A Review of the Peak-end Rule in Mental Health Contexts

Abstract

The peak-end rule, a memory heuristic in which the most emotionally salient part of an experience (i.e., peak) and conclusion of an experience (i.e., end) are weighted more heavily in summary evaluations, has been understudied in mental health contexts. The recent growth of intensive longitudinal methods has provided new opportunities for examining the peak-end rule in the retrospective recall of mental health symptoms, including measures often used in measurement-based care initiatives. Additionally, principles of the peak-end rule have significant potential to be applied to exposure-based therapy procedures. Additional research is needed to better understand the contexts in which, and persons for whom, the peak-end rule presents a greater risk of bias, to ultimately improve assessment strategies and clinical care.

1.1. Introduction

With a seemingly endless stream of information and stimuli presented to us, human memory cannot store and retain complete information from every historical moment—experiential memory declines quickly as a function of time and eventually transitions from feelings (i.e., episodic memory) to beliefs (i.e., semantic memory) [1]. Since we cannot perfectly recall our past our experiences, we often rely on heuristics to form our summary experiences and perceptions, though there are often discrepancies between recall and actual experience, referred to as the ‘memory-experience gap’ [2]. The peak-end rule refers to the memory heuristic in which the most emotionally salient part of an experience (i.e., peak) and the conclusion of an experience (i.e., end) are weighted more heavily in the summary evaluation than other timepoints, and even the overall length of an experience [3]. Peak affective experiences are believed to be significant in our memory because they serve as thresholds for our capacity to cope with an experience, whereas end-points can signify recency (i.e., more easily recalled as function of time) and/or represent a point in which an experience is complete and carries meaning about personal capacity, particularly in goal-directed experiences [4]. Experimental studies with a focus on pain have demonstrated that manipulating a condition by extending it with a less painful conclusion, such as ending a cold-pressor task with slightly warmer temperatures [5] or leaving a colonoscope in the rectum for a few minutes before complete removal [6], results in the experience being viewed overall as less painful. In other words, despite an equally painful procedure for these groups during the shared interval, the inclusion of additional discomfort at the end, albeit at a comparatively lower intensity level, resulted in a more positive summary evaluation than concluding the procedure earlier.

Given that our summary evaluations of an event or experience play a critical role in future decision making, the peak-end rule has been studied in various contexts, including healthcare [7,8], business [9], organizational behavior [10], interpersonal relationships/interactions [11–13], and pleasure/happiness [14,15]. A recent meta-analysis compared several heuristics including the peak-end average, peak-only, beginning/primacy, ending/recency-only, variability, and lowest intensity (trough), to the overall average mean of scores provided for the specified timeframe across numerous contexts [16]. Results indicated that the peak-end rule was robust to various conditions (e.g., positive or negative affect, summary evaluations or judgement and decision-making tasks, days or weeks, etc.) and outperformed the other heuristics with respect to correlation with the overall average effect. While these findings support the peak-end rule as a valid heuristic for assessing past experiences, some studies have indicated that clinical populations may differ from healthy controls in the application of the peak-end rule [17,18], which may present a risk for bias.

This review will highlight the applications of the peak-end rule in mental health contexts, with respect to both clinical assessment and in experimental applications and discuss the clinical implications and future directions for research on the peak-end rule in mental health.

1.2. Peak-end rule in Mental Health

1.2.1. Assessment Studies

Clinical assessments of mental health symptoms rely quite heavily on retrospective self-report of symptoms. To improve upon traditional, informal assessment approaches, many healthcare settings have advanced measurement-based care initiatives, in which symptom scores on standardized assessments are used to help formulate treatment plans and evaluate progress [19–21]. These assessments offer concrete, trackable data, and have high potential to improve clinical practices. Nevertheless, as these assessments are typically administered at intervals of weeks or months, they may be influenced by the peak-end rule. Improving our understanding for how individuals suffering from anxiety, depression, posttraumatic stress, or other mental health symptoms respond to these surveys could further enhance the value of measurement-based care.

In recent years, the ubiquity of smartphones has allowed researchers to apply intensive longitudinal methods, such as daily diary and ecological momentary assessments (EMA), to repeatedly assess various emotions and symptoms at shorter intervals, with the hope of reducing potential recall biases [22–24]. Importantly, these methods provide an opportunity to assess the peak-end rule in mental health symptoms by comparing daily-level EMA responses to traditional retrospective self-report assessments. Several studies have examined various potential biases, including the peak and end biases, in posttraumatic stress symptoms. For example, Kiefer et al. [25] conducted PTSD symptom assessments three times per day over thirty days with women experiencing interpersonal violence and found that the peak PTSD symptom scores were most closely aligned with the retrospective report of PTSD symptoms, whereas the average PTSD symptom scores during that period significantly differed from the retrospective score. Likewise, in a study by Schuler et al. [26], first responders to the World Trade Center attacks were assessed for PTSD symptoms three times a day over one week, and the symptoms on the worst day were more strongly aligned with the retrospective assessment of symptoms for the past week than the average of all EMA responses. Similarly, in a 28-day EMA study of post-9/11 male veterans by Decker et al. [27], peak symptoms were significantly more correlated to retrospective symptom reports than recency/end symptoms or average symptoms, and both peak and end symptoms significantly predicted retrospective PTSD symptoms over and above average EMA symptoms. In contrast, a study by Greene et al. [28] examined posttraumatic stress symptoms among a community sample exposed to rocket fire and found average daily symptoms over a 30-day period to correlate more strongly to retrospective symptom reports than peak or recent symptoms. Additionally, in a study of emergency department patients who experienced a criteria A trauma, the average symptom scores assessed over the subsequent eight weeks were better predictors of retrospective symptom reports than peak, end, or peak-end scores [29]. Notably, most of these studies separated peak and end scores in their evaluation of biases, but in a study of medical interns, depression scores were most strongly correlated to a variable that combined peak and end mood scores over a two-week period, compared to peak-only, end-only, or average daily scores [30].

Scharbert et al. [31] examined well-being in a EMA study that tested several components of the peak-end rule, including how the peak-end rule differed based on prompt for the retrospective assessment (e.g., “in general,” “in the last four weeks,” and a prompt asking about average responses to their EMAs), based on different emotions or states (e.g., happy, anxious, energy level), and using scores based on peak or most recent: 1) individual moments, 2) averages within a single day, and 3) averages within a single week. The influence of peak and end scores were lessened when prompts on retrospective prompts could activate thoughtful/reflective consideration (e.g., System 1 vs. System 2 [32]), and varied based on the valence of the emotion/state being assessed and the timeframe by which peak and end states were considered. Across these different EMA studies, variability in study length, frequency of assessment, and sample characteristics/populations make it difficult to draw firm or universal conclusions about the degree of potential peak-end biases in symptom responses—additional research using these intensive longitudinal methods is greatly needed to better specify the contexts, conditions, and timeframes in which the peak-end rule might generate bias. In summary, responses to commonly used retrospective symptom scales are likely influenced by peak and end states, with greater bias potential when respondents are asked to recall symptoms over longer periods of time.

1.2.2. Experimental Studies

In line with some of the earlier pain studies [5,6] that examined the impact of ending a procedure with an extended, lower-intensity condition, several studies have examined experiential anxiety with similar methods. A study by Müller et al. [33] evaluated anxiety ratings among young women viewing a clip of a horror movie, with one condition ending the clip at the most frightening moment, and the second condition extending the same clip to conclude at a moderately less-frightening endpoint. Those who viewed the shorter clip evaluated the clip as having produced greater overall anxiety than those viewing the extended version with the less-frightening endpoint. Müller et al. [34] extended these findings by placing adults in an anxiety induction shock paradigm that had differing phases (e.g., moderate threat, high threat). Those who ended the procedure during a ‘high threat’ phase retroactively reported greater anxiety/distress about the paradigm than those who ended under a moderate threat condition, despite the conditions being counterbalanced for threat levels. In a medical context, Hetou et al. [35] modified a urology procedure (cystoscopy) for first-time patients to conclude with a two-minute extension where the scope was left in the bladder unmanipulated, and retrospective anxiety and pain were significantly lower for those in the extension condition. Together, these studies suggest that anxiety may follow a similar course to pain and can be reduced in summary evaluations by extending procedures to end with a lower intensity. The potential influence of end-point anxiety on summary evaluations has significant implications for exposure-based therapies, where individuals intentionally confront situations that increase their anxiety. For example, a preliminary study by Williams [36] demonstrated feasibility for manipulating exposure stimuli for blood, injury, and injection phobia, whereby ending an exposure session with lower-intensity stimuli improved retrospective perceptions of distress from the session. In-session desensitization/habituation may not be necessary for exposure to be effective [37], and ending an exposure at a lower intensity would not necessarily undermine the inhibitory learning process taking place from confronting the ‘peak’ feared stimuli earlier in the session. Evaluating an exposure session more positively may have benefits for building self-efficacy and improving treatment adherence over time. Exposure therapies are used for a wide range of mental health conditions, including various anxiety disorders, obsessive-compulsive disorder, and posttraumatic stress disorder, and additional research into whether applications of the peak-end rule could be used to improve retention and/or enhance treatment would be highly valuable.

1.3. Conclusions

The peak-end rule has been understudied in mental health contexts despite significant implications for clinical assessments used in measurement-based care initiatives [19,21] and retention/adherence in exposure-based psychotherapies [38,39]. The existing research on the peak-end rule in mental health is too scant to provide conclusive recommendations, but certain trends offer opportunities for further investigation. A critical step is better understanding for whom the peak-end rule produces a potential problematic bias. Studies have demonstrated differences in recall bias based on clinical versus control conditions [17,40], age of participants [41,42], personality traits [43], and frequency of trauma exposure [25]. Certain personality features linked to depression and anxiety, such as dispositional negativity [44] or neuroticism [45], may also correspond to attentional biases underlying the peak-end rule. When applying a ‘better safe than sorry’ processing framework [46] to the peak-end rule, there may be adaptive value in emphasizing peak negative experiences/dangers and present threat levels to guide decision making. At the same time, this can become exaggerated in the context of PTSD or an anxiety disorder, where chronic rumination or worry might unduly amplify the influence of the peak-end features.

Importantly, the peak-end rule is just one heuristic among many potential recall biases, and even within the peak-end rule there may be individual differences with respect to the weights of the ‘peak’ versus ‘end’ components [31]. For instance, several studies assessing suicidal ideation have demonstrated inconsistencies between EMA and retrospective recall, where participants report no ideation retrospectively over weeks or months despite having reported ideation in EMAs [47,48]. At the same time, in an EMA study of young people with borderline personality disorder, peak EMA suicidal ideation scores correlated more strongly to retrospective reports of past-week suicidal ideation than average or recent EMA scores [49]. These assessment challenges raise difficult questions for how to respond to individuals when discrepancies exist [50] and highlight the need for a better understanding of these potential response biases.

Even if specific traits or conditions conferring greater peak-end bias were identified, there would be challenges for clinical application. Consider a scenario in which a patient reported two bad days over the course of two weeks in daily surveys, but due to a strong peak-end bias indicated at the session that the past two weeks had been awful. Should a provider make a significant medication/therapy change when 85% of the days (by the patient’s own daily report) were positive? Would patients feel invalidated when presented with this discrepancy between daily reports and summary evaluations, or could this discrepancy be used as a therapeutic opportunity for identifying and addressing potential cognitive biases? Given that the peak-end rule appears ecologically valid in most contexts [16], it would be important to examine how a potentially peak-end-biased summary report performs relative to average daily scores in projecting symptoms in the subsequent weeks. These “biased” summary evaluations should not be dismissed if retrospective perceptions correspond to more negative days ahead compared to what might have been expected from the daily-level data. With respect to applications of the peak-end rule in the therapeutic context, it would be worth exploring whether individuals with greater susceptibility to the peak-end bias respond more favorably to exposures that include a lower-intensity extension of the exposure condition. On the other hand, this extension might only benefit those less likely to ruminate on the peak negatives of the experience. These uncertainties indicate significant opportunities for mental health researchers to investigate and improve our understanding of how the peak-end rule can be used to improve assessment practices and enhance exposure-based therapies for individuals struggling with mental health problems.

Highlights.

• The peak-end rule has been relatively understudied in mental health contexts

• Peak-end rule may influence assessments used in measurement-based care initiatives

• Application of peak-end rule has the potential to improve exposure therapy outcomes

• Potential individual differences in peak-end bias warrant further investigation