The impact of self‐presentational intimacy and attachment on depletion of the self's regulatory resources

Abstract

The current work examined the self‐presentation–self‐regulation relationship utilizing a design in which participants disclosed self‐information during an actual interpersonal interaction. By comparison, prior work relied on assessing participants’ intention to disclose. Our findings showed that making self‐presentation disclosures during an interaction depleted participants’ self‐regulatory resources as a function of attachment style and intimacy level of disclosures. We discuss how using an interpersonal paradigm clarifies a self‐presentation rather than decision‐making interpretation of earlier work's results, how deciding what to disclose (i.e., intent) and actually making disclosures can be considered distinctly different acts, and how reversing the variable relationship (i.e., opposite to prior work) has theoretical importance because it examines the inverse question that is implied, but not tested or demonstrated in earlier work.

Keywords:

• Self‐presentation
• Self‐regulation
• Attachment

To manage self‐presentations people attend closely to the images they communicate and attentively monitor how they are coming across to others. This type of consciously controlled self‐presentation process requires not only cognitive exertion but also self‐regulatory resources (Vohs, Baumeister, & Ciarocco, 2005). The goal of the current study is to examine the relationship between self‐presentation disclosures and the self's regulatory resources.

Our conceptual understanding of self‐regulation emerges from the self‐regulatory resource model (SSR) where self‐regulation is characterized as the capacity to “override an initial impulse to engage in a particular behavior and replacing it with another response” (Schmeichel & Baumeister, 2004, p. 87). Evidence from the SSR model suggests that the self's regulatory resources are in limited supply and can be partially depleted by the exertion of self‐regulatory efforts. The typical finding of resource depletion occurs when participants in a depletion compared to non‐depletion condition exhibit diminished performance on a subsequent self‐regulatory activity (Baumeister, Bratslavsky, Muraven, & Tice, 1998; Baumeister & Heatherton, 1996; Baumeister, Vohs, & Tice, 2007; Muraven & Baumeister, 2000; Vohs & Heatherton, 2000). This indicates that regulatory capacity is not just reduced concurrently, but rather depletion effects continue after the initial self‐regulatory task is completed (i.e., the self's resources remain depleted for a period of time). Hence, because initial acts of self‐regulation in one area cause regulatory capacity to be reduced, subsequent self‐regulation efforts in other areas may not be as effective. However, this reduction in self‐regulation capacity is short‐term, with a person's regulatory strength eventually returning to its prior level (i.e., after the passage of some amount of time replenishment of resources occurs; Tyler & Burns, 2008).

As part of the social interaction process, self‐regulation intersects with self‐presentational efforts in that there are many occasions where people may deliberately alter their self‐presentation disclosures. People engage in such efforts as a means to influence their interpersonal relationships. That is, self‐presentation efforts are often meant to affect the thoughts, feelings, and behaviors of others as applied and related to the self‐presenter. Vohs and colleagues (2005) proposed that such controlled self‐presentations are “self‐regulatory in nature, where people continuously monitor their own behavior as well as feedback from others, and subsequently utilize this information to adjust their self‐presentations to successfully convey a desired image of the self” (p. 632).

Although research examining the relationship between self‐presentation disclosures and the self's regulatory resources is sparse, there are two studies (Vohs et al., 2005) relevant to the current work. The logic of these studies involved prior work on the appropriateness of self‐presentation disclosures. Specifically, Vohs et al. posited that people monitor and regulate the information they disclose to maintain an optimal level of self‐presentational disclosure, which frequently involves a moderate level of intimacy. This level is typically considered the most appropriate and is viewed as the most likable because when relationships are first developing moderate intimacy indicates an individual's desire to foster a closer relationship, but does so without inundating the other party with overly personal information (Collins & Miller, 1994). By comparison, if people reveal overly intimate information too early in a relationship they can be viewed as needy or desperate, or if they withhold personal information they can be perceived as cold or untrusting (Cozby, 1973). Thus, to the extent that people's self‐presentations are effectively regulated, their disclosures should reflect a relatively moderate level of intimacy. However, people's disclosures may diverge in either direction from this moderate level if their self‐regulatory resources are impaired.

Coupling this logic with attachment theory, Vohs et al. (2005) argued that depletion of self‐regulatory resources would differentially influence self‐presentations as a function of a person's particular attachment style. Attachment is typically characterized as the attitudes people have for their adult intimate relationships, often described as the emotional and psychological connectedness that human beings feel for each other (Collins & Miller, 1994; Collins & Read, 1990; Hazan & Shaver, 1987). Theorists suggest that the close relationships people experience during early childhood may shape the manner in which they relate to others in adulthood. In general, three primary attachment styles have been identified: avoidant, ambivalent, and secure. These classifications broadly characterize different ways that people may relate to others. For instance, evidence suggests that “avoidant” people disclose at an impersonal level of intimacy, “ambivalent” people disclose at a much higher level of intimacy, and “secure” people disclose at a more moderate level of intimacy (Mikulincer & Nachshon, 1991). Thus disclosing at a moderate level of intimacy may require avoidant and ambivalent people to exert self‐control, and if their capacity for self‐regulation is impaired (e.g., depleted) the likelihood of disclosing at an inappropriate level may increase. By comparison, secure people may typically favor disclosures that feature moderate intimacy, irrespective of depletion levels.

Consistent with these expectations, Vohs et al. (2005) found that when sufficient regulatory resources were unavailable (i.e., depleted resource condition) “avoidant” participants’ self‐presentations involved low‐intimacy disclosures, “ambivalent” participants’ self‐presentations involved high‐intimacy disclosures, and “secure” participants’ self‐presentations involved moderate‐intimacy disclosures. By comparison, when regulatory resources were available (i.e., non‐depleted resource condition) all participants’ self‐presentations involved the optimal level of moderate‐intimacy disclosures. Thus, while the norm for self‐presentation is to reveal moderately intimate information about oneself, being in a state of depleted resources may diminish people's capacity to do so, which at times may result in less than optimal self‐presentation disclosures.

Although these findings are valuable, it is important to point out that the measurement used for self‐presentation disclosures involved participants deciding which discussion items from the Relationship Closeness Induction Task to ostensibly discuss at a later point (RCIT; Sedikides, Campbell, Reeder, & Elliot, 1998). Hence the design did not involve an actual interpersonal interaction where participants shared personal information with another individual. Rather, after completing an initial resource depletion task, participants were told they would have a conversation with another student (the conversation never occurred). However, first they would be given time to decide and plan which of the RCIT topics they would like to discuss. This last activity essentially involved a decision‐making task, and we know from prior work that when people engage in decision making it can subsequently deplete the self's regulatory resources (Baumeister et al., 1998; Moller, Deci & Ryan, 2006; Vohs et al., 2008). This suggests that decision making per se may require at least some level of regulatory resources. Thus, with Vohs et al. (2005), it is possible that the initial depletion of resources impacted participants’ capacity for decision making, rather than their ability to control self‐presentation disclosures. Although we believe this interpretation unlikely, we utilize an interpersonal interaction design that provides us the opportunity to reduce decision making as much as possible. In other words, participants only see discussion items at the immediate onset of the interaction and are required to begin discussing the topics immediately. Thus we propose that using an interpersonal interaction design will help to theoretically clarify and strengthen the results from earlier work, as well as increasing ecological validity.

Moreover, even if Vohs et al.'s (2005) results actually demonstrated that the initial depletion of resources negatively impacts self‐presentation disclosures, the data do not provide evidence that making such disclosures during an interaction will also deplete the self's regulatory resources. As noted earlier, participants in Vohs et al.'s design were provided ample time to think about their responses. However, during interpersonal interactions people do not typically exhaust prolonged periods of time to ponder each response or to compare various response options. Rather, people's responses are more spontaneous and are likely to involve habitual or automatic responding, which has been shown to typically consume fewer resources (Paulhus, Graf, & Van Selst, 1989; Paulhus & Levitt, 1987; Vohs et al., 2005). Thus it is possible that making self‐presentation disclosures during an interaction may not require a significant level of regulatory resources. However, we expect this is not the case, and suggest that making self‐presentation disclosures during an actual interaction will indeed require some degree of the self's regulatory resources. As outlined earlier, people monitor their interaction partner (e.g., verbal and nonverbal feedback) and often use such information to regulate and adjust their self‐presentation disclosures. They may even use these social cues to suppress a particular disclosure; overriding it with a disclosure they feel more fitting to the audience. Hence it is likely that when people make self‐presentation disclosures during an interaction it will deplete some of their regulatory resources.

In addition, recall that Vohs et al. (2005) showed if participants are already initially depleted their subsequent capacity to self‐present at a moderate intimacy level may be impaired. This would suggest that if people engage in a social interaction when their resources are already depleted their self‐presentation efforts may suffer. With the current study we follow in the tradition of many self‐regulation studies and examine the opposite perspective: how making self‐presentation disclosures (during an interaction) may lead to depletion. Reversing the relationship has theoretical importance because it broadens the research scope by examining the inverse question that is implied, but not demonstrated, by Vohs et al.'s work. That is, does actually making certain self‐presentations consume the self's resources, hence leaving fewer resources available for subsequent regulatory tasks? This reversed perspective also has practical value in that people frequently engage in situations where they are faced with conflicting normative demands, which may require them to override and replace their typical self‐presentations with ones that better match the social setting. As we suggested earlier, such self‐presentational adjustments may result in deleting people's regulatory resources.

In short, we posit that using an experimental paradigm that involves an interaction provides the potential to clarify and strengthen Vohs et al.'s results, hence providing a clearer and more advanced theoretical understanding of the relationship between self‐presentation disclosures and the self's regulatory resources.

Overview of the Experiment

In the current experiment we crossed attachment style (avoidant vs ambivalent vs secure) with three self‐presentation disclosure conditions. In two self‐presentation conditions participants discussed either high‐ or low‐intimacy topics; in the third condition participants were free to determine the intimacy level of the topics. The dependent variable, self‐regulatory resources, was measured as the length of time participants worked on anagram puzzles, which is frequently used as a resource depletion measure (puzzles were made unsolvable by deleting a key letter) (Baumeister et al., 1998).

Because it may require regulatory resources to respond in a manner inconsistent with one's attachment style pattern, we predicted that avoidant people in the high‐intimacy condition (vs low‐intimacy) and ambivalent people in the low‐intimacy condition (vs high‐intimacy) would spend less time persisting on the anagram task. Secure people typically self‐present at a moderate‐intimacy level, thus for them to self‐present at a high‐ or low‐intimacy level may not be as incongruent with their typical response pattern compared to avoidant and ambivalent people. Hence the amount of time secure people in the high‐ and low‐intimacy conditions persist on the anagrams is expected to fall between avoidant and ambivalent people's persistence for both the high‐ and low‐intimacy conditions.

Moreover, as outlined earlier, when people have sufficient regulatory resources available their self‐presentations are typically aimed at revealing moderately intimate information. Thus, in the self‐presentation free condition we expect most participants to discuss topics that convey a moderate level of intimacy. For secure people, discussing moderate topics would represent their attachment style response pattern and as such should not deplete regulatory resources. However, as Vohs et al.'s (2005) research suggests, it should require some self‐regulatory resources for avoidant and ambivalent people to disclose at a moderate intimacy level. Thus, for avoidant and ambivalent people, discussing moderate topics should consume some self‐regulatory resources. Hence, in the self‐presentation free condition, secure people's performance on a subsequent regulatory task should be less impaired compared to avoidant and ambivalent people's performance. Moreover, as noted above, it should consume fewer regulatory resources for secure people to discuss moderate topics. By comparison, it should require more resources when secure people engage in atypical response patterns in the high‐ and low‐intimacy conditions. Thus we expect secure people's regulatory task performance in the self‐presentation free condition to be less impaired than secure people's performance in both the high‐ and low‐intimacy conditions.

Method

Participants and procedure

Students (N = 148) at a large university received extra course credit for participation. Two students registered for each session; one was randomly assigned as the target participant and to one of three self‐presentation conditions (high‐intimacy vs low‐intimacy vs free).

Target participants completed a questionnaire to measure their attachment style (adapted from Hazan & Shaver, 1987). These items require participants to select a description that best fits how they feel concerning their personal relationships (because of time constraints we used this briefer categorical measure rather than a longer multi‐item measure). The current study revealed the following pattern: avoidant (29.70%, n = 44), ambivalent (29.70%, n = 44), and secure (40.5%, n = 60).

The experimenter explained that the researchers were examining how people work together to complete tasks that involve communicating shared information. Participants were also told that these tasks are best done when people “know a little something about one another”, so first they would have a “get‐acquainted” conversation. During the conversation they would alternately talk about themselves; participants believed the target participant was randomly chosen to talk first (only one conversation occurred). The researcher also explained that it is often difficult to know what to talk about when initially meeting someone, so they would be provided with a series of questions to discuss.

These questions were adapted from the RCIT, which consists of 29 items ordered from low‐, to moderate‐, to high‐intimacy disclosure topics (Sedikides et al., 1998).1 In the self‐presentation high‐intimacy condition participants were given 10 high‐disclosure topics; in the self‐presentation low‐intimacy condition participants were given 7 low‐disclosure topics. For the self‐presentation free condition participants were given the full RCIT, but were told because of time constraints they could only talk about 10 topics (they circled 10 items; 12 of the 29 items have been empirically categorized as moderately intimate).

Participants engaged in a 10‐minute conversation (the experimenter was not in the room), after which the target participant was directed to another room where he/she worked on the self‐regulatory task (i.e., anagram puzzles). They were told to recombine the scrambled letters to derive a commonly identified word, to work on the puzzles for as long as they wanted, and to notify the experimenter when they wanted to stop. The experimenter left the room and surreptitiously timed how long the target participant worked on the anagrams. Those still working at 20 minutes were stopped and their time recorded as 20 minutes.

After finishing the self‐regulatory task target participants completed manipulation checks to assess intimacy disclosure perception (1, not personal at all and 7, very personal) and anagram difficulty (1, not difficult at all and 7, very difficult). Upon completion, participants were dismissed.

Results

Manipulation checks

Results from a 3 (self‐presentation: high‐intimacy vs low‐intimacy vs free) × 3 (attachment style: avoidant vs ambivalent vs secure) ANOVA showed the anagram task was equally difficult across all conditions (M_grand  = 6.82, SD = .45), with no significant effects, Fs < 1.60. An additional ANOVA indicated that the perception of intimacy disclosure levels differed significantly across self‐presentation conditions, F(2, 139) = 15.63, p < .001, η²= .18, suggesting that the manipulation was effective. Specifically, planned comparisons indicated the conversation was significantly more personal in the high‐intimacy condition (M = 4.92, SD = 1.32) compared to the free condition (M = 4.02, SD = 1.35), t(145) = 3.22, p < .01, d = .67. The free condition, however, was more personal than the low‐intimacy condition (M = 3.35, SD = 1.49), t(145) = 2.40, p < .05, d = .47. There were no other significant effects, Fs < 1.40.

Moreover, in the free condition 92.90% of the RCIT items were from the moderate pool (M = 9.29, SD = .74), and the number of moderate items did not significantly differ across attachment styles, M _avoidant = 9.53, SD = .64; M _ambivalent = 9.20, SD = .78; M _secure = 9.16, SD = .76; p > .30.

Primary analysis: Self‐regulatory task persistence

Results from a 3 (self‐presentation: high‐intimacy vs low‐intimacy vs free) × 3 (attachment style: avoidant vs ambivalent vs secure) ANOVA revealed a significant main effect for self‐presentation, F(2, 139) = 3.09, p < .05, η²= .04, which was qualified by a significant interaction F(4, 139) = 10.43, p < .001, η²= .23 (attachment factor was not significant, p > .15). For sake of brevity we focus on decomposing the interaction.

Specifically, planned comparisons showed that in the high‐intimacy condition avoidant participants’ task persistence was significantly less than secure participants, t(139) = 2.61, p < .01, d = 1.05, who in turn showed less persistence compared to ambivalent participants, t(139) = 2.04, p < .05, d = .64. In contrast, in the low‐intimacy condition ambivalent participants’ task persistence was significantly less than secure participants, t(139) = 2.45, p < .05, d = .89. Secure participants, however, showed less persistence on the task compared to avoidant participants, t(139) = 2.23, p < .05, d = .62. In the free condition avoidant participants’ task persistence was significantly less than secure participants, t(139) = 2.44, p < .05, d = 1.13; ambivalent participants’ persistence was also significantly less than secure participants, t(139) = 2.35, p < .05, d = .84. Moreover, in the high‐intimacy condition secure participants’ task persistence was significantly less than secure participants in the free condition, t(139) = 2.84, p < .01, d = .98; likewise for the low‐intimacy condition where secure participants’ task persistence was also significantly less than secure participants in the free condition, t(139) = 3.10, p < .01, d = 1.20. Secure participants’ task persistence did not significantly differ in the high‐ and low‐intimacy conditions, p > .80. See Table 1 for means and standard deviations.

Table 1. Task persistence as a function of self‐presentation disclosures and attachment style

		Self‐presentation
		High intimacy	Low intimacy	Free
Mean task persistence (seconds)	Attachment style	M (SD)	M (SD)	M (SD)
	Avoidant	404.40 (183.33)	819.43 (409.42)	653.80 (158.54)
	Ambivalent	818.80 (312.35)	397.93 (286.16)	662.27 (276.95)
	Secure	637.20 (255.89)	618.62 (198.41)	874.47 (227.01)
	Total	621.84 (298.85)	612.94 (331.53)	741.96 (245.75)
Higher means indicate greater persistence.

General Discussion

Other than Vohs et al. (2005), research examining the relationship between self‐presentation disclosures and self‐regulation is non‐existent. However, even with those early studies the design did not involve an interpersonal interaction where people actually disclosed personal information. Rather the design involved having initially depleted participants decide and plan what disclosure topics to ostensibly discuss. Thus it is possible that depleting participants’ resources subsequently influenced decision making rather than self‐presentation disclosures. In part the current study followed an interpersonal interaction design to reduce the issue of decision making as much as possible. Using an interpersonal design also helps to place the current effect into a social context, in which people engage in self‐presentation to influence the quality of their interpersonal relationships. In other words, when people self‐present they are often trying to influence how others might come to think, feel, and behave toward them (i.e., the self‐presenter). Also, as noted earlier, we used an interpersonal context because it is empirically unclear whether making self‐presentation disclosures during an interaction will actually require regulatory resources. Moreover, where Vohs et al. first depleted participants’ resources prior to assessing self‐presentational intentions, with our design we examined the opposite perspective. That is, participants first made self‐presentation disclosures (during an interaction) and then their regulatory resources were measured.

As outlined in the introduction, we propose that the current results offer relevant contributions to the literature in three related ways that extend prior work. In what follows we offer a brief summary of each point. First, although we noted that for reasons of automaticity making self‐presentations during interactions might plausibly require fewer resources, our data suggest otherwise. Specifically, the results show that, for some people, making certain self‐presentation disclosures (i.e., intimate or non‐intimate) during an interpersonal interaction consumes their regulatory resources. This suggests that if people suppress and override a disclosure as a function of the audience, it may deplete their subsequent available resources. This also raises the question: if an interaction calls for people to persistently enact this suppress‐override process, how are people able to maintain sufficient resources to do so? Clearly we are speculating, but the question broaches an interesting line of inquiry for future research. Moreover, by utilizing an interpersonal paradigm the present findings provide strength and clarity to Vohs et al. (2005), supporting the interpretation that indeed their findings were likely related to self‐presentation disclosures rather than decision‐making efforts. Indeed, if the effects were simply due to decision making we would expect persistence time in the self‐presentation free condition to be more negatively impacted; however, the results showed that this was not the case. In a related manner, recall that Vohs et al. (2005) asked participants to report their self‐presentational intentions. We suggest that people's predictions of how they intend to act in various social situations are generally not the best predictors of how they actually do act. Indeed, deciding what to disclose (i.e., intent) and actually disclosing the information to another person may be considered distinctly different acts (Sheppard, Hartwick, & Warshaw, 1988). Thus we posit that the current results derived from an interaction setting not only help to confirm the self‐presentation‐self‐regulation relationship, they also increase the generalizability and ecological validity of the findings. Finally, we outlined in the introduction that reversing the variable relationship (i.e., opposite to Vohs et al.) holds theoretical importance, in part, because it broadens the research scope. It does so by examining the inverse question that is implied, but not tested or demonstrated, in earlier work. That is, does actually making self‐presentations during an interaction consume resources: the current findings suggest the answer is yes. By comparison, earlier work does not directly answer this question. In addition, given that self‐presentational efforts may represent a rather ubiquitous form of social behavior, empirically documenting that it consumes regulatory resources has pragmatic value.

Although the findings support the general hypothesis, there are limitations. Utilizing a social interaction was important in extending previous research, however the setting was rather artificial. That is, participants were given a specific list of items to discuss during their conversation. With future work one would do well to assess whether the availability of regulatory resources influences the degree to which avoidant, ambivalent, and secure people freely discuss high‐, moderate‐, or low‐intimacy items in the absence of direct instructions. Such a design would advance both the validity and generalizability of the current findings. In addition, although the current experimental paradigm permitted a natural interaction, we did not actually measure or record participants’ self‐presentational behavior. This limitation offers an important area for future research, particularly when combined with a design in which participants are not provided with direct instructions or restrictions on how they should self‐present.

In conclusion, the present study provides convincing evidence expanding both theory and knowledge of the self‐presentation–self‐regulation relationship.

Notes

¹Abundant evidence (e.g., Heatherton & Vohs, 2000; Sedikides et al., 1998; Vohs & Heatherton, 2001) indicates that these topics vary in intimacy levels; thus the RCIT provides an empirically tested procedure to manipulate self-disclosure intimacy levels