Abstract
Can perceived responsiveness, the belief that meaningful others attend to and react supportively to core defining feature of the self, shape the structure of attitudes? We predicted that perceived responsiveness fosters open-mindedness, which, in turn, allows people to be simultaneously aware of opposing evaluations of an attitude object. We also hypothesized that this process will result in behavior intentions to consider multiple perspectives about the topic. Furthermore, we predicted that perceived responsiveness will enable people to tolerate accessible opposing evaluations without feeling discomfort. We found consistent support for our hypotheses in four laboratory experiments (Studies 1–3, 5) and a diary study (Study 4). Moreover, we found that perceived responsiveness reduces the perception that one’s initial attitude is correct and valid. These findings indicate that attitude structure and behavior intentions can be changed by an interpersonal variable, unrelated to the attitude itself.
Many of the attitudes people hold include both positive and negative aspects. For example, one can like the idea of jogging after work because it is good for health but dislike the fact that it interferes with family time. This phenomenon is called attitude ambivalence (e.g., Van Harreveld et al., 2015). Attitude ambivalence is usually experienced as aversive due to the evaluative conflict with which it is associated. This evaluative conflict includes feeling mixed and indecisive (Priester & Petty, 1996), and as it poses a threat to the self, people are motivated to resolve their ambivalent beliefs to restore cognitive consistency (e.g., Van Harreveld et al., 2009).
The threat to the self that arises from holding internally inconsistent beliefs has both interpersonal and intrapersonal roots. From an interpersonal perspective, people like to appear certain and consistent, so that expressing ambivalent attitudes may pose a threat to coherent self-presentation. Indeed, previous research has shown that ambivalent attitude holders face self-presentational concerns and are willing to express higher or lower levels of ambivalence, depending on their perception regarding the expectations of the social environment (Pillaud et al., 2013). More generally, from an intrapersonal perspective, cognitive inconsistency may insinuate a lack of coherence and decisiveness, which threatens one’s sense of self-worth (Steele, 1988). This threat activates the “psychological immune system,” which leads people to reinterpret situations in a biased way that reduces threat and is more consistent and coherent with prior attitudes (Clark & Wegener, 2013). Such re-interpretations are defensive in nature and preclude the individual having to change (Gilbert et al., 1998). Examples of such defensive biases are resisting attitude change or even bolstering initial attitudes (Lord et al., 1979), and dismissing information that suggests that present behavior poses a health risk (Kunda, 1987).
In this research, we examined the possibility that perceived responsiveness, an interpersonal variable unrelated to the attitude itself, can shape attitude structure and subsequent behavior intentions. We hypothesized that perceived responsiveness will make individuals more open-minded and will reduce defensive perceptions that their initial attitude is correct. As a result, perceived responsiveness should enable awareness of simultaneously opposing perspectives regarding the topic, yet this contradiction will be relatively more tolerable. Finally, we theorized that this process will enhance behavior intentions that reflect willingness to behave in a more open-minded manner.
Perceived responsiveness is defined as a process in which people believe that important others will attend to and react supportively to central defining features of the self (Reis et al., 2004). As a result perceived responsiveness helps people feel socially accepted and it lessens concerns about self-worth (Reis & Clark, 2013). The link between perceived responsiveness and self-worth is consistent with extensive evidence about the role of feedback from significant others in self-perception (Reis et al., 2004), and in particular with the sociometer model (Leary & Baumeister, 2000), which proposes that self-esteem is a barometer of people’s perception of how others value them. Thus, according to this model, self-worth increases when one feels accepted by the social environment and decreases when one feels rejected.
One important consequence of feeling valued by others is open-mindedness (Reis et al., 2018). Open-mindedness, defined as a willingness to consider multiple points of view, even those that oppose initial attitudes, enables people to process information in a broad manner (Reis et al., 2018). Defensiveness, in contrast, is usually conceptualized as the opposite of open-mindedness (e.g., Tjosvold et al., 1980). In our theorizing, perceived responsiveness should lessen defensiveness and thereby increase open-mindedness. This may occur because perceived responsiveness helps people feel valued and accepted, lessening the need to defend one’s own point of view. Existing evidence supports this supposition. For example, people who feel that their relationship partners are attentive and understanding have less defensive reactions to failure (Caprariello & Reis, 2011; Kumashiro & Sedikides, 2005) and display lesser needs to defensively self-enhance (Sedikides & Gregg, 2008). Also, perceived responsiveness reduces defensive biases such as the “better-than-average” effect and hindsight bias (Reis et al., 2018). However, none of these studies directly examined open-mindedness regarding attitudes. Hence, we hypothesized that perceived responsiveness should increase open-mindedness when people express their attitudes.
If perceived responsiveness makes people more attitudinally open-minded, they should be more willing to examine their attitude from multiple perspectives, including positions that oppose their initial attitude. As a result, they should become less certain that their initial attitude is the only correct attitude to hold. In terms of attitude structure, this process is defined as a reduction in attitude correctness, the subjective sense by which people believe their attitudes to be correct and valid (Petrocelli et al., 2007). Evidence for a link between levels of openness and attitude correctness comes from personality research that has found that people who are characterized by a closed-belief system also report greater confidence about their attitudes (Palmer & Kalin, 1991). Yet, to our knowledge, prior research has not established that open-mindedness can also lead to a diminished sense of attitude correctness.
When people perceive that others do not value their point of view, they typically try to bolster their original sense of correctness by searching for attitude-congruent information (Frey, 1986) or retrieving such information from memory (Kunda, 1990). This process eventually leads to holding one-sided attitudes (Heller et al., 1973). Hence, we theorized that if people perceive that their original attitude is less correct, but feel valued and appreciated by others, they would become aware of coexisting positive and negative attributes of the topic. Awareness of opposing aspects of an attitude objective refers to the extent to which they come to mind quickly and equally quickly and is termed simultaneous accessibility. Simultaneous accessibility is a prerequisite for objective-attitude ambivalence (Newby-Clark et al., 2002). Objective ambivalence is defined as the presence of both pros and cons about the same attitude object (DeMarree et al., 2011). 1
Previous research has shown that attitude certainty and objective ambivalence are negatively correlated (Luttrell et al., 2016). Moreover, Reis et al. (2018) found that perceived responsiveness enhances people’s recognition that their beliefs might be wrong and engenders greater receptivity to novel information that might be inconsistent with their existing self-views and beliefs. In terms of attitude structure, we focus on simultaneous accessibility. Simultaneous accessibility of opposing aspects of an attitude object increases subjective ambivalence (Newby-Clark et al., 2002), which is defined as the experience of an evaluative conflict and includes internal conflict, confusion, and mixed feelings about the topic (Priester & Petty, 1996).
Previous work has found that interpersonal variables that increase open-mindedness, such as the experience of high quality listening (Itzchakov et al., 2017, 2018; Itzchakov & Kluger, 2017a), attenuate the association between objective and subjective ambivalence. Yet, we sought to go one step further and examine whether perceived responsiveness can make opposing attitude-related thoughts tolerable while they are accessible. We propose that accessible internal inconsistencies will be experienced as less aversive when perceived responsiveness lessens the salience of threats to self-worth. Put differently, responsiveness should foster higher levels of simultaneous accessibility of opposing attitude-related thoughts without producing discomfort.
Finally, if perceived responsiveness increases the accessibility and tolerability of opposing positions, behavioral intentions should follow from attitudes. Attitude research characterizes ambivalent attitudes as weak attitudes (Thompson et al., 1995), which are therefore less predictive of specific behavior and behavior intentions (e.g., Armitage & Conner, 2000). However, ambivalence should increase intentions for behavioral exploration, that is, the behavior intentions of the ambivalent attitude holder, by definition, should also be two-sided. For example, if Jane is ambivalent about organ donation then she should be inclined to read newspaper articles advocating both in favor and against her initial attitude.
In summary, we tested the following hypotheses:
Individual Difference Moderators
Prior research has shown that the impact of perceived responsiveness varies from one individual to another (e.g., Birnbaum & Reis, 2012). We therefore decided to examine the influence of two individual difference moderators that directly concern people’s willingness to consider attitude-relevant information and therefore might be expected to moderate levels of open-mindedness after an experimental manipulation of responsiveness intended to affect their willingness to consider alternative points of view: need for cognitive closure (NFCC; Kruglanski et al., 1993) and need for cognition (NFC; Cacioppo & Petty, 1982). NFCC describes a general tendency to form unqualified judgments and to reach rigid decisions. People with a high NFCC tend to be intolerant of uncertainty and ambiguity (Webster & Kruglanski, 1994) and may resist the kind of openness that our manipulation is intended to engender. As a result, they may be less open-minded following the perceived-responsiveness manipulation.
As for NFC, people with a high NFC tend to engage in effortful thinking (Cacioppo & Petty, 1982). High NFCs can be expected to produce more thoughts related to priming manipulations than low NFCs, who may not engage in enough cognitive processing to make the prime-related thoughts accessible (Petty et al., 2008). Hence, we hypothesized that people high in the NFC will be more open-minded following the perceived-responsiveness manipulation, whereas people low in the NFC will show relatively smaller increases in open-mindedness following the manipulation.
Overview of Studies
We conducted five studies to test the research hypotheses. In Study 1 we asked participants to think about a difference of opinions with a meaningful other and manipulated high versus low perceived responsiveness from their friends. We then measured open-mindedness toward the issue and simultaneous accessibility of opposing positions. In Study 2 we aimed to conceptually replicate Study 1 using an attitude topic unrelated to relationships, added a control condition, and tested downstream effects of the responsiveness manipulation on behavioral intentions. Study 3 examined attitude correctness as an additional mediator and NFC and NFCC as moderators of the model. Moreover, we also considered affect as a potential alternative explanation for the hypothesized effects of perceived responsiveness (as perceived responsiveness might increase positive affect or reduce negative affect). Study 4 was a diary study that examined whether fluctuations in perceived responsiveness were associated with fluctuations in the attitudinal dependent variables. Finally, in Study 5 we examined the effect of perceived responsiveness of a specific person on attitude correctness and behavior intentions controlling for attachment security.
Study 1
Study 1 tested whether perceived responsiveness increases open-mindedness and whether increased open-mindedness creates awareness of opposing thoughts and emotions about an initial attitude expressed in a dispute with others (Hypotheses 1 and 3). Furthermore, we investigated whether perceived responsiveness allows people to acknowledge awareness of opposing aspects without feeling conflicted (Hypothesis 4).
Method
Participants
We computed the required sample size for power of at least 80%, assuming a medium effect size in a two-group between-participant design (Cohen’s d = .50). The required sample size was 128 participants (Faul et al., 2007). We recruited 150 undergraduates for course credit. Two participants did not complete the manipulation as instructed, and three participants had extreme response times on the simultaneous accessibility measure (more than 2 SDs from the sample mean) and their data were excluded from analyses. Hence, the final sample was composed of 145 participants (M age = 26.36, SD = 5.86, 58.7% female).
Procedure
Participants signed up for a study about attitudes in social interactions. Upon arriving, they signed a consent form and were presented instructions on a laboratory computer. First, participants were asked to think for 1 min about a difference of opinions with a person who is meaningful to them. Participants were then asked to briefly describe the event in 2 to 3 sentences.
Afterward, we randomly assigned participants to either a high or low perceived-responsiveness condition. We asked subjects to recall either two or eight kind or considerate things that other people had done for them in the past week. Participants were asked not to include the person that they had described previously. This manipulation relies on research showing that differences in the ease and difficulty of recalling events influence subsequent judgments (Schwarz, 1998). Participants in the high perceived-responsiveness condition were asked to recall two kind things that other people had done for them in the past week. Participants in the low perceived-responsiveness condition were asked to recall eight kind things, which should prime doubts about responsiveness, as recalling eight events is harder than recalling two events. This manipulation has been shown to be effective in manipulating perceptions of responsiveness (Reis et al., 2018).
Measures
In all studies, reliabilities and descriptive statistics for each variable are reported in the corresponding tables.
Perceived responsiveness (manipulation check)
We used the perceived partner responsiveness questionnaire (Reis et al., 2018), adapted to fit the present setting. This questionnaire consisted of 11 items, rated from 0 (not at all) to 10 (very much) on Likert-type scales. An example item is: “This week, other people made me feel like they valued my abilities and opinions.”
Open-mindedness
We measured open-mindedness using a standardized thought-listing task (Cacioppo & Petty, 1981). We first asked participants to list their thoughts about the issue selected at the beginning of the study (i.e., their difference of opinion with someone meaningful to them). We gave participants 2.5 min (monitored by computer) to complete up to eight thought-listing boxes. Two coders tabulated the categories that were used to calculate open-mindedness. The formula was:
This index ranged from 0 to 1. Higher numbers reflect greater open-mindedness.
Simultaneous accessibility of opposing evaluations
We calculated simultaneous accessibility by recording response latencies for a pair of split-semantic differential items asking separately about positive (negative) emotions and cognitions toward the participant’s attitude (Kaplan, 1972). The scales ranged from 1 (not at all) to 9 (very much). We controlled for individual differences in response latencies by standardizing the simultaneous accessibility scores on the focal attitude object (dispute with a meaningful other) using simultaneous accessibility scores from a control attitude object (going on vacation) (Fazio & Powell, 1997). This approach has been used in recent research (Itzchakov et al., 2020). Following Newby-Clark et al. (2002), we reciprocally transformed response latencies so that they became speed scores. We calculated simultaneous accessibility by squaring the slower speed score and dividing by the faster speed score. Higher scores indicate greater simultaneous accessibility of opposing aspects.
Objective ambivalence
Objective ambivalence was calculated by subtracting the absolute value of the difference between the positive (P) and negative (N) attributes of the Split-Semantic Differential Scale: (P + N)/2 −|P − N| (Thompson et al., 1995). Each item was anchored on a 9-point Likert-type scale. Higher scores indicate greater objective ambivalence.
Subjective ambivalence
We measured the subjective experience of evaluative conflict using a three-item scale (Priester & Petty, 1996) that asked about the extent to which participants felt (a) indecisive, (b) mixed, and (c) torn regarding their attitude. Items were anchored on a 9-point Likert-type scale.
Results and Discussion
Tables 1 and 2 present the correlations and reliabilities of all measures and means and SDs by conditions.
Correlations for Study 1.
Note. Reliabilities in parentheses; values in bold differ from 0 at p < .05; values in italics differ from p < .10.
Descriptive Statistics by Experimental Conditions for Study 1.
Manipulation check
Participants in the high perceived-responsiveness condition reported significantly greater perceived responsiveness t(143) = 4.80, p < .001, 95% CI = [0.84, 2.02], d = .80, indicating that the manipulation was effective.
Open-mindedness
Participants in the high perceived-responsiveness condition were significantly more open-minded about the attitude topic, t(143) = 2.87, p = .005, 95% CI = [0.05, 0.25], d = .48.
Simultaneous accessibility
The positive and negative aspects of the attitude object were significantly more accessible for participants in the high perceived-responsiveness condition than for participants in the low perceived-responsiveness condition, t(143) = 2.19, p = .03, 95% CI = [0.04, 0.76], d = .36.
Objective ambivalence
Participants in the high perceived-responsiveness condition reported significantly greater objective ambivalence than participants in the low perceived-responsiveness condition, t(143) = 2.11, p = .04, 95% CI = [0.06, 1.89], d = .35. This result, accompanied by the increase in simultaneous accessibility, indicates that when participants felt that other people had been more responsive toward them, they were simultaneously more aware of the positive and negative aspects of the attitude topic (increased accessibility) and reported greater mental presence of the contradictory aspects (increased objective ambivalence).
Subjective ambivalence
We observed no difference between the high and low perceived-responsiveness conditions on subjective-attitude ambivalence, t(143) = −0.30, p = .98, 95% CI = [−0.80, 0.78], d = −.05. To examine whether perceived responsiveness enables tolerance of contradictory evaluations, we compared the correlations of simultaneous accessibility and subjective-attitude ambivalence across experimental conditions. In the low perceived-responsiveness condition, this correlation was positive and significant, r = .47, p < .001, that is, higher simultaneous accessibility was accompanied by increased evaluative conflict. However, in the high perceived-responsiveness condition, this correlation was not significant, r = .14, p = .24. The difference between these correlations was significant, Z = 2.28, p = .029, Cohen’s Q = .37. This result supports the hypothesis that the responsiveness-induced simultaneous accessibility was experienced as tolerable. No difference emerged with regard to attitude importance t(143) = −0.14, p = .89, 95% CI = [−0.71, 0.61], d = −.02.
As our hypotheses involved multiple significance tests, it is possible that the family-wise error rate exceeded the nominal .05 level. To address this possibility, we applied a Bonferroni correction with an adjusted alpha of .017 (.05/3). The effect for open-mindedness remained significant, but the effect for simultaneous accessibility, which had a p-value of .04, did not. To address this inconsistency, we conducted a meta-analysis across our studies (described in the “General Discussion” section).
Mediation analysis
We tested the proposed mediation model using Model 4 in PROCESS (Hayes, 2017) with 5,000 bootstrapped samples. The direct effect of the manipulation on simultaneous accessibility was not significant b = 0.26, SE = .18, p = .16, 95% CI = [−0.10, 0.62]. The indirect effect was significant, b = 0.14, SE = .07, 95% CI [0.03, 0.29], which suggests that the effect of the perceived-responsiveness manipulation on simultaneous accessibility was mediated by open-mindedness. The total effect was also significant, b = 0.40, SE = .18, p = .03, 95% CI = [0.04, 0.76] (see Figure 1).

Mediation model of Study 1.
The results of Study 1 indicated that perceived responsiveness increased participants’ open-mindedness and made opposing aspects of their attitude more simultaneously accessible. Furthermore, responsiveness enabled participants to better tolerate inconsistencies regarding an attitude object. However, this experiment leaves certain key questions unanswered. First, as we did not include a control condition, our data do not indicate whether the effects on the dependent variables were attributable to the high, low, or both perceived-responsiveness conditions. Second, Study 1 did not examine downstream effects of perceived responsiveness on behavior intentions. Study 2 addressed these limitations.
Study 2
Study 2 had three goals. First, we wanted to conduct a preregistered 2 conceptual replication using a different attitude object, one that is unrelated to the responsiveness manipulation. Second, we added a control condition to determine whether high responsiveness, low responsiveness, or both drive the obtained effects. Finally, we tested downstream effects of perceived responsiveness on behavioral intentions. This study tested Hypotheses 1, 3, 4, and 5.
Method
Participants
We recruited 245 undergraduates for course credit. Of these participants, six did not complete the manipulation as instructed (e.g., did not write about kind or considerate events that others did for them), and 10 participants had extreme response times (more than 2 SD from the mean). Their data were excluded, yielding a final sample size of N = 229 (M age = 25.63, SD = 4.55, 66.4% female). This sample size has power of above .80 to detect a medium effect size (Cohen’s f = .25).
Procedure
Participants arrived at the laboratory to take part in a study about attitudes toward food. Instructions were presented on a laboratory computer. First, participants read a paragraph about genetically modified food (GMF; see Supplemental Material), after which they were asked to indicate whether their attitude about GMF was positive or negative and to write a short essay (4–8 sentences) advocating in favor of their attitude. The goal of this essay was to make the attitude topic salient and meaningful.
Afterward, we randomly assigned participants to a perceived responsiveness (low/high) or control condition. To strengthen the manipulation, we asked participants in the low perceived-responsiveness condition to list 10 kind and considerate events that others had done for them in the past week. Participants in the high perceived-responsiveness condition were asked to recall two kind things that others had done for them. Participants in the control condition were asked to describe the office or cubicle of a colleague. Afterwards, participants completed the same measures as in Study 1 (but now targeting GMF) and an additional measure of behavioral intentions. Finally, participants were debriefed.
Measures
We used the same measures, perceived-responsiveness manipulation check, subjective and objective-attitude ambivalence, and attitude importance, as in Study 1. All items were anchored on 9-point Likert-type scales.
Open-mindedness
We used a thought-listing task as in Study 1, here pertaining to GMF. Open-mindedness was coded as the ratio of the number of thoughts that were in a different direction to the initial attitude divided by the total amount of thoughts, similar to past research (Porter & Schumann, 2018). For example, a positive thought about GMF written by a participant whose initial attitude was negative was counted in the numerator. Before writing the essay, participants reported whether their attitude toward GMF was positive or negative. Open-mindedness was calculated using the number of positive thoughts listed as the numerator and vice versa.
Simultaneous accessibility
We calculated simultaneous accessibility using the same formula as in Study 1. The standardization object in Study 1 might have confounded the results as it shared some characteristics with the responsive manipulation, namely, both were positive and interpersonal. Therefore, in this study we changed the control attitude object used for standardization (a toothache).
Behavior intentions
We created a new measure to assess behavior intentions that capture openness to considering opposing points of view. This measure consisted of nine items (e.g., “To what extent are you willing to read a newspaper column or blog advocating against your attitude toward genetically modified food?”).
Results and Discussion
Tables 3 and 4 present the correlations and reliabilities for all measures and means and SDs by condition.
Correlations for Study 2.
Note. Reliabilities in parentheses; values in bold differ from 0 at p < .05; values in italics differ from p < .10.
Descriptive Statistics by Experimental Conditions for Study 2.
Manipulation check
An analysis of variance (ANOVA) indicated a main effect of the experimental manipulation, F(2, 226) = 28.27, p < .001,
Open-mindedness
An ANOVA indicated a significant effect of experimental conditions, F(2, 226) = 11.71, p < .001,
Simultaneous accessibility
An ANOVA indicated a significant main effect for experimental conditions, F(2, 226) = 5.18, p = .006,
Objective ambivalence
There was a main effect of the experimental manipulation on objective ambivalence, F(2, 226) = 9.49, p < .001,
Behavior intentions
An ANOVA indicated a significant main effect for experimental conditions, F(2, 226) = 8.93, p < .001,
Subjective ambivalence
There was no main effect of the experimental manipulation on subjective ambivalence, F(2, 226) = 1.29, p = .28,
Critically, as in Study 1, the association between simultaneous accessibility and subjective ambivalence varied by experimental conditions. The correlations in the low responsiveness and control conditions were positive and significant, r = .50 and .35, p < .001, p = .002, respectively. However, the correlation in the high responsiveness condition was not significant, r = .03, p = .77. The difference between the correlations in the high perceived-responsiveness and control conditions was significant, Z = 1.98, p = .047, Cohen’s Q = .34, as was the difference between the high and low perceived-responsiveness conditions, Z = 3.19, p = .001, Cohen’s Q = .52. Thus, responsiveness enabled participants to tolerate contradictions in their attitudes. No difference emerged with regard to attitude importance F(2, 225) = 1.02, p = .36,
As in Study 1, the hypothesized effects were also examined with Bonferroni correction, for α = .0125 (.05/4). All hypothesized effects remained significant.
Mediation analysis
We conducted serial-mediation analysis using Model 6 in PROCESS (Hayes, 2017), with 5,000 bootstrapped samples. As we could not assume linearity across three experimental conditions, we created two dummy variables and conducted two mediation analyses. The first dummy variable compared the high perceived-responsiveness (Dummy 1) to the control and low perceived-responsiveness conditions (Dummy 0). The second dummy variable compared the low perceived-responsiveness condition (coded as 1) to the control and high perceived-responsiveness conditions (coded as 0). 3 The direct effect was significant, b = 0.65, SE = .28, p = .02, 95% CI = [0.10, 1.20]. The indirect effect of the path: perceived responsiveness → open-mindedness → simultaneous accessibility → behavior intentions, was significant for Dummy 1, b = 0.04, SE = .03, 95% CI = [0.004, 0.11]. The total effect was significant, b = 1.00, SE = .28, p < .001, 95% CI = [0.46, 1.55] (see Figure 2A). The same pattern was observed when comparing low responsiveness to the control condition and to high responsiveness. All effects were significant. Specifically, the direct effect, b = −0.66, SE = .27, p = .02, 95% CI = [−1.20, −0.12], the indirect effect, b = −0.05, SE = .03, 95% CI = [−0.12, −0.01], and the total effect, b = −0.98, SE = .27, p < .001, 95% CI = [−1.52, −0.45] (see Figure 2B).

Serial mediation model of Study 2: (A) the effect of high perceived-responsiveness and (B) the effect of low perceived-responsiveness.
In sum, Study 2 replicated the results of Study 1 using an attitude topic that was distinct from the manipulation. Study 2 also extended the earlier study by showing that responsiveness-induced cognitive processes influence behavior intentions to act open-mindedly. Moreover, the addition of a control group allowed us to establish that both higher and lower responsiveness influence the dependent variables studied.
Studies 1 and 2 have several limitations. First, as they did not control for possible effects of the manipulation on affect, an alternative explanation might be that the experimental manipulation increased positive affect, which is known to facilitate open and flexible ways of thinking (Estrada et al., 1997; Fredrickson, 2001). Second, these studies do not speak to boundary conditions of the model (Hypotheses 6 and 7). Study 3 addressed these questions.
Study 3
Study 3 was a preregistered 4 conceptual replication with several goals. First, we examined whether affect might have been responsible for the manipulation-induced effects in Studies 1 and 2. Moreover, we investigated whether perceived responsiveness reduces attitude correctness and whether two individual difference variables, the NFC and the NFCC, moderate the effect of perceived responsiveness and open-mindedness on change in attitude structure. Study 3 tested Hypotheses 1 through 7.
Method
Participants
We recruited 200 undergraduates in exchange for course credit. This sample size has power of above .80 to detect the effect size obtained on behavior intentions in Study 2. Data from 33 participants were excluded: for not completing the manipulation or essay as instructed (n = 11), giving meaningless answers to the thought-listing task (n = 8), and for simultaneous accessibility scores that deviated by more than 2 SD from the mean (n = 14). Hence, the final sample size was N = 167 (M = 26.80, SD = 5.99, 55.7% female).
Procedure
The procedure was essentially identical to that of Study 2, except that the control condition was dropped. The individual difference variables were measured before the responsiveness induction.
Measures
We used the same measures, perceived-responsiveness manipulation check, subjective ambivalence, objective ambivalence attitude importance, open-mindedness, simultaneous accessibility, and behavior intentions, as in Study 2. All self-report items were anchored on 9-point Likert-type scales.
Attitude correctness
We measured attitude correctness using a three-item scale from Petrocelli et al. (2007), adapted to the current attitude topic. An example item is: “How certain are you that your attitude toward GMF is the correct attitude to have?”
Affect
We measured affect using the positive and negative affect schedule (PANAS; Watson et al., 1988). This scale consisted of 10 items that assess positive affect (e.g., “excited”) and 10 items that assess negative affect (e.g., “guilty”).
NFCC
The NFCC scale (Webster & Kruglanski, 1994) consists of 42 items. An example item is: “I dislike unpredictable situations.”
NFC
The NFC scale (Cacioppo & Petty, 1982) consists of 18 statements assessing people’s tendency to engage and enjoy effortful cognitive processes. An example item is: “I would prefer complex to simple problems.”
Results and Discussion
Tables 5 and 6 present the correlations and reliabilities of all measures and means and SDs by condition.
Correlations for Study 3.
Note. Reliabilities in parentheses; values in bold differ from 0 at p < .05; values in italics differ from p ≤ .10.
Descriptive Statistics by Experimental Conditions for Study 3.
Manipulation check
Participants in the high responsiveness condition reported significantly greater perceived responsiveness than participants in the low responsiveness condition, t(165) = 6.17, p < .001, 95% CI = [0.88, 1.71], d = .95.
Open-mindedness
Participants in the high responsiveness condition were significantly more open-minded than participants in the low perceived-responsiveness condition, t(165) = 5.99, p < .001, 95% CI = [0.11, 0.22], d = .93.
Simultaneous accessibility
Participants in the high responsiveness condition were significantly more aware of opposing aspects within the attitude than participants in the low responsiveness condition, t(165) = 2.49, p = .014, 95% CI = [0.004, 0.03] d = .39.
Objective ambivalence
Participants in the high responsiveness condition reported significantly higher objective ambivalence than participants in the low perceived-responsiveness condition, t(165) = 2.60, p = .01, 95% CI = [0.32, 2.32], d = .40.
Attitude correctness
Participants in the high responsiveness condition reported significantly lower attitude correctness than participants in the low perceived condition, t(165) = −4.37, p < .001, 95% CI = [−2.19, −0.83], d = −.68.
Behavior intentions
Participants in the high responsiveness condition reported greater intentions to act in a way that considered alternative points of view than participants in the low responsiveness condition, t(165) = 2.28, p = .02, 95% CI = [0.08, 1.09], d = .35.
Subjective ambivalence
There was no main effect of the experimental manipulation on subjective ambivalence, t(165) = 0.82, p = .41, 95% CI = [−0.32, 0.78], d = .01. Critically, the association between simultaneous accessibility and subjective ambivalence varied by condition. The correlation in the low responsiveness condition was positive and significant, r = .36, p < .001, whereas in the high responsiveness condition it was not significant, r = .08, p = .43. The difference between these two correlations was significant, Z = 2.07, p = .04, Cohen’s Q = .30. Thus, as in Studies 1 and 2, experiencing responsiveness enabled individuals to tolerate opposing aspects within their attitudes.
Affect
There was no difference between experimental conditions with regard to negative, t(165) = −0.86, p = .39, or positive affect, t(165) = −0.32, p = .75, d = −.05.
As in the previous studies we applied a Bonferroni correction with an adjusted α = .007 (.05/7). The effects for perceived responsiveness, open-mindedness, and attitude correctness remained significant. The main effects for simultaneous accessibility (p = .01) and behavior intentions (p = .02) were no longer significant. We address this difference in the General Discussion.
Mediation analysis
We conducted three mediation analyses. The first analysis was a serial-mediation analysis using Model 6 in PROCESS (Hayes, 2017) to examine the role of attitude correctness in the model: Perceived responsiveness → open-mindedness → attitude correctness → simultaneous accessibility → behavior intentions. We also controlled affect. The second and third analyses were serial moderated-mediation (Model 83 in PROCESS) to examine moderation by NFCC and NFC, respectively. Both analyses were conducted with 5,000 bootstrapped samples.
The direct effect was not significant, b = 0.14, SE = .14, p = .33, 95% CI = [−0.14, 0.42]. The indirect effect of the perceived-responsiveness manipulation on behavior intention via open-mindedness, attitude correctness, and simultaneous accessibility was significant, b = 0.01, SE = .008, 95% CI = [0.001, 0.03] (see Figure 3). The total effect was significant, b = 0.29, SE = .13, p = .02, 95% CI = [0.04, 0.54].

Serial mediation model of Study 3 controlling for affect.
In the second analysis, NFC significantly moderated the effect of the perceived-responsiveness manipulation on open-mindedness, b = 0.03, SE = .006, p < .001, 95% CI = [0.01, 0.04]. Specifically, perceived responsiveness increased open-mindedness more for high NFC individuals (1 SD above the mean), b = 0.16, SE = .02, p < .001, 95% CI = [0.11, 0.20], but not for low NFC individuals (1 SD below the mean), b = 0.02, SE = .02, p = .15, 95% CI = [−0.01, 0.06]. The direct effect was not significant, b = 0.14, SE = .14, p = .32, 95% CI = [−0.14, 0.42]. The conditional indirect effect was stronger for high NFC individuals, b = 0.02, SE = .01, 95% CI = [0.002, 0.06], than for low NFC individuals, b = 0.003, SE = .004, 95% CI = [−0.001, 0.01]. The index of moderated-mediation (Hayes, 2015) was significant, index = 0.004, SE = .003, 95% CI = [0.0002, 0.01] (see Figure 4).

Serial moderated mediation model of Study 3 with NFC.
The same pattern of results was obtained for the third mediation analysis with NFCC as a moderator. First, the NFCC × Responsiveness interaction on open-mindedness was significant, b = −0.02, SE = .005, p < .001, 95% CI = [−0.03, −0.01] (see Figure 5). Specifically, perceived responsiveness significantly increased open-mindedness for low NFCC individuals (1 SD below the mean), b = 0.10, SE = .02, p < .001, 95% CI = [0.06, 0.13], but not for high NFCC individuals (1 SD above the mean), b = 0.001, SE = .01, p = .94, 95% CI = [−0.03, 0.03]. The direct effect was not significant, b = 0.14, SE = .14, p = .32, 95% CI = [−0.14, 0.42]. The conditional indirect effect was stronger for low NFCC individuals, b = 0.01, SE = .01, 95% CI = [0.001, 0.04], than for high NFCC individuals, b = 0.000, SE = .002, 95% CI = [−0.004, 0.004]. The index of moderated mediation was significant, index = −0.003, SE = .002, 95% CI = [−0.009, −0.0002].

Serial moderated mediation model of Study 3 with NFCC.
In sum, Study 3 replicated the previous studies and added several novel insights. First, it ruled out positive and negative affect as potentially being the active ingredient in the model. Second, the perceived-responsiveness manipulation reduced attitude correctness, and the results of the mediation analysis suggest that this reduction may contribute to behavior intentions. Finally, Study 3 identified boundary conditions by showing that the NFCC can attenuate the effect of perceived responsiveness on open-mindedness and its downstream effects whereas the NFC can amplify its effect.
Study 4
Although Studies 1 to 3 provided internally valid tests of our hypotheses, they leave open the question of how well they represent processes occurring in spontaneous, everyday life (i.e., ecological validity). Therefore, we conducted a diary study to examine the focal associations in the context of natural activity. An important advantage of diary studies is that they enable researchers to track the flow of social processes within people’s everyday environment (Bolger & Laurenceau, 2013). In Study 4, we examined whether natural fluctuations in perceived partner responsiveness, over and above person averages, systematically predicted changes in the dependent variables. This study examined Hypotheses 1, 2, 3, and 5.
Method
Participants
Participants were MBA students enrolled in an Organizational Behavior course who completed the study as part of a class assignment. Participants (N = 147) completed the research questionnaires four times, every 3 days. A total of 24 participants answered the questionnaires three times. Following the preregistered protocol of Study 3, we excluded observations that were higher or lower by more than 2 SDs on the simultaneous accessibility measure (n = 25). 5 Hence, the final sample size was 147 participants and 539 observations. This sample has power above .80 to detect the average effect size of behavioral intentions from Studies 2 and 3 in a within-participants repeated measure design with four repetitions, assuming an average correlation of .50 between measurements.
Procedure
The study was administered in the context of a class assignment about interpersonal relationships at work. We used an interval contingent scheme in which the sampling interval was 3 days (Reis et al., 2014). First, we instructed participants to think and write about a conflict they had encountered at work. An example was: “my colleague is really slow in completing her share in our mutual assignments. This really upsets me but I’m not sure if I should go to my boss to tell about it as it will hurt our friendship.” Participants indicated their attitude about the conflict (positive/negative). Throughout the assignment, participants were asked to report their thoughts and feelings toward the same conflict they initially listed.
Participants completed the perceived-responsiveness questionnaire and the dependent measures four times, once every 3 days. We sent reminder emails in the morning of the third day of each round. At the end of the study, we debriefed participants.
Measures
Study 4 used the same measures as the previous studies. All self-report items were anchored on 9-point Likert-type scales. Open-mindedness again was calculated as the ratio of the number of thoughts that were opposite to the initial attitude divided by the total amount of thoughts.
Results and Discussion
Table 7 presents the descriptive statistics and inter-correlations (within and between) among study variables.
Correlations for Study 4.
Note. Correlations below the diagonal are within participants (Level 1, person-centered, N = 539); correlations above the diagonal are between participants (Level 2, person-mean, N = 147). Reliabilities in parentheses; values in bold differ from 0 at p < .05.
The data had a nested structure with reports (Level 1) nested within individuals (Level 2). Intra-class correlations for perceived responsiveness, open-mindedness, attitude correctness, simultaneous accessibility, and behavior intentions were: .57, .27, .36, .52 and .03, respectively. We first examined the first-order autoregressive correlations among study variables, none of which were significant. We therefore conducted the serial-mediation analysis using multilevel modeling in Mplus, which enables us to account for variance both within and between participants. We used person-mean centering for Level 1 variables. The model included random intercepts and fixed slopes and was estimated using maximum likelihood estimation. We used a diagonal error structure, which estimates residuals for each time-point separately.
We examined a Level 1 (within-person) mediation model where fluctuations in perceived responsiveness predicted fluctuations in behavioral intentions via fluctuations in open-mindedness, attitude correctness, and simultaneous accessibility. As Figure 6 shows, fluctuations in perceived responsiveness significantly predicted fluctuations in open-mindedness, b = 0.04, SE = .01, p < .001. Fluctuations in open-mindedness predicted fluctuations in attitude correctness above and beyond fluctuations in perceived responsiveness, b = −2.32, SE = .51, p < .001. Fluctuations in attitude correctness predicted fluctuations in simultaneous accessibility above and beyond fluctuations in perceived responsiveness and open-mindedness, b = −0.12, SE = .03, p < .001, and fluctuations in simultaneous accessibility predicted fluctuations in behavior intentions above and beyond fluctuations in perceived responsiveness, open-mindedness, and attitude correctness, b = 0.22, SE = .03, p < .001. The direct effect was not significant, b = 0.06, SE = .06, p = .301. The indirect effect from perceived responsiveness to behavior intentions was significant, b = 0.002, SE = .001, p = .043. All main effects remained significant with a Bonferroni correction of α = .0125. We also examined alternative models with attitude correctness in a different serial position in the model. The indirect effect with simultaneous accessibility and attitude correctness switched was not significant, b = 0.001, SE = .001, p = .09, as well as the indirect effect with the order of attitude correctness and open-mindedness reversed, b = 0.001, SE = .001, p = .11.

Serial multilevel mediation model of Study 4.
In sum, Study 4 replicated the previous experiments and increased ecological validity. The results indicate that 3-day within-person fluctuations in perceived responsiveness were associated with within-person fluctuations in behavior intentions via open-mindedness, attitude correctness, and simultaneous accessibility. More specifically, daily fluctuations in perceived responsiveness were associated with changes in attitude structure and behavior intentions regarding conflicts at work.
Study 5
Study 5 was conducted to test whether the observed changes in attitude structure and behavior intentions can be generalized to responsiveness from a single and specific person. Previous work on responsiveness has typically used inductions that included perceptions about a specific person (Caprariello & Reis, 2011). The second goal was to rule out attachment style as an alternative explanation for our results. Previous work has found that attachment style has similar effects to those found here for perceived responsiveness on variables such as curiosity, cognitive closure, and the extent to which individuals use existing knowledge in making social judgments (Mikulincer, 1997). This study was preregistered. 6
Method
Participants
We conducted a two-wave study through Prolific Academic. The first wave included 280 participants, and 214 completed the second survey. A total of 15 participants were excluded as they failed to follow instructions regarding the induction. Hence, the final sample size was 199 (M age = 35.44, SD = 11.75, 66.8% female). This sample has power of above .80 to detect the average effect size found for behavioral intentions in the previous studies.
Procedure
In the first wave, participants completed an attachment questionnaire and demographics. The second wave was administered 1 week afterwards. Participants were asked to briefly describe a difference of opinion with a significant other. These included romantic partners (56.8%), family members (26.1%), friends (14.5%), and colleagues (2.5%). Subsequently, they were randomly assigned to one of two responsiveness conditions. In the high responsiveness condition, participants were asked to list two kind or considerate things that the person with whom they had the dispute did for them in the past 3 days. Participants in the low responsiveness condition were asked to list 12 things. After 5 min participants were automatically advanced to the next page. Next, participants answered questions including the research variables. Finally, participants were debriefed and compensated.
Measures
All measures were anchored on 9-point Likert-type scales.
Attachment style
Attachment style was assessed with the Experiences in Close Relationships scale-short form (Wei et al., 2007). Secure attachment style was calculated as the sum of scale items.
Perceived responsiveness, attitude correctness, and behavior intentions
We used the same measures as in the previous studies. Items in the perceived-responsiveness manipulation check asked about the specific person (e.g., “In the recent three days, this person was responsive to my needs.”). One item in the behavior intention scale was dropped as it did not fit disputes with a specific person.
Results and Discussion
Tables 8 and 9 present the correlations and reliabilities of all measures and means and SDs by condition.
Correlations for Study 5.
Note. Reliabilities in parentheses; values in bold differ from 0 at p < .05.
Descriptive Statistics by Experimental Conditions for Study 5.
Main effects were analyzed using analyses of covariance (ANCOVAs) controlling for attachment security. Participants in the high responsiveness condition reported higher perceived responsiveness than participants in the low responsiveness condition, F(1, 196) = 23.09, p < .001,
In sum, Study 5 demonstrated that the effects of perceived responsiveness held when controlling for attachment security, allowing us to discount this alternative explanation. Furthermore, the present experiment increased generalizability in two ways. First, we replicated previous results using a responsiveness manipulation that asked about a specific person. Second, unlike Studies 1 to 4, which sampled undergraduate students who participated for course credit, the present study included participants with a larger age range (18–72). There were no interactions of attitude correctness or behavior intentions with attachment security (see Supplemental Material). Finally, secure attachment was associated with lower attitude correctness, r = −.33, p < .001, consistent with research showing that secure individuals are less prone than insecure individuals to defensive self-inflation. For example, Mikulincer (1997) found that secure individuals were able to maintain a stable view of the self by keeping more self-attributes, both positive and negative, in active memory. Insecure individuals, in contrast, were more likely to defensively reject negative self-relevant input. In our studies, claiming a higher degree of correctness would have implied certainty that the opposing point of view must be wrong, and would have required denying valid elements in that position.
Mini Meta-Analysis
In two of our studies the main effect on simultaneous accessibility became non-significant with a Bonferroni correction, as well as one main effect of behavior intention. Therefore, we conducted a mini-meta analysis (Goh et al., 2016). The results of a random-effect meta-analysis on simultaneous accessibility were: K = 3, and N = 618 were: d = .35, 95% CI = [0.18, 0.52], SE = .09, Z = 4.01, p < .001, with no heterogeneity across the studies, χ2(2) = 0.25, p = .88. The results of a random-effect meta-analysis on behavior intentions were: K = 3, and N = 672 were: d = .38, 95% CI = [0.21, 0.54], SE = .08, Z = 4.56, p < .001, χ2(2) = 0.06, p = .97. These results provide clear and strong support for Hypotheses 3 and 5.
General Discussion
Across five studies, we found that when people perceive others to be responsive, they become more open-minded, which in turn results in reduced attitude correctness (Studies 3–5) and increased simultaneous accessibility about both the positive and negative components of the attitude object. Moreover, people primed with high perceived responsiveness were better able to tolerate internal inconsistencies in their perceptions of the attitude object without feeling discomfort (i.e., subjective ambivalence). Finally, this process influenced their behavior intentions to act in an open-minded way. The current findings were obtained in controlled laboratory experiments, an online study, and a diary study, which together provide internal and external validity for our findings.
The current findings make several theoretical contributions to the literature. First, although previous work found that perceived partners’ responsiveness has various cognitive and emotional benefits for recipients (Reis et al., 2004), this research shows how it can also influence attitudes, and more so, attitudes unrelated to the people who are perceived to provide responsiveness. Some of the attitude objects were interpersonal (Studies 1, 4 and 5), while one attitude object, GMF (Studies 2 and 3), was completely impersonal. These findings indicate that several important features of attitude structure, such as attitude ambivalence and attitude correctness, are at least in part a direct reflection of the social environment. In this respect, our findings are consistent with previous work showing that people express more or less ambivalence depending on the perception that these expressions will gain positive social value for them (Pillaud et al., 2013).
Second, this work adds to and extends what is currently known about the link between affirming the self and attitude change. Existing research on self-affirmation shows that validating the self is one way to make attitudes more complex. For example, research about self-affirmation, which is defined as highlighting values or attributes that are favorable to the self (Sherman & Cohen, 2006), shows that it lessens self-serving biases (Steele, 1988) and allows people to better tolerate internal inconsistencies (Steele & Liu, 1983). Along with Reis et al. (2018), who showed that perceived partner responsiveness lessens self-serving biases, our studies show that interpersonal affirmations may have a similar effect. To be sure, although perceived responsiveness and self-affirmation share similarity in that they affirm a person’s self, they differ in the manner that they do so. In self-affirmation manipulations, people are their own source of validation (e.g., “What do I value”?), whereas in perceived-responsiveness manipulations, the source of validation is meaningful others. Thus, although responsiveness and self-affirmation research provide complementary perspectives, they should be considered conceptually distinct. Future studies should manipulate both constructs to determine if they have independent effects on attitude change. This possibility is interesting given that, to the best of our knowledge, self-affirmation has been tested within cognitive dissonance paradigms but not in attitude ambivalence research.
Our findings are consistent with early theorizing regarding the importance of unconditional positive regard (Rogers, 1980). Rogers theorized that when people feel that their communication partners are empathic and non-judgmental, they become more aware of previously unacknowledged aspects of self-views that contradict their conscious self-perceptions. Work about the effect of high quality listening on speakers’ attitudes found empirical support for Rogers’s theory (Castro et al., 2016; Itzchakov et al., 2017). Although responsiveness and listening are both forms of support, they differ in their level of abstraction, that is, responsiveness is a general and abstract term, whereas listening is associated with specific behaviors, such as paying attention (Ames et al., 2012). Future research integrating these different levels of abstraction will be useful.
In addition, Study 3 identified two important individual difference moderators of the effect of perceived partner responsiveness on open-mindedness. Specifically, we found that individuals with a high NFC were more likely to be open to alternative points of view following the experimental manipulation, whereas individuals with a high NFCC were less likely to be open. The former result suggests that perceived responsiveness may be effective as it allows effortful thinking about the attitude object. This interpretation is supported by the latter results. As individuals with a high NFCC tend to be intolerant of uncertainty and ambiguity (Webster & Kruglanski, 1994), our manipulation of perceived responsiveness may have been insufficient to induce them to think about the conflicting information in an open-minded way. Together, these results suggest that perceived partner responsiveness influences open-minded processing of attitude-inconsistent information primarily for people who are relatively willing to consider alternative points of view. It may take stronger manipulations of perceived responsiveness to influence those who are low in the NFC and high in the NFCC.
One final advantage of these studies is that throughout the studies, we ruled out alternative explanations for the effects of the perceived-responsiveness manipulation regarding attitude importance (Studies 1 and 2) and positive and negative emotions (Study 3).
Finally, this research also has practical implications. Recent public commentary has expressed concern about people’s lack of open-mindedness, and in particular, their unwillingness to expose themselves to information and perspectives that differ from their own views (e.g., Kolbert, 2017). Our research suggests that this tendency may reflect their perception of relatively low levels of responsiveness—that is, feeling understood and validated—in the social environment. This contention is consistent with long-standing ideas about the social-identity functions of attitudes and beliefs (Smith et al., 1956) and with research about the social functions of attitudes (Pillaud et al., 2013; Weaver & Bosson, 2011), and, more particularly, with the idea that people may hold fast to pre-existing beliefs when their sense of social support is relatively weak (Jost et al., 2009; Murray et al., 2020). If so, one way to enhance openness to diverging perspectives may be to provide responsive, empathic support prior to exposure to alternative positions (Itzchakov & Kluger, 2017b)
Limitations and Future Research
The current work has several limitations that should be noted. For one, the studies did not manipulate the mediators and thus we cannot be certain about the causal ordering of the variables in our mediation models, although we believe that the models tested are more theoretically plausible. The second limitation is that we used only a single manipulation of perceived responsiveness. Although other studies have found similar effects with varied manipulations, it remains to be shown that these other manipulations would also affect attitude structure. Another limitation to be considered in future work is whether individual differences related to perceived responsiveness (e.g., self-esteem, loneliness) might also moderate these effects.
The present research included only two-sided attitude topics. Hence, the results cannot be generalized to one-sided or extreme attitudes. Future research should explore this potential boundary condition. Another interesting avenue for future research would examine constructs related to, or affected by, responsiveness, such as positive evaluations of others, feelings of support, and perspective taking. Finally, future research should examine how actual (rather than perceived) responsiveness from a target person impacts attitudes and behavior.
Conclusion
People sometimes protect themselves from feeling insecure by resisting information that challenges their point of view. The current set of studies demonstrates that perceived responsiveness, an interpersonal variable unrelated to the attitude itself, can help people be more open-minded and tolerant of inconsistencies with regard to an attitude object. Many other social problems may be amenable to a similar approach—for example, prejudice and intergroup hostility, social inequality, and political divisiveness. In future research and applications, it will be important to further explore how strong and supportive interpersonal connections can be utilized to promote constructive approaches to the resolution of these and other societal conflicts.
Supplemental Material
Supplemental Material, Itzchakov_Online_Appendix - Perceived Responsiveness Increases Tolerance of Attitude Ambivalence and Enhances Intentions to Behave in an Open-Minded Manner
Supplemental Material, Itzchakov_Online_Appendix for Perceived Responsiveness Increases Tolerance of Attitude Ambivalence and Enhances Intentions to Behave in an Open-Minded Manner by Guy Itzchakov and Harry T. Reis in Personality and Social Psychology Bulletin
Supplemental Material
Supplementary_material - Perceived Responsiveness Increases Tolerance of Attitude Ambivalence and Enhances Intentions to Behave in an Open-Minded Manner
Supplementary_material for Perceived Responsiveness Increases Tolerance of Attitude Ambivalence and Enhances Intentions to Behave in an Open-Minded Manner by Guy Itzchakov and Harry T. Reis in Personality and Social Psychology Bulletin
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by a Grant Number 460/18 from the Israeli Science Foundation to the first author.
Notes
Supplemental Material
Supplemental material is available online with this article.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
