The Benefits and Obstacles to Perspective Getting

Damen, Debby; Pollmann, Monique M. H.; Grassow, Teri-Louise

doi:10.3389/fcomm.2021.611187

ORIGINAL RESEARCH article

Front. Commun., 28 May 2021

Sec. Psychology of Language

Volume 6 - 2021 | https://doi.org/10.3389/fcomm.2021.611187

This article is part of the Research Topic Perspective Taking in Language View all 18 articles

The Benefits and Obstacles to Perspective Getting

Debby Damen*

Monique M. H. Pollmann

Teri-Louise Grassow

TiCC, Department of Communication and Cognition, Tilburg University, Tilburg, Netherlands

In general, people tend to rely on egocentric projection when predicting others’ emotions, attitudes, and preferences. However, this strategy is less effective than the more obvious strategy of directly asking others what they feel, think, or desire (‘perspective getting’). In three experimental studies, we investigated how likely people are to ask for others’ perspectives, whether it leads to better predictions, and what factors impede perspective getting. In the first study, we let participants predict how happy another person would be with different money distributions. Only 26% of all people engaged in perspective getting, and it did not lead to better predictions. In the second study, we let people predict how expensive another person would think certain products are. The majority of people engaged in some form of perspective getting, but only 23% of all people did this thoroughly. Perspective getting did lead to better predictions. In the final study, we let people predict another person’s attitudes about a wide range of topics. Here, 70% of the people engaged in perspective getting and 12.5% did so thoroughly. Again, perspective getting led to better predictions. We found that confidence acted as a barrier for perspective getting. We also tested whether pointing out that perspective getting is the best strategy would increase perspective getting. We do not find a positive effect of this intervention. We discuss possible other interventions to increase people’s tendency to get rather than take perspective.

Introduction

How well do you know your partner or your friends? Do you think that you truly know their attitudes and their emotional reactions? Did you ask them about these things? Chances are that you did not and that you still feel confident about your ability to predict their beliefs and feelings. Oftentimes, however, your confidence will be misplaced. Your predictions will most likely be somewhat inaccurate because they are based on projection and too little on direct information.

Studies have repeatedly shown that we frequently overestimate our ability to infer what other people think, feel or desire, especially when we feel we truly know the other person (e.g., Swann and Gill, 1997; Scheibehenne et al., 2011; Eyal et al., 2018). Even though actively imagining what other people are thinking or experiencing should increase our interpersonal understanding of what goes on inside other people’s minds, research shows that we are rarely accurate about the inferences we make (e.g., Realo et al., 2003; Ames and Kammrath, 2004; Eyal et al., 2018). One of the reasons is that the ease by which our own beliefs and perceptions are accessible clouds our ability to acknowledge another person’s unique vantage point. Often, this causes us to overestimate the extent to which others view and evaluate the world similarly (e.g., Damen, et al., 2019a; Damen, et al., 2019b; Damen et al., 2020a; Keysar, et al., 1998a; Gilovich et al., 2000; Keysar et al., 2000; Keysar et al., 2003; Ross et al., 1977; Krueger and Clement, 1994).

Taking another person’s perspective is, therefore, a strategy of inferencing that is more prone to errors than other, more obvious, strategies such as actively inquiring about other peoples’ mental states. Strikingly, even though asking about other’s beliefs, feelings, and desires seems to be an obvious way to gain insight into other’s perspective, people do not feel more confident that getting their interlocutor’s perspective through conversation allows them to assess this person’s perspective better than when they take it (Eyal et al., 2018; see also; Zhou et al., 2017). Apparently, people do not see the added value of asking others about their perspectives. We wonder why this is the case. In this study, we set out to investigate how likely people are to get their interlocutor’s perspective. We investigate this question by examining people’s propensity to ask their interlocutors about their feelings (study 1), beliefs (study 2), and attitudes (study 3) before judging them. Additionally, we examine the obstacles people face when deciding to adopt a perspective getting strategy. In particular, we examine the extent to which people’s explicit knowledge that their perspective is not shared by others (study 2) and their confidence in their predictive accuracy (study 3) incites them to get rather than take perspective. We also investigate whether pointing out to people that perspective getting is the best strategy increases its use (study 3), and we explore the benefits this strategy has compared to perspective-taking on acquiring interpersonal accuracy.

Gaining Interpersonal Accuracy

Our ability to ascribe mental states onto others allows us to understand that our own view of the world might not be shared by those with whom we interact. This realization helps us to adjust our communication accordingly and helps us to collaborate, to empathize, and to build and sustain relationships. Yet, accurately perceiving another person’s thoughts and feelings is an extremely difficult task (Van Boven et al., 2000). The most obvious reason for this is that we cannot directly perceive the internal thoughts of others (Myers and Hodge, 2009). In fact, to guess what goes on inside other minds, we regularly employ different “tools” to gain this insight (for an overview, see Ames, 2005). For instance, we might employ evidence-based techniques, such as perceiving and interpreting others’ behavioral and emotional displays in context. In the absence of this direct behavioral “evidence”, we might use stored, stereotypical or idiosyncratic information about social categories to infer what others might think, feel, or do in a particular situation (e.g., Gopnik and Wellman, 1994; Galinsky et al., 2005; Kunda and Thagard, 1996). Alternatively, we might rely on our own beliefs and experiences to explain those of others, especially when we feel the object of our attention is highly similar to us (Kenny and Acitelli, 2001; Lerouge and Warlop, 2006). In other words, perceivers use their interpretations of others’ behaviors and emotions to infer other people’s mental states, or they use information about social categories (stereotyping), or about their own mental states (egocentric or social projection) to attribute to others (Ames, 2004). There is no clear consensus of which type of mentalizing strategy is used when and how. The only consensus there seems to be is that, whichever strategy is adopted, people generally strive to reach an accurate understanding of what drives other people (Ames, 2005).

One solution might be to use more reliable sources of information to guide our prediction of another person’s mental states (see also Ickes et al., 1990). Ample studies have evidenced that the more reliable (and relevant) information people receive about the other, the more reliable their inferences about this person’s perspective will be (Zhou et al., 2017). The most reliable source of information is almost always the target of interest. Even when people try to deceive, their deception can clue to others what actually transpires inside their minds. This means that people who directly experience or acquire information about another person’s actual thoughts, feelings, beliefs, or preferences are able to build a more reliable frame of reference that helps them to update their initial beliefs, and consequently, helps them to judge this person’s perspective more accurately (e.g., Camerer et al., 1989; Eyal et al., 2018; Marangoni et al., 1995; Porter et al., 2000; Thompson and De Harpport, 1994; Weingartner and Klin, 2005; West, 1996; see also volume of information in; Marangoni et al., 1995, and simulation in; Zhou et al., 2017). The most recent study showing the added value of this individuated “feedback” on interpersonal accuracy is the study by Eyal et al. (2018), who instructed married couples to get their partner’s perspective by conversing about each other’s attitudes before estimating them. The authors compared the interpersonal accuracy of these couples against the accuracy of the couples who were instructed to take their partner’s perspective by imagining themselves in their partner’s situation, and against a baseline condition in which couples did not receive any formal instructions. As expected, couples who talked about their attitudes were able to assess their partner’s perspective more accurately than those who took their partner’s perspective or participated in the control condition. Interestingly, couples who used the most effective inferencing strategy were not more confident about their predictive accuracy than others. In other words, couples overestimated the effectiveness of perspective taking compared to perspective getting, and, thus, had a misbelief about what the best strategy was to predict the attitudes of their partner. This misbelief could prevent people from selecting the most optimal strategy.

Obstacles to Perspective Getting

How likely are people to ask others for their views? Unfortunately, Eyal et al. (2018) did not test whether people spontaneously engage in perspective getting, nor were we able to find any other studies that did. The amount of information seeking in conversations may depend on the conversational setting, the conversational and interlocutors’ communicative goals, the number of conversational partners, and interlocutors’ non-verbal and verbal communicative reciprocity. Several fields show that people are able to extract information from another person for a certain explicit goal. For example, if a person is interested in the quality of a certain product, they may ask others what they think about it. If a person needs to negotiate a deal with another person, they may ask about their preferences (e.g., Schei et al., 2006). If a person is asked to find a way through a maze based on directions given by another person, they may ask clarification questions (Mills, 2014). In short, people do ask others for information. Although, even in these situations, the actual rate of questions is surprisingly low. What we are interested in are situations in which the only goal is to get a better understanding of what is on the other's mind. Do people directly ask others about their views to increase interpersonal accuracy? Research related to this question actually shows that in that case, people are not very likely to ask others for information. Below, we discuss several lines of research that give a first indication of whether people are inclined to seek information and, thus, to get rather than take perspective in conversation, even though this enumeration is, of course, not extensive.

First, research on negotiations shows the benefit of information sharing between interlocutors. When negotiators ask about their counterpart’s perspective, and their counterpart answers these queries truthfully, negotiators learn more about both the compatible and incompatible interests that are at stake, positively affecting the negotiation process and the integrativeness of the negotiated outcome (e.g., Pruit and Lewis, 1975; Thompson and Hastie, 1990; Weingart et al., 1993; Butler, 1999; Hyder et al., 2000; Adair and Brett, 2005; Schei et al., 2006; Elfenbein et al., 2010). However, dispositional and situational factors may affect the amount of information that is shared between negotiators. For instance, low perceived feelings of interpersonal trust (Butler, 1995; Butler, 1999), willingness to provide truthful information (Harsanyi, 1962; Bacharach and Lawler, 1981), personal accountability (De Dreu et al., 2006), egoistic motives (De Dreu et al., 2000), and the perception of power (Schei et al., 2006) all affect the extent to which negotiators seek and process information. These moderating factors cause that negotiators are not always inclined to seek or to provide relevant information, and this is especially unfortunate as information sharing is considered to be a reciprocal process (Thompson and Hastie, 1990; Weingart et al., 1993). It should be noted that these studies all focus on interlocutors with partly opposing interests. In our studies, we will focus on interlocutors whose task it is to predict the other’s perspective. So, the motives that hinder free information exchange in negotiations should play a smaller role here.

Another line of studies directs us to the doctor-patient communication setting. In these interactions, it should be obvious that the doctor asks their patient what is on their mind. However, several studies on how physicians talk to their patients show that not all patients are given the opportunity to explain their concerns, either because the physician does not start with an open question or because they interrupt their patient (on average within 23 s) (Beckman and Frankel, 1984; Marvel et al., 1999). Research has also shown that clinicians often do not check for understanding or assess their patient’s beliefs by asking whether the patient has more questions (White et al., 1994) and that doctors do not sufficiently acknowledge their patients’ emotional or social clues during an outpatient visit (Levinson et al., 2000). Most patients, therefore, are not able to voice their worries or their ideas, leading to misunderstanding (Barry et al., 2000). Hence, even in situations in which the other person is the best source of information, physicians do not engage in full perspective getting. In situations in which one’s own view is a better proxy, like when predicting the attitudes of a friend, perspective getting will probably even be less common.

The third set of studies can be found in the gift-giving literature and hints toward possible obstacles to perspective getting. In gift-giving, perspective getting would mean to ask the other person explicitly what kind of present they would like to receive. This seems like a good strategy, but people mistakenly believe that gifts chosen from a gift registry will be appreciated less than gifts the giver came up with themselves. Giving requested gifts is seen as less thoughtful and considerate by the giver (Gino and Flynn, 2011), and apparently, giving a gift from a registry undermines the giver’s motivation to show with the gift how well they know the other person (Ward and Broniarczyk, 2016). These findings show that in cases where predicting the other person’s perspective is seen as a challenge, people may feel like they should not engage in perspective getting because they feel it is not the right way to show their empathic skills.

Another challenge that might prevent people from getting another person’s perspective is their unawareness that their (private) perspective differs from someone else’s. One theoretical account in the perspective taking literature, for instance, claims that people who are aware that significant differences between themselves and the other exist are less likely to project their perspective onto this other person (e.g., Damen et al., 2019a; Decety and Sommerville, 2003; Mitchell, 2009; Mitchell et al., 2006; Mussweiler, 2003; Santiesteban et al., 2012; Savitsky et al., 2011; Todd et al., 2011; Simpson and Todd, 2017). Hence, people who operate on a false belief of similarity might not see the need for perspective getting, whereas those who are explicitly aware that their perspective is not shared by others might.

Finally, the perspective taking literature shows that people are often overconfident in their ability to predict others’ perspectives and that this overconfidence undermines their ability to learn and to improve (e.g., Einhorn and Hogarth, 1978; Eyal et al., 2018; Epley and Eyal, 2019; Ickes, 2003; Jacoby et al., 1994; Oskamp, 1965; Realo et al., 2003; Ryback, 1967; Swann and Gill, 1997; Thomas et al., 1997; see overconfidence effect in; Adams and Adams, 1961). People who are (over) confident in their perspective taking ability might, therefore, not feel the need to ask others for clarification or confirmation, making perspective getting less likely.

In general, the literature suggests that not all people will spontaneously engage in perspective getting. In the current paper, we present three studies in which we aim to test how likely people are to acquire information from the person they were trying to understand. We not only examined whether and the extent to which people inquired about their interlocutor’s feelings (study 1), thoughts (study 2), and attitudes (study 3), but also whether this perspective getting strategy led to more accurate predictions. We additionally focus on two proposed obstacles to perspective getting: people’s knowledge that their perspective is not shared by others (study 2), and their confidence in their predictive accuracy (study 3). We investigate the extent to which these obstacles influence the selection of the perspective getting strategy. In the final study, we also test whether pointing out to interlocutors that perspective getting is the best strategy increases its use.

Study 1

The first study was designed to investigate how many people use the option to ask the other person about their feelings and whether doing so is related to interpersonal accuracy. To this end, we created situations in which people differed in perspectives. Specifically, we asked people to predict how another person felt when this person received more or less money than themselves. There were four conditions in this study: perspective taking, receiving feedback, the option for perspective getting, and control. For this paper, we are only interested in how many people use the option for perspective getting, and whether this perspective getting increases interpersonal accuracy. We therefore only report the findings of this option for perspective getting condition and present the comparisons of the other conditions in the supplement material.

Method

Participants and Sample Size

The required sample size for a medium effect size was calculated for the full study (including all four conditions) using G*Power (version 3.1.9.2). This calculation showed that we would require at least 45 participants per experimental condition to obtain an alpha err probability of .05 and a power of 80. In total, 52 participants were randomly allocated to the option for perspective getting condition. However, 6 participants were excluded from the analyses because they experienced technical difficulties during the experiment. Our analyses are thus based on 46 participants in the option for perspective getting condition (26 women, 20 men, M_age = 22.52, age-range 18-42). Participants were recruited from the university campus, and they received course credits for their participation. The majority of participants was Dutch (N = 33, 71.7%). The 13 non-Dutch participants came either from other countries in Europe (N = 7, 58.3%), or from India (N = 1), Colombia (N = 1), Iran (N = 1), Vietnam (N = 1) and Thailand (N = 1). The majority had received a High School diploma (N = 24, 52.2%), followed by a Bachelor’s (N = 18, 39.1%) or Master’s degree (N = 4, 8.7%). The study was conducted in English, and all participants were proficient in the language of the study.

Procedure

Participants were invited to partake in a study that investigated their ability to predict another person’s feelings. Upon entering the lab, participants were seated in soundproof cubicles where they logged in into a computer to meet their partner via a synchronous, text-based chat (ChatPlat.com). This partner was a confederate of the study and followed a scripted text when responding to all participants’ queries. The dyads were not able to see each other, but could only communicate via the chat-box that was pulled up next to a questionnaire. The chat-box filled half of the screen and was, therefore, saliently present for participants to use to chat with their partner. Participants were deceived into believing that their partner was another student partaking in the study. The experimental leader instructed participants to start the experiment once all participants were set up. At the start of the experiment, participants were explicitly informed they should introduce themselves to their partner and that they could communicate with their partner at any time during the study.

The experiment was administered in Qualtrics™. First, participants answered demographic questions (age, gender, nationality, education level), followed by a practice round to introduce them to the procedure of the experiment. In the questionnaire, participants were exposed to thirty different scenarios that appeared in a randomized order in which the computer divided fictitious amounts of money between them and their partner. To elicit different emotions, the fictitious distributions were either advantageous for the participants (e.g., 7/3), for their partner (e.g., 3/7), or were equally beneficial for both (5/5). For each distribution, participants rated how happy they thought their partner was with the money distribution (1 = extremely unhappy, 7 = extremely happy), how confident they were in their prediction of the other’s happiness (1 = extremely unconfident, 7 = extremely confident), and how happy they personally were with the money distribution (1 = extremely unhappy, 7 = extremely happy). These scenarios were not tied in any way to the compensation participants got for participating, nor were participants led to believe they would.

Confederate Responses and Interpersonal Accuracy

The target’s (confederate) actual happiness ratings were collected prior to the study from 26 different participants who rated their own happiness score for each of the 30 money distributions (1 = extremely unhappy, 7 = extremely happy). For each distribution, participants also described in their own words how happy they were personally with the distribution. The answers to this question were used to produce a script the confederate could use in answering the participants’ queries in the main experiment. For each distribution, we averaged participants’ responses to construct a baseline score of happiness. Interpersonal accuracy was measured as the absolute mean difference between participants’ predicted ratings and these baseline scores, with smaller absolute differences representing larger interpersonal accuracy.

Results and Discussion

Only 12 out of the 46 participants (26%) engaged in some form of perspective getting. Out of this total, six people asked a general question, for example, whether the other person would be happier with an equal split or a distribution in which the other person got more. Four people asked about 1 or 2 specific distributions, and four people used a combination of those two strategies.

Those who engaged in perspective getting had an accuracy score of M = 0.93 (SD = 0.48) and those who did not had an accuracy score of M = 0.77 (SD = 0.32). The small sample size makes statistical testing unreliable, but even from the direction of the means it becomes clear that perspective getting did not increase accuracy in this case. We did run a bootstrapped t-test to be complete in our reporting, t(14.61) = 1.07, p = .302 [−0.14; 0.46], which showed that there was no statistical difference in accuracy between those who engaged in perspective getting and those who did not.

These results illustrate that people do not easily engage in perspective getting. In this particular setting, in which people also experienced themselves how the different money distributions make one feel, people may feel confident about using projection as a strategy. Projection is actually a good strategy in cases in which there is no reason to believe that the target is different from the predictor (Ames, 2005), so there was no urgent need to engage in perspective getting. In the next study, we therefore introduced a factor that would create a difference between predictor and target, namely the presence of the predictor’s explicit privileged knowledge.

Study 2

We made several changes in our second study. First, to test whether perspective getting leads to more interpersonal accuracy, we included a condition in which participants were explicitly instructed to engage in perspective getting and a control condition. Second, we used a scenario in which participants predicted how expensive a target would think different products were. This means that the perspective predictions are no longer about a target’s emotional reactions, but about a target’s price estimations. This was inspired by the negotiations literature in which correctly predicting the other parties’ ideas about the value of things is seen as an important factor to negation success (Chambers and De Dreu, 2014). Third, we introduced privileged knowledge, which is known to stand in the way of good predictions of naïve targets (e.g., Damen, et al., 2019b; Damen et al., 2020a; Epley et al., 2004; Keysar, 1994). In these studies, predictors with privileged knowledge find it hard to inhibit this knowledge during perspective-taking, causing them to overestimate the similarity between their own and the target’s perspective. Feedback that highlights this difference in perspectives, however, seems to decrease predictors’ overestimation of similarities in perspectives (Damen et al., 2021). Hence, we predicted that explicitly telling perceivers that their own knowledge was privileged to them and, thus, not known to the target, would increase the need for getting the target’s perspective.

Method

Participants and Design

For this study we used a 3 (Condition: instructed perspective getting, option for perspective getting, control) x 2 (Knowledge: no privileged knowledge, privileged knowledge) between-subjects design. For a medium effect size, the G*Power calculation (version 3.1.9.2) showed that we would require a sample size of at least 20 participants per experimental condition to obtain an alpha err probability of .05 and a power of .80. In total, we recruited 196 participants from the university campus to participate in a study in which they predicted another person’s beliefs (women = 126, men = 69, non-binary = 1; M_age = 22.80, age-range = 18–57). The majority of participants was Dutch (N = 136, 69.4%). The 60 non-Dutch participants came either from other countries in Europe (N = 55, 91.7%), or from Russian (N = 1), African (N = 1), Arabic (N = 1), Malay (N = 1), or Papiamentu (N = 1) speaking countries. The majority had received or followed a Bachelor’s degree (N = 167, 85.2%), followed by those who were following or had finished a Master’s (N = 17, 8.7%) or High School degree (N = 12, 6.1%). The study was conducted in English, and all participants were proficient in the language of the study. Participants were randomly distributed to one of the six conditions, resulting in 63 participants partaking in the instructed perspective getting conditions (30 received privileged knowledge), 66 in the option for perspective getting conditions (34 received privileged knowledge), and 67 in the control conditions (32 received privileged knowledge). Participants received course credits for their participation.

Procedure

Participants were invited to partake in a study in which they predicted their partner’s (a confederate named “Kim”) beliefs about the price of ten products that were supposedly up for sale in a web store. Kim is a gender-neutral name in the Netherlands, so by using this name we made sure that participants did not picture the other participant as having a certain gender. The products were presented to participants in a randomized order. Participants were seated in soundproof cubicles where they logged onto the computer to meet up with their partner for the first time via a synchronous, text-based chat (Chatplat.com). This chat box was pulled up next to the Qualtrics™ questionnaire during the entire duration of the study and filled half of the computer screen. Upon entering the questionnaire, participants received different instructions depending on the experimental condition to which they were assigned.

Perspective Getting Conditions

Participants in the instructed perspective getting conditions were told to contact their partner (the confederate) each time before estimating their partner’s price rating. The exact instructions were, “After seeing each advertised product, you will be asked several questions. Among these questions is what you think your partner thinks the price of the product is. Before answering that question, you have to talk to your partner using the chat function available. It is very important that you talk to your partner before you answer this particular question. Please, do not answer this question before you have spoken with your partner. When you ask your partner questions, you cannot ask directly for the price. Ask how he or she thinks or feels about the product.” Participants in the option for perspective getting conditions were explicitly instructed that they had the opportunity to converse with the other participant via the chat box. The exact instructions were, “You may contact the other participant. After seeing each product, you will be asked several questions about the product. Among these questions is what you think the other participant thinks the price of the item is. Before answering that question, you may talk to that other participant using the available chat function.” In both perspective getting conditions, participants also read: “When you ask something about a product, always mention which product you are talking about so you avoid miscommunication. You and your partner do not have to come to an agreement. Just ask what the other person thinks about the advertised product.” At the start of the experiment, participants in the perspective getting conditions were first instructed to introduce themselves to the other participant via the chat-box. The chat box was not available to participants in the control condition. These participants were only told that they had to estimate their partner’s price rating for several products.

Privileged Knowledge Conditions

In the privileged knowledge conditions, participants were additionally informed that they were going to receive inside information about the product’s actual prices. The exact instruction was, “We only tell you what the actual price of the products is. For every item that is for sale, you will receive the actual price of that product. Please note that ONLY YOU will receive this information. The other participant does NOT know the actual price of the products. Make sure you take a careful look at the product before answering the questions that follow it.” In addition, above each product, participants read again that the price presented above the item was privileged to them and not known to the other participant. All these additional instructions were absent in the condition in which participants did not receive this privileged information.

After having read the instructions, participants were asked to introduce themselves to their partner Kim, after which Kim introduced herself to the participants. The confederate introduced herself as a fellow student and informed the participants that her experience with estimating the prices of products was limited, and that she tended to look at the quality of the product before guessing its price. She ended her introduction by stating, “Sometimes the actual price matches my estimation, but sometimes it doesn’t”. Hereafter, participants were confronted with one practice trial to ensure all participants followed the condition-specific instructions. For each product, participants estimated their own and their partner’s price rating (a maximum of three digits were allowed), and how confident they were in their predictions of their partner’s perspective (1 = not confident, 11 = very confident).

Confederate Responses and Interpersonal Accuracy

The target’s (confederate) actual price ratings were collected prior to the study from one female participant who guessed the price of 15 unique accessories, electronics, and gadgets. After seeing each product, the participant additionally described in her own words what she thought of each product. The participant’s thoughts were used to create a script that the confederate used in answering participants’ queries in the main experiment. After collecting the target ratings, we selected 10 unique products to use in the main experiment. The prices for these products ranged from 20 to 95. Participants’ interpersonal accuracy was measured as the absolute mean difference between participants’ predicted price rating and the target’s actual price rating (smaller absolute differences representing larger accuracy).

Results and Discussion

Descriptive Statistics

The descriptive statistics are presented in Table 1. The accuracy scores were not normally distributed, and the variances were somewhat unequal. However, since there were at least 30 observations in every condition, the ANOVA should be fairly robust against these violations. To be sure, we also calculated the parameter estimates with robust standard errors to see whether the results are the same.

TABLE 1

TABLE 1. Means and standard deviations for accuracy scores (absolute mean difference score between predicted and actual price estimates) in Study 2.

Frequency of Perspective Getting

First, we were interested in how many people in the optional perspective getting condition actually used the chat function. Of the 66 people in that condition, only 15 (23%) asked the confederate something about all items. 25 (38%) people did not use the function at all or only for one item. Whether or not someone engaged in perspective getting did not depend on the privileged knowledge condition, χ²(1) = 0.20, p = .655, nor did the number of items asked about depend on the privileged knowledge condition (Mann-Whitney U = 567, p = .765). Those who had privileged knowledge, and were expected to have more difficulty in taking the other person’s perspective, did not engage in more perspective getting to compensate for the difference in perspectives.

The Effect of Perspective Getting on Accuracy

We first conducted a factorial ANOVA to test whether the perspective getting condition (instructed, optional, none) and privileged knowledge (yes, no) influenced the accuracy with which the target’s price estimates were predicted. We were not interested in the privileged knowledge condition for this analysis, so we included this factor mainly to see whether there was an interaction effect or whether the effect of perspective getting can be tested independently. Both perspective getting (F(2, 190) = 4.17, p = .017), and privileged knowledge (F(1, 190) = 8.61, p = .004) had main effects, but there was no significant interaction between perspective getting and privileged knowledge, F(2, 190) = 0.77, p = .46 (also no interaction parameter with robust standard errors was significant). We therefore conducted the more focused one-way ANOVA next with perspective getting as the sole independent variable. Given the unequal variances, we report the Welch statistic. For the contrasts, the comparison group was the condition in which no perspective getting was possible. There was a significant overall effect of the perspective getting condition on accuracy (F(2, 123.56) = 4.37, p = .015). In the control condition, the predicted price estimation was on average 36.73 euros (SD = 17.69) away from the target’s actual price estimation. In the instructed perspective getting condition, the difference was not significantly smaller (M = 31.96, SD = 16.10, t(127.87) = 1.61, p = .110), but in the optional perspective getting condition it was (M = 29.08, SD = 11.58, t(114.02) = 2.96, p = .004). We did not formulate a hypothesis for the difference in accuracy between the instructed and optional perspective getting conditions, so we did not perform a contrast for this. A post-hoc test with Bonferroni correction showed that they did not significantly differ (p = .864).

It is somewhat surprising that people who were instructed to get the other person’s perspective do not perform better than people in the control condition. It could be that the instructions felt artificial and that people were more comfortable with the option to get perspective. In our next study, we therefore use a more natural instruction. Instead of asking participants to get perspective for each item separately, we allow participants to discuss all items at once.

Next, we focused on the optional condition and tested whether engaging in more perspective getting (i.e. asking about more items) is related to accuracy. Because the number of items that was asked about was not normally distributed, we calculated the Spearman rank correlation. We found that the more items someone asked about, the more accurate that person was (there was a lower discrepancy between the predicted and the actual price estimates), ρ = −0.277, p = 0.024. The 95% bootstrapped confidence interval of the correlation coefficient ranged from -49 to -0.20. This is fairly wide due to the relatively large standard deviation in the data. Still, we can be quite confident that people who ask the target about their opinion about a product (not even the direct price estimation) are better atn predicting the target’s price estimation than people who do not do this.

Study 3

In our final study, we made another set of changes. First, instead of emotions or thoughts, we now let people predict a target’s attitudes. Second, instead of making predictions about an unknown target, the target was now a friend of the participant. Third, we tested whether increasing people’s awareness of the benefit of using the perspective getting strategy to predict their friend’s attitudes increases the use of this strategy. We expected this because even though people sometimes discount advice they receive, if it comes from a credible source they often follow it (Bonaccio and Dalal, 2006). To investigate people’s propensity to engage in perspective getting, and to explore the strategy’s impact on interpersonal accuracy, we conceptually replicated Eyal et al. (2018) design. Instead of romantic couples, we invited befriended dyads to partake in a study that was conducted online via an online video-conferencing tool. As in Eyal et al. (2018), we asked one party to predict their friend’s attitude on 21 opinion statements (Hoch, 1987). Besides allocating dyads to an instructed perspective getting and control condition, we also allocated dyads to a condition in which they had the option to converse with one another before predicting the target’s attitudes. Moreover, half of these dyads were made aware of the benefits of perspective getting over perspective-taking strategies. We expected that perspective getting would positively affect interpersonal accuracy. More specifically, we tested the hypothesis that people who engage in perspective getting (whether they are instructed or given the option to do so) will predict more statements accurately than people who cannot engage in perspective getting (control). Furthermore, we expect that making people aware of the benefits of perspective getting increases the use of perspective getting, and that, per consequence, people will predict more statements correctly. Finally, we test the hypothesis that people who are more confident that they can predict their friend’s attitudes are less likely to engage in perspective getting.

Method

Participants and Sample Size

The effect sizes found in Eyal et al. (2018) varied between medium and high effects. Eyal et al. (2018) allocated 25 participants to each of the experimental conditions, and our G*Power calculation showed that we would require at least 19 (high effect) to 45 (medium effect) participants per experimental condition to obtain an alpha err probability of .05 and a power of 80. In total, we were able to invite 160 befriended dyads to partake in a study that was conducted online via the online video-conferencing tool Zoom (Predictors: 49 male, 111 female, M_age = 24.96, age-range 18-57; Targets: 63 male, 97 female, M_age = 25.38, age range 13-68). On average, predictors reported being friends with their partner for a period of 8 years and 2 months (SD = 7.77 years). When dyads signed up for the study, we randomly allocated them to one of the four conditions. Our analyses are based on 37 participants in the control, 41 in the instructed perspective getting condition, 40 in the option for perspective getting with awareness intervention condition, and 42 in the option for perspective getting without awareness intervention condition. This means that per condition, we allocated at least 17 to 20 persons more per experimental condition than Eyal et al. (2018). Participants were recruited through various social media channels of the researchers. The majority of participants was Dutch (N = 142, 88.8%). The 18 non-Dutch participants came either from other countries in Europe (N = 15, 9.4%), or from Russian (N = 1), Indonesian (N = 1), or Vietnamese (N = 1) speaking countries. The majority had completed an academic education (N = 79, 49.4%), followed by a higher professional (N = 37, 23.1%), a vocational (N = 24, 15.0%), or a pre-university education (N = 16, 10%). Other education levels were senior general secondary school (N = 2), secondary school (N = 1), and undefined (N = 1). This study was conducted either in English or in Dutch, depending on the type of language participants felt most comfortable to converse in. To ensure participants had sufficient time to read the study’s information and sign their consent, dyads received an information letter in which they were assigned the role of either the predictor or the target one day for their participation. This also ensured that the time the dyads were left alone in the video-call would not be used to read the instructions. Predictors were informed that the next day, during the experiment, they were going to estimate their friend’s (the target) attitudes on 21 opinion statements (Hoch, 1987) that were attached to the email.

Procedure and Measurements

Dyads met each other and the experimenter in an online video-call that was administered in Zoom. Instead of conversing through a chat platform, dyads now conversed via video-call. In Zoom, the experimenter ensured that dyads could hear and see one another, that they were not seated in the same room, and that they were preferably logged on from another location before continuing the experiment. Then, the experimenter informed those participants who had been randomly assigned the role of the predictor that they were going to predict their friend’s attitude on the 21 declarative statements that were attached to their invitation email. These statements covered a wide range of topics, including politics, personality, nutrition, and social issues that were taken from Hoch (1987), and adapted to students’ current lifestyle and situation (e.g., “I like to pay contactless for everything I buy”, “The corona crisis is the biggest pandemic in history around the world today”). Hereafter, each participant received a personalized link to Qualtrics™ that they opened after their sound was muted and their webcams were turned off in Zoom. In this questionnaire, dyads gave their informed consent for recording their audio in Zoom. Subsequently, we measured predictors’ confidence in their perspective-taking ability by asking them to indicate: “how many of these statements do you think you will predict correctly?” (0–21). After that, they were instructed to return to the Zoom conversation.

Perspective Getting Conditions

Once back in the Zoom, dyads received further instructions based on the condition to which they had been randomly assigned. The instructions for both the instructed perspective getting and control conditions were directly replicated from Eyal et al. (2018). In the instructed perspective getting condition, predictors were explicitly told to take approximately 5 min to ask their friend about the extent to which they agreed or disagreed with each of the 21-opinion statements. Dyads received the following instructions, “In this experiment, we will examine your ability to accurately perceive your friend’s opinions to the 21 opinion statements. Please take approximately 5 min to ask your friend about the extent to which they agree or disagree with each of the topics on the list, trying to get a sense of the range of your friend’s opinions. Your partner might strongly agree with some statements, somewhat agree with others, and they may strongly disagree with others. Once you have done that, we would like you to use the information you got from your friend to predict the extent to which your friend would agree or disagree with these statements. Please open the list with the 21 opinion statements. I (the experimenter) will now leave the video room for 5 min and give you both time to discuss.” Next, the experimenter left the video call by muting her audio and turning her camera off for approximately 5 min to give the participants the opportunity to discuss the statements and engage in a conversation with each other. After 5 min, the experimenter rejoined the video call and instructed the participants to mute their audio and turn their camera off to make sure that consultation between the participants was avoided when they re-entered the online questionnaire. The explicit instructions to discuss the target’s opinion on each of the 21 statements were absent in the option for perspective getting conditions, and in the control condition. In the option for perspective conditions, the experimenter also left the video call by muting her audio and turning off her camera for approximately 5 min to give the friends the opportunity to engage in a conversation with each other. In the control condition, participants did not receive this 5-min opportunity to discuss the target’s attitudes. Instead, dyads were informed that “In this experiment, we will examine your (the predictor’s) ability to accurately perceive your friend’s (the target’s) opinions to the 21 opinion statements you already received in the information mail. We would like for you to use whatever strategy you think is best to predict your friend’s opinion to each of the statements.”

Awareness Intervention

The awareness intervention was administered only in the option for perspective getting conditions, and only half of the dyads received the intervention before they had the opportunity to discuss the target’s attitudes. In both option for perspective getting conditions, participants were informed about the different perspective-taking strategies people generally use when estimating another person’s opinion. They received the following instruction, “In this experiment, we will examine your (the predictor’s) ability to accurately judge your friend’s (the target’s) attitude on 21 opinion statements. You can use different strategies to predict your friend’s opinion. Let us summarize some of the strategies with an example using apples. Imagine you want to predict whether your friend prefers red apples over green ones. To infer your friend’s opinion, you can either 1) Use your own preferences to predict those of your friend: “I prefer red apples over green apples, so I think my friend will also prefer the red ones over the green ones too”; 2) Apply general knowledge about what other people who are similar to your friend like: “Most people prefer red apples, so I think my friend is most likely to prefer red apples over green ones too”; 3) Let situational circumstances guide your prediction: “Most of the apples in the supermarket are red, so I think my friend will prefer the red apples over the green ones”; or 4) Directly ask your friend which type of apples he or she prefers”. In the awareness condition, participants were additionally informed of the benefit of perspective getting (strategy 4) over the other three, “Now, we know from previous research that not all of these strategies help us to understand our friend’s perspective. The main reason for this is that, for the first three summarized strategies, we often overestimate the extent to which our friend thinks (or feels) like we do or like a stereotypical group does. Hence, if we use these strategies, we often fail to appreciate that our friend can have unique, different preferences. Therefore, research has shown that the most effective strategy to judge our friend’s perspective is to directly ask our friend about their opinions before we try to predict them”. After these instructions, the experimenter left the dyads alone for 5 min after saying and stressing they could use whatever strategy they thought was best to get a sense of their friend’s attitudes.

After 5 min (or immediately after the general instruction in the control), participants were redirected to the questionnaire. The experimenter made sure that participants’ audio was muted and that their webcams were turned off before they re-entered the questionnaire. The predictors predicted their friend’s (the target) opinions on each of the 21 opinion statements (1 = strongly disagree, 7 = strongly agree), followed by how confident they were about the accuracy of their prediction (1 = not at all confident, 11 = extremely confident). After predicting their friend’s attitudes, predictors reported their own attitude for each of the 21 statements in succession (1 = strongly disagree, 7 = strongly agree). At the end of the survey, predictors indicated: “how many of the statements do you think you predicted correctly?” (0–21), after which they indicated the length of their friendship with the target (in years and months), and filled in their demographics. Finally, participants were debriefed and thanked for their participation.

Coding of Perspective Getting

The conversations were transcribed and coded by three coders. For each statement, it was coded whether the predictor asked the target about their attitudes or not. All data were coded by two coders and disagreements were solved by discussion. There were no cases in which this could not be accomplished.

Interpersonal Accuracy

As in the previous studies, interpersonal accuracy was measured as the absolute mean difference between predictors’ estimation of a target’s attitude and the target’s actual attitude, with smaller absolute mean differences representing larger accuracy.

Results and Discussion

Frequency of Getting Perspective

First, we were interested in how many people in the optional perspective getting conditions asked their friend about the statements. Of the 80 people in those conditions, 56 (70%) engaged in perspective getting by actively asking information about at least one statement. Ten people (12.5%) asked something about all statements. On average, participants asked about 8.75 of the 21 statements. Remember that it was made very obvious to people that they were allowed to talk about the statements, and they were left alone in the online meeting. It is therefore quite remarkable that still, 30% of the people did not start talking about the statements, and that, on average, not even half of the statements were discussed and checked.

Next, we tested whether the awareness intervention increased the use of perspective getting. We conducted a chi-square test with the intervention (yes/no) as independent variable, and perspective getting (yes/no) as dependent variable. There was no significant relationship between the intervention and whether people engaged in perspective getting, χ²1) = 0.95, p = 0.329. We also conducted a t-test with the intervention (yes/no) as independent variable, and the number of statements asked about as dependent variable. The number of statements was not normally distributed, so the p-value may not be reliable, and we will rely on the bootstrapped confidence intervals to decide whether there is a significant difference. We found that people who were informed that perspective getting is the best strategy asked about 10.33 (SD = 7.92) items, and people who did not get that information asked about 7.18 (SD = 7.88) items. This difference was not significant (t(78) = 1.78, p = 0.078, d = 0.40), as the confidence interval ranged from −0.4 to 6.5. Thus, our intervention, in which we told participants that perspective getting is a good strategy, did not significantly increase perspective getting.

The Effect of Perspective Getting on Accuracy

The descriptive statistics of the accuracy scores are presented in Table 2. All assumptions for parametrical testing were met. We first conducted a one-way ANOVA to test whether the perspective getting condition (instructed, optional (with intervention), optional (without intervention), control) influenced the accuracy with which the friend’s attitudes were predicted.

TABLE 2

TABLE 2. Means and standard deviations for accuracy scores (absolute mean difference score between predicted and actual attitudes) in Study 3.

Based on our hypotheses, we conducted three planned contrasts: 1) the three perspective getting conditions against the control condition, 2) the instructed condition against the optional conditions, and 3) the optional condition with intervention against the optional condition without intervention. There was a significant overall effect of the type of condition on accuracy (F(3, 156) = 12.53, p < .001). The first contrast was significant, t(156) = 4.81, p < .001, ds = 0.90. People who could engage in perspective getting were more accurate than people who could not engage in perspective getting. The second contrast was also significant, t(156) = 3.54, p = 0.001, ds = 0.68. In line with our hypothesis, we found that people who were instructed to engage in perspective getting were more accurate than people who had the option to engage in perspective getting. The third contrast was not significant, t(156) = 1.37, p = 0.173, ds = 0.30, showing that there were no significant differences in interpersonal accuracy between the people who were made aware of the benefit of perspective getting and those who were not.

Next, we focused on the optional condition and checked whether engaging in more perspective getting is related to differences in accuracy. Because the number of statements that was asked about was not normally distributed to performed a Spearman rank correlation. We found that people are more accurate (the discrepancy between their prediction and their friend’s actual attitude is smaller) if they ask about more statements, ρ = −0.70, p < .001. The 95% bootstrapped confidence interval of the correlation coefficient ranged from −0.78 to −0.60. This is a relatively narrow interval and the lower limit is far from zero. Based on this we can be quite confident that people who ask their friend about a larger number of their attitudes are better at predicting these attitudes than people who use this opportunity to a lesser extent.

Confidence as a Barrier to Perspective Getting?

Finally, we tested whether people who are more confident that they can predict their friend’s attitudes are less likely to engage in perspective getting. Again, we only did this for people in the optional condition. We used the estimation given by the participant at the start of the study of how many items they would answer correctly as independent variable and the number of statements that they asked about as dependent variable. Because the latter variable was far from being normally distributed (with two peaks at 0 and 21), we conducted a Spearman’s rank correlation test. We found a significant negative correlation, ρ = −0.22, p = 0.047, 95% CI [−0.44, −0.01], indicating that people who thought that they knew more of their friend’s attitudes asked about fewer of the statements. Confidence was not related to accuracy, ρ = 0.04, p = 0.70, 95% CI [−0.19, 0.28], so the lower engagement in perspective getting was not justified. This null effect is intriguing because it suggests that there is another factor that is related to confidence and accuracy. If confidence is related to perspective getting, and perspective getting is related to accuracy, confidence should also be directly related to accuracy. There must be another variable that cancels out the effect. This could be, for example, actual knowledge. If actual knowledge is positively related to confidence and accuracy its effect would be opposite to the effect of perspective getting, which is negatively related to confidence and positively to accuracy. In future studies, it may therefore be interesting to include actual knowledge.

Insights From the Conversations

When coding the conversations, we noticed that there was a certain reluctance to engage in perspective getting. Some people thought that it was too straightforward to simply ask the other person about their attitudes, “The easiest now is that I ask all questions, but I think that is a little bit stupid.” Other people thought perspective getting was redundant. “I knew all of this before you said it” or “Ok, do we need to discuss more or not? Because I think that I can predict the rest of it pretty well.” Moreover, others claimed that they just did not want to know, but wanted to do the prediction on their own:

P: “After the corona crisis we should keep working from home if possible.” I was wondering, what would you think about this?

T: Yes, do you want to know?

P: Yes, I don’t know actually. I don’t want to know it, but it is a question that I can’t … No, I don’t want to know. I will answer it for you.

P: Or are we supposed to…. Are we the typical case that we are so naive and think we can predict each other? That even if we sit together for 5 min and can ask all questions, that we still don’t do it? That we still think we know better. I think that’s what we are.

What we also noticed is that dyads took ample time to discuss the validity of each of the previously introduced techniques to gauge another person’s perspective. In addition, even the targets were often very confident that their predictor would be able to take their perspective on their own:

T: But you can accurately guess my opinions, right?

P: Yes, absolutely. But, look, she (the experimenter) wants me to use a technique. Though, I think I can answer all those questions for you. But, I have to use a strategy, so I'm going to do it with one of those four techniques.

T: But with the latest technique, then you can just ask me what I would do?

P: Yes, I think so, but let's not do that.

T: No.

Most predictors even used projection as a strategy while discussing the items with their friend:

P: “I would rather live in a big city than a small town.” Yes. “The university has provided good education for homeschooling during the corona crisis.” That's an ambiguous question. It depends on the professors and everything. But, okay, in general.

T: If you want to ask me, I could tell you what I think.

P: I think you would agree.

T: Alright. Okay.

Another Example

P: “I easily make contact with other people.” Yes, absolutely. “Information from advertisements helps me make a better purchasing decision.” Ehm…

T: Hey, I have to answer that.

These quotes show that there are many reasons why people do not engage in perspective getting, even after it was pointed out to them that it is the best strategy.

General Discussion

In this study, we investigated the likelihood of people getting their interlocutor’s perspective. We examined this by investigating people’s proneness to ask about their interlocutor’s feelings (study 1), beliefs (study 2), and attitudes (study 3) before predicting them. Furthermore, we explored whether the perspective getting strategy increases interpersonal accuracy and examined the barriers to the use of perspective getting. In this line of research, we only examined explicit perspective differences (study 2) and one’s confidence (study 3) as possible reasons for people to (mis)select the most appropriate strategy to judge other people’s perspectives. However, the list of obstacles to perspective getting might be extensive, and offers, therefore, an important starting point for future studies.

Frequency of Perspective Getting

Our findings showed that perspective getting is a technique that people do not readily use. Results from the first study indicate that when predicting another person’s feelings about money distributions, the majority of people will rely on their own judgment rather than asking the other person. Moreover, those who did engage in some form of perspective getting did not do so thoroughly enough to be beneficial to interpersonal accuracy. Similar results were found in the second study. When predicting another person’s price estimates, the majority of people did engage in some form of perspective getting, but only 23% of all people did this thoroughly by asking their interlocutor’s perspective for all of the items. In the third study, participants predicted their friend’s attitudes. In this final study, we found that, although the majority of people did engage in perspective getting, again only a small number did this thoroughly. On average, people inquired about their friend’s opinion on less than half of all the statements they were able to discuss and check.

In general, our findings are in line with previous research suggesting that people will often choose their own intuition rather than other, non-intuitive alternatives when gauging others’ interpretations (Denes-Raj and Epstein, 1994; Kirkpatrick and Epstein, 1992, see also “intuition bias” in; Simmons and Nelson, 2006). People seem to prefer using their own judgments rather than using more effective (and more obvious) methods of gaining interpersonal accuracy, such as getting the other person’s perspective by asking about it.

But why is perspective getting not an attractive strategy? There are several explanations for our findings that people do not use it thoroughly. The first one lies in the methodology of the studies. In two of the studies, confederates were used to represent either an average target person (study 1) or an actual person (study 2) whose perspective participants were asked to predict. There are pros and cons to each method used. The disadvantage of the ‘average response’ is that the prediction is not about an actual person. We took away that concern by using an actual person in study 2, and the confederate in that study used that person's actual responses to respond to all participants’ queries. Still, the setup in the first two studies is a bit artificial. Also because in study 2 participants were not allowed to ask for the price directly. That is why we used real friends and real conversations in the final study. It is still not a complete ‘real-life’ situation, because one rarely has to know another person’s exact attitudes. This setup may have felt like a game to participants and perspective getting may have felt like cheating. To take away that concern, we stressed that perspective getting is a valid strategy, but one explanation for the incomplete use of it may still be that it did not ‘feel right’.

The second reason people may be reluctant to use non-intuitive techniques could be that they are not aware of the benefits other strategies have on increasing interpersonal accuracy. Eyal et al. (2018) found that people overestimate the effectiveness of perspective taking compared to perspective getting and that they, therefore, have a misbelief about which strategy was better for predicting the attitudes of their partner. Eyal et al. (2018) suggest that this misconception may be preventing people from selecting the most optimal strategy–getting a person’s perspective. We tested whether pointing out that perspective getting is the best strategy would increase its use, but we found no indication for this. Thus, people’s reluctance toward using perspective getting cannot be overcome by simply informing people of its merits.

Interestingly, results from our study indicate that even when people do engage in perspective getting they do not do so in depth. This suggests there may be a limit on how willing people are to ask continuously about other people’s perspectives. Fiske and Taylor (1984) suggest that people are, by design, “cognitive misers”. Previous research shows that we naturally use ‘cognitive shortcuts’, and although these shortcuts sometimes lead us to wrong answers, they tend to provide useful answers often enough to be beneficial and utilized by default in most circumstances (Judd and Park, 1993). Although interpersonal understanding is an important concept in close relationships and a predictor of relationship satisfaction (Reis, et al., 2017), it is oftentimes not based on having actual knowledge about the other person (Pollmann and Finkenauer, 2009). It is often sufficient to give the partner the feeling that they are understood by making validating statements. Therefore, it may not always be necessary to try to achieve a perfect understanding of others’ feelings and attitudes. In situations like the ones in our studies, perspective getting may take too much effort in relation to a relatively little gain, but in situations in which accurate understanding is more important, like in conflict situations, it may be more valuable. Future studies should investigate whether people are more likely to engage in perspective getting if more is at stake.

Another explanation for our findings may be found in the gift-giving literature. Asking what gift someone would like, or simply buying a gift from a registry, is not only thought of as less thoughtful by the giver, but also diminishes the giver’s opportunity to show how well they know the other person (Gino and Flynn, 2011; Ward and Broniarczyk, 2016). Perhaps predicting another person’s perspective is seen as a challenge, causing people to avoid perspective getting as they feel it is the wrong way to show their empathic skills. It could be suggested that our findings reflect those found in gift-giving literature. In study 3, insights from the conversations showed that some people did not want to know their interlocutor’s perspective, but wished to predict it on their own. It could be argued that the ability to guess someone’s opinion accurately could be viewed as more beneficial to the relationship than simply asking - similar to the way people feel about predicting what gift would suit the receiver rather than asking what they want. Our research did not ask follow-up questions about why people used a specific method. Future research may wish to explore the reasons people give for not wishing to engage in perspective getting. After all, a good way to understand the mind of others is to simply ask. This could be done through qualitative research and investigating if there is a common theme or reason that people give when they do not ask their interlocutor about their perspective.

Finally, the setup of our study may have reduced perspective getting. Our first and second study were conducted in English, which was not the native language of our participants. Processing a second language is more effortful than processing a native language (e.g. Ardila, 2003). It may be that our participants’ working memory capacity was reduced and there was less capacity left to engage in perspective getting. Relatedly, research on the foreign language effect shows that decisions made in a foreign language differ from those made in the native language (Keysar et al., 2012), although these findings remain somewhat unclear (Hayakawa et al., 2019). So the results of Study 1 and Study 2 may underestimate people’s tendency to engage in perspective getting. It should be noted, however, that all our participants were fluent in English and they use it on a daily basis because their study program is English. Also, in the third Study participants were allowed to have the conversation in Dutch, and we still find low to moderate levels of perspective getting there. In any case, more studies are needed to investigate how likely people are to get perspective in different context and how we can stimulate it.

The Effects of Perspective Getting on Accuracy

The findings from our studies (2 and 3) showed that people who engaged in perspective getting were more accurate at predicting others’ price estimates (study 2) and attitudes (study 3) than those who did not engage in perspective getting. Our findings thus support previous research proposing that perspective getting leads to more interpersonal accuracy than other intuitive techniques (Eyal et al., 2018).

We did find a different effect of instructed perspective getting on interpersonal accuracy between study 2 and 3. In study 2, instructed perspective getters were not more accurate in their predictions than those who were given the option to do so. This in contrast to study 3, where predictors instructed to get their friend’s perspective for all statements were more accurate than those who could choose to discuss the statements. We speculate that this difference occurred because of the somewhat artificial chat-based perspective getting instructions in study 2. Whereas predictors were instructed to get the target’s perspective for each item separately in study 2, dyads in study 3 could freely discuss each attitudinal statement at the start of the experiment. Of course, this assumption can only be tested by comparing different perspective getting procedures in one experiment.

In our study, interactions took place online via synchronized chat conversations and online video meeting platforms. This produced a unique insight into the effectiveness of different strategies to obtain interpersonal accuracy without face-to-face interactions. Our research shows that perspective getting is still effective in leading to more accurate results, even when communication is done via online modes, where the transmission of non-verbal and social cues (study 1 and study 2), or direct eye-contact (study 3) is limited. We encourage future researcher to examine how effects of perspective getting on accuracy are related to communicative settings in which interlocutors are able to rely on a variety of cues to build interpersonal report, and simultaneously or consequently, interpersonal understanding (e.g., Damen, et al., 2020b).

Confidence as a Barrier to Perspective Getting

In this study, we also tested whether people who are more confident that they could predict their friend’s attitudes would be more or less likely to engage in perspective getting (study 3). As expected, the results showed that people who were confident they knew more of their friend’s attitudes asked about fewer of the statements. However, confidence was not related to accuracy, so the lower engagement in perspective getting was not justified. This finding supports previous research that indicated that people are often overconfident in their ability to predict others’ perspective, and that this overconfidence prevents the willingness to learn or to improve (Eyal et al., 2018; Epley and Eyal, 2019; Ickes, 2003; Jacoby et al., 1994; Oskamp, 1965; Realo et al., 2003; Ryback, 1967; Swann and Gill, 1997; Thomas et al., 1997; see overconfidence effect in; Adams and Adams, 1961). Therefore, this study shows that (over) confidence acts as a barrier to perspective getting, and the less people asked, the less accurate their inferences about their interlocutor’s perspective were. What we still do not know is whether perspective getting reduces interlocutors’ overconfidence once they learn more about their communicative partner’s perspective. Further studies could examine how likely people are to get their interlocutor’s perspective after receiving feedback that makes them lose confidence in their ability to accurately judge other minds.

Conclusion

This study has revealed that although directly asking one’s interlocutor about their perspective is a good strategy to increase interpersonal accuracy, this perspective getting is not readily adopted. Most people do not ask others in detail about others’ perspectives, not even when they are clearly given the option to do so, or are explicitly advised on the benefits of doing so. This hesitation to use the perspective getting strategy highlights that there may be many barriers to perspective getting and interpersonal accuracy as a whole.

Data Availability Statement

The datasets presented in this study can be found here: https://osf.io/xt3n6/?view_only=df80ccd4ea3b4eccaaa2e97c0078e91f.

Ethics Statement

The studies involving human participants were reviewed and approved by the Ethical Review Board (ERB) of the Tilburg School of Humanities of Tilburg University. The participants provided their written informed consent to participate in this study.

Author Contributions

All authors contributed to writing the manuscript. DD planned the studies and supervised the data collection. MP analyzed the data. T-LG contributed to the data collection. All authors approved the final version of the paper.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgments

We would like to thank Giordano Tuvo, Saya Hoelasie, Tessa Guntlisbergen, Yvette Laros, Lotte Boone, and Kim Schneider for their significant input while conducting (parts of) the studies that are presented in this paper.

Supplementary Material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcomm.2021.611187/full#supplementary-material

References

Adair, W. L., and Brett, J. M. (2005). The Negotiation Dance: Time, Culture, and Behavioral Sequences in Negotiation. Organ. Sci. 16, 33–51. doi:10.1287/orsc.1040.0102