Next Article in Journal
Risk-Averse and Self-Interested Shifts in Groups in Both Median and Random Rules
Next Article in Special Issue
Strategic Information Suppression in Borrowing and Pre-Lending Cognition: Theory and Evidence
Previous Article in Journal
Differential Game-Theoretic Models of Cournot Oligopoly with Consideration of the Green Effect
Previous Article in Special Issue
Misperception and Cognition in Markets
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Review

Motivated Memory in Economics—A Review

by
Andrea Amelio
1 and
Florian Zimmermann
1,2,*
1
Department of Economics, University of Bonn, 53113 Bonn, Germany
2
Institute on Behavior and Inequality (briq), 53113 Bonn, Germany
*
Author to whom correspondence should be addressed.
Games 2023, 14(1), 15; https://doi.org/10.3390/g14010015
Submission received: 29 October 2022 / Revised: 23 January 2023 / Accepted: 25 January 2023 / Published: 31 January 2023
(This article belongs to the Special Issue Economics of Motivated Beliefs)

Abstract

:
Motivated reasoning refers to the idea that people hold certain beliefs about themselves or the world due to their desire to do so, rather than striving for accuracy. This type of belief formation can lead to overconfidence and polarization, as well as facilitate immoral behavior at both the individual and collective levels. One of the supply-side mechanisms for motivated reasoning is motivated memory, or the selective retrieval of past experiences or information based on self-serving criteria. In this article, we review the still young economics literature on motivated memory. Summarizing both theoretical and empirical work, we highlight the key results this literature has produced. We also discuss open questions and potentially exciting avenues for future research in this area.

1. Introduction

The standard economics view on belief formation is that decision-makers strive to hold beliefs that are as accurate as possible given the decision environment. In contrast, the notion of motivated reasoning posits that decision-makers want to hold certain beliefs about themselves or the world (e.g., Kunda [1]; Bénabou and Tirole [2], Bénabou and Tirole [3], Bénabou and Tirole [4], Bénabou and Tirole [5]; Köszegi [6]). Motivated reasoning can give rise to overconfidence (e.g., Malmendier and Tate [7], Malmendier and Tate [8]) and partisan polarization (Kahan [9]), and can facilitate moral transgression at the individual (e.g., Babcock et al. [10]; Konow [11]; Dana et al. [12]; Haisley and Weber [13]; Gneezy et al. [14]; Di Tella et al. [15]) and collective (Bénabou [16]) levels.
A central question in the literature on motivated beliefs is how self-deception operates in the presence of feedback. To put it differently, how do people manage to preserve self-servingly biased beliefs in situations where unbiased information is available? Bénabou and Tirole [2] have termed the cognitive strategies that people employ to deceive themselves the “supply side” of motivated beliefs.1 Perhaps the most obvious strategy to maintain a positive self-view is to avoid potentially unpleasant information altogether (e.g., Dana et al. [12]; Golman et al. [22]). When information cannot be avoided, individuals might engage in asymmetric information processing, placing more weight on positive than negative signals (e.g., Eil and Rao [23]; Sharot [24]; Möbius et al. [25]). In multiple-self settings, the current self can use current actions to signal a certain self-view to future selves (see Bénabou and Tirole [26,27], Hong et al. [28]). Yet another supply-side mechanism emphasizes the role of motivated memory (Bénabou and Tirole [2]). Motivated memory can be defined as the selective retrieval of past experiences or information based on self-serving criteria. Hence, motivated memory can be seen as a supply-side manifestation of motivated reasoning, specifically featuring selective retrieval as an ego-preserving mechanism.2
In this article, we aim to review and assess the economics literature on motivated memory.3 While this literature, to some extent, is still in its infancy, we nonetheless believe it is useful to take stock and document and discuss empirical as well as methodological regularities. We begin by outlining the seminal theoretical framework by Bénabou and Tirole [2] to provide a formal illustration of the main intuitions in this literature. We then briefly summarize a set of more recent theoretical works that feature motivated memory. The main focus of this review is on the existing empirical work on motivated memory in economics. This literature is partially aimed at testing theoretical predictions but also studies different shades of motivated memory mechanisms. We first consider laboratory evidence on motivated memory. This evidence ranges from prosocial decision-making to ego-relevant domains such as IQ to investment behavior. In a second step, we then review field evidence on selective recall. Having summarized the existing empirical literature, we then take a step back and (i) try to identify the key takeaways and regularities from the literature and (ii) discuss methodological considerations related to the study of motivated recall. In the last section, we conclude by highlighting some open research questions.

2. Theoretical Work

2.1. Bénabou and Tirole’s Framework

A key challenge for the concept of motivated reasoning is how self-deception is even logically possible. After all, a decision-maker who tries to manipulate her views and beliefs needs to take action with the goal of self-deception. Hence, one might argue that self-deception is an impossibility, since the decision-maker must know why she took certain actions. Philosophers and psychologists have debated this question for decades. In this section, we present a theoretical framework based on Bénabou and Tirole [2] and Bénabou and Tirole [27], which are the first works in economics to formally present the concept of motivated memory. This framework formalizes how self-deception via selective recall can be effective, even if the decision-maker knows that she sometimes purposefully forgets things, providing a conceptual structure with which to approach and interpret motivated memory.
Consider a present-biased agent with imperfect memory, facing a decision problem such as picking a level of effort e R + to exert in some task or project. Exerting effort is costly, with costs described by the cost function c ( e ) , and the agent’s returns on her effort V ( e , θ ) being a function of her personal skill or abilities θ Θ . 4 The time horizon of the decision problem is split into three key moments, that is, t { 0 , 1 , 2 } . At t = 0 , the agent can observe a noisy signal σ * Σ about her skill level θ .5 The agent knows the signal generating process, that is she is aware of p ( σ θ ) for each σ and θ . At t = 1 , the agent decides how much effort e to exert in the task, incurring the related cost c ( e ) , and at t = 2 the utility from the exerted effort realizes.
The agent’s memory is imperfect, in that at t = 1 she retrieves the signal σ ^ : with probability λ , σ * is forgotten, meaning no information is recalled, despite having observed some signal about θ in the previous period. Formally,
σ ^ = σ * , λ , 1 λ ,
with ∅ representing the case of no signal being retrieved from memory. Additionally, it assumed that the agent at t = 0 , or Self 0, may manipulate the probability of forgetting the signal λ . The incentive to manipulate memory stems from the fact that the agent’s preferences are dynamically inconsistent. Formally, the agent’s utility at t = 0 is
U 0 ( e ) = c ( e ) + E σ * [ V 0 ( e , θ ) ] ,
while her utility from t = 1 perspective is
U 1 ( e ) = c ( e ) + E σ ^ [ V 1 ( e , θ ) ] .
Hence, for any given expectation about θ , the optimal effort from Self 0’s perspective is different than that from Self 1’s perspective. Moreover, it is assumed that V 0 ( · ) and V 1 ( · ) are such that:
e 0 * ( θ ) = arg max e R + V 0 ( e , θ ) > arg max e R + V 1 ( e , θ ) = e 1 * ( θ ) ,
for any θ Θ . Hence, Self 0 will have the incentive to increase the recall probability of positive signals and decrease that of negative signals. Self 0’s strategy can then be described as a mapping assigning a recall probability λ to any possible signal realization σ , that is λ : Σ [ 0 , 1 ] . When Self 1 recalls the signal, that is σ ^ = σ * , her beliefs will just follow from combining her prior belief about θ and p ( σ * θ ) for each θ through Bayes’ rule. On the other hand, when observing ∅, Self 1 may still update her beliefs. This is because Self 1 knows that she may have forgotten the signal and that, possibly, σ . Moreover, she is aware of Self 0’s incentives, and will not necessarily take σ ^ = at face value.
Given this framework, the papers we discuss in the next section study the existence and features of Bayes–Nash equilibria and how the set of existing equilibria changes with problem parameters. Of particular interest are corner equilibria, namely full-disclosure and babbling or self-deception equilibria, in which, respectively, Self 0 always and never truthfully reports negative signals.

2.2. Other Motivated Memory Models

Chew et al. [31] extend Bénabou and Tirole [2]’s theoretical framework, enriching the signals space: Self 0 can observe and also send a good (G) signal about her skill level, in addition to a bad (B) one and the absence of signal (∅). Hence, in addition to repressing negative memory (amnesia, using the authors’ terminology), Self 0 can also send a good signal in the absence of information (delusion). As opposed to Bénabou and Tirole [2], in Chew et al. [31], the probability of misreporting the signal is not strictly increasing in the level of present bias, as the presence of delusion decreases the value of amnesia, given that present bias is large enough to sustain a delusion equilibrium. Moreover, Chew et al. [31] show that self-deception equilibria always exist if the ego-dependent component of utility is large enough, implying that when beliefs have a strong impact on utility, self-deception is also sustainable for agents who do not suffer from present bias.
Hagenbach and Koessler [32] also present a setting related to Bénabou and Tirole [2]’s framework, but focus on studying how different classes of psychological utilities affect the existence of a full-disclosure equilibrium. Moreover, the decision-maker depicted in Hagenbach and Koessler [32] does not suffer from present bias, hence no dynamic inconsistency issues arise: the trade-off between perfect memory and forgetting arises because of belief-based utility, as in Chew et al. [31]. Finally, the authors extend their model by adding exogenous memory failures: with some probability, Self 1 will not be able to remember the state of the world, even though Self 0 decided to recall. It is shown how the absence of full-disclosure equilibrium is generally possible in this setting, and under which additional assumptions such equilibrium exists for the classes of utilities considered.
Fudenberg et al. [33] theoretically study imperfect memory in a different framework from Bénabou and Tirole [2]. Fudenberg et al. [33] illustrate a model featuring an agent who only recalls a subset of past experiences and, crucially, study the dynamics and long-run behavior of the agent’s beliefs and actions. The recalled experiences depend on a collection of memory functions and on the currently observed information. The agent is naive about her imperfect memory, updating beliefs and selecting actions as if the recalled experiences are the only ones that occurred. Fudenberg et al. [33] show how in their framework it is possible to generate, among other biases, motivated memory, and how, in this setting, the agent’s long-run beliefs will be skewed in an ego-boosting direction and hence her actions will be also sub-optimal and biased in that direction. Additionally, the authors note how an agent with a misspecified model is observationally equivalent to an agent with ego-boosting memory, but how their memory-based framework predicts a different impact of changes in information provision. For example, unlike the misspecification case, their framework predicts that providing jointly positive and negative information about different, unrelated, and ego-relevant dimensions, increases the likelihood that the agent retrieves that specific experience, inducing the long-term beliefs about the ego-decreasing dimension to be more precise.

3. Empirical Evidence

In the following, we review the empirical evidence on motivated memory. To provide a systematic overview of existing work, we group papers along two dimensions: (i) laboratory vs. field and (ii) decision environment, e.g., prosocial behavior versus investment choice. This allows us to distill potential methodological or conceptual differences along those dimensions. We begin with controlled lab evidence from the context of prosocial behavior. Table 1 provides a summary of several methodological aspects of all the empirical papers reported in this section, both laboratory and field studies.

3.1. Motivated Memory and Prosocial Behavior

A first block of evidence in the literature explores the role of motivated memory in the domain of prosocial behavior. The underlying intuition is that most people want to think of themselves as following high moral standards. At the same time, people sometimes succumb to moral transgressions. This creates a tension between desired self-view and actual behavior, which can be resolved by the motivated recall of past prosocial behavior.
Li (2013): The first paper that investigates empirically the role of motivated memory in the economics literature is Li [34]. The article investigates experimentally how the recall accuracy of social interactions depends on a player’s own behavior. Additionally, it investigates the dynamics of how actions by others are perceived over time.6
The experiment is divided into two phases: (i) a trust game and (ii) a recall phase. Participants in the first phase are not aware that they will take part in the second phase. The first mover in the game can choose to not trust (end the game) or to trust (allow the other player to play). Given that player 1 trusted her, player 2 can be kind (split the pay-off evenly) or not kind (keep all the pay-off). There were three main treatments, manipulating the delay between the first and the second phases: the recall phase could take place immediately after the first phase, after seven days, or after forty-three days. In the recall phase, participants are asked to recall their and others’ decisions in the game. For all but a control treatment, the recall is financially incentivized, granting participants a fixed amount of money for each correct answer. Moreover, participants rate the kindness of the recalled choices and are asked to guess the share of players playing each action. Finally, the two players played a dictator game.
The author finds that participants who received unkind acts are less likely to recall what the other player did and also to correctly recall their own actions. Participants who decided to be not kind in the first phase seem to change their perception of their action kindness level, as the delay increases. Relatedly, when there is no delay between the two phases of the experiment, participants’ beliefs about the share of players picking each action are biased towards their experience (e.g., a player who received an unkind act tends to overestimate the share of players choosing to be not kind). However, this difference seems to decay over time below the significance threshold.
Carlson et al. (2020): Carlson et al. [35] focus on motivated memory in dictator games and tackle the following questions. First, they investigate the direction and the extent of memory errors, using a continuous measure, the split of the pie in percentage points. Second, they study how the relationship between subjective fairness norms and actions impacts the extent and direction of memory errors. Finally, they study how active as opposed to forced choices differ in triggering motivated memory.
In total, the authors conducted five experiments, two in the laboratory and three online. In the first experiment, participants took five allocation decisions on how to split an amount of money between themselves and an anonymous partner. Afterward, they answered demographic questions and completed a numeracy test. Then, in an announced recall task, participants would be asked to report the average share of the pie assigned to the anonymous partner in the previous stage of the experiment. The other experiments then implement incremental changes to the first experiment. The second experiment is identical to the first, except that participants’ fairness norms are elicited before the dictator game rounds. The third experiment features a different, more fine-grained, incentivization scheme for the recall phase. While in the first two experiments, participants were rewarded if their recall was within 10 % of their actual average generosity, in the third experiment the bonus was an increasing function of the accuracy at the 1 % level. The two final experiments are related and aimed at investigating the third question. In the fifth experiment, different from the previous ones, after the fairness norm elicitation, participants are informed that instead of deciding the splits themselves, the computer implements a randomly drawn choice from a participant of experiment number four.
The authors provide evidence that, on average, participants commit self-serving memory errors. Recalled generosity tends to be biased upwards. Additionally, the authors show that fairness norms matter in predicting motivated memory errors. Independently of the actual generosity level, violators (participants who were less generous than their reported norm) have positive memory errors on average, while upholders (participants who on average were more generous than the reported norm) do not. Finally, concerning the role of responsibility, participants’ self-serving memory errors seem to fade out when choices were forced and hence real responsibility is absent.
Saucet and Villeval (2019): Saucet and Villeval [36] focus on the role of responsibility for motivated memory distortions. They do so in the context of dictator games. Furthermore, they study how financial incentives and the level of prosociality 7 impact recall at the intensive margin: the object of the recall exercise is not binary, which allows measuring how these parameters modulate the magnitude and the direction of recall errors.
The experiment is split into four parts: (i) 12 binary dictator games, (ii) a filler task, (iii) a recall task, and (iv) a memory assessment task. In the baseline condition, participants are assigned to the roles of dictator or receiver. In a control condition, the dictator’s role is played by a computer. Each dictator game has a selfish and an altruistic option. Option A is fairer than option B if the receiver’s pay-off for A is higher than for B, regardless of how the pie is split. During the recall task, which is not announced beforehand, participants are asked to recall the amount allocated to the receiver (or to themselves in one treatment). Unlike Li [34], Saucet and Villeval [36] do not vary the delay between the dictator games and the recall task. Instead, they make all participants undergo the same filler task for eight minutes.
Dictators are found to have a higher recall accuracy when they picked the fair option. Interestingly, this difference is not present when allocations are made by the computer. The effect also does not seem to be explained or moderated by participants’ education level or performance in an additional, unrelated, memory task. This is an important step towards a causal interpretation where past behavior is indeed the driver of recall accuracy. We discuss this further at the end of this section. Moreover, the authors provide evidence that partially supports the hypothesis of incentives playing a role in determining recall accuracy: incentives significantly increase the share of correctly recalled amounts by dictators, but only for those who picked the fair option.
Galeotti et al. (2020): Galeotti et al. [37] focus on a different aspect, namely the mechanisms behind incorrect memories following antisocial behavior. The authors argue that unethical amnesia, defined as imprecise memory of morally dubious past actions, is due to instrumental reasons rather than self-image concerns. In other words, they hypothesize that faulty memory, in their setting, is caused by the need to justify future immoral decisions as opposed to (or on top of) the preservation of self-image.
This is shown through an experiment divided into two stages. First, participants undergo twenty sessions of a wheel game. Afterward, they take part in a recall stage. The time elapsed between the first and the second stage is three weeks. The wheel game is structured as follows. Participants are shown a wheel divided into six sections. Their task is to pick in their mind one of those sections. After they picked, each section is associated with a number from 1 to 6. Participants have to report the number associated with the section they picked, and pay-off increases in the reported number. Hence, potentially, participants could always state to have picked the section of the wheel with the largest number and receive the largest possible pay-off. In the recall stage, participants have to recall how many times they reported each outcome in the twenty wheel game sessions. To disentangle hedonic and instrumental memory errors, the authors present two treatments. In the Instrumental treatment, at the end of the session, right before the memory stage starts, participants are told that they may give up part of their pay-off and are encouraged to do so if they cheated in the first stage. In the Hedonic treatment, instead, participants are not given this possibility. Hence, memory errors in the Instrumental treatment may also depend on the desire to perpetuate unethical behavior, while in the Hedonic treatment, any memory error should not derive from strategic considerations, as participants are not given the chance to change their pay-off. Additionally, for both Hedonic and Instrumental treatments, there is a corresponding control condition, which is identical to the respective treatment, except for the fact that participants can not cheat in the wheel game, as they are required to click on the screen, as opposed to picking the section of the wheel in their mind.
To illustrate the results, the authors provide a type classification of participants: participants in treatment conditions are classified as dishonest if the distribution of reported outcomes in the first part is right-skewed and significantly different from a uniform with a 3.5 mean. First, the authors report no significant difference in memory errors comparing honest participants from the treatment conditions and participants from control conditions. Comparing memory errors from the pooled control conditions with those from dishonest participants in Instrumental and in Hedonic, memory errors in Instrumental are significantly higher. No such effect is found for Hedonic. This provides direct evidence for the hypothesis that the main drivers of memory errors, in this framework, are instrumental motives.
Interim Summary: Taken together, the bulk of existing evidence in the context of prosocial behavior suggests that people self-servingly recall prosocial acts better than antisocial ones. A challenge in this literature, however, is to obtain causal evidence. The key reason this concern arises is that actual moral behavior is of course endogenous and hence might relate to factors that correlate with recall accuracy. For instance, it is conceivable that more prosocial people, in general, have a better memory8 Saucet and Villeval [36] actually measure recall performance in an unrelated task and show that their results remain unaffected when controlling for recall performance, strengthening a causal interpretation of their results. As we will see in the next sections, a key focus in the domains of investment behavior and IQ is to establish causality in the relationship between feedback and recall accuracy.

3.2. Motivated Memory and Financial Decisions

A second decision context where motivated memory has been studied is financial investments. Intuitively, investors want to think of themselves as being good at what they do. At the same time, some investments fail, creating a motivation to selectively recall past investment outcomes.
Gödker et al. (2022): In a series of controlled laboratory experiments, Gödker et al. [40] provide experimental evidence of systematic memory bias in the domain of financial decisions. In this context, the authors aim to relate memory, beliefs, and reinvestment choices.
The authors introduce the following paradigm. The experiment is split into four stages: (i) first investment and observation of returns (ii) memory elicitation (iii) belief elicitation, and (iv) second investment choice.9 In the first stage, participants allocate an initial endowment between a risky asset and a safe bond. The stock can be good or bad, with the good stock being more likely to generate positive returns. Afterward, participants observe a set of noisy signals about the stock quality and, with different degrees of delay, are asked to recall the returns realizations and to provide their assessment of the probability of the stock being good or bad. The delay between the first and the second stage of the experiment can be either zero or one week. This allows the authors to isolate the effect of memory from the role of any other factor that may impact the recollection through encoding (e.g. attention, salience). The authors argue that any difference in recall performance between the conditions with different delays is to be attributed to memory, as any other factor which may influence information acquisition or processing is held constant.
Participants who decide to invest in the stock are found to disproportionately recall positive returns and, as a consequence, hold more optimistic beliefs about the stock being good. Moreover, the level of memory bias exhibited by participants predicts their probability of making an incorrect re-investment decision in the third phase of the experiment, showing how memory bias is connected not only to beliefs but also to investment choices. The authors also study how stake size has an impact on the size of the memory distortion and show that memory bias disappears on average when the incentives for accurate recall are substantially increased.

3.3. Motivated Memory and IQ

Another context where motivated recall has been studied is IQ test performance. Most people like to think of themselves as being smart, in particular relative to others. Hence, the presence of unbiased feedback about IQ test performance might create a motive to remember good news and misremember bad.
Chew et al. (2020): In a first attempt to empirically study memory about IQ-related feedback, Chew et al. [31] propose and implement an experiment that closely follows the theoretical framework they present in the paper (see Section 2.1 of this review). The goal of the experiment is to document three memory error types (i) positive amnesia (not remembering a bad signal) (ii) delusion (remembering a signal that was never observed) and (iii) positive confabulation (remembering a good signal after observing a bad one) and relate these errors to the degree of present bias exhibited by participants.
The main task in the experiment is Raven’s IQ test, with an additional recall phase. The delay between the two stages of the experiments is stated to be “several months”, which suggests that different participants experienced different delays in this order of magnitude. In the recall phase, participants are shown six questions and the related correct answers and may choose one of the following: (a) My response was correct; (b) My response was incorrect; (c) I didn’t see this question; (d) I don’t remember. Four out of the six questions were drawn from the actual test, while two were not present in it.
The results show that participants tend to exhibit positive rather than negative memory errors. In other words, positive amnesia, delusion, and confabulation are relatively more frequent than the negative corresponding phenomena and are relevant in terms of incidence, ranging from 25 % to 50 % . Additionally, it is shown that the likelihood of exhibiting positive false memories increases in present bias, while this relationship does not hold for positive amnesia.
Zimmermann (2020): Zimmermann [41] takes a different approach and studies the dynamics of updating from feedback about IQ test performance. He tries to map out how immediate updating differs from updating in the long run. In a second step, he then asks how selective recall can explain the dynamics of motivated beliefs, and what the limits of motivated memory are.
In the experiment, participants take an IQ-related test. Participants then state prior beliefs about their relative test performance. They then receive noisy feedback on relative test performance. The noise component in feedback is crucial, as it creates exogenous variation in whether subjects obtained good or bad news, hence allowing for the identification of a causal effect of feedback on belief dynamics and recall accuracy. In a between-subject design, participants then (unannounced) state posterior beliefs, either immediately after obtaining the feedback or one month later. In a separate treatment, instead of stating posterior beliefs one month after the feedback, subjects are asked to recall the feedback. In additional treatment variations, (i) the belief elicitation one month later is announced and (ii) incentives for recall accuracy are substantially increased.
Zimmermann [41] finds that while in the short-run beliefs respond to both positive and negative feedback, over time, beliefs after negative feedback move back to prior levels. Positive feedback instead has a more long-lasting effect. This dynamic pattern seems well-explained by selective recall. After one month, participants more accurately recall positive feedback compared to negative feedback. Importantly, these effects are substantially mitigated when (i) future belief elicitation is announced and (ii) incentives for recall are high.

3.4. Field Evidence on Motivated Memory

While the majority of existing studies on selective recall are laboratory-based, in the following, we review the existing work that studies motivated memory in the field.
Huffman et al. (2022): Huffman et al. [42] are the first to study the role of motivated memory in a field setting. The context of the study is managerial overconfidence and how it responds to feedback. More specifically, the authors study persistent overconfidence of in-store managers and whether such overconfidence can be explained through motivated memory.
They found an interesting setting where managers have access to a ranking concerning their performances and are asked to: (i) state a prediction of the quintile of their ranking for the following quarter and (ii) recall their past ranking quintile.
Managers are shown to be overall overconfident. This overconfidence is persistent despite the existence of feedback. Managers also exhibit upward memory bias. While managers’ memory is precise when recalling positive past performances, memory errors arise and increase as the rank decreases. This provides a possible explanation for the persistence of memory bias.
Müller (2022): Müller [43] provides field evidence on motivated memory in the domain of fertility desires, combining a large Kenyan panel covering more than ten years with experimental methods.
He shows how participants’ recollection of their past fertility desires is a function of their actual fertility rate. In particular, participants tend to overestimate their past desires, as on average they had more children than stated ten years before. Additionally, financial incentives do not seem to improve memory performances for fertility rates, while they seem to have a significant impact on ego-neutral recall questions, pointing towards mechanisms of selective forgetting.
Müller [43] also reports evidence concerning information avoidance in this domain. Survey participants are randomly assigned to different conditions in which they were to learn their past fertility desires at the end of the survey, with or without financial incentives to do so. The results show that financial incentives impact the take-up rate only for participants who did not exceed their past fertility desires: for participants with excess fertility rates financial incentives seem to have no significant effect.
Roy-Chowdhury (2022): Roy-Chowdhury [44] provides additional field evidence on motivated memory in the domain of school performances and relates it to prior beliefs and preferences. The analysis is based on longitudinal data, collected biannually, of teenagers’ memories of their mathematics grades.
Similarly to Huffman et al. [42] and Müller [43], the author shows how recall of past performances is imperfect, with memory errors being skewed towards ego-enhancing recalls: on average, students recall better grades than the correct ones, and students with worse grades exhibit on average more memory errors. Roy-Chowdhury [44] additionally exploits the different lags between the observation of the actual grade by the children and the moment in which the survey was conducted each semester, that is the recall date. The Fall wave of the survey asked about grades that the children observed approximately five months before, compared to the three weeks of the Spring wave. This way, the author can perform a within-subject analysis, showing how the evidence on motivated memory is robust to within-individual estimation.
The paper also reports evidence of memory distortion impacting the dynamics of self-confidence levels. Self-confidence shows high degrees of autocorrelation, but also depends greatly on recalled grades, and not on actual grades, suggesting that memory distortion plays a role in feasibly keeping high self-confidence levels. The author interprets these dynamics in the light of self-persuasion motives: students distort their memory of their past grades to impact their self-confidence level, which in turn may impact future academic performance.
Finally, Roy-Chowdhury [44] estimates a structural model which decomposes the probability of recalling a specific grade, given the actual grade, into two components, along the lines of Bénabou and Tirole [2]: the ego benefit of recalling a high grade and a cost associated to distorting memory. The parameters associated with these factors are significant, supporting the idea of strategic memory manipulation, with students with a stated preference in math, or higher self-confidence, exhibiting higher gains from memory distortions.
Sial et al. (2023): In the domain of past gym attendance, Sial et al. [45] show how biased memory seems to modulate the level of present bias naivete. The authors recruited 1292 participants with a gym membership, having access to their attendance data. Additionally, participants completed a survey, encompassing, crucially, their recall of past attendance, their attendance expectations for the following weeks, and their willingness to pay for different incentivization mechanisms and commitment devices.
The paper reports three main results. First, individuals exhibit a significant memory bias concerning their gym attendance, with this bias being linked to overestimation of their future attendance (consistently with Müller [43] and Huffman et al. [42]). Second, individuals with larger memory bias tend to have a lower willingness to pay for future attendance incentives. Third, memory bias is orthogonal to the level of present bias but is associated with a higher level of naivete.10 Note that Sial et al. [45] do not explicitly refer to the memory bias they identify as motivated memory. However, motivated memory is mentioned as a plausible underlying mechanism.

4. Discussion—Taking Stock

The literature on motivated memory is still in its infancy. Nonetheless, in the following, we will highlight empirical patterns that have emerged and discuss both conceptual and methodological aspects related to the study of motivated memory:
1.
Motivated memory has by now been found in a rather broad set of contexts, ranging from investment behavior to prosocial decision-making. What distinguishes these contexts is the object of and the reason for self-deception. This highlights that studies of motivated memory always jointly test the existence of motivated reasoning and selective recall as a supply-side mechanism. This provides an empirical challenge since an implicit assumption in empirical studies of selective recall is that subjects do have a motive for self-deception to begin with. The same challenge is of course also present in the study of other supply-side mechanisms. In fact, part of the reason why the literature on asymmetric updating has produced somewhat mixed results might be the joint testing of motivated reasoning plus asymmetry in updating.11
2.
Despite the pervasive evidence for motivated memory in different contexts, the literature has also identified important limits to selective recall. In particular, two patterns emerge: (i) substantial increases in stake size as in Gödker et al. [40] and Zimmermann [41] mitigate motivated memory and instead induce people to accurately remember both good and bad news; (ii) the announcement of a future recall task diminishes motivated memory, see Coffman et al. [49] and Zimmermann [41]. In other words, when future recall tasks are anticipated, asymmetries in recall appear to vanish. In sum, the prevalence of motivated memory seems to depend heavily on features of the decision environment.
3.
A methodological pattern that emerges from existing work on motivated memory is that the role of selective recall is identified via delay between action taken or feedback provided and subsequent memory elicitation. Two comments seem in order: (i) so far, no consensus has emerged on the delay that is necessary for selective recall to emerge. In most existing studies, delay ranges between one and several weeks; (ii) implementing memory constraints via delay is different from the typical approach in studies of non-motivated recall, where memory constraints are made binding via interference (similar memory entries compete and interfere with each other, making retrieval of a specific memory cue less likely).12
4.
A further methodological pattern seems to be that, in most studies, the recall comes as a surprise. This again is in contrast with studies of non-motivated memory, where typically the entire experiment, including future recall elicitation is explained in advance. Importantly, as discussed, motivated memory appears to be mitigated when future recall tasks are announced.
5.
One empirical challenge is to establish causality in selective recall. Studies that have attempted to identify causality predominantly exploited noisy feedback provision (e.g., Gödker et al. [40] and Zimmermann [41]), where participants obtain noisy feedback about their own test performance on investment success, and the noise component allows for a causal identification of selective recall. In action-based paradigms such as prosocial behavior, causality is more difficult to establish, but causal interpretations have been bolstered by adding controls for memory performance (Saucet and Villeval [36]).
6.
In terms of the measurement of memory, the literature has mostly used incentivized recall paradigms, where participants receive a reward if recall accuracy is high. In terms of question format, different approaches have been followed (similar to some of the leading approaches in memory research, see Kahana [53]’s book): (i) in recognition paradigms, respondents are asked to recall past behavior or feedback received and are provided with several possible answers. The task then is to pick the correct one; (ii) free recall paradigms are similar in that subjects are also asked to recall past behavior or feedback received. However, they are not provided with a list of possible answers, but need to recall freely; (iii) in open texts paradigms, instead of being asked to recall specific events (e.g. past action or feedback), subjects are asked to describe certain aspects of an experiment or a decision environment. All these formats can be incentivized.13

5. Concluding Remarks

In this review, we summarized the still small but growing economics literature on motivated memory. While clear limits to selective recall appear to exist, the evidence suggests the prevalence of selective recall in a broad set of domains. In other words, motivated memory seems to be an important input into the development and maintenance of motivated belief patterns.
In the following, we will try to point to a few holes in the literature that might provide promising directions for future research. First, until now, the literature on motivated and non-motivated memory essentially moved in parallel with no overlap, both conceptually and in terms of methodology. Bringing these strands of the literature together might be fruitful in many different ways. For instance, the literature on non-motivated recall has identified a series of key features of how memory operates, such as associativeness and interference (Bordalo et al. [54]; Enke et al. [51]), and has studied implications, for instance for the reaction to news or stereotypical thinking. It would be interesting if motivated agents were able to exploit these key patterns in memory to self-servingly bias what they recall. Second, as this review highlights, there is until now a clear dominance of laboratory-based work on selective recall. While the laboratory has many advantages for the study of memory, more field work will be needed to assess the importance and limits of motivated memory for economic applications. Third, the demand side of motivated memory remains partially unexplored. In this direction, we see two promising avenues. The first is to better understand the determinants of the value of motivated memory, that is, which factors impact the demand for motivated memory. Carlson et al. [35], Saucet and Villeval [36] and Gödker et al. [40] provide evidence on how responsibility or active choices may impact the values of the demand side, but there may be more relevant factors at play. The second is to study individual heterogeneity in demand for motivated beliefs and separate it from heterogeneity in the ability to repress unpleasant information.14 Fourth, and finally, until now, the bulk of empirical evidence is in a reduced form. In light of the existing theoretical work on motivated memory, it might be fruitful to follow more structural approaches that allow for more direct tests of model predictions and for the identification of model parameters.

Author Contributions

All authors contributed to conceptualization, methodology, review validation, writing, draft preparation, and editing. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy—EXC 2126/1-390838866. This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (ERC Starting Grant: 948424—MEMEB).

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Acknowledgments

Review article prepared for special issue in Games on “Economis of Motivated Beliefs”. For valuable comments and discussions we thank Katrin Goedker, Maximilian Mueller, Charlotte Saucet, Paul Smeets and Marie Claire Villeval.

Conflicts of Interest

The authors declare no conflict of interest.

Notes

1
In turn, the “demand side” (Bénabou and Tirole [3]) of self-deception is concerned with why people might want to hold motivated beliefs. Several motives have been suggested. Köszegi [6] and Brunnermeier and Parker [17] formalize decision-makers who derive consumption utility from being optimistic about themselves and/or the future. Bénabou and Tirole [2] emphasize the motivational value of optimistic beliefs. Schwardmann and van der Weele [18] argued that people first need to deceive themselves in order to deceive others. More generally, the literature on social signaling emphasizes the role of (stated) optimism as a social signal (see Burks et al. [19]; Charness et al. [20]; Ewers and Zimmermann [21]).
2
With a slight abuse of terminology, we use motivated memory and motivated recall interchangeably.
3
We do not review related contributions from other fields such as neuroscience and psychology. For excellent overviews, see for instance Schacter [29] or Baddeley et al. [30].
4
In Bénabou and Tirole [2]’s paper, the skill parameter θ is more specifically interpreted as the probability of success of the project. This is just a normalization of any other dense set Θ R .
5
Σ represents the set of possible signals the agent may observe at t = 1 . As signals represent the agent’s information set in a given period, this includes the empty set (∅).
6
Recall of behavior of others might also be subject to motivated reasoning. People might want to believe that the world is just (Bénabou and Tirole [3]) and accordingly might have too rosy recollections of how they were treated in the past.
7
The level of prosociality is captured in absolute terms. This differs from Carlson et al. [35], who investigate the role of personal fairness norms. In contrast to Saucet and Villeval [36]’s results on prosociality, Carlson et al. [35] find no effect of absolute fairness on recall.
8
There is mixed evidence on the relationship between prosocial behavior and IQ. For example, Han et al. [38] report evidence in favor of this relation only for tasks with higher complexity. Guo et al. [39], instead, report a significant relationship between IQ and self-report measures of pro-social behavior.
9
Points (ii) to (iv) were elicited in random order.
10
To derive this result, the authors estimate a structural model featuring quasi-hyperbolic discounting and a parameter that allows for different degrees of naivete.
11
Eil and Rao [23] test how individuals acquire and process information in ego-relevant contexts, intelligence and beauty, showing evidence for asymmetric updating. Möbius et al. [25] report similar evidence of asymmetric updating and conservatism in the domain of beliefs concerning relative performance in an IQ test. Ertaç [46] also reports evidence of asymmetric updating, but in the opposite direction: participants react more to bad news about their performance, compared to good news. Sharot [24] show how people update their beliefs asymmetrically, reacting more to positive than negative news, relating this asymmetry to previous expectations and to variations in neural coding of positive and negative news. Schwardmann and van der Weele [18] reproduce results concerning asymmetric updating in the domain of intelligence tests, showing that such asymmetry towards positive self-image is fostered by motives to persuade others. Coutts [47] reports evidence of asymmetry and conservatism in belief updating also in financially relevant and not value-relevant domains, arguing that asymmetric updating in ego-relevant domains could also be not due to motivated reasoning. Barron [48] investigates updating in a financial domain, reporting no significant evidence of asymmetric updating.
12
See, for example, Graeber et al. [50], Enke et al. [51] and Bordalo et al. [52].
13
Open text messages can be more difficult to incentivize. In Zimmermann [41], subjects receive a reward if a group of RAs judges the text as an accurate description of the object of interest.
14
With the current paradigms, it would be difficult to distinguish the demand value of motivated memory and the ability to forget information. How to disentangle these two elements is an open empirical challenge.

References

  1. Kunda, Z. The Case for Motivated Reasoning. Psychological Bulletin 1990, 108, 480–498. [Google Scholar] [CrossRef] [PubMed]
  2. Bénabou, R.; Tirole, J. Self-Confidence and Personal Motivation. Q. J. Econ. 2002, 117, 871–915. [Google Scholar] [CrossRef]
  3. Bénabou, R.; Tirole, J. Belief in a Just World and Redistributive Politics. Q. J. Econ. 2006, 121, 699–746. [Google Scholar] [CrossRef] [Green Version]
  4. Bénabou, R.; Tirole, J. Identity, Morals, and Taboos: Beliefs as Assets. Q. J. Econ. 2011, 126, 805–855. [Google Scholar] [CrossRef]
  5. Bénabou, R.; Tirole, J. Mindful Economics: The Production, Consumption, and Value of Beliefs. J. Econ. Perspect. 2016, 30, 141–164. [Google Scholar] [CrossRef] [Green Version]
  6. Köszegi, B. Ego Utility, Overconfidence, and Task Choice. J. Eur. Econ. Assoc. 2006, 4, 673–707. [Google Scholar] [CrossRef] [Green Version]
  7. Malmendier, U.; Tate, G. CEO Overconfidence and Corporate Investment. J. Financ. 2005, 60, 2661–2700. [Google Scholar] [CrossRef] [Green Version]
  8. Malmendier, U.; Tate, G. Who makes acquisitions? CEO overconfidence and the market’s reaction. J. Financ. Econ. 2008, 89, 20–43. [Google Scholar] [CrossRef] [Green Version]
  9. Kahan, D.M. Ideology, motivated reasoning, and cognitive reflection. Judgm. Decis. Mak. 2013, 8, 407–424. [Google Scholar] [CrossRef]
  10. Babcock, L.; Loewenstein, G.; Issacharoff, S.; Camerer, C. Biased Judgments of Fairness in Bargaining. Am. Econ. Rev. 2013, 85, 1337–1343. [Google Scholar]
  11. Konow, J. Fair Shares: Accountability and Cognitive Dissonance in Allocation Decisions. Am. Econ. Rev. 2000, 90, 1072–1091. [Google Scholar] [CrossRef] [Green Version]
  12. Dana, J.; Kuang, J.X.; Weber, R.A. Exploiting moral wiggle room: Experiments demonstrating an illusory preference for fairness. Econ. Theory 2007, 33, 67–80. [Google Scholar] [CrossRef]
  13. Haisley, E.C.; Weber, R. Self-serving interpretations of ambiguity in other-regarding behavior. Games Econ. Behav. 2010, 68, 614–625. [Google Scholar] [CrossRef]
  14. Gneezy, U.; Saccardo, S.; Serra-Garcia, M.; van Veldhuizen, R. Bribing the Self. Games Econ. Behav. 2020, 120, 311–324. [Google Scholar] [CrossRef]
  15. Di Tella, R.; Perez-Truglia, R.; Babino, A.; Sigman, M. Conveniently Upset: Avoiding Altruism by Distorting Beliefs about Others’ Altruism. Am. Econ. Rev. 2015, 105, 3416–3442. [Google Scholar] [CrossRef]
  16. Bénabou, R. Groupthink: Collective Delusions in Organizations and Markets. Rev. Econ. Stud. 2013, 80, 429–462. [Google Scholar] [CrossRef] [Green Version]
  17. Brunnermeier, M.K.; Parker, J.A. Optimal Expectations. Am. Econ. Rev. 2005, 95, 1092–1118. [Google Scholar] [CrossRef]
  18. Schwardmann, P.; van der Weele, J. Deception and self-deception. Nat. Hum. Behav. 2019, 3, 1055–1061. [Google Scholar] [CrossRef] [Green Version]
  19. Burks, S.V.; Carpenter, J.P.; Goette, L.; Rustichini, A. Overconfidence and Social Signalling. Rev. Econ. Stud. 2013, 80, 949–983. [Google Scholar] [CrossRef]
  20. Charness, G.; Rustichini, A.; van de Ven, J. Self-confidence and strategic behavior. Exp. Econ. 2018, 21, 72–98. [Google Scholar] [CrossRef]
  21. Ewers, M.; Zimmermann, F. Image and Misreporting. J. Eur. Econ. Assoc. 2015, 13, 363–380. [Google Scholar] [CrossRef]
  22. Golman, R.; Hagmann, D.; Loewenstein, G. Information Avoidance. J. Econ. Lit. 2017, 55, 96–135. [Google Scholar] [CrossRef] [Green Version]
  23. Eil, D.; Rao, J.M. The Good News-Bad News Effect: Asymmetric Processing of Objective Information about Yourself. Am. Econ. J. Microecon. 2011, 3, 114–138. [Google Scholar] [CrossRef] [Green Version]
  24. Sharot, T. The optimism bias. Curr. Biol. 2011, 21, 941–945. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  25. Möbius, M.; Niederle, M.; Niehaus, P.; Rosenblat, T. Managing Self-Confidence: Theory and Experimental Evidence. Manag. Sci. 2022, 68, 7793–8514. [Google Scholar] [CrossRef]
  26. Bénabou, R.; Tirole, J. Incentives and Prosocial Behavior. Am. Econ. Rev. 2006, 96, 1652–1678. [Google Scholar] [CrossRef] [Green Version]
  27. Bénabou, R.; Tirole, J. Willpower and Personal Rules. J. Political Econ. 2004, 112, 848–886. [Google Scholar] [CrossRef]
  28. Hong, F.; Huang, W.; Zhao, X. Sunk Cost as a Self-Management Device. Manag. Sci. 2018, 65, 1949–2443. [Google Scholar] [CrossRef]
  29. Schacter, D.L. Searching for Memory: The Brain, the Mind, and the Past; Basic Books: 1996. Available online: https://psycnet.apa.org/record/1996-97906-000 (accessed on 31 December 2022).
  30. Baddeley, A.; Eysenck, M.W.; Anderson, M.C. Memory; Routledge: Oxford, UK, 2020; Available online: https://www.taylorfrancis.com/books/edit/10.4324/9780429449642/memory-alan-baddeley-michael-eysenck-michael-anderson (accessed on 31 December 2022).
  31. Chew, S.H.; Huang, W.; Zhao, X. Motivated False Memory. J. Political Econ. 2020, 128, 3913–3939. [Google Scholar] [CrossRef]
  32. Hagenbach, J.; Koessler, F. Selective Memory of a Psychological Agent. Eur. Econ. Rev. 2022, 142, 104012. [Google Scholar] [CrossRef]
  33. Fudenberg, D.; Lanzani, G.; Strack, P. Selective Memory Equilibrium. Working Paper, SSRN. 2022. Available online: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4015313 (accessed on 2 March 2022).
  34. Li, K.K. Asymmetric memory recall of positive and negative events in social interactions. Exp. Econ. 2013, 16, 248–262. [Google Scholar] [CrossRef]
  35. Carlson, R.W.; Maréchal, M.A.; Oud, B.; Fehr, E.; Crockett, M.J. Motivated misremembering of selfish decisions. Nat. Commun. 2020, 11, 2100. [Google Scholar] [CrossRef]
  36. Saucet, C.; Villeval, M.C. Motivated memory in dictator games. Games Econ. Behav. 2019, 117, 250–275. [Google Scholar] [CrossRef] [Green Version]
  37. Galeotti, F.; Saucet, C.; Villeval, M.C. Unethical amnesia responds more to instrumental than to hedonic motives. Proc. Natl. Acad. Sci. USA 2020, 117, 25423–25428. [Google Scholar] [CrossRef]
  38. Han, R.; Shi, J.; Yong, W.; Wang, W. Intelligence and Prosocial Behavior: Do Smart Children Really Act Nice? Curr. Psychol. 2012, 31, 88–101. [Google Scholar] [CrossRef]
  39. Guo, Q.; Sun, P.; Cai, M.; Zhang, X.; Song, K. Why are smarter individuals more prosocial? A study on the mediating roles of empathy and moral identity. Intelligence 2019, 75, 1–8. [Google Scholar] [CrossRef]
  40. Gödker, K.; Jiao, P.; Smeets, P. Investor Memory. Working Papers. 2022. Available online: https://ssrn.com/abstract=3348315 (accessed on 30 August 2022).
  41. Zimmermann, F. The Dynamics of Motivated Beliefs. Am. Econ. Rev. 2020, 110, 337–361. [Google Scholar] [CrossRef] [Green Version]
  42. Huffman, D.; Raymond, C.; Shvets, J. Persistent Overconfidence and Biased Memory: Evidence from Managers. Am. Econ. Rev. 2022, 112, 3141–3175. [Google Scholar] [CrossRef]
  43. Müller, M.W. Selective Memory around Big Life Decisions. Working Papers. 2022. Available online: https://fass.nus.edu.sg/ecs/wp-content/uploads/sites/4/2022/08/MuellerMW_SelectiveMemory.pdf (accessed on 30 August 2022).
  44. Roy-Chowdhury, V. Self-Confidence and Motivated Memory Loss: Evidence from Schools. Cambridge Working Papers in Economics 2213, Faculty of Economics, University of Cambridge. 2022. Available online: https://ideas.repec.org/p/cam/camdae/2213.html (accessed on 30 August 2022).
  45. Sial, A.Y.; Sydnor, J.R.; Taubinsky, D. Biased Memory and Perceptions of Self-Control. Working Paper 30825, National Bureau of Economic Research. 2023. Available online: https://www.nber.org/system/files/working_papers/w30825/w30825.pdf (accessed on 30 August 2022).
  46. Ertaç, S. Does self-relevance affect information processing? Experimental evidence on the response to performance and non-performance feedback. J. Econ. Behav. Organ. 2011, 80, 532–545. [Google Scholar] [CrossRef] [Green Version]
  47. Coutts, A. Good news and bad news are still news: Experimental evidence on belief updating. Exp. Econ. 2019, 22, 369–395. [Google Scholar] [CrossRef] [Green Version]
  48. Barron, K. Belief updating: Does the ‘good-news, bad-news’ asymmetry extend to purely financial domains? Exp. Econ. 2021, 24, 31–58. [Google Scholar] [CrossRef] [Green Version]
  49. Coffman, K.B.; Ugalde Araya, P.; Zafar, B. A (Dynamic) Investigation of Stereotypes, Belief-Updating, and Behavior. Working Paper 29382, National Bureau of Economic Research. 2021. Available online: https://www.nber.org/system/files/working_papers/w29382/w29382.pdf (accessed on 31 August 2022).
  50. Graeber, T.; Roth, C.; Zimmermann, F. Stories, Statistics and Memory. Working Paper. 2022. Available online: https://thomasgraeber.com/assets/papers/Stories_Statistics_Memory.pdf (accessed on 31 December 2022).
  51. Enke, B.; Schwerter, F.; Zimmermann, F. Associative Memory and Belief Formation. Working Paper 26664, National Bureau of Economic Research. 2022. Available online: https://www.nber.org/system/files/working_papers/w26664/w26664.pdf (accessed on 31 December 2022).
  52. Bordalo, P.; Conlon, J.J.; Gennaioli, N.; Kwon, S.Y.; Shleifer, A. Memory and Probability*. Q. J. Econ. 2022, 138, 265–311. [Google Scholar] [CrossRef]
  53. Kahana, M.J. Foundations of Human Memory; Oxford University Press: Oxford, UK, 2012; Available online: https://global.oup.com/academic/product/foundations-of-human-memory-9780195333244?cc=cn&lang=en& (accessed on 31 December 2022).
  54. Bordalo, P.; Gennaioli, N.; Shleifer, A. Memory, Attention, and Choice. Q. J. Econ. 2020, 135, 1399–1442. [Google Scholar] [CrossRef]
Table 1. Empirical papers summary table. The Delay column reports the delay between the elicitation of the object to be recalled and the recall task. Announced Recall states whether the recall task was announced to participants beforehand. Continuous Recall Object concerns the type of information that participants are asked to retrieve from memory: a binary information (e.g. the amount was fair/unfair) or a continuous one (e.g. a specific amount, a percentage). The Incentivized Recall column provides information about whether the memory task was financially incentivized. Finally, the Causal Identification column concerns the presence, or absence, of an exogenous manipulation of motivated behavior which allows for establishing causality by design.
Table 1. Empirical papers summary table. The Delay column reports the delay between the elicitation of the object to be recalled and the recall task. Announced Recall states whether the recall task was announced to participants beforehand. Continuous Recall Object concerns the type of information that participants are asked to retrieve from memory: a binary information (e.g. the amount was fair/unfair) or a continuous one (e.g. a specific amount, a percentage). The Incentivized Recall column provides information about whether the memory task was financially incentivized. Finally, the Causal Identification column concerns the presence, or absence, of an exogenous manipulation of motivated behavior which allows for establishing causality by design.
DelayAnnounced RecallContinuous Recall ObjectIncentivized RecallCausal Identification
Li (2013)Treatment dependent: 0, 7 and 43 days
Carlson et al. (2018)Participant dependent: ∼5 min
Chew et al. (2018)Months (not further specified)Not specified
Saucet and Villleval (2019)8 minTreatment dependent: ✓/✗
Galeotti et al. (2020)3 weeks
Zimmermann (2020)Treatment dependent: No Delay or 1 monthTreatment Dependent: ✓/✗
Gödker et al. (2021)Treatment dependent: 0 and 1 week
Huffman et al. (2021)3 months
Müller (2021)10 years
Roy-Chowdhury (2022)Treatment dependent: 5 months and 3 weeks
Sial et al. (2023)1 to 20 weeks
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Amelio, A.; Zimmermann, F. Motivated Memory in Economics—A Review. Games 2023, 14, 15. https://doi.org/10.3390/g14010015

AMA Style

Amelio A, Zimmermann F. Motivated Memory in Economics—A Review. Games. 2023; 14(1):15. https://doi.org/10.3390/g14010015

Chicago/Turabian Style

Amelio, Andrea, and Florian Zimmermann. 2023. "Motivated Memory in Economics—A Review" Games 14, no. 1: 15. https://doi.org/10.3390/g14010015

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop