Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Cognition and Emotion

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/pcem20

Social impression formation and depression:


examining cognitive flexibility and bias

Wisteria Deng, Tyrone D. Cannon & Jutta Joormann

To cite this article: Wisteria Deng, Tyrone D. Cannon & Jutta Joormann (2023) Social impression
formation and depression: examining cognitive flexibility and bias, Cognition and Emotion,
37:1, 137-146, DOI: 10.1080/02699931.2023.2165043

To link to this article: https://doi.org/10.1080/02699931.2023.2165043

View supplementary material

Published online: 06 Jan 2023.

Submit your article to this journal

Article views: 303

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=pcem20
COGNITION AND EMOTION
2023, VOL. 37, NO. 1, 137–146
https://doi.org/10.1080/02699931.2023.2165043

BRIEF ARTICLE

Social impression formation and depression: examining cognitive


flexibility and bias
Wisteria Deng , Tyrone D. Cannon and Jutta Joormann
Department of Psychology, Yale University, New Haven, CT, USA

ABSTRACT ARTICLE HISTORY


Depression is associated with a bias toward negative interpretations of social Received 26 September 2022
situations and resistance to integrating evidence consistent with positive Revised 18 December 2022
interpretations. These features could contribute to social isolation by generating Accepted 30 December 2022
negative expected value for future social interactions. The present study examined
KEYWORDS
potential associations between depressive symptoms and positive (i.e. trust and Cognitive inflexibility; social
liking) and negative (i.e. distrust and disliking) social impression formation of impression; interpersonal
individuals who previously appeared in positive or negative contexts. Participants relationship; context
(N = 213) completed the Interpretation Inflexibility Task and were subsequently insensitivity
asked to provide social impression ratings of characters from each scenario type of
the task (i.e. positive and negative) as well as characters not previously
encountered. In examining social impression formation, higher severity of
depressive symptoms was associated with higher negative social impression
ratings regardless of scenario outcome, as well as lower positive social impression
ratings, but only for characters who previously appeared in positive contexts.
Those higher in depression also rated novel characters as significantly more
unlikeable and untrustworthy and to an equivalent degree as the characters
previously encountered. These findings suggest a role of negative interpretation
bias and inflexibility in contributing to negative evaluations of potential social
interaction partners in depression.

One of depression’s most pernicious features is reluc- social interaction are devalued. That is, in those with
tance to engage in activities, in particular social inter- depression, social stimuli, rather than predicting posi-
action, that are potentially pleasurable. Social tive reward value, may be predictive of negative
isolation, in turn, likely contributes to the mainten- reward value, thereby inhibiting engagement (Silk
ance of depressed mood, as in a vicious cycle (Hu et al., 2012). In this formulation, the key question
et al., 2022). Given the potential salutary effects of becomes, what factors drive the prediction of nega-
social interaction, understanding the mechanisms tive reward value of social interaction in those with
underlying low social interest and engagement in depression?
depression is crucial for developing interventions Prior work has shown that when interpreting
that could harness interactions with others in the ambiguous emotional stimuli, individuals with
service of elevating mood and maintaining euthymia. depression exhibit strong initial biases (priors) for
Viewed within a reinforcement learning/predictive negative interpretations (Joormann & Gotlib, 2010)
modelling framework, reduced social interest and and reduced integration of evidence disconfirming
engagement in depression can be conceptualised as negative interpretations, while also showing
reflecting a state in which the rewarding qualities of intact (or even enhanced) integration of evidence

CONTACT Wisteria Deng wisteria.deng@yale.edu Department of Psychology, Yale University, 2 Hillhouse Avenue, New Haven, CT
06520, USA
Supplemental data for this article can be accessed online at https://doi.org/10.1080/02699931.2023.2165043.
© 2023 Informa UK Limited, trading as Taylor & Francis Group
138 W. DENG ET AL.

disconfirming positive interpretations (Deng et al., (Radke et al., 2014), they may be more likely to main-
2022; Everaert et al., 2018). How these biases affect tain both a negative view of self and negative expec-
social impression formation and social behaviour in tations of future social interactions.
depression has not yet been determined empirically. The impression formation process, especially the
In general, negative interpretations of past inter- judgment of another person’s character, relies
actions frequently prompt the formation of negative heavily on prior knowledge about how that person
impressions about people involved in those inter- acts in interpersonal contexts (Rogers, 2018).
actions (Denrell, 2005). Such negative biases in However, because depressed individuals show a
social perception may be especially accentuated in negative interpretation bias for both negative and
depression (Silk et al., 2012), contributing to social positive emotional stimuli (Feeser et al., 2013) as
withdrawal and perpetuating negative social well as an asymmetric deficit in resolving ambiguous
impressions and depressed mood (Derntl et al., scenarios toward more positive interpretations (Deng
2011). Indeed, studies examining social functioning et al., 2022; Everaert et al., 2018), they are more likely
in depression have suggested an importance of to retain negative evaluations of prior interpersonal
understanding biased impression formation in contexts and thereby negative expectations of
depression (Haim-Nachum & Levy-Gigi, 2021). It has future interactions with others who were present in
been theorised that even as other people display those contexts, regardless of whether the scenarios
good intentions and initiate social interaction with involved objectively positive or negative outcomes.
positive outcomes, depression-prone individuals The present study utilised the Interpretation
may maintain initial negative impressions and experi- Inflexibility Task as a platform for examining the
ence difficulty forming positive impressions (Hames relationships between depression, interpretation
et al., 2013). While a theoretical connection between bias and flexibility, and social impression formation.
biased social impression formation and interpretation After rating interpretations of initially ambiguous
inflexibility in depression has been established, social situations that resolved toward a negative or
empirical research is needed to test this proposed a positive outcome, participants viewed faces of ancil-
relationship. lary characters from the scenarios (as well as new
If negative interpretation bias and inflexibility in faces not previously encountered) and rated the char-
evaluating ambiguous socioemotional contexts con- acters’ trustworthiness and likelihood of compliment-
tribute to a negative expected value of future social ing them in a future interaction, as well as the inverses
interaction, they are likely to do so by associative pro- of these characteristics, likelihood of plotting against
cesses in which individuals present in those contexts them and disliking them. In particular, we aimed to
come to be viewed in a negative light. Specifically, test if there is an interaction effect between
the perception of other people’s trustworthiness depression severity and IIT scenario emotional
and liking can affect one’s expectations regarding valence (positive, negative and foils) on social
whether future social interactions with them will be impression ratings. Based on the foregoing, we pre-
experienced as rewarding (Denrell, 2005; Soygüt & dicted that both depressive symptoms and interpret-
Savaşir, 2001). Individuals higher in depression ation inflexibility would be associated with more
expect less complimentary responses from others in negative social impressions (i.e. perceiving distrust
friendly situations and rate the expected responses and dislike from others), regardless of whether the
of others as more undesirable than those lower in scenario resolved to a positive or negative outcome.
depression (Soygüt & Savaşir, 2001). Perceiving We also expected to replicate previous findings of a
oneself as unlikable may result in the anticipation of bias and difficulties revising the bias (i.e. inflexibility)
social rejection and the subsequent need for exces- among individuals with depressive symptoms.
sive reassurance seeking in depression (Stewart &
Harkness, 2017), whereas perceiving others as
untrustworthy interferes with recognising others’ Method
positive feedback and revising negative social
Participants
impressions accordingly (Li et al., 2019). As individuals
with depressive symptoms question the credibility of Participants (N = 213) were recruited via Amazon’s
positive feedback and disregard other positive social Mechanical Turk (MTurk), an online crowdsourcing
cues (e.g. perceiving happy faces as less engaging) platform that provides access to a large and diverse
COGNITION AND EMOTION 139

sample for mental health research studies (demo- pictures that make up the scenario (panels 1a-1c)
graphics: see Supplement Table 1). Participation in reveal more information about the scenario, culminat-
this study was restricted to MTurk users who were ing in panel 1c, which encourages a positive interpret-
18 years or older and lived in the United States. ation. An example negative outcome scenario of the
IIT is also shown in Figure 1 (panels 2a-2c). Each of
the three pictures (panels 2a-2c) reveals more infor-
Data quality
mation about the scenario, culminating in panel 2c,
Following recommendations for research using which prompts a negative interpretation. For each
crowdsourced samples, the study only recruited of the three pictures, four possible interpretations of
MTurk users who had a history of providing good- the depicted scenario were presented to participants,
quality responses. Participants were required to who were asked to rate the plausibility of each
have completed at least 500 MTurk studies and to interpretation. Interpretations include 1 Absurd, 2
have had their work approved (vs. rejected) in 98% Lures, and 1 True interpretation for each scenario.
of the studies they completed previously on MTurk. For example, the IIT scenario depicted in Figure 1(a,
In addition, three questions were included to discrimi- b), and 1c is followed by an absurd interpretation
nate attentive from inattentive participants. These (“People discuss the smell of the field”), two lure
questions were presented at random intervals and interpretations (“People stop you from starting a
participants were required to answer all three cor- fight” and “People are making fun of you”), and a
rectly. Consistent with previous research, one partici- true interpretation (“People celebrate what a great
pant was excluded from all analyses because they player you are”). Participants were instructed to
completed the survey in less than 60% of the pro- imagine each scenario as if they were an observer in
jected time (< 27 min). The projected time was deter- the situation and could see it through their own
mined by taking the average time of completion from eyes. It is also worth highlighting that while there is
a previous MTurk study using the same task (Deng a “true” interpretation for every scenario, whether par-
et al., 2022). Research using a similar approach (e.g. ticipants have endorsed the “true” interpretation at
requiring a history of good quality response) has the last stage is less important. Instead, the task was
demonstrated that MTurk data are comparable to designed to capture how the likelihood ratings
those collected in the laboratory. change for each statement across the stages of
disambiguation.
Measures
Social impression formation
Interpretation inflexibility task After completing the IIT, participants were asked to
The Interpretation Inflexibility Task (IIT) is a validated, rate the trustworthiness of selected IIT characters
picture-based task assessing individuals’ belief inflexi- and to rate how much they thought these characters
bility and interpretation bias in social and emotional like them. These ratings were acquired 30 min after
scenarios (Deng et al., 2022). The IIT consists of 24 the IIT. As a manipulation check, participants were
scenarios depicting interpersonal situations, each is also asked if they recognised the characters as
gradually revealed to the respondents in three having appeared before (in the IIT).
stages: with 80% of the photo blurred, 20% of the In this task, participants were shown a total of 24
photo blurred and the original photo with no blurring faces, 18 of which were previously shown IIT charac-
effects. The blurred region was selected with the goal ters whose faces were cropped from their individual
of obscuring the emotional valence of a given scen- IIT scenarios (see Figure 2), 9 from each emotional
ario. By gradually reducing the percentage of the valence condition (i.e. positive vs. negative outcome
photo that is blurred, the respondent receives more scenarios). The chosen characters were never the
information that may help to resolve the initially main character that was blurred in the IIT. Instead,
ambiguous situation. they appeared in a context that was gradually disam-
Two types of scenarios were included in the IIT to biguated and became increasingly emotional. To
examine whether interpretation inflexibility differed control for depression-related baseline levels of trust-
according to the valence of the new evidence. An worthiness and liking, 6 foils were included showing
example positive outcome scenario is shown in novel faces that had not been part of the IIT. The 6
Figure 1 (top). In this example, each of the three foils were stock photos from the same series (i.e.
140 W. DENG ET AL.

Figure 1. Example IIT scenarios, with gradual unblurring to achieve the disambiguation of emotional information.

produced by the same artist on the same website) as negative characteristics rather than positive as in the
the IIT pictures but had not been selected as a task first two). Scores of positive and negative social
scenario during the IIT development. impressions were calculated respectively by aver-
For each of the faces, participants were asked to aging items 1 and 2 (i.e. liking and trustworthiness
imagine being in a social interaction with the for positive), items 3 and 4 (i.e. dislike and distrust
depicted character, and rate on a 7-point Likert scale for negative) together.
(ranging from 1, “very unlikely” to 7, “very likely”)
how likely it is to (1) receive a compliment from the Beck depression inventory-II (BDI-II)
character; (2) trust the character; (3) be disliked by The BDI-II (Beck et al., 1996) is a well-established 21-
the character and (4) be plotted against by the charac- item self-report measure of depressive symptom
ter. The latter two ratings were used to control for the severity over the past two weeks. Participants rate
valence of question framing (i.e. emphasising the degree to which they have experienced each

Figure 2. Example Post-IIT social impression item.


COGNITION AND EMOTION 141

symptom on a four-point scale (from 0 to 3). The BDI-II calculate the interpretation flexibility index for each
has overall good reliability and validity (Richter et al., scenario.
1998). The internal consistency of the BDI-II in this
study was ωtotal = .96. This measure was included
because previous research using the IIT suggests Interpretation flexibility index

that depressive symptom severity is related to inflexi- 
 (bias score stage 3 − bias score stage 2)2 +

bility of interpretations above and beyond interpret-  (bias score stage 2 − bias score stage 1)2
ation bias (Deng et al., 2022). =
2

Procedure
The analytic plan aimed to address two primary
All participants gave informed consent in accord- objectives: (a) replicate previous findings of negative
ance with the Yale University Institutional Review bias and inflexibility in depression, associating
Board. Participants completed a survey which depression severity with difficulty endorsing positive
began with demographic questions followed by outcomes, independent of initial interpretation bias;
the IIT items, which were presented in randomised (b) test the relation between depressive symptoms
order. Participants then completed the BDI-II and and social impression formation (i.e. measured by
other individual differences measures, before the perception of others’ trustworthiness and liking),
giving ratings based on their impression of social especially how such relations are affected by prior
characters. Upon completion of the survey, partici- exposure to these characters in a positive or negative
pants were debriefed and received remuneration social context (compared to the foils). First, to
(6 USD). examine whether previously observed relations
between interpretation inflexibility and depression
replicate in this sample, multiple regression models
Data analysis
were built testing whether interpretation bias and
Based on participants’ performance during the IIT, the inflexibility uniquely explained variation in depressive
following computational approach was employed to symptoms. For each scenario type, regression models
capture (a) initial interpretation bias and (b) within- were tested separately with depression as the depen-
person revision of prior interpretations (Deng et al., dent variable. In each model, interpretation bias and
2022). An interpretation bias score was calculated at inflexibility (negative or positive, depending on the
each of the three stages for the 24 IIT scenarios, by scenario outcomes) were simultaneously entered
taking the additive inverse of the true statement into the regression equation as the independent
plausibility rating divided by the average of the variables.
endorsement ratings for the two lures. In particular, For the second part of the study, impression
the initial interpretation bias was represented by the ratings for characters previously shown in the IIT scen-
bias score at stage 1 (i.e. the first picture of each IIT arios were categorised according to valence – positive
scenario). Using the interpretation bias scores at impression (trustworthiness and liking) vs. negative
each stage, the interpretation flexibility index was cal- impression (distrust and dislike). In addition,
culated by taking the Root Mean Square of Successive emotional valence of the IIT scenarios was taken
Differences (RMSSD) that captures moment-to- into account (faces taken from positive vs. negative
moment fluctuations in interpretation bias as a outcome scenarios of the IIT vs. foils). Mixed effects
proxy for positive or negative interpretation inflexi- regression models were built with social impression
bility (as derived from scenarios with positive or nega- ratings as the dependent variable, depression severity
tive emotional valences). Higher RMSSD values (as measured by the BDI scores) as a between-subject
represent high moment-to-moment variability and factor, and valence of the ratings (positive vs. nega-
suggest flexibility in revising an initial (biased) tive), as well as valence of IIT scenario outcomes (posi-
interpretation. By contrast, low RMSSD values rep- tive vs. negative) as within-subject factors.
resent low moment-to-moment variability and Assumptions of homoscedasticity and normality of
reflect inflexibility in revising an initial interpretation residuals were met for all analyses. Collinearity stat-
bias based on novel information provided by the IIT istics were within acceptable limits (VIF’s < 1.21, Toler-
scenarios. The following formula was applied to ance’s > .83).
142 W. DENG ET AL.

Results for both positive and negative outcome scenarios,


flexibility predicts depressive symptoms even after
Sample characteristics
controlling for variation in biases (Table 2).
Participants’ BDI-II scores (M = 10.05, SD = 11.87) rep-
resented varying symptom severity across the spec-
Depressive symptoms and social impression
trum: 147 respondents reported minimal (range: 0–
ratings
13), 16 reported mild (range: 14–19), 25 reported
moderate (range: 20–28), and 25 reported severe A mixed effects model was built with impression
(range: 29–58) depressive symptoms. The observed ratings as the dependent variable, valence of
BDI-II scores enabled this study to examine the impression (positive/negative) and IIT scenarios (posi-
relations between social impression formation and tive/negative) as the within-subject variables, and
depressive symptoms along broad continua of depressive symptoms as the between-subject inde-
symptom severity. For this study, the BDI-II scores pendent variable. All three main effects were signifi-
are included as a continuous variable in all analyses. cant, with more negative social impressions
associated with depression, negative framing of the
character rating, and negative outcome scenarios.
Interpretation inflexibility in depression
Depressive symptoms also interacted significantly
Replicating prior work, zero-order correlations with each of the within-subject factors. Given that
showed that depression severity was significantly cor- there was a trend for a three-way interaction effect
related with higher negative interpretation bias and among depression severity, valence of impression
lower positive interpretation bias. In addition, ratings, and valence of IIT scenario outcomes (t =
depression severity was associated with interpret- −1.91, SE = .01, p = .057; Table 3), we focused on
ation inflexibility, but only the correlation with endor- decomposing this higher-order effect for purposes
sing positive outcomes was significant, not for of display and interpretation.
negative outcomes (Table 1). Figure 3 plots the relationships between depress-
In regression models with both interpretation bias ive symptoms and character ratings by valence of
and inflexibility predicting symptoms of depression, scenario outcomes, separately for positive and

Table 1. Correlation matrix between depression and belief updating indices.


2 3 4 5 6 7 8 9
1. Depression -.18** −.12 −.24** .23** −.11 −.01 .33** .30**
2. Positive IFI − .54** .02 −.15* .15* −.42** −.10 −.21**
3. Negative IFI - −.12 −.08 −.01 −.30** −.22** −.13
4. PIB - −.37** .19** −.20** .21** −.22**
5. NIB - −.16* .22** −.09 .17*
6. Liking and trust (positive outcome scenario) - −.10 .79** 0.04
7. Liking and trust (negative outcome scenario) - .07 .85**
8. Dislike and distrust (positive outcome scenario) - .01
9. Dislike and distrust (negative outcome scenario) -
Note. * p < .05; ** p < .01. PIB = positive interpretation bias, NIB = negative interpretation bias, IFI = interpretation flexibility index.

Table 2. Regression models examining the relation among depressive symptoms, interpretation bias and inflexibility.
Dependent variables Component b SEb β t p-value R2
Depressive symptoms
Constant 12.98 5.72 2.27 .024 .125
Flexibility (P) −3.86 1.65 −.16 −2.34 .020
PIB −8.77 3.37 −.18 −2.60 .010
NIB 6.87 3.39 .14 2.03 .044
Constant 12.22 5.72 2.14 .034 .118
Flexibility (N) −2.66 1.33 −.13 −2.00 .047
PIB −9.54 3.42 −.20 −2.79 .006
NIB 7.19 3.39 .15 2.12 .035
Note. Flexibility (P) = interpretation flexibility index from positive outcome scenarios, Flexibility (N) = interpretation flexibility index from nega-
tive outcome scenarios, PIB = positive interpretation bias, NIB = negative interpretation bias.
COGNITION AND EMOTION 143

Table 3. Effects of depressive symptom severity, emotion valence of bias as covariates in the regression model). No signifi-
impression ratings (Positive vs. Negative) and of IIT scenario cant interactions were found between impression
outcomes on total ratings of social impression.
ratings or depression severity and participants’
Dependent variable: ratings of social impression
memory of the characters. In other words, our
Predictors Estimates CI p
finding on the interaction between depression sever-
(Intercept) 6.29 5.79–6.78 <0.001
Depressive symptoms −0.05 −0.08 – −0.02 0.001 ity and impression ratings is not dependent on the
Valence of impression −2.18 −2.48 – −1.87 <0.001 differential recognition of the characters.
ratings
Valence of IIT scenario −0.79 −1.02 – −0.57 <0.001
outcomes Discussion
Depressive symptoms * 0.04 0.02–0.06 <0.001
Valence of This study had two aims: (a) replicating previous
impression ratings
Depressive symptoms * 0.01 −0.00–0.03 0.065 findings that show a bias and difficulties revising the
Valence of IIT bias (i.e. inflexibility) among individuals with depress-
scenario ive symptoms; (b) examining the relation between
Valence of impression 0.49 0.35–0.63 <0.001
ratings * Valence of depressive symptoms and social impression for-
IIT scenario mation for characters that previously appeared in
Depressive symptoms * −0.01 −0.02–0.00 0.057 positive or negative contexts (or were foils). We repli-
Valence of
impression ratings * cated prior work in showing that both bias and inflexi-
Valence of IIT bility, in both positive and negative outcome
scenario scenarios, predicted depressive symptoms. In examin-
Random Effects
σ 2
0.63 ing social impression formation, higher severity of
τ00 subject 0.27 depressive symptoms was associated with higher
ICC 0.30 negative social impression ratings for all characters,
N subject 213
Observations 1278 as well as lower positive social impression ratings
Marginal R2 / 0.242 / 0.469 (only for characters appeared in positive contexts).
Conditional R2 While inflexibility in both positive and negative
outcome scenarios predicted depressive symptoms
negative impression ratings. Overall, compared with in regression models that also account for interpret-
those lower in depression, individuals with higher ation biases, the zero-order correlation was only sig-
depressive symptoms gave significantly higher nificant between depressive symptoms and
ratings of negative characteristics (distrust and inflexibility in adapting to positive outcomes. Previous
dislike), regardless of scenario outcome. This pattern work has shown that depression is more consistently
is attenuated and inverted for ratings of positive associated with difficulties revising existing beliefs
characteristics (trustworthiness and liking), on which and adapting to positive (vs. negative) outcomes
more severe depressive symptoms were associated (Deng et al., 2022) – which is in line with our
with lower positive ratings, but only significantly so findings. More work is needed to examine positive
for characters who previously appeared in positive and negative belief revisioning respectively, especially
contexts. Stated another way, whereas the positive given the relevance of emotional context in studying
and negative character ratings of those lower in inflexibility in depression. On a similar note, while
depression were graded according to scenario depression severity was linked with both diminished
valence (e.g. higher positive ratings and lower positive impressions for characters from a positive
negative ratings for characters from scenarios with context and exaggerated negative impressions for
positive outcomes), character ratings for those all characters, the association between depressive
higher in depression showed no such gradations; symptoms and increased negative social impressions
rather, regardless of scenario outcome, ratings of is especially consistent across studies (Flett et al.,
those higher in depression converged toward the 2014; Pearson et al., 2010). Prior work showed a scen-
profile of foils (not previously encountered) who ario-based negative bias in depression, which
were viewed as dislikeable and likable, untrustworthy prompts an individual to interpret social scenarios
and trustworthy, to equivalent degrees. All as negative regardless of their outcome. Adding on
findings remained significant adjusting for initial to the existing literature, this study revealed that
interpretation bias (i.e. including initial interpretation negative social impressions persist despite prior,
144 W. DENG ET AL.

Figure 3. Interaction effects between depression severity and IIT scenario emotional valence (positive, negative and foils) on social impression
ratings. Regression lines are plotted at predicted values from mixed effect regressions of character ratings by depression, impression valence,
and scenario context, with standard errors of predicted values, adjusting for initial interpretation bias.

positive contexts the depicted characters were seen social situations (Van De Walle & Six, 2014). While
in. The finding on social impression formation furthers the ability to form positive social impressions may
our understanding of negative bias in depression – be at least partially intact in individuals with depress-
such bias not only exists in scenario-specific interpret- ive symptoms (Segarra et al., 2016), they are also sus-
ations, but also generalises to form poor social ceptible to forming negative impressions about these
impressions of the characters, an interpretation sup- characters. The competing beliefs about other charac-
ported by the equivalency of negative ratings for ters and conflicting expectations of interpersonal
foils (characters not previously encountered) and encounters may accentuate the uncertainty of social
characters from both positive and negative outcome outcomes among individuals with depressive symp-
scenarios for those higher in depression. toms. Coupled with the intolerance of uncertainty in
Furthermore, in responding to the novel characters depression (i.e. responding to uncertain situations
(i.e. foils) and characters from negative contexts, there with worry and rumination), mixed impressions of
was only a significant association between depressive others can discourage individuals from future social
symptoms and increased negative ratings (e.g. distrust interactions, perpetuate repetitive negative thoughts
and dislike), not decreased positive ratings (e.g. trust and hamper social functioning (Carleton et al., 2012).
and liking). This finding supports existing research The study is not without limitations. While partici-
that posits trust and distrust as two distinct concepts pants were prompted to imagine interacting with
(Van De Walle & Six, 2014). In other words, positive characters depicted in the task, certain scenarios
and negative social impressions are not just two may be less relatable to some participants, creating
ends of the same continuum, but two distinct con- potential difficulty to imitate real-life social inter-
cepts that may coexist and create uncertainty to actions with the characters. While the characters
COGNITION AND EMOTION 145

selected display a wide range of emotions to taper the allocation in depression (Treadway, 2016), the inflexi-
impacts of bias against specific emotions (e.g. bias bility in social interpretations and impression for-
against a face displaying sadness) on impression mation may also be related to a higher subjective
ratings, the valence of the stimuli (i.e. the emotions cost of the cognitive effort involved in belief
the characters are showing) may still confound the revisioning.
findings. Future work is needed to establish a baseline In sum, our findings highlight that individuals with
impression rating for each stimulus in the general depressive symptoms experience difficulty not only in
population. In addition, the design of the current endorsing scenario-specific positive interpretations,
study did not allow for examining impression updat- but also in integrating positive contextual information
ing over time. Longitudinal studies can tap into the and revising negative social impressions. Our findings
persistence or stability of these social impressions underline the impacts of negative belief inflexibility
with follow-up measurements after a greater time on social impression formation, noting the need for
lapse. In the same vein, future in-person studies can challenging negative social impressions as a potential
explore the valence of social impressions when indi- target for intervention.
viduals are in an emotion-laden context, such as a
threatened state. Moreover, while the present study
Disclosure statement
measured impression ratings for non-IIT characters
(i.e. foil ratings), the limited number of foils may No potential conflict of interest was reported by the author(s).
hinder the interpretability of foil ratings as the base-
line level of social impressions. Future research may Funding
collect impression ratings of more foil characters
before participants complete the IIT, so as to establish The author(s) reported there is no funding associated with the
work featured in this article.
a baseline level of positive vs. negative impressions of
novel characters.
In addition to cognitive bias and inflexibility in ORCID
belief updating, individual differences in personality Wisteria Deng http://orcid.org/0000-0001-6563-7772
may play a crucial role in facilitating the transition Tyrone D. Cannon http://orcid.org/0000-0002-5632-3154
from event-specific interpretations to generalised
social impressions. Higher levels of self-esteem, for
example, may influence one’s expectations of other References
characters’ action in social settings, prompting expec- Beck, A. T., Steer, R. A., Ball, R., & Ranieri, W. F. (1996). Comparison
tations of compliments and less anticipation of dislike of Beck Depression Inventories-IA and-II in Psychiatric
(Orth et al., 2014). Further exploration of individual Outpatients. Journal of Personality Assessment, 67(3), 588–
597. http://dx.doi.org/10.1207/s15327752jpa6703_13
differences as moderators between depression and Carleton, R. N., Mulvogue, M. K., Thibodeau, M. A., McCabe, R. E.,
negative social impressions can provide novel Antony, M. M., & Asmundson, G. J. (2012). Increasingly certain
targets for behavioural interventions (e.g. building about uncertainty: Intolerance of uncertainty across anxiety
self-efficacy as a means to encourage social and depression. Journal of Anxiety Disorders, 26(3), 468–479.
engagement). https://doi.org/10.1016/j.janxdis.2012.01.011
Deng, W., Everaert, J., Creighton, M., Bronstein, M. V., Cannon, T.,
While numerous interventions for depression aim & Joormann, J. (2022). Developing a novel assessment of
at bolstering positive anticipations and modifying interpretation flexibility: Reliability, validity and clinical impli-
negative expectations about the self and the future cations. Personality and Individual Differences, 190, 111548.
(Horwitz et al., 2017), a focus on reevaluating negative https://doi.org/10.1016/j.paid.2022.111548
social impressions of others could be beneficial. Denrell, J. (2005). Why most people disapprove of me:
Experience sampling in impression formation. Psychological
Specifically, early intervention for individuals with Review, 112(4), 951–978. https://doi.org/10.1037/0033-295X.
subclinical depressive symptoms could involve chal- 112.4.951
lenging their anticipation of social rejections, building Derntl, B., Seidel, E. M., Eickhoff, S. B., Kellermann, T., Gur, R. C.,
a sense of trust in others as well as training social skills Schneider, F., & Habel, U. (2011). Neural correlates of social
to perceive others’ intention with less negative bias approach and withdrawal in patients with major depression.
Social Neuroscience, 6(5-6), 482–501. https://doi.org/10.1080/
(e.g. emotion labelling, mindful observations) 17470919.2011.579800
(Elsayed et al., 2021; Seshadri et al., 2021). Considering Elsayed, N. M., Vogel, A. C., Luby, J. L., & Barch, D. M. (2021).
the motivational deficits and aberrant cognitive effort Labeling emotional stimuli in early childhood predicts
146 W. DENG ET AL.

neural and behavioral indicators of emotion regulation in late longitudinal study of Mexican-origin youth. Developmental
adolescence. Biological Psychiatry: Cognitive Neuroscience and Psychology, 50(2), 622–633. https://doi.org/10.1037/
Neuroimaging, 6(1), 89–98. https://doi.org/10.1016/j.bpsc. a0033817
2020.08.018 Pearson, K. A., Watkins, E. R., & Mullan, E. G. (2010). Submissive
Everaert, J., Bronstein, M. V., Cannon, T. D., & Joormann, J. (2018). interpersonal style mediates the effect of brooding on
Looking through tinted glasses: Depression and social future depressive symptoms. Behaviour Research and
anxiety are related to both interpretation biases and Therapy, 48(10), 966–973. https://doi.org/10.1016/j.brat.
inflexible negative interpretations. Clinical Psychological 2010.05.029
Science, 6(4), 517–528. https://doi.org/10.1177/ Radke, S., Guths, F., Andre, J. A., Muller, B. W., & de Bruijn, E. R.
2167702617747968 (2014). In action or inaction? Social approach-avoidance ten-
Feeser, M., Schlagenhauf, F., Sterzer, P., Park, S., Stoy, M., dencies in major depression. Psychiatry Research, 219(3), 513–
Gutwinski, S., Dalanay, U., Kienast, T., Bauer, M., Heinz, A., 517. https://doi.org/10.1016/j.psychres.2014.07.011
Strohle, A., & Bermpohl, F. (2013). Context insensitivity Richter, P., Werner, J., Heerlein, A., Kraus, A., & Sauer, H. (1998).
during positive and negative emotional expectancy in On the Validity of the Beck Depression Inventory.
depression assessed with functional magnetic resonance Psychopathology, 31(3), 160–168. http://dx.doi.org/10.1159/
imaging. Psychiatry Research: Neuroimaging, 212(1), 28–35. 000066239
https://doi.org/10.1016/j.pscychresns.2012.11.010 Rogers, K. B. (2018). Do you see what I see? Testing for individual
Flett, G. L., Besser, A., & Hewitt, P. L. (2014). Perfectionism and differences in impressions of events. Social Psychology
interpersonal orientations in depression: An analysis of vali- Quarterly, 81(2), 149–172. https://doi.org/10.1177/
dation seeking and rejection sensitivity in a community 0190272518767615
sample of young adults. Psychiatry: Interpersonal and Segarra, N., Metastasio, A., Ziauddeen, H., Spencer, J., Reinders,
Biological Processes, 77(1), 67–85. https://doi.org/10.1521/ N. R., Dudas, R. B., Arrondo, G., Robbins, T. W., Clark, L.,
psyc.2014.77.1.67 Fletcher, P. C., & Murray, G. K. (2016). Abnormal frontostriatal
Haim-Nachum, S., & Levy-Gigi, E. (2021). To be or not to be activity during unexpected reward receipt in depression
flexible: Selective impairments as a means to differentiate and schizophrenia: Relationship to anhedonia.
between depression and PTSD symptoms. Journal of Neuropsychopharmacology, 41(8), 2001–2010. https://doi.
Psychiatric Research, 136, 366–373. https://doi.org/10.1016/j. org/10.1038/npp.2015.370
jpsychires.2021.02.015 Seshadri, A., Orth, S. S., Adaji, A., Singh, B., Clark, M. M., Frye, M.
Hames, J. L., Hagan, C. R., & Joiner, T. E. (2013). Interpersonal pro- A., McGillivray, J., & Fuller-Tyszkiewicz, M. (2021).
cesses in depression. Annual Review of Clinical Psychology, 9 Mindfulness-Based cognitive therapy, acceptance and com-
(1), 355–377. https://doi.org/10.1146/annurev-clinpsy- mitment therapy, and positive psychotherapy for major
050212-185553 depression. American Journal of Psychotherapy, 74(1), 4–12.
Horwitz, A. G., Berona, J., Czyz, E. K., Yeguez, C. E., & King, C. A. https://doi.org/10.1176/appi.psychotherapy.20200006
(2017). Positive and negative expectations of hopelessness Silk, J. S., Davis, S., McMakin, D. L., Dahl, R. E., & Forbes, E. E.
as longitudinal predictors of depression, suicidal ideation, (2012). Why do anxious children become depressed teen-
and suicidal behavior in high-risk adolescents. Suicide and agers? The role of social evaluative threat and reward proces-
Life-Threatening Behavior, 47(2), 168–176. https://doi.org/10. sing. Psychological Medicine, 42(10), 2095–2107. https://doi.
1111/sltb.12273 org/10.1017/S0033291712000207
Hu, H. X., Jiang, S. Y., Shan, H. D., Chu, M. Y., Lv, Q. Y., Yi, Z. H., Lui, Soygüt, G., & Savaşir, I. (2001). The relationship between inter-
S. S. Y., Cheung, E. F. C., & Chan, R. C. K. (2022). Negative personal schemas and depressive symptomatology. Journal
belief-updating bias for positive daily life events in individ- of Counseling Psychology, 48(3), 359–364. https://doi.org/10.
uals with schizophrenia and social anhedonia. Cognitive 1037/0022-0167.48.3.359
Neuropsychiatry, 27(4), 237–254. https://doi.org/10.1080/ Stewart, J. G., & Harkness, K. L. (2017). Testing a revised interper-
13546805.2021.2014309 sonal theory of depression using a laboratory measure of
Joormann, J., & Gotlib, I. H. (2010). Emotion regulation in excessive reassurance seeking. Journal of Clinical
depression: Relation to cognitive inhibition. Cognition & Psychology, 73(3), 331–348. https://doi.org/10.1002/jclp.
Emotion, 24(2), 281–298. https://doi.org/10.1080/ 22338
02699930903407948 Treadway, M. T. (2016). Current topics in behavioral neuro-
Li, C., Liu, D., & Dong, Y. (2019). Self-esteem and problematic sciences. Current Topics in Behavioral Neurosciences, 27,
smartphone use among adolescents: A moderated mediation 337–355. https://doi.org/10.1007/7854_2015_400
model of depression and interpersonal trust. Frontiers in Van De Walle, S., & Six, F. (2014). Trust and distrust as distinct
Psychology, 10, 2872. https://doi.org/10.3389/fpsyg.2019. concepts: Why studying distrust in institutions is important.
02872 Journal of Comparative Policy Analysis: Research and
Orth, U., Robins, R. W., Widaman, K. F., & Conger, R. D. (2014). Is Practice, 16(2), 158–174. https://doi.org/10.1080/13876988.
low self-esteem a risk factor for depression? Findings from a 2013.785146

You might also like