (M ⫽ 2.97, SE ⫽ .06), p ⫽ .01, or allowed us to rule out potential alternative explanations concerning
suppressed neutral faces (M ⫽ 2.98, SE ⫽ .05), p ⫽ .05. Differ- whether interocular suppression was being achieved via different
ences in the personality ratings of neutral faces paired with sup- mechanisms for trials within the concurrent condition relative to
pressed smiling faces and suppressed neutral faces were in the the affect before target and affect after target conditions in Studies
predicted direction, but did not reach significance, p ⫽ .73. 1 and 2 (e.g., binocular rivalry vs. CFS; Tsuchiya, Koch, Gilroy, &
Finally, the omnibus repeated-measures ANOVA also revealed Blake, 2006).
a main effect for timing condition on personality ratings, F(2,
110) ⫽ 11.27, p ⬍ .001, ␩p2 ⫽ .170. A post hoc Fischer’s LSD test Method
revealed that, for all affect conditions, seen neutral faces were
rated less positively in the affect before target condition (M ⫽ All procedures and materials were approved by the Northeastern
2.92, SE ⫽ .05) than the concurrent condition (M ⫽ 2.97, SE ⫽ University Institutional Review Board.
.06), p ⬍ .001, or the affect after target condition (M ⫽ 2.96, Participants. Target sample size for Study 3 was determined
SE ⫽ .06), p ⫽ .003. Personality ratings did not differ significantly by conducting a power analysis in GⴱPower (Faul et al., 2007)
between the concurrent and affect after target conditions, p ⫽ .17. using the effect size of the critical Affect Condition ⫻ Timing
Condition interaction from Study 2. This power analysis revealed
that in order for an interaction with an effect size of ␩2 ⫽ 0.070 to
Discussion
be detected (80% chance) with significance at the ␣ ⫽ .05 level, a
Across two studies, we see evidence that seen neutral target sample of at least 87 participants would be required. We recruited
faces are evaluated more positively when paired with suppressed 100 undergraduate students from Northeastern University (40
positive affective information than when paired with suppressed males, 60 females; Age 18 – 40, Mage ⫽ 19.43, SDage ⫽ 2.54) who
negative affective information. Critically, this effect is only sig- participated for course credit. All participants reported normal or
nificant when the suppressed affective information is presented corrected-to-normal visual acuity and were naïve to the experi-
concurrent with the seen neutral target face and not when the mental hypotheses. Individuals wearing glasses were excluded
neutral target face is presented before or after the affective infor- because glasses interfere with the proper function of the visual
mation. This pattern of results demonstrates that affect can be apparatus (mirror stereoscope). Prior to data analysis, two partic-
integrated into the experience of the seen neutral target face in real ipants were removed because they experienced difficulty using the
time and not just influence evaluations of it post hoc. stereoscope. One additional subject was removed during analyses
A major strength of these studies is the inclusion of the trial- for Study 3 because he or she had so few valid trials (13/540) that
by-trial measure of conscious visual awareness, which allows us to not all necessary cells for the analyses had valid responses.
directly assess the possibility of interocular fusion on all trials Experimental tasks. The procedure and experimental tasks
across all conditions. Specifically, the seen neutral face and the for Study 3 were nearly identical to that of Study 1, with the
suppressed affective face were always of opposite gender, and we exception that, within the primary task (the person perception
asked participants to report the gender of the face they saw or to task), the neutral target face presented to the dominant eye flashed
 AFFECTIVE REALISM 795

instead of being presented for a steady 200 ms on all trials (i.e., it 3.1
was presented twice for 50 ms separated by a Mondrian-type *
3.05
image for 50 ms; See Figure 1).
3

Average Personality Rating

Results
2.95
As in Studies 1 and 2, each participant’s three personality Smile
2.9 Neutral
ratings of the neutral target face (likable, trustworthy, reliable)
Scowl
were averaged together to create a single personality rating mea- 2.85
sure for each trial of the person perception task. We again found
very strong and consistent internal reliability across the three 2.8
ratings within each timing condition in Study 3 (affect after target
2.75
condition: ␣ ⫽ .96; concurrent condition: ␣ ⫽ .95; affect before
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

target condition: ␣ ⫽ .95). Separate analyses on the individual 2.7

This document is copyrighted by the American Psychological Association or one of its allied publishers.

personality ratings revealed the same general pattern of results and Affect Before Target Concurrent Affect After Target
can be found in the online supplemental materials (Tables S5 and Timing Condition
S6).
Objective awareness task. One participant in Study 3 did not Figure 4. Average personality rating by affect and timing conditions for
complete the objective awareness task because the experimental Study 3. Error bars represent ⫾SE (between-subjects). ⴱ p ⬍ .05.
session ran over the allotted time (and the participant opted not to
stay to complete it). Of the participants who did complete the
objective awareness task, 13 participants were able to correctly
guess the orientation of the suppressed face on more than 61.67% concurrent condition, as predicted, a post hoc Fischer’s LSD test
of the trials (better than chance, p ⬍ .05, two-tailed), demonstrat- revealed that seen neutral faces were rated significantly less pos-
ing that the majority of participants did not demonstrate evidence itively when paired with suppressed scowling faces (M ⫽ 2.81,
of objective awareness of the suppressed affective faces. These SE ⫽ .05) than with suppressed smiling faces (M ⫽ 2.97, SE ⫽
participants were included in all analyses, but the general pattern .05), p ⫽ .001, or with suppressed neutral faces (M ⫽ 2.92, SE ⫽
of results for Study 3 did not change whether we included (or .05), p ⫽ .007. Differences in the personality ratings of neutral
excluded) these participants. faces paired with suppressed smiling faces and suppressed neutral
Conscious visual awareness. All trials of the person percep- faces were in the predicted direction, but did not reach signifi-
tion task in which perceivers selected the gender of the suppressed cance, p ⫽ .21.
face or the “don’t know” option were excluded from analyses In addition, the omnibus repeated-measures ANOVA again re-
(31.7% of trials).2 To examine whether conscious visual awareness vealed a significant main effect of affect condition on personality
varied across affect and timing conditions, we conducted a 3 ratings, F(2, 192) ⫽ 5.51, p ⫽ .005, ␩p2 ⫽ .054. As predicted, a
(affect condition: suppressed smile, suppressed neutral, suppressed post hoc Fischer’s LSD test revealed that, across all timing con-
scowl) ⫻ 3 (timing condition: affect after target, concurrent, affect ditions, seen neutral faces were rated significantly less positively
before target) repeated-measures ANOVA with number of in-
when paired with suppressed scowling faces (M ⫽ 2.83, SE ⫽ .05)
cluded trials as the dependent variable. This analysis revealed that
than with suppressed smiling faces (M ⫽ 2.92, SE ⫽ .05), p ⫽
the number of included/excluded trials did not differ significantly
.009, or suppressed neutral faces (M ⫽ 2.88, SE ⫽ .05), p ⫽ .04.
across affect conditions, F(2, 192) ⫽ 2.37, p ⫽ .10, ␩p2 ⫽ .024, and
Differences in the personality ratings of neutral faces paired with
there was no significant interaction between timing and affect
condition on the number of included trials, F(4, 384) ⫽ 0.54, p ⫽ suppressed smiling faces and suppressed neutral faces were in the
.71, ␩p2 ⫽ .006. There was, however, a significant difference in the predicted direction, but did not reach significance, p ⫽ .09.
number of included trials across timing conditions, F(2, 192) ⫽ Finally, the omnibus repeated-measures ANOVA also revealed
19.60, p ⬍ .001, ␩p2 ⫽ .170. A post hoc Fischer’s LSD test revealed a main effect for timing condition on personality ratings, F(2,
that there were significantly more included trials, on average, 192) ⫽ 3.57, p ⫽ .03, ␩p2 ⫽ .036. A post hoc Fischer’s LSD test
within the affect before target conditions (M ⫽ 43.35, SE ⫽ 1.60) revealed that, for all affect conditions, seen neutral faces were
compared with both the affect after target conditions (M ⫽ 39.94, rated significantly more positively in the concurrent condition
SE ⫽ 1.71), p ⬍ .001, and the concurrent conditions (M ⫽ 39.67, (M ⫽ 2.90, SE ⫽ .04) than the affect after target condition (M ⫽
SE ⫽ 1.73), p ⬍ .001, which did not differ, p ⫽ .61. 2.86, SE ⫽ .05), p ⫽ .02. Personality ratings within the affect
Person perception task. In Study 3, the influence of the before target condition (M ⫽ 2.87, SE ⫽ .04) did not differ
suppressed affective faces on personality ratings once again dif- significantly from those in either the concurrent condition, p ⫽ .07,
fered significantly across timing conditions, F(4, 384) ⫽ 3.87, p ⫽ or the affect after target condition, p ⫽ .36.
.004; ␩p2 ⫽ .039 (see Figure 4). As in Studies 1 and 2, the effect of
affect condition on personality ratings was only significant within
2
the concurrent condition, F(2, 192) ⫽ 8.09, p ⬍ .001, ␩p2 ⫽ .078; We believe the larger rate of trial-by-trial exclusions in Study 3 is due
to the more rapid, flashing presentation of the neutral target face across all
we found no significant effect in the affect after target condition, trial types. Participants may have selected the “don’t know” option much
F(2, 192) ⫽ 1.43, p ⫽ .24, ␩p2 ⫽ .015, or the affect before target more frequently because they were not able to identify the gender of the
condition, F(2, 192) ⫽ 2.41, p ⫽ .09, ␩p2 ⫽ .025. Within the rapidly presented neutral target face.
 796 WORMWOOD ET AL.

Discussion within both the affect before target and concurrent conditions. Instead,
results are more consistent with the affective realism hypothesis.
Replicating Studies 1 and 2, we found that seen neutral faces The present findings are consistent with recent empirical work
were evaluated more positively when presented concurrent with demonstrating that one’s affective state may influence how positive or
suppressed affectively positive stimuli than when presented con- negative a neutral target face looks to the perceiver in a very literal
current with suppressed affectively negative stimuli and that eval- way (Siegel et al., 2018): neutral faces were perceived as looking
uations of neutral target faces did not differ on trials where the more smiling when presented concurrent with suppressed affectively
suppressed affective faces were not presented concurrent with the positive stimuli and as looking more scowling when presented con-
neutral target faces. Moreover, we again saw no evidence of current with suppressed affectively negative stimuli. Thus, affective
greater subjective awareness of the suppressed affective stimuli realism may involve changes in actual perception. What is unclear is
within the concurrent timing conditions compared with the offset whether the mechanism is a type of sensory integration (similar to
timing conditions. Thus, even with the neutral target faces flashing multimodal processing) or some other process. We believe our data
dynamically on all trials, we find consistent support for the affec- lend support to the idea that affect, at least in the case of affective
tive realism hypothesis: affect can influence the experience of a
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

realism effects, may be acting like a “sixth sense” that is integrated

neutral target stimulus in real time, not just the evaluation of that
This document is copyrighted by the American Psychological Association or one of its allied publishers.

with other sensory processing (e.g., vision or hearing), similar to the

target stimulus post hoc. multimodal processing that has been repeatedly demonstrated among
the “traditional” senses. For example, visual stimuli can shape sound
General Discussion perception (e.g., the McGurk effect; McGurk & MacDonald, 1976).
Moreover, past research has shown that sensory events are more likely
Across all three studies, the patterns of results provide consistent to be perceived as coming from the same source when they occur in
evidence that affective realism can be distinguished from more gen- close temporal proximity, irrespective of physical location (Calvert,
eral affective misattribution. Whereas the affective misattribution Brammer, & Iversen, 1998; Spence, 2007). In a similar way, affective
hypothesis would predict that the effect of the suppressed affective feelings may bind with the visual properties of a neutral face, causing
faces on personality ratings of seen neutral faces would be present the person to literally be seen as more or less trustworthy, reliable, and
across all timing conditions, results instead reveal that the effect was likeable when the feelings and the face occur in close temporal
only significant when affective information was presented at the same proximity.
time as the target stimulus. That is, neutral faces were experienced Although our results demonstrate that affective realism is sensitive
more positively (i.e., as more trustworthy, likable, and reliable) when to very small time differences between the presentation of a target
presented at the same time as suppressed smiling faces, and were stimulus and incidental affective information (i.e., 100 ms), they do
experienced less positively when presented at the same time as sup- not show that this close temporal coupling is necessary to produce
pressed scowling faces. These findings suggest that affective feelings affective realism. That is, we must be careful about drawing conclu-
are being integrated into the experience of the target face. Thus, our sions concerning the null results in our offset timing conditions (when
findings support the philosophical conjecture that affect is a property the affective face was presented before the target or after the target).
of conscious experience. There are, in fact, many examples where It is not clear whether nonsignificant differences in these timing
affect is perceived as a property of objects and people in the world, conditions reflect a true null effect or simply a weaker effect of the
similar to the way we typically perceive color as a property of objects affective information that the current studies are not sufficiently
in the world (e.g., Danziger et al., 2011; Leibovich et al., 2016; powered to observe. Significant but weaker effects in the nonconcur-
Veltkamp et al., 2008). In the studies we report here, participants did rent timing conditions could provide evidence that affective realism
not appear to use negative/positive feelings to guide post hoc judg- weakens as a function of timing offset, or it could suggest that, within
ments of another person; they experienced the person as more nega- a single task, affect may influence person perception via multiple
tive/positive. separable mechanisms. Nevertheless, the significant interaction dem-
Furthermore, our findings demonstrate that affective realism and onstrates support for the affective realism hypothesis by showing that
affective misattribution are empirically distinguishable. To be clear, the relative timing is important for this particular social judgment
not all instances of affective misattribution are necessarily affective effect; a tight temporal coupling (⬍100 ms) between target and
realism. Our studies show, however, that some subset of instances affective stimuli significantly enhances the influence of affect on
previously understood as affective misattribution may, in fact, be person perception, suggesting that affect can indeed influence the
better explained by this more specific causal relationship. Interest- experience of target stimuli in real-time. However, it would be inter-
ingly, our results also cannot be explained as another common phe- esting to directly explore the nuances of these offset timing conditions
nomenon, affective priming, which would occur when the unseen in future work. For example, a future study could employ a greater
affective stimuli were presented prior to the seen target stimuli. The number of timing conditions where the timing off-sets are manipu-
timing offsets used in these studies (100 –150 ms) are well within the lated in smaller increments (e.g., 5 ms). Such a design would allow us
normal range of timings used in typical priming manipulations. For to plot the strength of the effect as a continuous function of the timing
example, it is common to have a subliminal affective prime precede off-set in both the affect before and affect after target conditions.
a target stimulus by 125 ms (e.g., Gawronski & Ye, 2014; Payne et al., Results would reveal, not only the boundary conditions of affective
2005) and subliminal primes have been shown to influence judgments realism, but to some extent would also inductively reveal the timing
of target stimuli even when they precede it by as much as 1,500 ms boundaries of a single conscious event (as opposed to two distinct
(Payne et al., 2005). If the present effects were merely an example of events).
subliminal priming, we should have seen a significant effect of the Our findings highlight several interesting avenues for future re-
suppressed affective faces on personality ratings of seen neutral faces search. The suppressed facial configurations utilized in the present
 AFFECTIVE REALISM 797

study (i.e., smiling, scowling) have been shown in prior work to Barrett, L. F., & Simmons, W. K. (2015). Interoceptive predictions in the
influence judgments of the personality traits we assessed (see, e.g., brain. Nature Reviews Neuroscience, 16, 419 – 429. http://dx.doi.org/10
Campellone & Kring, 2013, on trust-related evaluations). Future work .1038/nrn3950
could examine whether affective realism is robust to less directly Barrett, L., & Wormwood, J. B. (2015, April 19). When a gun is not a gun.
relevant affective stimuli (e.g., when using suppressed faces with The New York Times, p. SR9.
Bartlett, M. Y., Condon, P., Cruz, J., Baumann, J., & Desteno, D. (2012).
different positive and negative facial configurations (e.g., fear, pride)
Gratitude: Prompting behaviours that build relationships. Cognition and
or suppressed affective stimuli that are not faces). In addition, future
Emotion, 26, 2–13. http://dx.doi.org/10.1080/02699931.2011.561297
research could develop and test suppression paradigms to allow for
Bartlett, M. Y., & DeSteno, D. (2006). Gratitude and prosocial behavior:
the simultaneous presentation of a consciously perceived visual stim- Helping when it costs you. Psychological Science, 17, 319 –325. http://
ulus (e.g., a neutral target face) and a suppressed affective stimulus to dx.doi.org/10.1111/j.1467-9280.2006.01705.x
spatially distinct portions of the visual field. Although we excluded Baumann, J., & Desteno, D. (2010). Emotion guided threat detection:
trials in the present analyses where participants reported consciously Expecting guns where there are none. Journal of Personality and Social
perceiving the suppressed affective face or a blending of the neutral Psychology, 99, 595– 610. http://dx.doi.org/10.1037/a0020665
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

target face and suppressed affective face, a paradigm that involved Baumann, J., & DeSteno, D. (2012). Context explains divergent effects of
This document is copyrighted by the American Psychological Association or one of its allied publishers.

separating the images spatially could more completely rule out any anger on risk taking. Emotion, 12, 1196 –1199. http://dx.doi.org/10
possibility that interocular fusion contributed to the strength of the .1037/a0029788
present findings. Bowling, A., & Lovegrove, W. (1980). The effect of stimulus duration on
the persistence of gratings. Perception & Psychophysics, 27, 574 –578.
http://dx.doi.org/10.3758/BF03198688
Conclusion Bowling, A., & Lovegrove, W. (1981). Two components to visible persis-
tence: Effects of orientation and contrast. Vision Research, 21, 1241–
Reality is not a direct perception of what is out there in the 1251. http://dx.doi.org/10.1016/0042-6989(81)90229-7
world (for discussions, see Barrett, 2017; Barrett & Bliss- Calvert, G. A., Brammer, M. J., & Iversen, S. D. (1998). Crossmodal
Moreau, 2009; Lambie & Marcel, 2002). As perceivers, peo- identification. Trends in Cognitive Sciences, 2, 247–253. http://dx.doi
ple’s experience of the world is meaningfully shaped by the .org/10.1016/S1364-6613(98)01189-9
affect they feel in real time. The fact that we perceive the world Campellone, T. R., & Kring, A. M. (2013). Who do you trust? The impact
differently, depending on how we feel, has wide-reaching im- of facial emotion and behaviour on decision making. Cognition and
plications for real-world situations of great import. For exam- Emotion, 27, 603– 620. http://dx.doi.org/10.1080/02699931.2012
ple, the affective realism hypothesis sheds new light on cases of .726608
military and police violence in which intense emotions appear Chanes, L., & Barrett, L. F. (2016). Redefining the role of limbic areas in
cortical processing. Trends in Cognitive Sciences, 20, 96 –106. http://dx
to actively shape the way military and law enforcement person-
.doi.org/10.1016/j.tics.2015.11.005
nel perceive threats around them, and may lead to deadly errors
Clore, G. L., Gasper, K., & Garvin, E. (2001). Affect as information. In
in judgment when experience is amplified or otherwise dis-
J. P. Forgas (Ed.), Handbook of affect and social cognition (pp. 121–
torted by affect (for a discussion, see Barrett & Wormwood, 144). Mahwah, NJ: Erlbaum.
2015). Research on affective realism stands to fundamentally Clore, G. L., & Huntsinger, J. R. (2007). How emotions inform judgment
alter the way in which scientists understand perception and its and regulate thought. Trends in Cognitive Sciences, 11, 393–399. http://
impact on decision making, with important implications for dx.doi.org/10.1016/j.tics.2007.08.005
how we train and evaluate individuals who must act rapidly Danziger, S., Levav, J., & Avnaim-Pesso, L. (2011). Extraneous factors in
under pressure in dangerous situations. judicial decisions. Proceedings of the National Academy of Sciences of
the United States of America, 108, 6889 – 6892. http://dx.doi.org/10
.1073/pnas.1018033108
References DeSteno, D., Bartlett, M. Y., Baumann, J., Williams, L. A., & Dickens, L.
(2010). Gratitude as moral sentiment: Emotion-guided cooperation in
Anderson, E., Siegel, E., White, D., & Barrett, L. F. (2012). Out of sight
economic exchange. Emotion, 10, 289 –293. http://dx.doi.org/10.1037/
but not out of mind: Unseen affective faces influence evaluations and
a0017883
social impressions. Emotion, 12, 1210 –1221. http://dx.doi.org/10.1037/
Dolman, C. P. (1919). Tests for determining the sighting eye. American
a0027514
Balcetis, E., & Dunning, D. (2006). See what you want to see: Motivational Journal of Ophthalmology, 2, 867. http://dx.doi.org/10.1016/S0002-
influences on visual perception. Journal of Personality and Social Psy- 9394(19)90258-3
chology, 91, 612– 625. http://dx.doi.org/10.1037/0022-3514.91.4.612 Efron, R. (1970). The minimum duration of a perception. Neuropsycholo-
Barrett, L. F. (2017). How emotions are made: The secret life of the brain. gia, 8, 57– 63. http://dx.doi.org/10.1016/0028-3932(70)90025-4
New York, NY: Houghton Mifflin-Harcourt. Faul, F., Erdfelder, E., Lang, A. G., & Buchner, A. (2007). GⴱPower 3: A
Barrett, L. F., & Bar, M. (2009). See it with feeling: Affective predictions flexible statistical power analysis program for the social, behavioral, and
during object perception. Philosophical Transactions of the Royal Soci- biomedical sciences. Behavior Research Methods, 39, 175–191. http://
ety of London, Series B: Biological Sciences, 364, 1325–1334. http://dx dx.doi.org/10.3758/BF03193146
.doi.org/10.1098/rstb.2008.0312 Fink, W. H. (1938). The dominant eye: Its clinical significance. Archives
Barrett, L. F., & Bliss-Moreau, E. (2009). Affect as a psychological of Ophthalmology, 19, 555–582. http://dx.doi.org/10.1001/archopht
primitive. In M. P. Zanna (Ed.), Advances in experimental social psy- .1938.00850160081005
chology (Vol. 41, pp. 167–218). Burlington, MA: Academic Press. Garg, N., Inman, J. J., & Mittal, V. (2005). Incidental and task-related
Barrett, L. F., & Russell, J. A. (2015). An introduction to psychological affect: A re-inquiry and extension of the influence of affect on choice.
construction. In L. F. Barrett & J. A. Russell (Eds.), The psychological Journal of Consumer Research, 32, 154 –159. http://dx.doi.org/10.1086/
construction of emotion (pp. 1–17). New York, NY: Guilford Press. 426624
 798 WORMWOOD ET AL.

Garg, N., Wansink, B., & Inman, J. J. (2007). The influence of incidental Phelps, E. A., Ling, S., & Carrasco, M. (2006). Emotion facilitates per-
affect on consumers’ food intake. Journal of Marketing, 71, 194 –206. ception and potentiates the perceptual benefits of attention. Psycholog-
http://dx.doi.org/10.1509/jmkg.71.1.194 ical Science, 17, 292–299. http://dx.doi.org/10.1111/j.1467-9280.2006
Gasper, K., & Clore, G. L. (2002). Attending to the big picture: Mood and .01701.x
global versus local processing of visual information. Psychological Ramsøy, T. Z., & Overgaard, M. (2004). Introspection and subliminal
Science, 13, 34 – 40. http://dx.doi.org/10.1111/1467-9280.00406 perception. Phenomenology and the Cognitive Sciences, 3, 1–23. http://
Gawronski, B., & Ye, Y. (2014). What drives priming effects in the affect dx.doi.org/10.1023/B:PHEN.0000041900.30172.e8
misattribution procedure? Personality and Social Psychology Bulletin, Redelmeier, D. A., & Baxter, S. D. (2009). Rainy weather and medical
40, 3–15. http://dx.doi.org/10.1177/0146167213502548 school admission interviews. Canadian Medical Association Journal,
Gruber, J., Siegel, E. H., Purcell, A. L., Earls, H. A., Cooper, G., & Barrett, 181, 933–933. http://dx.doi.org/10.1503/cmaj.091546
L. F. (2016). Unseen positive and negative affective information influ- Schwarz, N., & Clore, G. L. (1983). Mood, misattribution, and judgments
ences social perception in bipolar I disorder and healthy adults. Journal of well-being: Informative and directive functions of affective states.
of Affective Disorders, 192, 191–198. Journal of Personality and Social Psychology, 45, 513–523. http://dx
Hirshleifer, D., & Shumway, T. (2003). Good day sunshine: Stock returns .doi.org/10.1037/0022-3514.45.3.513
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

and the weather. The Journal of Finance, 58, 1009 –1032. http://dx.doi Siegel, E. H., & Stefanucci, J. K. (2011). A little bit louder now: Negative
.org/10.1111/1540-6261.00556 affect increases perceived loudness. Emotion, 11, 1006 –1011. http://dx
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Isen, A. M., & Levin, P. F. (1972). Effect of feeling good on helping: .doi.org/10.1037/a0024590
Cookies and kindness. Journal of Personality and Social Psychology, Siegel, E. H., Wormwood, J. B., Quigley, K. S., & Barrett, L. F. (2018).
21, 384 –388. http://dx.doi.org/10.1037/h0032317 Seeing what you feel: Affect drives visual perception of structurally
Johnson, E. J., & Tversky, A. (1983). Affect, generalization, and the neutral faces. Psychological Science, 29, 496 –503. http://dx.doi.org/10
perception of risk. Journal of Personality and Social Psychology, 45, .1177/0956797617741718
20 –31. http://dx.doi.org/10.1037/0022-3514.45.1.20 Spence, C. (2007). Audiovisual multisensory integration. Acoustical Sci-
Keltner, D., & Lerner, J. S. (2010). Emotion. In S. T. Fiske, D. T. Gilbert, ence and Technology, 28, 61–70. http://dx.doi.org/10.1250/ast.28.61
& G. Lindzey (Ed.), Handbook of social psychology (5th ed., Vol. 1, pp. Teachman, B. A., Stefanucci, J. K., Clerkin, E. M., Cody, M. W., &
317–352). Hoboken, NJ: Wiley. Proffitt, D. R. (2008). A new mode of fear expression: Perceptual bias in
Kring, A. M., Siegel, E. H., & Barrett, L. F. (2014). Unseen affective faces height fear. Emotion, 8, 296 –301. http://dx.doi.org/10.1037/1528-3542
influence person perception judgments in schizophrenia. Clinical .8.2.296
Psychological Science, 2, 443– 454. http://dx.doi.org/10.1177/ Tsuchiya, N., & Koch, C. (2005). Continuous flash suppression reduces
2167702614536161 negative afterimages. Nature Neuroscience, 8, 1096 –1101. http://www
Lambie, J. A., & Marcel, A. J. (2002). Consciousness and the varieties of .nature.com/neuro/journal/v8/n8/suppinfo/nn1500_S1.html
emotion experience: A theoretical framework. Psychological Review, Tsuchiya, N., Koch, C., Gilroy, L. A., & Blake, R. (2006). Depth of
109, 219 –259. http://dx.doi.org/10.1037/0033-295X.109.2.219 interocular suppression associated with continuous flash suppression,
Leibovich, T., Cohen, N., & Henik, A. (2016). Itsy bitsy spider? Biological flash suppression, and binocular rivalry. Journal of Vision, 6, 6. http://
Psychology, 121, 138 –145. http://dx.doi.org/10.1016/j.biopsycho.2016 dx.doi.org/10.1167/6.10.6
.01.009 van Ulzen, N. R., Semin, G. R., Oudejans, R. R., & Beek, P. J. (2008).
Lerner, J. S., & Tiedens, L. Z. (2006). Portrait of the angry decision maker: Affective stimulus properties influence size perception and the Ebbing-
How appraisal tendencies shape anger’s influence on cognition. Journal haus illusion. Psychological Research, 72, 304 –310. http://dx.doi.org/
of Behavioral Decision Making, 19, 115–137. http://dx.doi.org/10.1002/ 10.1007/s00426-007-0114-6
bdm.515 Veltkamp, M., Aarts, H., & Custers, R. (2008). Perception in the service of
Loewenstein, G., & Lerner, J. S. (2003). The role of affect in decision goal pursuit: Motivation to attain goals enhances the perceived size of
making. In R. Davidson, H. Goldsmith, & K. Scherer (Eds.), Handbook goal-instrumental objects. Social Cognition, 26, 720 –736. http://dx.doi
of affective science (pp. 619 – 642). Oxford, UK: Oxford University .org/10.1521/soco.2008.26.6.720
Press. Winkielman, P., Berridge, K. C., & Wilbarger, J. L. (2005). Unconscious
McGinnis, S. M., Brickhouse, M., Pascual, B., & Dickerson, B. C. (2011). affective reactions to masked happy versus angry faces influence con-
Age-related changes in the thickness of cortical zones in humans. Brain sumption behavior and judgments of value. Personality and Social
Topography, 24, 279 –291. http://dx.doi.org/10.1007/s10548-011- Psychology Bulletin, 31, 121–135. http://dx.doi.org/10.1177/
0198-6 0146167204271309
McGurk, H., & MacDonald, J. (1976). Hearing lips and seeing voices. Winkielman, P., Zajonc, R. B., & Schwarz, N. (1997). Subliminal affective
Nature, 264, 746 –748. http://dx.doi.org/10.1038/264746a0 priming resists attributional interventions. Cognition and Emotion, 11,
Murphy, S. T., & Zajonc, R. B. (1993). Affect, cognition, and awareness: 433– 465. http://dx.doi.org/10.1080/026999397379872
Affective priming with optimal and suboptimal stimulus exposures. Wormwood, J. B., Lynn, S. K., Feldman Barrett, L., & Quigley, K. S.
Journal of Personality and Social Psychology, 64, 723–739. http://dx (2016). Threat perception after the Boston Marathon bombings: The
.doi.org/10.1037/0022-3514.64.5.723 effects of personal relevance and conceptual framing. Cognition and
Niedenthal, P. M. (1990). Implicit perception of affective information. Emotion, 30, 539 –549. http://dx.doi.org/10.1080/02699931.2015
Journal of Experimental Social Psychology, 26, 505–527. http://dx.doi .1010487
.org/10.1016/0022-1031(90)90053-O Zadra, J. R., & Clore, G. L. (2011). Emotion and perception: The role of
Oliver, G., Wardle, J., & Gibson, E. L. (2000). Stress and food choice: A affective information. WIREs Cognitive Science, 2, 676 – 685. http://dx
laboratory study. Psychosomatic Medicine, 62, 853– 865. http://dx.doi .doi.org/10.1002/wcs.147
.org/10.1097/00006842-200011000-00016
Payne, B. K., Cheng, C. M., Govorun, O., & Stewart, B. D. (2005). An
inkblot for attitudes: Affect misattribution as implicit measurement. Received July 6, 2017
Journal of Personality and Social Psychology, 89, 277–293. http://dx Revision received May 29, 2018
.doi.org/10.1037/0022-3514.89.3.277 Accepted June 8, 2018 䡲