Content uploaded by Michele Vicovaro
Author content
All content in this area was uploaded by Michele Vicovaro on Jun 09, 2022
Content may be subject to copyright.
1
© 2022, American Psychological Association. This paper is not the copy of record and
may not exactly replicate the final, authoritative version of the article. Please do not
copy or cite without authors' permission. The final article will be available, upon
publication, via its DOI: 10.1037/xhp0001036
2
Towards The Boundaries of Self-Prioritization: Associating The Self With Asymmetric Shapes
Disrupts The Self-Prioritization Effect
Michele Vicovaro1, Mario Dalmaso2, & Marco Bertamini1,3
1 Department of General Psychology, University of Padova, Italy
2 Department of Social and Developmental Psychology, University of Padova, Italy
3 Department of Psychology, University of Liverpool, UK
Correspondence concerning this article should be addressed to:
Michele Vicovaro
Department of General Psychology, University of Padova, via Venezia 8, 35131 Padova, Italy
Email: michele.vicovaro@unipd.it
Phone: +39 049 8276602
3
Abstract
Individuals tend to prioritize self-relevant information over other-relevant information. Converging
empirical evidence indicates that stimuli that are arbitrarily associated with the self are processed
more efficiently than stimuli that are arbitrarily associated with stranger identities. In the present
study, we tested if a salient perceptual feature (i.e., presence or absence of symmetry) can modulate
this self-prioritization effect. In particular, we wanted to know how the valence of symmetry would
integrate or interfere with the self. Under one condition, participants were asked to associate the self
with symmetric shapes and a stranger with asymmetric shapes, whereas, under another condition, the
association was inverted (i.e., self-asymmetry/stranger-symmetry). The two conditions were
manipulated within participants (Experiment 1, laboratory-based) or between participants
(Experiment 2, online). Participants classified a randomly generated shape (symmetric vs.
asymmetric) and a label (you vs. stranger) as either matching or nonmatching with the previously
learned association. In both experiments, a clear self-prioritization effect emerged in the self-
symmetry/stranger-asymmetry condition whereas, strikingly, no evidence of a self-prioritization effect
emerged at all in the opposite condition. The results suggest that the self-prioritization effect is not
mandatory and can be modulated by the valence of the stimuli with which self and stranger are
associated.
Keywords: Self-prioritization; Symmetry; Valence; Perceptual matching; Self-identification
4
Towards The Boundaries of Self-Prioritization: Associating The Self With Asymmetric Shapes
Disrupts The Self-Prioritization Effect
Due to the limits of the cognitive system, only a small part of the incoming information can be
processed efficiently. To successfully adapt to the environment, humans must prioritize important
information, and empirical studies suggest that self-relevant information tends to be prioritized over
other-relevant information, a phenomenon known as the self-prioritization effect (for reviews, see
Cunningham & Turk, 2017; Sui & Humphreys, 2015a).
Sui et al. (2012) provided an elegant empirical demonstration of the self-prioritization effect.
In their main experiment, there was a learning phase followed by a matching task. In the learning
phase, participants were instructed to associate themselves, a friend, and a stranger with three
arbitrary geometric shapes (i.e., they were provided with the following instruction: “in this experiment
you are a circle, a friend is a triangle, and a stranger is a square”). Then, on each trial of the matching
task, one of the three shapes (i.e., circle, tringle, or square) and one of the three labels (i.e., you,
friend, or stranger) were simultaneously presented on the screen for 100 ms. Participants had to
indicate if the shape-label pair was correct (consistent with one of the learned associations; e.g.,
square + stranger) or incorrect (inconsistent with the previously learned associations; e.g., square +
you). A remarkable pattern of results emerged. Responses were faster and more accurate on trials in
which the label you was paired with the self-related shape (e.g., circle + you), than on trials in which
any of the other shape-label associations were presented (e.g., square + you, square + stranger,
triangle + friend). Control experiments showed that this pattern of results did not depend on greater
familiarity, concreteness, or grammatical salience of the you label with respect to the other labels
(Schäfer et al., 2017; Sui et al., 2012; Woźniak & Koblich, 2019; but see Wade & Vickery, 2017).
The authors suggested that self-related information is processed more efficiently—at a perceptual
level—than the friend- and stranger-related information (see also Liu & Sui, 2016; Sui et al., 2015;
Sui & Humphreys, 2015b; but see Stein et al., 2016). Nevertheless, the nature of the self-prioritization
effect is still debated. For instance, in addition to perception, attention and memory contribute to the
phenomenon, because self-related stimuli tend to attract more attentional resources (Dalmaso et al.,
2019; Humphreys & Sui, 2016; Macrae et al., 2018; Sui & Rotshtein, 2019; Zhao et al., 2015; but see
Siebold et al., 2016) and form more stable memory traces (e.g., Reuther & Chakravarthi, 2017) than
stimuli related to both friends and strangers.
The self-prioritization effect is robust and generalizable. For instance, the effect is present in
different cultures (Jiang et al., 2019), and it also emerges when only the identities of self and stranger
are used in the learning and matching tasks (i.e., the identity of friend is not strictly necessary; e.g.,
Stein et al., 2016). Moreover, the effect occurs even when, instead of being associated with simple
geometric shapes, the self and the stranger are arbitrarily associated with stimuli such as Gabor
patches varying in orientation (Stein et al., 2016), motion directions (Frings & Wentura, 2014),
musical instruments (Schäfer et al., 2015), sounds (Schäfer et al., 2016), vibrotactile stimulations
(Schäfer et al., 2016), or even unfamiliar neutral faces (Payne et al., 2017; Woźniak & Knoblich,
2019). The effect also occurs when self and stranger are associated with conceptual categories rather
than with specific objects (e.g., the broad categories of triangles and circles, which include triangles
and circles varying in size or color; Schäfer et al., 2015; Sui et al., 2014).
A prioritization effect has also been documented for items that belong to the self. For
instance, when different categories of items (e.g., pens and pencils) are arbitrarily assigned to either
the self or another individual, the self-owned items are responded to faster than the items owned by
the other (Golubickis et al., 2018, 2019; see also Constable et al., 2019; Cunningham, 2008).
Interestingly, this self-ownership advantage disappears when the self-owned items are presented
outside a symbolic space associated arbitrarily with the self (McPhee et al., 2021; Strachan et al.,
2020), and it reverses when participants are informed that the friend-owned items are more likely to
appear than the self-owned items (Falbén et al., 2020). This suggests that “…the processing advantage
for owned objects is something that can be modulated by the context in which it is embedded.”
(Strachan et al., 2020, p. 795).
5
Valence and the self-prioritization effect
The mechanisms underlying the self-prioritization effect have been the target of extensive empirical
work and theoretical debate. Sui et al. (2012) found that stimuli associated with relatively high
monetary values were prioritized with respect to stimuli associated with low monetary values. The
analogy between self- and reward-related prioritization appears to suggest that the relationship with
the self may act as a form of reward (i.e., self-related stimuli would be more rewarding than stranger-
related stimuli; see also Humphreys & Sui, 2015). However, later studies also highlighted some
structural differences between self-related and reward-related prioritization. For instance, Sui and
Humphreys (2015c) found no correlation between the magnitudes of the two effects, and Sui and
Humphreys (2015b) found that the relationship with the self could favor the integration of the stimuli
both at a perceptual and at a conceptual level, whereas the relationship with a high reward could favor
the integration of the stimuli only at a conceptual level (i.e., not at a perceptual level; see also Sui et
al., 2015). This appears to suggest that the self-prioritization effect is at least partially independent
from reward-related prioritization effects.
A possible key for the interpretation of the self-prioritization effect is the general advantage
for the processing of positive valence stimuli over neutral or negative valence stimuli (Sui &
Humphreys, 2015b; Sui et al., 2016). An otherwise neutral stimulus may acquire a positive or a
negative valence because of its association with the self or with a stranger, respectively. Consistently
with this hypothesis, Sui et al. (2016) hypothesized that negative mood can reduce the positive
emotional response elicited by self-related stimuli, and found indeed a stronger self-prioritization
effect when participants were in a neutral mood compared to when they were in a negative mood (but
see Qian et al., 2020). Hu et al. (2020) had participants associate neutral shapes with the good part of
the self, the good part of a stranger, the bad part of the self, and the bad part of a stranger. The results
showed that the prioritization effects were driven not only by self-identification but also by valence.
Indeed, the shapes associated with a good feature of the self (e.g., the morally good aspect of the
responder) were prioritized over the shapes associated with a bad feature of the self, and the shapes
associated with a good feature of the stranger were prioritized over the shapes associated with a bad
feature of the stranger. Therefore, prioritization effects appear to be driven by positive valence above
and beyond self-identification.
A promising strategy for exploring of the possible relationship between valence and self-
prioritization is testing whether the self-prioritization effect is modulated by the valence of the stimuli
with which the self and the stranger are associated (Constable et al., 2021; Golubickis et al., 2021;
McIvor et al., 2021). For instance, suppose that the self-prioritization effect is enhanced when the self-
related information has positive valence and the other-related information has negative valence,
compared to when the self-related information has negative valence and the other-related information
has positive valence. Converging evidence indicates that healthy adults tend to have a positive bias for
the self and a negative bias for the stranger (Taylor & Brown, 1988). A modulation effect of valence
on self-prioritization would be functional to keep a positive bias for the self, as it would mean that
self-related information with a positive valence is more strongly prioritized than self-related
information with a negative valence. Moreover, if self and stimulus valence produce faster responses
because of a shared underlying mechanism (i.e., faster responses for positive valence), then when they
are combined, the effects should be additive. Alternatively, the self-prioritization effect might be
impervious to the valence of the self- and other-related information, which would mean that the self-
prioritization effect is inflexible and mandatory and that it is unrelated to the cognitive processes
underlying the positive bias for the self.
In Constable et al.’s (2021) first experiment, half of the participants were asked to associate
themselves with a happy face (positive valence) and a stranger with a sad face (negative valence),
whereas the other half of the participants were asked to perform the opposite association. A stronger
self-prioritization effect emerged for the self-happy/stranger-sad association than for the self-
sad/stranger-happy association, which suggests that associating the self with a negative valence
stimulus can reduce the magnitude of the self-prioritization effect. In apparent contrast with these
results, McIvor et al. (2021) found that the self-prioritization effect was unaffected by the valence of
emotional faces (i.e., happy, neutral, or sad) appearing inside self-related geometric shapes. However,
this null effect can be due to the fact that participants had to respond to the geometric shapes rather
than to the emotional faces, and therefore the valence of the emotional faces was irrelevant to the task.
6
Support for the hypothesis that valence can modulate the self-prioritization effect also emerged from
the results of Golubickis et al.’s (2021) ownership categorization task. Participants were presented
with posters showing either pleasant (positive valence) or unpleasant (negative valence) scenarios.
Half of the participants were informed that they owned two pleasant posters and a closely related
friend owned two unpleasant posters, whereas the other half of the participants were presented with
the opposite association. Then, in an ownership categorization task, participants had to classify the
posters as either owned-by-self or owned-by-friend. A robust self-ownership prioritization effect
emerged when the posters owned-by-self had positive valence and the posters owned-by-friend had
negative valence, whereas no self-ownership prioritization effect emerged in response to the opposite
association. The results of this study appear to indicate that associating the self with negative valence
stimuli can disrupt the self-prioritization effect.
It is worth highlighting that, both in Constable et al.’s (2021) first experiment and in the study
by Golubickis et al. (2021), the associations between stimuli and identities were probably not as
arbitrary as those in the original study by Sui et al. (2012), in which the identities were associated
with abstract geometric shapes. Indeed, healthy individuals tend to seek positive emotions and
pleasant scenarios and avoid negative emotions and unpleasant scenarios, which means that they are
probably more familiar with happy faces and pleasant scenarios than with sad faces and unpleasant
scenarios (see also Constable et al., 2021). In other words, the associations between the self and
pictures of happy faces/pleasant scenarios may activate strong and privileged associations stored in
the long-term memory which, in turn, could explain the results obtained by Constable et al. (2021,
Experiment 1) and by Golubickis et al. (2021).
Does the valence of the stimuli modulate the self-prioritization effect even when the stimuli
are unrelated to previously learned associations with the self? A positive answer to this question
would support the hypothesis of a deep link between valence and self-prioritization that may affect
arbitrary newly learned associations. An approach that should minimize the possible effects of
previously learned associations with the self would be that of associating self and stranger identities
with stimuli that are not obviously related to these identities in everyday life experience. In this
regard, Sui and Humphreys (2015d) presented participants with shapes related to themselves, friends,
and strangers that varied in size. A stronger self-prioritization effect emerged when self-related shapes
were presented as relatively large, compared to when they were medium or small. According to the
authors, this may have emerged because of well-known motivational biases favoring large shapes,
which reflects a positive relationship between size and valence (see also Schubert et al., 2009). In
Constable et al.’s (2021, Experiment 2), the lightness of the stimuli was manipulated: Half of the
participants associated themselves with a lighter geometric shape and a stranger with a darker
geometric shape, whereas the other half of the participants performed the opposite association. The
idea was that lighter and darker shapes would have positive and negative valence, respectively. The
results were somewhat mixed, as an effect of association type emerged when the perceptual difference
in lightness between the lighter and darker shape was small, whereas no effect of association type
emerged when the difference was perceptually large. In sum, the potential role of the valence of
abstract stimuli in shaping the self-prioritization effect is unclear and still largely unknown.
Self, symmetry, and valence: An overview of the present study
In the present study, we varied the visual properties of the stimuli associated with self and stranger.
Instead of size (Sui & Humphreys, 2015d) or lightness (Constable et al., 2021, Experiment 2), we
manipulated a visual property that is more consistently related to valence, that is, symmetry. Indeed,
several studies in experimental aesthetics suggest that visual symmetry can shape the perceived
valence of otherwise neutral stimuli. For instance, Makin et al. (2012) found that symmetric figures
tend to be implicitly associated with positive attributes, whereas asymmetric figures tend to be
implicitly associated with negative attributes (see also Bertamini et al., 2013; Pecchinenda et al.,
2014). Moreover, symmetric stimuli are generally preferred over asymmetric stimuli (e.g., Cardenas
& Harris, 2006; Eisenman, 1967), and symmetry was listed as a key principle of aesthetics by
Ramachandran and Hirstein (1999, "Symmetry, of course, is also aesthetically pleasing" p. 27). All
this suggests that symmetry and asymmetry are polarized concepts associated with positive and
negative valence, respectively, and this recalls the polarization that can also be observed for the
concepts of self (positive valence) and stranger (negative valence; e.g., Greenwald & Farnham, 2000;
7
Taylor & Brown, 1988). To prevent the possible influence of previously learned associations with the
self, the stimuli in our experiments were abstract symmetric and asymmetric stimulus configurations
composed of random dots. Moreover, a different configuration of symmetric and asymmetric random
dots was presented on each trial of the perceptual matching task.
We designed two experiments to explore the influence of symmetry on the self-prioritization
effect. In both experiments, there were two conditions: either the self was associated with symmetric
shapes and the stranger with asymmetric shapes (i.e., the self-symmetry association) or vice versa
(i.e., the self-asymmetry association). We tested two specific pre-registered hypotheses (see the Open
Practices statement for further details), which can be summarized as follows.
1) We expected that, in matched trials, the difference between the response times for self- and
stranger-related shapes (i.e., the self-prioritization effect as operationalized by Sui et al., 2012) should
be larger in the self-symmetry condition (i.e., when the self is associated with positive valence stimuli
and the stranger with negative valence stimuli) than in the self-asymmetry condition. In other words,
consistently with the hypothesis that the self-prioritization effect is modulated by the valence of the
stimuli with which self and stranger are associated, we expected a stronger self-prioritization effect in
the self-symmetry condition than in the self-asymmetry condition.
2) According to the polarity correspondence principle (Proctor & Cho, 2006), stimuli
characterized by similar valence would associate with each other more easily than stimuli
characterized by different valence. Therefore, in matched trials, responding correct to stimuli of the
self-symmetry association (i.e., label you and a symmetric shape, and stranger and an asymmetric
shape), should be easier than responding correct to stimuli of the opposite association. In
nonmatching trials, responding incorrect to stimuli of the self-symmetry association (i.e., label you
and an asymmetric shape, or stranger and symmetric shape), should be easier than responding
incorrect to stimuli of the opposite association. In sum, in matching as well as in nonmatching trials,
responses should be faster and more accurate for the self-symmetry association than for the self-
asymmetry association (i.e., a main effect of association type).
It is worth noting that the self-prioritization effect cannot be understood as a polarity effect,
because in the self-prioritization effect there is a speeding up of responses to self-related items, not a
general speeding up of responses to congruent pairs. Therefore, our two hypotheses are independent
of each other. In other words, we hypothesize that the manipulation of symmetry/asymmetry can
modulate the self-prioritization effect (Hypothesis 1) and/or produce a polarity effect (Hypothesis 2).
Experiment 1
Methods
Sample size
The determination of the sample size was based on the following considerations. To be considered of
theoretical interest, the difference between the response times (RTs) for the self-symmetry association
and the RTs for the self-asymmetry association should lead, at least, to a medium effect size (d = -
0.5). The same should hold true for response accuracies. A power analysis showed that, for a paired-
sample one-tailed t-test with α = .05, β = .80, and d = -0.5, sufficient power would be reached with N
= 26.14. For the sake of parsimony, we decided to test 30 participants.
Participants
Thirty participants (Mean age = 24.67 years, SD = 7.05 years, 14 males) voluntarily participated in
exchange for course credits. All of them were naive to the purpose of the experiment and reported
normal or corrected-to-normal vision.
Stimuli and Apparatus
The participants sat in a dimly lit room at a distance of about 57 cm from a 15.5’’ computer screen.
The screen background was grey. The experiment was created and run through PsychoPy3 (Pierce et
al., 2019). On each trial, the stimuli were randomly generated by the program. Thus, no configuration
was shown more than once, avoiding any effect of familiarity (Bertamini et al., 2013; Makin et al.,
2012; Pecchinenda et al., 2014). Symmetric and asymmetric shapes were patterns comprising 64
8
white dots (diameter = 0.4 cm, about 0.4 degrees of visual angle), randomly distributed in a 10-cm
circular area (~10°), with a minimum distance of 0.25 cm (~0.25°) between dots. For the symmetric
shapes, the random dots distribution was constrained to be symmetric with respect to the horizontal
and vertical axes. This constraint was absent in the case of asymmetric shapes. Examples of
symmetric and asymmetric shapes are depicted in Figure 1.
Procedure
Before starting the experiment, participants read and signed the written informed consent form
approved by the Ethics Committee for Psychological Research at the University of Padova (protocol
number 3455, February 10th, 2020).
The experiment was divided into two blocks, and the order counterbalanced across
participants. The two blocks corresponded to the two types of association (i.e., self-symmetry and
self-asymmetry); in each block, there was a learning phase followed by a matching task (see Figure
1). Wang et al. (2016) found that the self-prioritization effect emerges even when shape-identity
associations are manipulated within participants. Therefore, participants should be able to switch from
associating the self with symmetric shapes and the stranger to asymmetric shapes in one block to
doing the opposite in the other block.
In the learning phase, participants were informed that they would receive instructions that
they had to read and memorize. Then, a screen was presented for 40 s, showing the relevant shape-
identity associations (i.e., for the self-symmetry association: “In this experiment, you are the
symmetric shapes, and a stranger is the asymmetric shapes”; for the self-asymmetry association: “In
this experiment, you are the asymmetric shapes, and a stranger is the symmetric shapes”; see Figure 1,
identity-shape association frame). Three randomly generated small-size symmetric shapes and three
randomly generated small-size asymmetric shapes (diameter = 5 cm) were presented below the
corresponding sentence. Then, the matching task started. Each trial started with a central black
fixation dot which was presented for 500 ms (Figure 1, fixation frame). This was followed by the
synchronous presentation, at the center of the screen, of a symmetric or asymmetric shape and one of
two labels (black Arial font, height 0.5 cm), which could be ‘YOU’ or ‘STRANGER’ (in Italian: ‘TU’
or ‘ALTRO’, respectively; these labels had been used in previous self-prioritization studies involving
Italian participants; e.g., Dalmaso et al., 2019; Stein et al., 2016). The shape-label pair disappeared
after 100 ms (Figure 1, match frame). After that, a blank screen appeared (Figure 1, blank screen
frame), and participants had to press the ‘A’ or the ‘L’ key to indicate whether the shape-label
association was correct or incorrect (timeout: 1500 ms). The key-response category association was
counterbalanced across participants. Visual feedback (black Arial font, height 0.5 cm) was then
presented at the center of the screen for 500 ms, which could be the word ‘OK’ if a correct response
was provided, ‘NO’ if an incorrect response was provided, and ‘TOO SLOW’ (in Italian: ‘TROPPO
LENTO’) if participants did not respond before timeout (Figure 1, feedback frame).
Consistently with the paradigm of Sui et al. (2012), the trials in the matching task can be
divided based on two orthogonal factors, that is, shape-label matching (matched vs. nonmatching) and
type of shape (self-related vs. stranger-related). In the self-symmetric block, the self-related shapes
were symmetric, and the stranger-related shapes were asymmetric, whereas the opposite was true in
the self-asymmetric block. Each experimental block had 240 trials, according to the following design:
2 matching [matched vs. nonmatching] × 2 shape [self-related vs. stranger-related] × 60 repetitions.
Each experimental block was preceded by 24 practice trials.
9
Figure 1. Procedures and stimuli employed in experiments 1 and 2. Panel A shows the learning phase
in which participants were asked to create an association between identity (i.e., self vs. stranger) and
shape (i.e., symmetric vs. asymmetric). Panel B shows a match trial in which the label “you” is
presented with a symmetric stimulus, and a correct response is provided (the feedback states: “ok”).
Panel C shows a match trial in which the label “stranger” is presented with an asymmetric stimulus,
and a wrong response is provided (the feedback states: “no”). Please note that stimuli are not drawn to
scale.
Results
Missed responses (1.69% of trials) were excluded and not analyzed due to their low frequency. Errors
(i.e., wrong responses; 19.17%) and the RTs of correct responses (79.14%) were analyzed separately.
Correct responses with RTs faster than 200 ms (i.e., anticipated responses; 2.68% of correct
responses) were also eliminated (see also Sui et al., 2012).
As shown in Figure 2, the patterns of results for RTs of correct responses (panels A and B)
and for errors (panels C and D) are similar. For brevity, only the main results that emerged in RTs and
errors analyses are reported, whereas the full results are reported in Appendix A.
RTs of correct responses
The RTs of correct responses were analyzed through a three-way within-participant ANOVA with
factors association type (self-symmetric vs. self-asymmetric), matching (matched vs. nonmatching),
and shape (self-related vs. stranger-related)
1
. Here and in the following analyses, bi-directional paired-
sample Bayesian t-tests were also performed to compare the likelihood of the null hypothesis of a zero
difference between the RTs for the self- and the stranger-related shapes and the likelihood of the
alternative hypothesis of a positive or a negative difference. Separate JSZ tests were performed on
matched and nonmatching trials using the BayesFactor package (Morey & Rouder, 2018), within the
R environment (R Core Team, 2021). The effect size was assumed to be 0 under the null hypothesis
and a Cauchy distribution centred on zero with a scale parameter of √2/2 under the alternative.
Bayesian t-tests were performed and reported only for comparisons that were particularly important
for the main experimental hypotheses.
Our first hypothesis was that the self-prioritization effect would be stronger for the self-
symmetry association than for the self-asymmetry association. The hallmark of the self-prioritization
effect is the significant matching × shape interaction, due to faster RTs for the self- than for the
stranger-related shape in matching trials only (i.e., not in nonmatching trials; see Sui et al., 2012). The
results showed that the association type × matching × shape interaction was significant [F(1,29) =
1
Shapiro-Wilk normality tests showed that the distribution of the RTs was not significantly different
from normal in five out of the eight cells of the experimental design. Only for two cells of the design the
skewness coefficient was smaller than -1. Overall, the data distributions appear to be sufficiently close to normal
to allow the use of ANOVA.
10
46.17, p < .001, η2G = .035], suggesting that, consistently with our hypothesis, the matching × shape
interaction was modulated by association type.
Figure 2. Results of Experiment 1. Boxplots representing the RTs of correct responses longer than
200 ms (panels A and B) and the percentage of errors (panels C and D) for the self-symmetry
association (left column) and for the self-asymmetry association (right column). Thick boxplots
represent symmetric shapes, which correspond to self-related shapes in the self-symmetry association
and to stranger-related shapes in the self-asymmetry association.
Two separate ANOVAs with factors matching and shape were then conducted for the two
association types. For the self-symmetry association, the shape × matching interaction was significant
[F(1,29) = 54.24, p < .001, η2G = .064]. In matched trials, a large difference emerged between the RTs
for the self-related shape (M = 602 ms, SE = 18 ms) and the RTs for the stranger-related shape [M =
763 ms, SE = 30 ms; t(29) = -9.12, p < .001, d = -1.7; BF10 > 1000]. Instead, no significant difference
emerged in nonmatching trials [self-related shape: M = 729 ms, SE = 29 ms; stranger-related shape: M
= 739 ms, SE = 28 ms; t(29) = -0.98, p = .34, d = 0.18; BF01 = 3.32 ± .01%]. These results show a
clear self-prioritization effect for the self-symmetry association (see also Figure 2A). On the contrary,
no evidence of a self-prioritization effect emerged for the self-asymmetry association: The shape ×
matching interaction was significant [F(1,29) = 10.12, p = .003, η2G = .012]; however, in matched
trials, there was no significant difference between the RTs for the self-related shape (M = 716 ms, SE
= 22 ms) and the RTs for the stranger-related shape [M = 708 ms, SE = 26 ms; t(29) = 0.49 p = .63, d
= 0.09; BF01 = 4.60 ± .01%]. A significant difference emerged instead for nonmatching trials [self-
related shape: M = 718 ms, SE = 22 ms; stranger-related shape: M = 770 ms, SE = 28 ms; t(29) = -
4.78, p < .001, d = -0.89; BF10 = 511.48 ± .0%]. The latter result suggests that rejecting the
nonmatching self-symmetry pairs was more difficult than rejecting the nonmatching stranger-
asymmetry pairs (see also Figure 2B).
Our second hypothesis was that, consistently with the polarity correspondence principle,
responses would be faster for the self-symmetry association than for the self-asymmetry association.
11
This hypothesis is not supported by the results, as the main effect of association type was not
significant [F(1,29) = 1.9, p = .18, η2G = .005; BF01 = 2.18 ± .01%]
2
.
The inspection of individual data showed that 14 out of the 20 outliers in Figures 2A and 2B
(defined as the data points above the third quartile of the distribution plus 1.5 times the interquartile
range, or below the first quartile of the distribution minus 1.5 times the interquartile range) were due
to the responses of three participants (i.e., participants 5, 7, and 9 in the dataset on OSF), who had fast
mean RTs (M = 421 ms, SE = 36 ms; sample mean = 718 ms, SE = 23 ms), and high mean
percentages of errors (M = 46.4%, SE = 1.6; sample mean = 19.2%, SE = 2.3). All the main results
were replicated by an analysis conducted after the exclusion of the data of these responders from the
original dataset (see the Supplementary analyses on OSF at the following link:
https://doi.org/10.17605/OSF.IO/FE3JW).
Errors
Figures 2C and 2D show that the distribution of the percentage of errors tended to be positively
skewed. Shapiro-Wilk normality tests confirmed this impression: the distributions were significantly
different from normal in seven out of the eight cells of the experimental design. A skewness
coefficient larger than one was observed for four cells of the experimental design. Due to these
deviations from normality, the percentage of errors was analyzed through a mixed-effect logit model
(Jaeger, 2008) with association type, matching, shape, and the interactions as fixed effects and the by-
subject intercept as random effect.
The main effect of association type was statistically significant [χ2(1) = 71.15, p < .001; BF10
= 28.6 ± .0%], due to a smaller percentage of errors for the self-symmetry association (M = 17.4%, SE
= 2.4) than for the self-asymmetry association (M = 20.9%, SE = 2.2). This appears to suggest a
polarity correspondence effect for errors, although the effect was not observed for the RTs. The
association type × matching × shape interaction was significant [χ2(1) = 43.29, p < .001], therefore we
applied two separate mixed-effect logit models on the two association types, with matching, shape,
and the interaction as fixed effects and the by-subject intercept as random effect. For the self-
symmetry association, the shape × matching interaction was significant [χ2(1) = 51.6, p < .001].
Pairwise comparisons for the mixed-effect logit model showed that the percentage of errors was
significantly smaller in matched trials with the self-related shape (M = 11.2%, SE = 3.1) than in
matched trials with the stranger-related shape (M = 25.2%, SE = 2.5; z = 11.59, p < .001; BF10 >
1000). No significant difference emerged in nonmatching trials [self-related shape: M = 17.3%, SE =
3.0; stranger-related shape: M = 19.6%, SE = 3.0; z = 1.89, p = .23; BF01 = 2.11 ± .01%]. These results
confirm the self-prioritization effect in the case of the self-symmetry association (see also Figure 2C).
On the contrary, no evidence of a self-prioritization effect emerged for the self-asymmetry
association: The shape × matching interaction was significant [χ2(1) = 4.57, p = .03]; however, the
differences between the self- and the stranger-related shape were not significant, neither in matched
trials (self-related shape: M = 20.8%, SE = 2.5; stranger-related shape: M = 18.1%, SE = 2.4; z = -
2.26, p = .11; BF01 = 2.99 ± .0%) nor in nonmatching trials (self-related shape: M = 20.3%, SE = 2.6;
stranger-related shape: M = 20.9%, SE = 2.5; z = 0.75, p = .88; BF01 = 4.35 ± .0%). This pattern of
results confirms the lack of a self-prioritization effect for the self-asymmetry association (see also
Figure 2D). All the main results were replicated by an analysis conducted after the exclusion of the
data from participants 5, 7, and 9 (see the Supplementary analyses on OSF).
In sum, the results of Experiment 1 show a clear self-prioritization effect in the case of the
self-symmetry association, and the complete lack of the effect in the case of the self-asymmetry
association. Experiment 2 will test the robustness and generalizability of these results using a
between-participant manipulation of the type of association.
2
The power analysis was based on a unidirectional alternative hypothesis, namely, that the RTs for the
self-symmetry association were faster than the RTs for the self-asymmetry association. However, the p value for
the main effect of association type refers to a bi-directional alternative hypothesis. This p value, divided by two,
corresponds to the correct unidirectional p value, which is still nonsignificant (p = .053).
12
Experiment 2
Methods
Sample size
Using the same logic as in Experiment 1, we hypothesized at least a medium effect size (d = -0.5) for
the difference between the response times (RTs) for the self-symmetry association and those for the
self-asymmetry association. Unlike Experiment 1, here the type of association was manipulated
between participants. A power analysis showed that, for an independent-sample one-tailed t-test with
α = .05, β = .80, and d = -0.5, sufficient power would be reached with N = 50.15 per group. For the
sake of parsimony, we decided to test 52 participants per group (i.e., 104 participants in total).
Participants
One hundred and four participants (Mean age = 22.72 years, SD = 6.28 years, 14 males) voluntarily
participated in exchange for course credits. All of them were naive to the purpose of the experiment
and reported normal or corrected-to-normal vision. None of them had participated in Experiment 1.
Stimuli and Apparatus
Everything was identical to Experiment 1, with the following exceptions. The experiment was
delivered online through Pavlovia, which is known to provide reliable behavioral data (Bridges et al.,
2020). For technical reasons we could not use the online procedure of Experiment 1 to randomly
generate symmetric and asymmetric shapes. Instead, an offline procedure was used to pre-generate a
large number of pictures of symmetric and asymmetric random dot patterns with labels ‘YOU’ (in
Italian: ‘TU’) or STRANGER (in Italian: ‘ALTRO’) appearing at the center. On each trial of the
matching task, one of these pictures was randomly presented at the center of the screen. Picture size
(900 × 500 pixels) was adapted to the size and the resolution of the screen. From the observer’s
perspective, these stimuli were virtually identical to those of Experiment 1.
Procedure
Before starting the experiment, participants read the informed consent form approved by the Ethics
Committee for Psychological Research at the University of Padova (protocol number 3455, February
10th, 2020), and then gave their consent to participate through a response key.
Everything was identical to Experiment 1, except that the type of association was manipulated
between participants. Half of the participants were randomly assigned to the self-symmetry
association, and the other half to the self-asymmetry association. As in Experiment 1, the key-
response category association was counterbalanced across participants. Participants were presented
with 240 experimental trials (2 matching [matched vs. nonmatching] × 2 shape [self-related vs.
stranger-related] × 60 repetitions), which were preceded by 24 practice trials.
Results
Data were analyzed as in Experiment 1. Missed responses (3.67% of trials) were excluded and not
analyzed due to their low frequency. Errors (i.e., wrong responses; 24.88%) and the RTs of correct
responses (71.45%) were analyzed separately. Correct responses with RTs faster than 200 ms (1.49%
of correct responses) were also eliminated.
As shown in Figure 3, the patterns of results for RTs of correct responses (panels A and B)
and for errors (panels C and D) are similar. For brevity, only the main results of the RTs analysis and
the percentage of errors analysis are reported, whereas full results are in Appendix B. The RTs of
correct responses were analyzed through a three-way mixed ANOVA with association type (self-
symmetric vs. self-asymmetric) as a between-participant factor and matching (matched vs.
nonmatching) and shape (self-related vs. stranger-related) as within-participant factors
3
.
3
Shapiro-Wilk normality tests showed that the distribution of the RTs was significantly different from
normal in seven out of the eight cells of the experimental design. However, a closer inspection of the
distributions showed that these were similar to those of Experiment 1. A skewness coefficient smaller than -
1was observed for only two cells of the experimental design.
13
The results replicate those of Experiment 1. The main effect of association type was not
significant [F(1,102) = 0.22, p = .64, η2G = .002; BF01 = 4.37 ± .02%], whereas the association type ×
matching × shape interaction was significant [F(1,102) = 5.93, p = .017, η2G = .001]. As for the self-
symmetry association, the shape × matching interaction was significant [F(1,51) = 22.46, p < .001,
η2G = .009]. For matched trials, a large difference emerged between the RTs for the self-related shape
(M = 750 ms, SE = 17 ms) and the RTs for the stranger-related shape [M = 867 ms, SE = 24 ms; t(51)
= -9.52, p < .001, d = -1.3; BF10 > 1000]. A significant difference also emerged in nonmatching trials
[self-related shape: M = 834 ms, SE = 21 ms; stranger-related shape: M = 891 ms, SE = 24 ms; t(51) =
-6.07, p < .001, d = -0.84; BF10 > 1000], but the magnitude of this difference was clearly smaller as
compared to matched trials (i.e., 57 ms vs. 117 ms; see also Figure 3A). As for the self-asymmetry
association, the shape × matching interaction was not significant [F(1,51) = 0.83, p = .37, η2G < .001],
indicating no self-prioritization effect (see also Figure 3B). Despite the nonsignificant interaction, for
the sake of comparison with the results of Experiment 1, the results of the pairwise comparisons
between the RTs for the self- and stranger-related shapes are also reported. These were not significant,
neither in matched trials (self-related shape: M = 820 ms, SE = 20 ms; stranger-related shape: M = 835
ms, SE = 22 ms; t(51) = -1.32, p = .19, d = -0.18; BF01 = 2.92 ± .0%) nor in nonmatching trials (self-
related shape: M = 869 ms, SE = 21 ms; stranger-related shape: M = 872 ms, SE = 21 ms; t(51) = -
0.31, p = .76, d = -0.04; BF01 = 6.33 ± .0%).
The inspection of individual data showed that 13 out of the 15 outliers in Figure 3A were due
to the responses of five participants (i.e., participants 1, 7, 18, 19, and 41 in the dataset on OSF).
These participants had fast mean RTs (M = 478 ms, SE = 26 ms; sample mean = 835 ms, SE = 21 ms),
and high mean percentages of errors (M = 46.3 %, SE = 1.3; sample mean = 24.4%, SE = 1.7). The
three outliers in Figure 3B were due to the responses of one participant (i.e., participant 56), who also
had fast mean RTs (384 ms; sample mean = 849 ms, SE = 20 ms) and high mean percentages of errors
(M = 51.7 %; sample mean = 25.4%, SE = 1.7). All the main results were replicated by an analysis
conducted after the exclusion of the data from participants 1, 7, 18, 19, 41, and 56 (see the
Supplementary analyses on OSF).
Errors
The percentages of errors were analyzed through a three-way mixed ANOVA with association type
(self-symmetric vs. self-asymmetric) as a between-participant factor and matching (matched vs.
nonmatching) and shape (self-related vs. stranger-related) as within-participant factors
4
.
The main effect of association type was not significant [F(1,102) = 0.17, p = .68, η2G = .001;
BF01 = 4.46 ± .02%]. The association type × matching × shape interaction was significant [F(1, 102) =
30.0, p < .001 η2G = .026]. As for the self-symmetry association, the shape × matching interaction was
significant [F(1,51) = 86.65, p < .001, η2G = .030]. For matched trials, there was a significant
difference between the self-related shape (M = 13.4%, SE = 2.0) and the stranger-related shape [M =
35.6%, SE = 1.6; t(51) = -11.59, p < .001, d = -1.6; BF10 > 1000]. A smaller but significant difference
also emerged in nonmatching trials [self-related shape: M = 22.0%, SE = 2.1; stranger-related shape:
M = 26.5%, SE = 1.9; t(51) = -4.37, p < .001, d = -0.61; BF10 = 360 ± .0%]. As for the self-asymmetry
association, the shape × matching interaction was not significant [F(1,51) = 0.83, p = .37, η2G < .001],
indicating no self-prioritization effect (see also Figure 3D). No significant differences emerged
between self- and stranger-related shapes, neither in matched trials (self-related shape: M = 25.3%, SE
= 1.9; stranger-related shape: M = 22.4%, SE = 1.9; t(51) = 1.32, p = .23, d = 0.18; BF01 = 2.92 ± .0%)
nor in nonmatching trials (self-related shape: M = 27.9%, SE = 2.5; stranger-related shape: M =
25.9%, SE = 2.0; t(51) = 1.05, p = .30, d = 0.15; BF01 = 3.95 ± .0%). All the main results were
replicated by an analysis conducted after the exclusion of the data from participants 1, 7, 18, 19, 41,
and 56 (see the Supplementary analyses on OSF).
4
The distribution of the percentage of errors tended to be positively skewed. Shapiro-Wilk normality
tests showed that the distributions were significantly different from normal in seven out of the eight cells of the
experimental design. However, a skewness coefficient larger than 1 was observed for only one cell of the
experimental design. Therefore, the distributions appear to be sufficiently close to normal to allow the use of
ANOVA.
14
Figure 3. Results of Experiment 2. Boxplots representing the RTs of correct responses longer than
200 ms (panels A and B) and the percentage of errors (panels C and D) for the self-symmetry
association (left column) and for the self-asymmetry association (right column). Thick boxplots
represent symmetric shapes, which correspond to self-related shapes in the self-symmetry association
and to stranger-related shapes in the self-asymmetry association. Overall, both the percentage of
errors and the RTs are higher and more dispersed than in Experiment 1. This might be due to the
between-subject manipulation of association type (i.e., the practice effects were probably reduced
compared to Experiment 1), and to the fact that Experiment 2 was an online study rather than a
laboratory study.
General Discussion
In this work, we explored whether the self-prioritization effect (Sui et al., 2012) is modulated by the
valence of the shape to which identities (self vs. stranger) are associated with. In particular, we
presented participants with shapes differing in terms of the presence or absence of symmetry (i.e.,
symmetric shapes vs. asymmetric shapes). According to our pre-registered hypotheses, different
claims were tested (see the Open Practices statement for further details).
The matching task was not easier overall in the case of the self-symmetry association than the
self-asymmetry association. Based on the general principle of polarity correspondence (Proctor &
Cho, 2006), one of our hypotheses predicted that the type of association would have affected RTs and
errors in matching as well as in nonmatching trials. Specifically, it could be predicted that, in matched
trials, responding correct to stimuli of the self-asymmetry association (i.e., label you and an
asymmetric shape, or stranger and a symmetric shape), could be more difficult than responding
correct to stimuli of the opposite association (i.e., you-symmetric shape and stranger-asymmetric
shape). The results (see Tables 9 and 10 in Appendix C) showed that it was indeed more difficult
responding correct to incongruent you-asymmetry pairs than to congruent you-symmetry pairs;
however, at odds with polarity correspondence, the results also showed that it was more difficult
responding correct to congruent stranger-asymmetric pairs than to incongruent stranger-symmetric
pairs. Moreover, based on the principle of polarity correspondence, it could also be predicted that, in
15
nonmatching trials, responding incorrect to stimuli of the self-asymmetry association (i.e., label you
and a symmetric shape, or stranger and an asymmetric shape), could be more difficult than
responding incorrect to stimuli of the opposite association (i.e., you-asymmetric shape and stranger-
symmetric shape). However, the results showed that this was not the case (see Tables 11 and 12 in
Appendix C).
The type of association had, instead, a surprisingly strong influence on the self-prioritization
effect. Whereas a standard self-prioritization effect emerged for the self-symmetry association, the
effect did not emerge in the case of the self-asymmetry association. Therefore, even if the results of
previous studies suggest that the self-prioritization effect is a robust phenomenon that occurs across a
range of stimuli and experimental situations (e.g., Frings & Wentura, 2014; Fuentes et al., 2016;
Payne et al., 2017; Schäfer et al., 2015; 2016; Stein et al., 2016; Woźniak & Knoblich, 2019), the
results of our two experiments suggest that there are definite boundaries and constraints to the
phenomenon itself (see also Constable et al., 2021; Falbén et al., 2020; Golubickis et al., 2021;
McPhee et al., 2021; Strachan et al., 2020).
Previous studies suggest that symmetry and asymmetry are polarized concepts characterized
by positive and negative valence, respectively (Bertamini et al., 2013; Makin et al., 2012;
Pecchinenda et al., 2014). The different pattern of results emerging for the self-symmetry and the self-
asymmetry associations suggests that the self-prioritization can be disrupted when the self-relevant
information has negative valence, and the stranger-relevant information has positive valence.
Importantly, the effects of valence on self-prioritization do not appear to be bounded to associations
that recall (or that are in conflict with) previously learned long-term associations with the self
(Constable et al., 2021, Experiment 1; Golubickis et al., 2021), but would also extend to arbitrary
newly learned associations involving unfamiliar stimuli. Through an analysis of RTs performed with a
drift diffusion model, Golubickis et al. (2021) have recently concluded that positive valence stimuli
associated with the self are processed more efficiently, at a perceptual level, than stranger-related
stimuli. Based on these results, it can be speculated that self-related symmetric shapes are processed
faster, at a perceptual level, than symmetric or asymmetric stranger-related shapes. The processing
advantage for self-related stimuli characterized by positive valence may facilitate learning and
recalling new associations between the self and stimuli with positive valence, with respect to new
associations between the self and stimuli with negative valence. This flexibility of the self-
prioritization effect may have an important adaptive value. Indeed, not prioritizing negative valence
self-relevant information might be functional to keeping a positive bias for the self, which would
favor subjective well-being (Taylor & Brown, 1988). For instance, if a student fails to pass an exam in
which most other students succeed, avoiding the association with this negative valence information
might be useful in keeping negative emotions under control.
It is also worth mentioning that valence typically co-varies with several perceptual and
conceptual properties of the stimuli. Symmetric and asymmetric shapes make no exception. Besides
having more positive valence, symmetric shapes are also processed more fluently at a perceptual level
and are perceived as simpler and more arousing compared to asymmetric shapes (Bertamini et al.,
2013; Makin et al., 2012; Pecchinenda et al., 2014). Additionally, symmetry and asymmetry also
differ in terms of conceptual specificity, in that symmetry refers to a specific and well-defined
property, whereas asymmetry generically refers to the lack of this property. Lastly, symmetry is more
salient than asymmetry at a perceptual level (e.g., Bertamini et al., 2013). Therefore, although
converging evidence indicates that valence plays a crucial role in self-prioritization (Constable et al.,
2021; Golubickis et al., 2021; Hu et al., 2020; Sui & Humphreys, 2015d; Sui et al., 2016), it cannot be
excluded that, in our two experiments, the self-prioritization effect could be (also) modulated by any
of these features of the stimuli. Future studies should seek to disentangle the valence from these
correlated dimensions. For instance, to disentangle valence from perceptual salience, researchers may
reverse the typical relationship between these two variables, testing if highly salient stimuli
characterized by negative valence are prioritized over less salient stimuli characterized by positive
valence.
As a final note, we also point out that a third, non-pre-registered hypothesis, could be the
presence of a systematic advantage for the processing of symmetric shapes. If true, this would lead to
faster and more accurate responses to all symmetric shapes, than to asymmetric shapes, independently
of self-prioritization and polarity correspondence. This hypothesis would be consistent with the idea
16
that symmetry is salient at a perceptual level (e.g., Bertamini et al., 2013). The results of both
experiments are inconsistent with this additional hypothesis. Had the hypothesis been correct, then,
independently of identity (i.e., self or stranger) trials showing symmetric shapes would have been
responded faster and more accurately than trials showing asymmetric shapes. However, a facilitation
effect of symmetry emerged in matched trials but not in nonmatching trials (see Appendix C).
To conclude, the results of our two experiments indicate that the self-prioritization effect can
be flexibly modulated by specific features of self- and other-relevant information. We suggest that
valence might play a key role for the self-prioritization effect, as negative self-relevant information
would not be prioritized over positive other-relevant information. This would reflect a general
tendency of the cognitive system to keep a positive bias for the self and a negative bias for the
stranger (see also Taylor & Brown, 1988).
17
Appendix A
Experiment 1: RTs
Table 1. Results of the 2 (association type) × 2 (matching) × 2 (shape) ANOVA on the RTs of correct
responses of Experiment 1.
Effect
F value
p
ηG2
Association type
F (1,29) = 1.9
.178
.005
Matching
F (1,29) = 36.51
<.001*
.022
Shape
F (1,29) = 33.81
<.001*
.036
Association type × Matching
F (1,29) = 4.52
.042*
.001
Association type × Shape
F (1,29) = 31.65
<.001*
.013
Matching × Shape
F (1,29) = 13.56
<.001*
.006
Association type × Matching × Shape
F (1,29) = 46.17
<.001*
.035
Note. Symbol * indicates a statistically significant effect (p < .05).
Table 2. Results of the 2 (matching) × 2 (shape) ANOVAs on the RTs of correct responses of
Experiment 1, for the self-symmetry and the self-asymmetry associations.
Self-symmetry association
Self-asymmetry association
Effect
F value
p
ηG2
Effect
F value
p
ηG2
Matching
F (1,29) = 39.29
<.001*
.031
Matching
F (1,29) = 15.11
<.001*
.014
Shape
F (1,29) = 66.30
<.001*
.082
Shape
F (1,29) = 3.72
.06
.006
Matching × Shape
F (1,29) = 54.24
<.001*
.064
Matching × Shape
F (1,29) = 10.12
.003*
.012
Note. Symbol * indicates a statistically significant effect (p < .05).
18
Experiment 1: Percentage of errors
Table 3. Results of the mixed-effect logit model for the percentage of errors in Experiment 1.
Effect
χ2 value
p
Association type
χ2 (1) = 71.15
<.001*
Matching
χ2 (1) = .35
.55
Shape
χ2 (1) = 5.35
.02*
Association type × Matching
χ2 (1) = 20.61
<.001*
Association type × Shape
χ2 (1) = 97.71
<.001*
Matching × Shape
χ2 (1) = 4.65
.03*
Association type × Matching × Shape
χ2 (1) = 43.29
<.001*
Note. Symbol * indicates a statistically significant effect (p < .05).
Table 4. Results of the mixed-effect logit model for the percentage of errors in Experiment 1, for the
self-symmetry and the self-asymmetry associations.
Self-symmetry association
Self-asymmetry association
Effect
χ2 value
p
Effect
χ2 value
p
Matching
χ2 (1)= 31.39
<.001*
Matching
χ2 (1) = .32
.57
Shape
χ2 (1)= 134.29
<.001*
Shape
χ2 (1) = 5.1
.02*
Matching × Shape
χ2 (1) = 51.6
<.001*
Matching × Shape
χ2 (1) = 4.57
.03*
Note. Symbol * indicates a statistically significant effect (p < .05).
19
Appendix B
Table 5. Results of the 2 (association type) × 2 (matching) × 2 (shape) ANOVA on the RTs of correct
responses of Experiment 2.
Effect
F value
p
ηG2
Association type
F (1,102) = 0.22
.64
.002
Matching
F (1, 102) = 115.92
<.001*
.025
Shape
F (1, 102) = 68.98
<.001*
.025
Association type × Matching
F (1, 102) = 1.59
.21
<.001
Association type × Shape
F (1, 102) = 45.37
<.001*
.016
Matching × Shape
F (1, 102) = 14.47
<.001*
.004
Association type × Matching × Shape
F (1, 102) = 5.93
.017*
.001
Note. Symbol * indicates a statistically significant effect (p < .05).
Table 6. Results of the 2 (matching) × 2 (shape) ANOVAs on the RTs of correct responses of
Experiment 2, separately for the self-symmetry and the self-asymmetry associations.
Self-symmetry association
Self-asymmetry association
Effect
F value
p
ηG2
Effect
F value
p
ηG2
Matching
F (1,51) = 86.65
<.001*
.030
Matching
F (1,51) = 38.77
<.001*
.020
Shape
F (1,51) = 93.51
<.001*
.074
Shape
F (1,51) = 1.56
.22
.001
Matching × Shape
F (1,51) = 22.46
<.001*
.009
Matching × Shape
F (1,51) = 0.83
.37
<.001
Note. Symbol * indicates a statistically significant effect (p < .05).
20
Experiment 2: Percentage of errors
Table 7. Results of the 2 association type × 2 matching × 2 shape ANOVA on the percentage of
errors of Experiment 2.
Effect
F value
p
ηG2
Association type
F (1,102) = 0.17
.68
.001
Matching
F (1, 102) = 2.22
.14
.002
Shape
F (1, 102) = 31.35
<.001*
.034
Association type × Matching
F (1, 102) = 3.07
.08
.003
Association type × Shape
F (1, 102) = 66.37
<.001*
.070
Matching × Shape
F (1, 102) = 24.12
<.001*
.020
Association type × Matching × Shape
F (1, 102) = 30.0
<.001*
.026
Note. Symbol * indicates a statistically significant effect (p < .05).
Table 8. Results of the 2 matching × 2 shape ANOVAs on the percentage of errors of Experiment 2,
separately for the self-symmetry and the self-asymmetry associations.
Self-symmetry association
Self-asymmetry association
Effect
F value
p
ηG2
Effect
F value
p
ηG2
Matching
F (1,51) = 0.04
.84
<.001
Matching
F (1,51) = 4.40
.04*
.010
Shape
F (1,51) = 160.84
<.001*
.20
Shape
F (1,51) = 2.30
.14
.007
Matching × Shape
F (1,51) = 62.44
<.001*
.094
Matching × Shape
F (1,51) = 0.14
.71
<.001
Note. Symbol * indicates a statistically significant effect (p < .05).
21
Appendix C
Table 9. Results of the t-tests exploring the effects of association type on the RTs of matched trials in
Experiments 1 and 2
Experiment 1
Experiment 2
Comparison
t value
p
d
Effect
t value
p
d
(you, symmetric)
vs.
(you, asymmetric)
t(29) = -6.78
<.001
*
-1.24
(you, symmetric)
vs.
(you, asymmetric)
t(98.9) =
-2.69
.008*
-0.75
(stranger, symmetric)
vs.
(stranger, asymmetric)
t(29) = -3.26
.003*
-0.6
(stranger, symmetric)
vs.
(stranger, asymmetric)
t(101.1) =
-0.96
.34
-0.27
Note. Here and in the following tables, for Experiment 1 the results refer to paired-sample two-sided
t-tests, whereas for Experiment 2 they refer to independent-sample two-sided t-tests with Welch’s
correction for the degrees of freedom. Positive (negative) ts and ds indicate that the RTs for the first
pair in the column Condition were slower (faster) than the RTs for the second pair. The symbol *
indicates a statistically significant effect (p < .05). The results show that, in matched trials of both
experiments, the RTs for the congruent (you, symmetric) pair were significantly faster than the RTs
for the incongruent (you, asymmetric) pair. However, inconsistently with a polarity correspondence
principle, in Experiment 1 the RTs for the incongruent (stranger, symmetric) pair were significantly
faster than the RTs for the congruent (stranger, asymmetric) pair, whereas no significant difference
emerged in Experiment 2.
22
Table 10. Results of the t-tests exploring the effects of association type on the percentage of errors of
matched trials in Experiments 1 and 2
Experiment 1
Experiment 2
Comparison
t value
p
d
Effect
t value
p
d
(you, symmetric)
vs.
(you, asymmetric)
t(29) = -3.31
.003
*
-0.6
(you, symmetric)
vs.
(you, asymmetric)
t(101.9) =
-4.33
<.001
*
-1.2
(stranger, symmetric)
vs.
(stranger, asymmetric)
t(29) = -2.77
.009
*
-0.51
(stranger, symmetric)
vs.
(stranger, asymmetric)
t(100) = -5.27
<.001
*
-1.46
Note. Positive (negative) ts and ds indicate that the percentage of errors for the first pair in the column
Condition were larger (smaller) than the percentage of errors for the second pair. The symbol *
indicates a statistically significant effect (p < .05). The results show that, in matched trials of both
experiments, the percentage of errors for the congruent (you, symmetric) pair was significantly
smaller than the percentage of errors for the incongruent (you, asymmetric) pair. However,
inconsistently with a polarity correspondence principle, in both experiments, the percentage of errors
for the incongruent (stranger, symmetric) pair was significantly smaller than that for the congruent
(stranger, asymmetric) pair.
23
Table 11. Results of the t-tests exploring the effects of association type on the RTs of nonmatching
trials in Experiments 1 and 2
Experiment 1
Experiment 2
Comparison
t value
p
d
Effect
t value
p
d
(you, symmetric)
vs.
(you, asymmetric)
t(29) = 1.63
.12
0.3
(you, symmetric)
vs.
(you, asymmetric)
t(100.3) =
-0.62
.54
-0.09
(stranger, symmetric)
vs.
(stranger, asymmetric)
t(29) = 0.56
.58
0.1
(stranger, symmetric)
vs.
(stranger, asymmetric)
t(101.9) = -1.2
.23
-0.17
Note. The results show that, inconsistently with a polarity correspondence principle, there was no
difference between the RTs for congruent and incongruent pairs in the nonmatching trials of both
experiments.
Table 12. Results of the t-tests exploring the effects of association type on the percentage of errors of
nonmatching trials in Experiments 1 and 2
Experiment 1
Experiment 2
Comparison
t value
p
d
Effect
t value
p
d
(you, symmetric)
vs.
(you, asymmetric)
t(29) = 0.66
.52
0.12
(you, symmetric)
vs.
(you, asymmetric)
t(101.6) = -
0.23
.82
0.03
(stranger, symmetric)
vs.
(stranger, asymmetric)
t(29) = -1.55
.13
-0.28
(stranger, symmetric)
vs.
(stranger, asymmetric)
t(99.1) = -1.76
.08
-0.24
Note. The results show that, inconsistently with a polarity correspondence principle, there was no
difference between the percentage of errors for congruent and incongruent pairs in the nonmatching
trials of both experiments.
24
Funding
This research was supported by a grant from MIUR (Dipartimenti di Eccellenza DM 11/05/2017 n.
262) to the Department of General Psychology, University of Padova, and by an individual grant
(DPSS-SID2019) to M.D.
Open Practices Statement
Raw data and supplementary analyses are available at the following link:
https://doi.org/10.17605/OSF.IO/FE3JW.
The pre-registered hypotheses and methods are available at the following link:
https://aspredicted.org/blind.php?x=ts2m5i
References
Bertamini, M., Makin, A., & Rampone, G. (2013). Implicit association of symmetry with positive
valence, high arousal and simplicity. i-Perception, 4(5), Article 317-327.
https://doi.org/10.1068/i0601jw
Bridges, D., Pitiot, A., MacAskill, M. R., & Peirce, J. W. (2020). The timing mega-study:
comparing a range of experiment generators, both lab-based and online. PeerJ, 8,
e9414. https://doi.org/10.7717/peerj.9414
Cárdenas, R. A., & Harris, L. J. (2006). Symmetrical decorations enhance the attractiveness of faces
and abstract designs. Evolution and Human Behavior, 27, 1–18.
https://doi.org/doi:10.1016/j.evolhumbehav.2005.05.002
Constable, M. D., Becker, M. L., Oh, Y. I., & Knoblich, G. (2021). Affective compatibility with the
self modulates the self-prioritisation effect. Cognition & Emotion, 35(2), 291–304.
https://doi.org/10.1080/02699931.2020.1839383
Constable, M. D., Welsh, T. N., Huffman, G., & Pratt, J. (2019). I before U: Temporal order
judgements reveal bias for self-owned objects. Quarterly Journal of Experimental
Psychology, 72(3), 589–598. https://doi.org/10.1177/1747021818762010
25
Core Team (2021). R: A language and environment for statistical computing. R Foundation for
Statistical Computing, Vienna, Austria. URL https://www.R-project.org/
Cunningham, S. J., & Turk, D. J. (2017). Editorial: a review of self-processing biases in cognition.
Quarterly Journal of Experimental Psychology, 70(6), 987-995.
https://doi.org/10.1080/17470218.2016.1276609
Cunningham, S. J., Turk, D. J., Macdonald, L. M., & Macrae, C. N. (2008). Yours or mine?
Ownership and memory. Consciousness and Cognition, 17(1), 312–318.
https://doi.org/10.1016/j.concog.2007.04.003
Dalmaso, M., Castelli, L., & Galfano, G. (2019). Self-related shapes can hold the eyes. Quarterly
Journal of Experimental Psychology, 72(9), 2249-2260.
https://doi.org/10.1177/1747021819839668
Eisenman, R. (1967). Complexity–simplicity: I. Preference for symmetry and rejection of complexity.
Psychonomic Science, 8, 169–170. https://doi.org/10.3758/BF03331603
Falbén, J. K., Golubickis, M., Wischerath, D., Tsamadi, D., Persson, L. M., Caughey, S., Svensson, S.
L., & Macrae, C. N. (2020). It's not always about me: The effects of prior beliefs and stimulus
prevalence on self-other prioritisation. Quarterly Journal of Experimental Psychology, 73(9),
1466–1480. https://doi.org/10.1177/1747021820913016
Frings, C., & Wentura, D. (2014). Self-priorization processes in action and perception. Journal of
Experimental Psychology: Human Perception and Performance, 40(5), 1737-1740.
https://doi.org/10.1037/a0037376
Fuentes, L. J., Sui, J., Estévez, A. F., & Humphreys, G. W. (2016). The differential outcomes
procedure can overcome self-bias in perceptual matching. Psychonomic Bulletin & Review,
23(2), 451-458. https://doi.org/10.3758/s13423-015-0895-3
Golubickis, M., Falbén, J. K., Cunningham, W. A., & Macrae, C. N. (2018). Exploring the self-
ownership effect: Separating stimulus and response biases. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 44(2), 295-306.
https://doi.org/10.1037/xlm0000455
26
Golubickis, M., Ho, N. S. P., Falbén, J. K., Mackenzie, K. M., Boschetti, A., Cunningham, W. A., &
Macrae, C. N. (2019). Mine or mother’s? Exploring the self-ownership effect across
cultures. Culture and Brain, 7, 1–25. https://doi.org/10.1007/s40167-018-0068-0
Golubickis, M., Ho, N., Falbén, J. K., Schwertel, C. L., Maiuri, A., Dublas, D., Cunningham, W. A.,
& Macrae, C. N. (2021). Valence and ownership: object desirability influences self-
prioritization. Psychological Research, 85(1), 91–100. https://doi.org/10.1007/s00426-019-
01235-w
Greenwald, A. G., & Farnham, S. D. (2000). Using the implicit association test to measure self-
esteem and self-concept. Journal of Personality and Social Psychology, 79(6), 1022-1038.
https://doi.org/10.1037//0022-3514.79.6.1022
Hu, C.-P., Lan, Y., Macrae, N., & Sui, J. (2020). Good me bad me: prioritization of the good-self
during perceptual decision-making. Collabra: Psychology, 6(1), Article 20.
https://doi.org/10.1525/collabra.301
Humphreys, G. W., & Sui, J. (2015) The salient self: Social saliency effects based on self-bias,
Journal of Cognitive Psychology, 27(2), 129-140,
https://doi.org/10.1080/20445911.2014.996156
Humphreys, G. W., & Sui, J. (2016). Attentional control and the self: the Self-Attention Network
(SAN). Cognitive Neuroscience, 7(1-4), 5-17.
https://doi.org/10.1080/17588928.2015.1044427
Jiang, M., Wong, S. K. M., Chung, H. K. S., Sun, Y., Hsiao, J. H., Sui, J., & Humphreys, G. W.
(2019). Cultural orientation of self-bias in perceptual matching. Frontiers in Psychology, 10,
1469. https://doi.org/10.3389/fpsyg.2019.01469
Liu, M., & Sui, J. (2016). The interaction between social saliency and perceptual saliency. Quarterly
Journal of Experimental Psychology, 69(12), 2419-2430.
https://doi.org/10.1080/17470218.2015.1120330
Macrae, C. N., Visokomogilski, A., Golubickis, M., & Sahraie, A. (2018). Self-relevance enhances
the benefits of attention on perception. Visual Cognition, 26(7), 475-481.
https://doi.org/10.1080/13506285.2018.1498421
27
Makin, A. D. J., Pecchinenda, A., & Bertamini, M. (2012). Implicit affective evaluation of visual
symmetry. Emotion, 12(5). 1021-1030. https://doi.org/10.1037/a0026924
McIvor, L., Sui, J., Malhotra, T., Drury, D., & Kumar, S. (2021). Self-referential processing and
emotion context insensitivity in major depressive disorder. European Journal of
Neuroscience, 53(1), 311-329. https://doi.org/10.1111/ejn.14782
McPhee, A. M., Constable, M., Saccone, E. J., & Welsh, T. N. (2021) The influence of location,
ownership, and the presence of a coactor on the processing of objects. Canadian Journal of
Experimental Psychology, 75(4), 362-373. https://doi.org/10.1037/cep0000232
Morey, R. D., & Rouder, J. N. (2018). BayesFactor: Computation of Bayes Factors for common
designs. R package version 0.9.12-4.2. https://CRAN.R-project.org/package=BayesFactor
Payne, S., Tsakiris, M., & Maister, L. (2017). Can the self become another? Investigating the effects
of self- association with a new facial identity. Quarterly Journal of Experimental Psychology,
70(6), 1085-1097. https://doi.org/10.1080/17470218.2015.1137329
Pecchinenda, A., Bertamini, M., Makin, A. D. J., & Rampone, G. (2014). The pleasantness of visual
symmetry: always, never or sometimes. PLoS ONE, 9(3), e92685.
https://doi.org/10.1371/journal.pone.0092685
Peirce, J., Gray, J. R., Simpson, S., MacAskill, M., Höchenberger, R., Sogo, H., Kastman, E., &
Lindeløv, J. K. (2019). PsychoPy2: experiments in behavior made easy. Behavior Research
Methods, 51, 195–203. https://doi.org/10.3758/s13428-018-01193-y
Proctor, R. W., & Cho, Y. S. (2006). Polarity correspondence: a general principle for performance of
speeded binary classification tasks. Psychological Bulletin, 132(3), 416–442.
https ://doi.org/10.1037/0033-2909.132.3.416.
Qian, H., Wang, Z., Li, C., & Gao, X. (2020). Prioritised self-referential processing is modulated by
emotional arousal. Quarterly Journal of Experimental Psychology, 73(5), 688–697.
https://doi.org/10.1177/1747021819892158
Ramachandran, V. S., & Hirstein, W. (1999). The science of art: a neurological theory of aesthetic
experience. Journal of Consciousness Studies, 6(6–7), 15–51.
28
Reuther, J., & Chakravarthi, R. (2017). Does self-prioritization affect perceptual processes? Visual
Cognition, 25(1-3), 381-398. https://doi.org/10.1080/13506285.2017.1323813
Schäfer, S., Wentura, D., & Frings, C. (2015). Self‑prioritization beyond perception. Experimental
Psychology, 62(6), 415-425. https://doi.org/10.1027/1618-3169/a000307
Schäfer, S., Wesslein, A- K., Spence, C., Wentura, D., & Frings, C. (2016). Self‑prioritization in
vision, audition, and touch. Experimental Brain Research, 234(8), 2141-2150.
https://doi.org/10.1007/s00221- 016-4616-6
Schubert, T. M., Waldzus, S., & Giessner, S. R. (2009). Control over the association of power and
size. Social Cognition, 27(1), 1-19. https://doi.org/10.1521/soco.2009.27.1.1
Siebold, A., Weaver, M. D., Donk, M., & van Zoest, W. (2016). Social salience does not transfer to
oculomotor visual search. Visual Cognition, 23(8), 989–1019.
doi:10.1080/13506285.2015.1121946
Stein, T., Siebold, A., & van Zoest, W. (2016). Testing the idea of privileged awareness of self-
relevant information. Journal of Experimental Psychology: Human Perception and
Performance, 42(3), 303-307. https://doi.org/10.1037/xhp0000197
Strachan, J., Constable, M. D., & Knoblich, G. (2020). It goes with the territory: Ownership across
spatial boundaries. Journal of Experimental Psychology: Human Perception and
Performance, 46(8), 789–797. https://doi.org/10.1037/xhp0000742
Sui, J., He, X., & Humphreys, G. W. (2012). Perceptual effects of social salience: evidence from self-
prioritization effects on perceptual matching. Journal of Experimental Psychology: Human
Perception and Performance, 38(5), 1105-1117. https://doi.org/10.1037/a0029792
Sui, J., & Humphres, G. W. (2015a). The integrative self: how self-reference integrates perception and
memory. Trends in Cognitive Sciences, 19(12), 719-728.
https://doi.org/10.1016/j.tics.2015.08.015
Sui, J., & Humphreys, G. W. (2015b). More of me! Distinguishing self and reward bias using
redundancy gains. Attention, Perception, & Psychophysics, 77, 2549-2561.
https://doi.org/10.3758/s13414-015-0970-x
29
Sui, J., & Humphreys, G. W. (2015c). The interaction between self-bias and reward: evidence for
common and distinct processes. Quarterly Journal of Experimental Psychology, 68(10), 1952-
1964. https://doi.org/10.1080/17470218.2015.1023207
Sui, J., & Humphreys, G. W. (2015d). Super-size me: self biases increase to larger stimuli.
Psychonomic Bulletin & Review, 22, 550-558. https://doi.org/10.3758/s13423-014-0690-6
Sui, J., & Rotshtein, P. (2019). Self-prioritization and the attentional system. Current Opinion in
Psychology, 29, 148–152. https://doi.org/10.1016/j.copsyc.2019.02.010
Sui, J., Ohrling, E., & Humphreys, G. W. (2016). Negative mood disrupts self- and reward-biases in
perceptual matching. Quarterly Journal of Experimental Psychology, 69(7), 1438–1448.
https://doi.org/10.1080/17470218.2015.1122069
Sui, J., Yankouskaya, A., & Humphreys, G. W. (2015). Super-capacity me! Super-capacity and
violation of race independence for self- but not reward-associated stimuli. Journal of
Experimental Psychology: Human Perception and Performance, 41(2), 441-452.
https://doi.org/10.1037/a0038288
Sui, J., Sun Y., Peng, K., & Humphreys, G. W. (2014). The automatic and the expected self:
separating self and familiarity biases effects by manipulating stimulus probability. Attention,
Perception, & Psychophysics, 76, 1176-1184. https://doi.org/10.3758/s13414-014-0631-5
Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: a social psychological perspective on
mental health. Psychological Bulletin, 103(2), 193-210.
https://doi.org/10.1037/0033-2909.103.2.193
Wade, G. L., & Vickery, T. J. (2017). Self-relevance effects and label choice: Strong variations in
label-matching performance due to non-self relevant factors. Attention, Perception, &
Psychophysics, 79, 1524-1534. https://doi.org/10.3758/s13414-017-1307-8
Wang, H., Humphreys, G. W., & Sui, J. (2016). Expanding and retracting from the self: gains and
costs in switching self-associations. Journal of Experimental Psychology: Human Perception
and Performance, 42(2), 247-256. https://doi.org/10.1037/xhp0000125
Woźniak, M., & Knoblich, G. (2019). Self-prioritization of fully unfamiliar stimuli. Quarterly Journal
of Experimental Psychology, 72(8), 2110-2120. https://doi.org/10.1177/1747021819832981
30
Zhao, S., Uono, S., Yoshimura, S., & Toichi, M. (2015). Self make-up: the influence of self-
referential processing on attention orienting. Scientific Reports, 5, Article 14169.
https://doi.org/10.1038/srep14169
A preview of this full-text is provided by American Psychological Association.
Content available from Journal of Experimental Psychology Human Perception & Performance
This content is subject to copyright. Terms and conditions apply.
