Exploiting the guilt aversion of others: do agents do it and is it effective?

Abstract

The general idea of guilt aversion is that agents may be motivated to avoid letting others down, even at the expense of their own material payoff. Several experimental studies have documented behavior that is consistent with agents exhibiting guilt averse motivations in social interactions. However, there are strategic implications of guilt aversion, which can impact economic outcomes in important ways, that have yet to be explored. I introduce a game that admits the possibility for agents to induce guilt upon others in a manner consistent with the method posited by Baumeister et al. (Psychol Bull 115:243–267, 1994). This game enables me to experimentally test whether agents attempt to exploit the guilt aversion of others by inducing guilt upon them, and whether agents are actually susceptible to this exploitation. Additionally, the design enables me to test whether agents exhibit higher degrees of trust when they are given such an opportunity to exploit the guilt aversion of others. The data suggest that agents do not attempt to fully exploit the guilt aversion of other agents by inducing guilt upon them; however, the data suggest that agents would have been susceptible to guilt induction.

Appendices

Appendix A: Application of B&D model of guilt

In this Appendix, I formalize the arguments put forth in Sect. 2.3 regarding the connection between the posited method by BSH (1994) of how guilt is induced and the formal model of guilt developed by B&D. In doing so, I first provide a brief outline of the B&D model of guilt, followed by its application to $\Gamma _{\mathrm{PPT}}$. I then derive conditions under which the BSH (1994) method is consistent with predictions of the B&D model of guilt. I conclude by showing that the BSH (1994) method can be supported as an equilibrium in $\Gamma _{\mathrm{PPT}}$ under the framework of the B&D model. I conclude by briefly discussing the recent extension of the B&D model developed by Attanasi et al. (2015) that incorporates incomplete information about guilt sensitivities, and how incomplete information about guilt sensitivities can impact behavioral predictions in $\Gamma _{\mathrm{PPT}}$.

1.1 B&D model of guilt applied to $\Gamma _{\mathrm{PPT}}$

Before I proceed in formally applying the B&D model of guilt to $\Gamma _{\mathrm{PPT}},$ I first provide a general overview of the B&D model of simple guilt.³⁰ What follows is only a simplified outline of the model. Interested readers should refer to B&D for the full, technical presentation of the model including illustrative examples. Informally, the model posits that agents suffer disutility, in the form of guilt, from failing to live up to others’ expectations. This is captured by modeling an agent’s utility as a function of his/her own material payoffs and the extent to which he/she let other agents down.

Formally, simple guilt is modeled by specifying a utility function for player i given by:

$$\begin{aligned} u_{i}^{\mathrm{SG}}=m_{i}-\underset{j\ne i}{\sum }\theta _{ij}\cdot D_{j} \end{aligned}$$

(Simple Guilt Utility)

In this expression, $m_{i}$ represents player i’s material payoff, and $\underset{j\ne i}{\sum }\theta _{ij}\cdot D_{j}$ represents player i’s disutility from simple guilt. The latter component is composed of two pieces. The first, $\theta _{ij}$, is an exogenously given constant that measures player i’s sensitivity of feeling guilty toward player j. The second, $D_{j}$, represents the amount by which player i lets player j down, as a result of player i’s strategy. $D_{j}=E_{j}-m_{j}$ is expressed as the difference between the material payoff that player j was expecting, $E_{j}$, and the material payoff that player j actually receives, $m_{j}.$ $E_{j}$ itself is a function of player j’s strategy, and player j’s vector of “first-order” beliefs regarding the strategies of the other players. Note, player i does not actually observe $D_{j}$, as it is a function of player j’s first-order beliefs. Therefore, it is assumed that player i maximizes the expected value of $u_{i}^{\mathrm{SG}}$, given player i’s first-order beliefs regarding player j’s strategy, and player i’s “second-order” belief regarding player i’s first-order beliefs.

In proceeding with the application of the B&D model of simple guilt to $\Gamma _{\mathrm{PPT}}$, I derive the guilt that Player B would experience from choosing Left in $\Gamma _{\mathrm{PPT}}$ at each of the three possible conveyance states (histories): (i) $X=0$ was conveyed, which I denote $C^{0}$, (ii) $X=6$ was conveyed, which I denote $C^{6}$, and (iii) the value of X was not conveyed, which I denote $C^{N}$. A strategy for Player B is a probability distribution over Player B’s possible actions, {Left, Right}, at each of the three possible conveyance states. In deriving this guilt that Player B would suffer from choosing Left, it is necessary to first derive Player A’s material expectation. Before Player A makes her initial In or Out decision, Player A forms an initial first-order belief regarding Player B’s strategy, which can be represented as the probabilities that Player B would choose Right at each of the three conveyance states; I denote this vector of probabilities as ${\varvec{\alpha }}_{A}=(\Pr ({ Right }|C^{0}),\Pr ({ Right }|C^{N}),\Pr ({ Right }|C^{6})).$ For a given strategy, Player A forms an initial expectation, which I denote $E_{A},$ weighted over ${\varvec{\alpha }}_{A}$ and moves by Nature, of her material payoff.

Player B experiences disutility from simple guilt when he chooses a strategy that yields a payoff to Player A that is lower than $E_{A}$. However, Player B does not observe $E_{A}.$ Therefore, Player B must form an expectation of $E_{A},$ conditional on the conveyance state. I define this conditional expectation of $E_{A}$ as $E_{B}|h$ where $h\in \{C^{0},C^{N},C^{6}\}$, which is implicitly a function of Player B’s conditional second-order beliefs regarding ${\varvec{\alpha }}_{A}$. The guilt that Player B would suffer from choosing Left is proportional to the difference between $E_{B}|h$ and $m_{A},$ where $m_{A}$ is the material payoff that Player A actually receives as a result of Player B’s Left decision. Let $\theta _{B}\ge 0$ denote Player B’s sensitivity to feeling guilty. Below is the amount of simple guilt that Player B would suffer from choosing Left at each of the three possible conveyance states.

• $X=0$ was conveyed: Because the conveyance is credible, Player B knows that if he chooses Left, Player A will receive a payoff of $m_{A}=0$. Player B’s expectation of Player A’s expectation is $E_{B}|C^{0}.$ By choosing Left, Player B will suffer disutility from guilt equal to $\theta _{B}\cdot (E_{B}|C^{0}-0).$ Thus, Player B’s utility from choosing Left, after $X=0$ was conveyed, is equal to:

  $$\begin{aligned} 6-\theta _{B}\cdot (E_{B}|C^{0}-0) \end{aligned}$$

• $X=6$ was conveyed: Because the conveyance is credible, Player B knows that if he chooses Left, Player A will receive a payoff of $m_{A}=6$. Player B’s expectation of Player A’s expectation is $E_{B}|C^{6}.$ By choosing Left, Player B will suffer disutility from guilt equal to $\theta _{B}\cdot (E_{B}|C^{6}-6).$ Thus, Player B’s utility from choosing Left, after $X=0$ was conveyed, is equal to:

  $$\begin{aligned} 6-\theta _{B}\cdot (E_{B}|C^{6}-6) \end{aligned}$$

• Value of X not conveyed: If the value of X is not conveyed, then Player B must think about the expected material payoff that Player A would receive if he chooses Left. Let ${\widehat{m}}_{A}=E_{B}[m_{A}|C^{N}]$ denote this expectation. As I have previously shown (see Sect. 2.1), ${\widehat{m}}_{A}\in [1,5]$. Player B’s expectation of Player A’s expectation is $E_{B}|C^{N}.$ By choosing Left, Player B will suffer disutility from guilt equal to: $\theta \cdot (E_{B}|C^{N}-{\widehat{m}}_{A})$ where ${\widehat{m}}_{A}\in [1,5].$ Thus, Player B’s utility from choosing Left, after the value of X was not conveyed, is equal to:

  $$\begin{aligned} 6-\theta _{B}\cdot (E_{B}|C^{N}-{\widehat{m}}_{A})\quad \hbox {where } {\widehat{m}}_{A}\in [1,5] \end{aligned}$$

1.2 Guilt induction in $\Gamma _{\mathrm{PPT}}$ via the B&D model of guilt

Based on the BSH (1994) method for how an agent can induce guilt upon another, it was derived that Player A could attempt to induce guilt upon Player B in $\Gamma _{\mathrm{PPT}}$ by choosing to Convey $X=0$ and Not Convey $X=6$. In what follows, I derive conditions under which this is consistent with predictions of the B&D model of simple guilt. From the previous section, we derived that the disutilities, from feeling guilt, that Player B would suffer from choosing Left at each of the three possible conveyance states are:

• $X=0$ was conveyed: $\theta _{B}\cdot (E_{B}|C^{0}-0)$

• $X=6$ was conveyed: $\theta _{B}\cdot (E_{B}|C^{6}-6)$

• Value of X not conveyed: $\theta _{B}\cdot (E_{B}|C^{N}-{\widehat{m}}_{A})$ where ${\widehat{m}}_{A}\in [1,5]$

According to B&D, Player B would suffer more disutility from choosing Left when $X=0$ was conveyed, compared to when the value of X was not conveyed, when:

$$\begin{aligned} \theta _{B}\cdot (E_{B}|C^{0}-0)\ge \theta _{B}\cdot (E_{B}|C^{N}-\widehat{m }_{A})\Longrightarrow E_{B}|C^{0}\ge E_{B}|C^{N}-{\widehat{m}}_{A} \end{aligned}$$

(Condition 1)

Similarly, Player B would suffer more disutility from choosing Left when the value of X was not conveyed, compared to when $X=6$ was conveyed, when:

$$\begin{aligned} \theta _{B}\cdot (E_{B}|C^{N}-{\widehat{m}}_{A})\ge \theta _{B}\cdot (E_{B}|C^{6}-6)\Longrightarrow E_{B}|C^{N}-{\widehat{m}}_{A}\ge E_{B}|C^{6}-6 \end{aligned}$$

(Condition 2)

Essentially, Condition 1 states that Player B’s second-order belief of Player A’s expectation after $X=0$ is conveyed, $E_{B}|C^{0},$ is not too much lower than Player B’s second-order belief of Player A’s expectation after the value of X is not conveyed, $E_{B}|C^{N}.$ Similarly, Condition 2 states that Player B’s second-order belief of Player A’s expectation after the value of X is not conveyed, $E_{B}|C^{N},$ is not too much lower than Player B’s second-order belief of Player A’s expectation after $X=6$ is conveyed, $E_{B}|C^{6}.$ Conditions 1 and 2 would certainly be satisfied if we assumed that Player B does not update his belief of Player A’s expectation (i.e., $E_{B}|C^{0}=E_{B}|C^{N}=E_{B}|C^{6}$), which would be satisfied in an equilibrium. However, making the assumption that $E_{B}|C^{0}=E_{B}|C^{N}=E_{B}|C^{6}$ is clearly stronger than is needed for Condition 1 and 2 to be satisfied.

If Conditions 1 and 2 are satisfied, then the B&D model predicts that Player B would feel more guilt from choosing left after $X=0$ was conveyed and the value of X was not conveyed, compared to when the value of X was not conveyed and $X=6$ was conveyed, respectively. Therefore, if Condition 1 and 2 hold, then according to the B&D model, Player A can induce guilt upon Player B by choosing to Convey $X=0$ and Not Convey $X=6$, which is consistent with the BSH (1994) method for inducing guilt.

1.3 Guilt induction as an equilibrium of $\Gamma _{\mathrm{PPT}}$

Next, I show that effective guilt induction in $\Gamma _{\mathrm{PPT}}$ can be supported as sequential equilibrium (SE) of $\Gamma _{\mathrm{PPT}}$ under the framework of B&D.³¹ The intuition behind this rests in the fact that in an SE, an assessment (profile of behavioral strategies and conditional hierarchical beliefs) will be consistent. Battigalli and Dufwenberg (2009) show that in equilibrium, “players never change their beliefs about the conditional beliefs that the opponents would hold at each h (history)” (pp. 16). Thus, in equilibrium, there is no belief updating. It follows that $E_{B}|C^{0}=E_{B}|C^{N}=E_{B}|C^{6}$, which implies that Conditions 1 and 2 from above would be satisfied in an equilibrium. In this equilibrium analysis, I will assume that there is complete information regarding Player B’s guilt sensitivity, $\theta _{B}$. While this assumption may be unrealistic, and not a feature of the design, it is common in the extant literature related to belief-dependent motivations (e.g., Charness and Dufwenberg 2006, 2011; Dufwenberg et al. 2013); furthermore, it allows me to focus on the motivation of this analysis, which is to establish that guilt induction can be supported as an equilibrium, without the information complexities that would otherwise need to be addressed. I also assume that Player A is not guilt averse, i.e., Player A’s guilt sensitivity is zero.³²

Guilt induction by Player A in $\Gamma _{\mathrm{PPT}}$ is characterized by the choice to Convey $X=0$ and Not Covey $X=6$. Therefore, the strategy (In, Convey $X=0$, Not Convey $X=6$) for Player A is consistent with attempted guilt induction. Effective guilt induction is characterized by Player B choosing Right as a response to the guilt induction by Player A. The following two strategies for Player B are consistent with responding in kind to Player A’s guilt induction: (Right $\vert C^{0}$, Right $\vert C^{N}$, Left $\vert C^{6}$), and (Right $\vert C^{0}$, Left $\vert C^{N}$, Left $\vert C^{6}$). Thus, the strategy profiles ((In, Convey $X=0$, Not Convey $X=6$), (Right $\vert C^{0}$, Right $\vert C^{N}$, Left $\vert C^{6}$)), and ((In, Convey $X=0$, Not Convey $X=6$), (Right $\vert C^{0}$, Left $\vert C^{N}$, Left $\vert C^{6}$)) are the candidate equilibrium profiles for effective guilt induction in $\Gamma _{\mathrm{PPT}}$. In Claims 1 and 2 below, I show that each of these strategy profiles can be supported as an equilibrium of $\Gamma _{\mathrm{PPT}}$.

Claim 1

The strategy profile ((In, Convey $X=0$, Not Convey $X=6$), (Right $\vert C^{0}$, Right $\vert C^{N}$, Left $\vert C^{6}$)) can be supported as a SE of $\Gamma _{\mathrm{PPT}}$ for $\theta _{B}\in [\frac{2}{5}, \frac{1}{2}]$.

To verify that this strategy profile is an equilibrium, we need to check that neither player has a profitable deviation. For Player A, this is rather trivial. By following the equilibrium strategy, Player A earns a payoff of 10, which is the highest payoff of the game. Therefore, Player A has no profitable deviation. For Player B, we need to consider deviations at each of the possible conveyance states. Given consistent beliefs in equilibrium, we have that ${\varvec{\alpha }}_{A}={\varvec{\beta }}_{B}=(1,1,0), E_{A}=E_{B}|h=10\ \forall h\in \{C^{0},C^{N},C^{6}\},$ and ${\widehat{m}}_{A}=5.$ Player B will not deviate to Right $\vert C^{6}$ so long as: $6-\theta _{B}\cdot [10-6]\ge 4\Leftrightarrow \theta _{B}\le \frac{1}{2}.$ Player B will not deviate to Left $\vert C^{N}$ so long as: $4\ge 6-\theta _{B}\cdot [10-5]\Leftrightarrow \theta _{B}\ge \frac{2}{5}.$ Similarly, Player B will not deviate to Left $\vert C^{0}$ so long as: $4\ge 6-\theta _{B}\cdot [10-0]\Leftrightarrow \theta _{B}\ge \frac{1}{5}$ which is satisfied if $\theta _{B}\ge \frac{2}{5}$.

Claim 2

The strategy profile ((In, Convey $X=0$, Not Convey $X=6$), (Right $\vert C^{0}$, Left $\vert C^{N}$, Left $\vert C^{6}$)) can be supported as a SE of $\Gamma _{\mathrm{PPT}}$ for $\theta _{B}\in [\frac{2}{7} ,1]$.

Again, to verify that this strategy profile is an equilibrium, we need to check that neither player has a profitable deviation. For Player A, playing the equilibrium strategy yields an expected payoff of 7; therefore, Player A cannot profitably deviate to Out. Player A would not deviate and Not Convey $X=0$, which would result in a payoff of 0 compared to a payoff of 10 from following the equilibrium strategy to Convey $X=0$. Player A is indifferent between Convey $X=6$ and Not Convey $X=6.$ Therefore, Player A has no profitable deviation from the prescribed equilibrium strategy. For Player B, we need to consider deviations at each of the possible conveyance states. Given consistent beliefs in equilibrium, we have that ${\varvec{\alpha }}_{A}={\varvec{\beta }}_{B}=(1,0,0), E_{A}=E_{B}|h=7\ \forall h\in \{C^{0},C^{N},C^{6}\},$ and ${\widehat{m}}_{A}=5.$ Player B will not deviate to Right $\vert C^{6}$ so long as: $6-\theta _{B}\cdot [7-6]\ge 4\Leftrightarrow \theta _{B}\le 2$. Player B will not deviate to Right $\vert C^{N}$ so long as: $6-\theta _{B}\cdot [7-5]\ge 4\Leftrightarrow \theta _{B}\le 1.$ Similarly, Player B will not deviate to Left $\vert C^{0}$ so long as: $4\ge 6-\theta _{B}\cdot [7-0]\Leftrightarrow \theta _{B}\ge \frac{2}{7}$.

1.4 Incomplete information regarding sensitivity to guilt

As I stated in the previous section, equilibrium Claims 1 and 2 assume common knowledge of Player B’s sensitivity to guilt, $\theta _{B}.$ In a recent paper, Attanasi et al. (2015) have extended the B&D model to allow for incomplete information regarding guilt sensitivities in a 2-player trust game.³³ A formal application of their model to $\Gamma _{\mathrm{PPT}}$ would require extensive theoretical preliminaries regarding the formulation of hierarchies of beliefs for each player via “type” structures, which is beyond the scope of this paper and would not add a proportionate increase in the main conclusions drawn. In addition, because I do not elicit any information about beliefs or types in the experiment (as the sessions were run prior the Attanasi et al. 2015 paper), I would be unable to derive precise predictions of the incomplete information model. As a result, I forgo a formal equilibrium analysis of $\Gamma _{\mathrm{PPT}}$ under incomplete information; rather, I appeal to insights from the Attanasi et al. (2015) paper to provide some general discussion regarding how incomplete information about types could impact behavior in $\Gamma _{\mathrm{PPT}}.$ ³⁴

The main difference between the complete information and incomplete information extension of the guilt aversion model (as is most relevant to the discussion that follows) is that in an incomplete information setting where Player As and Player Bs are matched at random (as in the current experimental procedure), it is likely that participants hold heterogeneous and dispersed beliefs (Attanasi et al. 2013, p. 18). The important behavioral implication is that under complete information, there is likely to be much more polarization of behavior because the common knowledge of $\theta _{B}$ enables coordination. Specifically, if $\theta _{B}$ is small enough then Player B will not be susceptible to guilt induction and will choose Left, and Player A will choose Out (the standard payoff maximizing equilibrium outcome); on the other hand, if $\theta _{B}$ is sufficiently large then Player B will be susceptible to guilt induction by Player A, and Player A could increase his/her payoff by choosing In and then inducing guilt (consistent with Claims 1 and 2 above).

However, with incomplete information about $\theta _{B}$ (a reasonable assumption with regard to Player As in the experiment) more dispersion in beliefs and mis-coordination can arise, which can possibly lead to interesting patterns of behavior in $\Gamma _{\mathrm{PPT}}$. Namely, many Player As could have held beliefs that Player Bs were not sensitive to guilt (i.e., low $\theta _{B}$), which may have motivated them to choose Out, which could explain why there was only a marginal 14 % increase in In rates between $\Gamma _{\mathrm{PPT}}$ and $\Gamma _{\mathrm{UPT}}.$ Additionally, for those Player As that did choose In (for other possible reasons besides attempting to exploit the guilt aversion of Player B), they would not be motivated to convey $X=0$ and/or not convey $X=6$ as a means of inducing guilt, as it would not be effective given their beliefs about $\theta _{B}$ (which is consistent with the analysis of the questionnaire data in Table 4); these Player As may opt, for example, to Convey $X=6$ to display honesty to Player B, consistent with the discussion in Sect. 4.3. On the other extreme, some Player As could have believed that Player Bs were very sensitive to feeling guilt (i.e., high $\theta _{B}$), such that Player B would be motivated to choose Right regardless of the conveyance state; in this case, these Player As would, again, not necessarily be motivated to convey $X=0$ and/or not convey $X=6$ as a means of inducing guilt, as it would not be necessary to motivate Player B to choose Right.

If we assume that Player As have incomplete information about Player B’s guilt sensitivity, $\theta _{B},$ then multiple equilibria can arise in $\Gamma _{\mathrm{PPT}}$, as in the case with perfect information. In particular, the “selfish” equilibrium exists that results in Player A always choosing Out since Player B would always choose Left, if Player Bs have sufficiently low $\theta _{B}$s.³⁵ In addition, by placing some restrictions on the distribution of $\theta _{B}$ (e.g., assuming $\theta _{B}\in [\frac{2}{7},1]$), equilibria that correspond to effective guilt induction (those specified in Claims 1 and 2 above) can also arise. However, with incomplete information regarding $\theta _{B}$, players do not have the ability to coordinate based on $\theta _{B}$; hence, it is much more likely to observe heterogeneity in behavior because of the dispersion in beliefs (as shown by Attanasi et al. 2013). As a result, the fact that participants playing the role of Player A posses incomplete information regarding Player B’s sensitivity to guilt could explain some of the observed patterns in the data; namely, why only marginally more Player As chose In in $\Gamma _{\mathrm{PPT}}$, as well as the conveyance behavior inconsistent with attempted guilt induction, where Player As choose to Not Convey $X=0$ and Convey $X=6$.

Appendix B.1: Copy of experimental instructions

1.1 Sample instructions: PPT treatment

Welcome and thank you for participating. Your participation is VOLUNTARY, and you may leave at any time. Feel free to raise your hand and ask questions at any time, and you may refer back to these instructions at any time during the session. Please remain seated and quiet for the remainder of the session. All decisions are to be completed individually and interaction with other participants is strictly PROHIBITED. Thank you for your cooperation.

Each person will receive a $5 show-up payment for participating. In addition, you can receive additional compensation based on the decision(s) that are made in the decision task described below. After the task is complete, you will be privately paid the amount of money you have earned. Upon completions of the decision task, please remain quietly seated in your carrel until you have been paid.

1.1.1 The decision task

You will be participating in a 2-person decision task. Each person will be randomly and anonymously paired with another person in the lab. In each of the 2-person decision making pairs, one person will be randomly assigned the role of PLAYER A and the other person will be randomly assigned the role of PLAYER B. You will remain in your assigned role for the entire session. The earnings of each Player will depend on the decision(s) he/she makes, and/or the decision(s) of the Player with whom they are paired. A brief outline of steps of the decision task will first be provided, followed by a detailed description of each step and the corresponding earnings for Player.

Step 1: PLAYER A begins by first choosing IN or OUT

• If PLAYER A chooses OUT, the task ends

• If PLAYER A chooses IN, then the task proceeds to Step 2

Step 2: PLAYER A might privately learn some payoff information that was initially unknown to both players. If PLAYER A does learn the information, the PLAYER A will then have an opportunity to convey the information to PLAYER B. Then the task will proceed to Step 3.

Step 3: PLAYER B chooses between RIGHT or LEFT, and the task ends

A full description of each step and the corresponding earnings for Player follows:

• 1. PLAYER A first chooses between IN or OUT.

  • If PLAYER A chooses OUT, then the decision task ends. PLAYER A will receive $6 and PLAYER B will receive $2.

  • If PLAYER A chooses IN, the task proceeds to step (2) where PLAYER A might privately learn the unknown information, and then have an opportunity to convey that information to PLAYER B. After step (2), the task will proceed to step (3) where PLAYER B will then be asked to decide between RIGHT or LEFT.

• 2. I postpone the details about the information that PLAYER A can possible learn, and convey to PLAYER B until after step (3) is described. Describing step (3) first will help clarify step (2).

• 3. If PLAYER A chooses IN, at step (1), then PLAYER B must choose between RIGHT or LEFT.

  • If PLAYER B chooses RIGHT, then the decision task ends. PLAYER A will receive $10 and PLAYER B will receive $4.

  • If PLAYER B chooses LEFT, then the decision task ends and PLAYER A will receive $X and PLAYER B will receive $6. There is a 50 % chance that X $=$ $0 and a 50 % chance that X $=$ $6. That is, X $=$ $0 and X $=$ $6 are equally likely.

NOTE: When the decision task begins, neither PLAYER A nor PLAYER B knows the value of X. Therefore, PLAYER A does not know the value of X when he/she decides between IN or OUT in step (1).

Back to the description of step (2):

• 2. If PLAYER A chooses IN, there is an 80 % chance that PLAYER A will privately learn the value of X, and a 20 % chance that PLAYER A will not privately learn the value of X.

  • If PLAYER A does learn the value of X (80 % chance), then PLAYER A must then decide whether or not to convey the value of X to PLAYER B before PLAYER B makes his/her decision in step (3).

    • If PLAYER A does convey the value of X, then PLAYER B will know the value of X before he/she decides between RIGHT or LEFT in step (3).

    • If PLAYER A does not convey the value of X, then PLAYER B will not know the value of X before he/she decides between RIGHT or LEFT in step (3).

  • If PLAYER A does not learn the value of X (20 % chance), then PLAYER A will not have an opportunity to convey the value of X to PLAYER B. The task will proceed to step (3) where PLAYER B will then choose between RIGHT or LEFT without knowing the value of X.

If PLAYER A chose to convey the value of X, then when PLAYER B makes his/her decision in step (3), the following message will appear on PLAYER B’s screen:

“PLAYER A has chosen to convey the value of X to you”

“The value of X is: [actual value of X]”

If either (a) PLAYER A did not learn the value of X after choosing IN, or (b) PLAYER A did learn the value of X but chose not to convey the value of X, then when it is time for PLAYER B to make his/her decision in step (3), the following message will appear on PLAYER B’s screen:

“The value of X remains unknown”

Payoff Table:

The table below summarizes the earnings of each Player for each of the possible outcomes in the decision task:

DECISION OUTCOME	Earnings of PLAYER A	Earnings of PLAYER B
PLAYER A chooses OUT	$6	$2
PLAYER A first chooses IN and then:
PLAYER B chooses RIGHT	$10	$4
PLAYER B chooses LEFT	$X	$6
There is a 50 % chance X = 0 and a 50 % chance X = 6

Each person will participate in this decision making task ONE time. After the task has ended, the decision(s) of each Player and the corresponding earnings of each Player will be revealed to both Players. Additionally, the value of X will be revealed to both PLAYER A and PLAYER B regardless of the decisions made in the task. You will then be asked to fill out a short questionnaire that will take about 3 min to complete. Your answers to the questionnaire are confidential and will not be shared with any other participants. After completion of the questionnaire, an Experimenter will then come by and privately pay you your total experimental earning which equals your earnings from the decision task PLUS the $5 show-up payment. After you have been paid, you may quietly exit the lab.

1.2 Appendix B.2: Sample screen shots

Player A’s Initial In/Out Decision

Player A’s Conditional Convey/Not Convey Decision

Player B’s Conditional Left/Right Decision (X Known)

Player B’s Conditional Left/Right Decision (X Unknown)