Behavioral Obligation and Information Avoidance

Asking the typical person why he or she avoids some piece of information will likely yield a response that can be summarized in one word: fear. But what do people fear? A recent review proposed that people do not necessarily fear the information itself, but rather the cognitive, affective, and behavioral consequences of learning the information [1]. The present study focuses on how the behavioral consequences of information lead to avoidance. To this end, we focus on one possible motivation for avoidance: learning the information might obligate undesired action.

Autonomy Concerns

We contend that people are motivated to avoid information that obligates undesired action at least in part because such information threatens a basic need for autonomy (i.e., the ability to determine one’s personal actions). People experience autonomy when they can select their behavior based on internal motivations (e.g., personal desires and values) rather than external motivations (e.g., guilt and obligation [2]). Information can threaten autonomy by obligating people to engage in a behavior that is not internally motivated. For example, a physician may tell a patient at high risk for diabetes to make diet changes such as giving up fried foods. The patient may perceive these changes as a threat to autonomy because they represent unwanted behaviors that are not internally motivated.

Researchers have characterized gaining and sustaining autonomy as a fundamental human desire [3], as an intrinsically motivated goal [4], as a key motive underlying behavior and cognition [5], and as necessary for well-being [6, 7]. Because a sense of autonomy is such a central aspect of the self, people naturally react defensively to threats to their autonomy. For example, people often respond to messages they perceive as constraining their behavior by behaving in ways that are directly counter to the message [8, 9]. By responding defensively, people reassert their autonomy.

Autonomy and Avoidance

In the present study, we examined a different type of defensive responding: information avoidance. We propose that people sometimes avoid learning information if they believe that the information could require them to engage in an undesired behavior, which is behavior that is “difficult, inconvenient, demanding, expensive, or unpleasant” [1] (p. 343). Although we know of no experimental evidence for heightened avoidance of information that may obligate undesired behavior, several self-report studies provide evidence consistent with this notion. When asked why they were avoiding genetic testing for an unborn child, expectant parents who had previously given birth to a child with a genetic defect cited fears that the results would prompt them to abort the pregnancy [10]. Further, in separate surveys of South African sex workers [11] and Belgian immigrants [12], many reported that they avoided HIV screening because a positive test result would obligate them to take undesired action. For the sex workers, the undesired action was giving up a source of revenue. For the immigrants, the undesired action was reporting their HIV status to the government and thus jeopardizing their citizenship application. Finally, the primary reason a sample of Nigerian women with a suspicious lump in their breast reported for not visiting a physician was fear that they would be required to have a mastectomy [13].

A limitation of these studies is that they rely on self-reports, which are often inaccurate [14]. Lacking is experimental evidence demonstrating greater avoidance when the behavior obligated by the information is highly (as opposed to moderately) undesirable. To address this gap, we conducted three studies in which we manipulated obligation experimentally by varying the undesirability of the behavioral consequences of the information. In all three studies, participants believed they could learn their risk for an enzyme deficiency based on their responses to a risk calculator. In Study 1, participants learned that anyone who received feedback indicating a high risk for the deficiency would be required to undergo a non-embarrassing (low obligation) or highly embarrassing (high obligation) follow-up examination. In Study 2, participants received no information about treatment (low obligation) or learned that treatment entailed taking pills for an unspecified period of time (high obligation). In Study 3, participants learned that treatment entailed taking pills for 2 weeks (low obligation) or for the rest of their life (high obligation). In all studies, we hypothesized that people would decline to learn risk information more often when doing so would obligate more undesirable behavior in response.

Of course, we are not the first to examine people’s tendency to avoid information. Research on selective exposure has long shown that people prefer to receive information that is consistent with their attitudes over information that is inconsistent [15]. However, whereas the selective exposure literature focuses on the choice between attitude-consistent and attitude-inconsistent information, we address the larger issue of the choice between receiving information and not receiving information. Moreover, researchers largely explain the preference for attitude-consistent information in terms of dissonance theory—people experience dissonance (an unpleasant arousal state) in response to inconsistency between their cognitions or behaviors [16]. Research has also proposed that defensive reactions, like avoidance, may be an attempt to regulate negative emotions [17] or to maintain personal self-worth [18, 19]. However, none of these accounts can easily explain why obligation might cause people to avoid information. Specifically, although information that obligates unwanted action may threaten autonomy, it does not clearly create cognitive inconsistency (i.e., dissonance), produce high negative emotion, or threaten self-worth. In short, the research we describe goes beyond research on selective exposure and defensive processing.

Study 1

Methods

Participants

Participants were 112 undergraduate women (M age = 19.0, SDage = 1.2) who participated in partial fulfillment of a research participation requirement.

Design and Procedure

When participants arrived for the experiment, an experimenter dressed in medical scrubs escorted them to work stations and told them that they would complete a survey for the university hospital assessing risk for a newly discovered disease. The experimenter then left the room and the computer-guided participants through the remaining procedures. After consenting to participate, participants watched one of two informational videos about TAA deficiency [20, 21], a (fictitious) condition modeled after endometriosis that ostensibly produces a problem with the body’s ability to process nutrients and that can lead to pain, infertility, and other physical complications. Participants in the high obligation condition learned that definitive testing for TAA deficiency required a cervical examination, whereas participants in the low obligation condition learned that definitive testing required a cheek swab.

We chose to use a fictitious disease rather than an existing one for several reasons. First, factors such as personal knowledge, family history, and perceived risk factors can dramatically influence people’s thinking about diseases, and we wished to eliminate the influence of these factors on people’s decision making. Second, using a fictitious a disease allowed us to hold constant disease characteristics such as severity, likelihood, controllability, and treatability, all of which could overwhelm the effect of our experimental manipulation of obligation on decision making. Third, using a fictitious disease allowed us to tailor the pertinent risk factors and the onset of symptoms so that they would seem relevant to a healthy sample of people. Finally, using a fictitious disease made it possible for us to manipulate obligation and thereby examine experimentally its effect on decision making, something that would be impossible with real diseases. In sum, using an invented disease permitted examination of our hypothesis in a carefully controlled situation.

Eighteen pilot participants indicated on an eight-item index the extent to which they viewed the two conditions as threatening to their autonomy. Example items include, Learning that I am at high risk would… (a) require me to engage in undesired behavior, (b) obligate me to spend time in a way I don’t want to, and (c) restrict my freedom to act as I wish. Participants responded to each item from 1 = strongly disagree to 7 = strongly agree (α for both conditions > .90). The pilot participants reported that the cervical exam posed a greater threat to autonomy (M = 4.9, SD = 1.4) than did the cheek swab (M = 4.3, SD = 1.1), t (17) = 2.37, p = .03, d = .56Footnote 1.

After watching the video, participants completed a (fictitious) risk calculator and then read that the computer could assess their lifetime risk for TAA deficiency based on their responses. They also read that if they chose to learn their risk and the calculator indicated that they were at high risk, they would be legally obligated by the state to undergo definitive testing for TAA deficiency within the next 2 weeks. Next, all participants chose between receiving and not receiving their risk-calculator feedback. Finally, participants completed items assessing “How serious of a condition is TAA deficiency?” (1 = not at all serious; 7 = very serious) and “How worried [they were] about developing TAA deficiency?” (1 = not at all worried; 7 = very worried). The experimenter then fully debriefed participants, probed for suspicion about the true nature of the study, and explained that TAA was a fictitious disease.

All procedures were reviewed and approved by the university Institutional Review Board. We carefully developed the procedures, including information about TAA and the instructions that the state would require definitive testing for high-risk participants, to ensure they were credible to participants. Supporting the credibility of our procedures, only four participants across all three studies reported doubts about the credibility of the information we provided. Data from these participants were removed prior to analysis.

Results and Discussion

As evident in Fig. 1, more participants opted to avoid learning their risk in the high obligation (cervical exam) condition (66 %) than in the low obligation (cheek swab) condition (45 %), X 2 (1, 112) = 5.20, p = .02, Φ = .22. These findings show experimentally that people avoid information that could potentially obligate undesired behavior.

Fig. 1
figure 1

Information avoidance

Full size image

We found no effect of condition on the extent to which participants worried about developing TAA deficiency or viewed the condition as serious, ts(110) < .24, ps > .81, r < .03, indicating that we did not inadvertently manipulate worry or perceived seriousness. Further, neither of these variables predicted avoidance behavior (rs < .11) suggesting that our effects are not due to differences in worry or perceived seriousness.

Study 2

Study 1 suggests that people are more inclined to avoid learning their risk for a medical condition when learning their risk could require an unpleasant examination. Of course, a cervical exam and a cheek swab differ in many ways, not all of which represent a threat to autonomy. For example, some women may want to know their risk but would prefer a cervical exam to be conducted by their own physician, rather than an unknown physician at the university hospital. In addition, a cervical exam is more invasive and unpleasant than having one’s cheek swabbed. To establish the robustness of our effect, we used a different manipulation of obligation in Study 2.

Methods

Participants

Participants were 82 undergraduate women (M age = 19.5, SDage = 1.4) who participated in partial fulfillment of a research participation requirement.

Design and Procedure

Study 2 was identical to Study 1 with two exceptions. First, we told all participants that definitive testing for TAA deficiency involved a cheek swab. Second, we varied the treatment obligation. Participants learned that treatment for TAA deficiency required enrolling in a pill regimen of undefined length (high obligation condition) or they learned nothing about treatment (low obligation condition). Thus, the low obligation conditions in Studies 1 and 2 are identical (i.e., they both require a cheek swab and do not describe treatment).

Pilot participants (N = 34) evaluated the high and low obligation conditions using the same eight-item index as in Study 1 (α for both conditions > .91). They reported that the pill regimen posed a greater threat to autonomy (M = 4.5, SD = 1.7) than did the unspecified treatment (M = 4.1, SD = 1.3), t (33) = 2.26, p = .03, d = .39.

Results and Discussion

As evident in Fig. 1, more participants opted to avoided learning their risk in the high obligation condition (67 %) than in the low obligation condition (43 %), X 2 (1, 82) = 4.14, p = .04, Φ = .23.

As in Study 1, we found no effect of condition on the extent to which participants worried about developing TAA deficiency or viewed the condition as serious, ts(80) < .66, p > .50, r < .08, indicating again that we did not inadvertently manipulate worry or seriousness. Further, neither of these variables predicted avoidance behavior (rs < .12) suggesting that our effects are not due to differences in worry or seriousness.

Study 3

Studies 1 and 2 demonstrate a robust increase in risk feedback avoidance in response to a threat to autonomy. However, both studies patterned TAA deficiency after endometriosis. Thus, we did not include men. In Study 3, we tested the generalizability of our findings by including men, by varying further our manipulation of obligation, and by changing the nature of the disease. Doing so ensures that our effect generalizes to men and is not specific to the symptom constellation associated with endometriosis.

Methods

Participants

Participants were 66 undergraduates (43 women, M age = 18.4, SDage = .74) who participated in partial fulfillment of a research participation requirement.

Design and Procedure

Study 3 was identical to Study 2 with three exceptions. First, the sample included men. Second, we modeled the symptoms of TAA deficiency after type 2 diabetes (e.g., exhaustion, inability to process certain foods, weight gain) rather than endometriosis, thereby making the symptoms relevant to men and allowing us to test the generality of our findings to a different symptom constellation. Third, we varied behavioral obligation by manipulating the duration of treatment for TAA deficiency. Participants learned that treatment for TAA deficiency required taking a pill daily either for 2 weeks (low obligation condition) or for the rest of one’s life (high obligation condition).

Once again, pilot participants (N = 34) evaluated the high and low obligation conditions using the same eight-item index (αs > .95) used in Studies 1 and 2. The pilot participants reported that a lifetime pill regimen posed a greater threat to autonomy (M = 4.6, SD = 1.7) than did a 2-week pill regimen (M = 3.8, SD = 1.7), t (33) = 4.33, p = .01, d = .74.

Results and Discussion

Because preliminary analyses revealed no difference in avoidance between men (30 %) and women (39 %) across conditions, X 2 (1, N = 66) = .50, p = .48, Φ = .09, we analyzed their responses together. As evident in Fig. 1, more participants opted to avoid learning their risk in the high obligation condition (52 %) than in the low obligation condition (21 %), X 2 (1, N = 66) = 6.55, p = .01, Φ = .32. Interestingly, levels of avoidance were lower in Study 3 than in Studies 1 and 2, perhaps because the symptoms of TAA in Study 3 seemed milder (exhaustion, inability to process certain foods and weight gain vs. pain, infertility, and physical complications).

As in Studies 1 and 2, we found no effect of condition on the extent to which participants worried about developing TAA deficiency or viewed the condition as serious, ts(64) < 1.7, p > .10, r < .11, indicating once again that we did not inadvertently manipulate worry or seriousness. Further, neither of these variables predicted avoidance behavior (rs < .22) suggesting that our effects are not due to differences in worry or perceived seriousness.

General Discussion

The results of three studies confirmed that the more onerous the behavior demanded by information, the more people were inclined to avoid the information. Specifically, more participants avoided learning their results from a risk calculator when bad news required a follow-up test that was described as more invasive (Study 1), when bad news obligated undergoing an unpleasant treatment regimen (Study 2), and when the duration of treatment was longer (Study 3).

Our results add to the small but growing experimental literature showing that people sometimes manage threatening information by proactively avoiding it [1, 22, 23]. The research is novel in that it demonstrates experimentally that people avoid information that may obligate undesired behavior, and our findings highlight the important role of autonomy concerns in information avoidance. Participants in our study were more inclined to avoid information when learning the information threatened their autonomy by obligating undesired behavior. As such, our results are the first to reveal that threats to autonomy can not only provoke psychological reactance [8, 9] but also prompt avoidance of health information.

Importantly, by replicating our effect using three different manipulations of obligation, we rule out several alternative explanations for our findings. In Study 1, one could argue that it was the noxiousness of the behavior rather than a threat to autonomy that lead to greater avoidance in the high obligation condition. However, Studies 2 and 3 replicated the effects of Study 1 using a different manipulation of obligation that did not involve undergoing noxious behavior. In Study 2, one could argue that it is a lack of information rather than low obligation that is reducing avoidance. However, the low obligation conditions in Studies 1 and 3 had explicit information about what behavior was required, suggesting that it was not simply a lack of information that produced our effects. Finally, we found no differences between the low and high obligation conditions in reports of worry or judgments of seriousness of the disease in any of our studies, suggesting that our manipulations did not inadvertently influence how serious participants found the disease or how worried they were about developing it. Further, we found no relationship between worry or seriousness and avoidance, suggesting that our manipulation, rather than worry or perceived seriousness, was responsible for the effects we observed.

We are careful to acknowledge limitations in our findings. For instance, it remains unknown whether behavioral obligations might influence avoidance of other types of information, such as whether one’s relationship partner is cheating, which can force an unwanted confrontation or breakup, or whether one has a sexually transmitted disease, which can force a change in personally preferred sexual practices. Further, obligation is just one of several factors that can lead to avoidance of information. Indeed, people are more inclined to avoid learning risk when a disease is described as uncontrollable [22, 23] or when it will produce undesired affect or challenge a cherished belief [1]. In many contexts, there may be multiple factors that make information threatening and prompt avoidance. For example, a diagnosis may not only obligate an unwanted response, but also make people anxious or challenge a view of self as healthy. In such instances, removing or reducing the obligation may not eliminate avoidance.

Overall, our study offers an important, initial step in understanding obligation as a motive for information avoidance. We show that people avoid information in response to multiple threats to autonomy. Likely, these types of threats to autonomy underlie avoidance of regular pelvic checkups and cervical cancer screenings for women who find Pap tests invasive and unpleasant [24], as well as avoidance of type 2 diabetes testing for people worried about the major lifestyle changes and unpleasant daily treatment necessitated by a diagnosis [25]. These types of avoidance represent a public health concern when early-stage detection of disease is critical for disease course and control [26]. Our results suggest that health messages attempting to emphasize the importance of screening or treatment should not overemphasize the behaviors required to diagnose or treat a disease, as doing so may increase psychological reactance and inadvertently decrease screening.