Accepted Manuscript

The two faces of cooperation: On the unique role of HEXACO Agreeableness

for forgiveness versus retaliation

Benjamin E. Hilbig, Isabel Thielmann, Sina A. Klein, Felix Henninger

PII: S0092-6566(16)30100-3
DOI: http://dx.doi.org/10.1016/j.jrp.2016.08.004
Reference: YJRPE 3593

To appear in: Journal of Research in Personality

Received Date: 24 June 2016

Revised Date: 13 August 2016
Accepted Date: 17 August 2016

Please cite this article as: Hilbig, B.E., Thielmann, I., Klein, S.A., Henninger, F., The two faces of cooperation: On
the unique role of HEXACO Agreeableness for forgiveness versus retaliation, Journal of Research in Personality
(2016), doi: http://dx.doi.org/10.1016/j.jrp.2016.08.004

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Running Head: HEXACO Agreeableness 1

The two faces of cooperation: On the unique role of HEXACO Agreeableness for forgiveness

versus retaliation

Benjamin E. Hilbig*

University of Koblenz-Landau and

Max Planck Institute for Research on Collective Goods

Isabel Thielmann

University of Koblenz-Landau

Sina A. Klein

University of Koblenz-Landau

Felix Henninger

University of Koblenz-Landau and

Max Planck Institute for Research on Collective Goods

* Correspondence:

Cognitive Psychology Lab, Department of Psychology, University of Koblenz-Landau

Street address: Fortstraße 7, D-76829 Landau, Germany

Phone: +49 (0)6341 280 34234, Fax: +49 (0)6341 280 34240
Running Head: HEXACO Agreeableness 2

ABSTRACT

Cooperation requires a tendency for fairness (versus exploitation) and for forgiveness (versus

retaliation). Exactly these tendencies are distinguished in the HEXACO model of personality

which attributes the former to Honesty-Humility (HH) and the latter to Agreeableness (AG).

However, empirical dissociations between these basic traits have primarily supported the

substantial and unique role of HH, whereas the picture for AG has remained somewhat

inconclusive. To overcome limitations of prior studies, we introduce an economic paradigm, the

uncostly retaliation game, to more conclusively test the unique role of AG for forgiveness versus

retaliation. In two fully incentivized experiments, we found that AG (and not HH) indeed

negatively predicts retaliation decisions in the face of prior exploitation. Furthermore, the results

confirm that the paradigm provides a more direct measure of retaliation (beyond individual

payoff-concerns and social preferences such as inequality aversion) than previous measures and

that it may thus serve future investigations into the reactive aspect of cooperation.

Keywords: Cooperation; fairness; forgiveness; HEXACO; Agreeableness; Honesty-Humility;

economic games
Running Head: HEXACO Agreeableness 2

ABSTRACT

Cooperation requires a tendency for fairness (versus exploitation) and for forgiveness (versus

retaliation). Exactly these tendencies are distinguished in the HEXACO model of personality

which attributes the former to Honesty-Humility (HH) and the latter to Agreeableness (AG).

However, empirical dissociations between these basic traits have primarily supported the

substantial and unique role of HH, whereas the picture for AG has remained somewhat

inconclusive. To overcome limitations of prior studies, we introduce an economic paradigm, the

uncostly retaliation game, to more conclusively test the unique role of AG for forgiveness versus

retaliation. In two fully incentivized experiments, we found that AG (and not HH) indeed

negatively predicts retaliation decisions in the face of prior exploitation. Furthermore, the results

confirm that the paradigm provides a more direct measure of retaliation (beyond individual

payoff-concerns and social preferences such as inequality aversion) than previous measures and

that it may thus serve future investigations into the reactive aspect of cooperation.

Keywords: Cooperation; fairness; forgiveness; HEXACO; economic games

Running Head: HEXACO Agreeableness 3

INTRODUCTION

Cooperation and pro-social behavior are vital pillars of societal functioning and

commonly considered an essential aspect of human nature (Bowles & Gintis, 2011). Whereas the

evolution of cooperation originally constituted a puzzle to researchers from diverse disciplines,

seminal research revealed that cooperation can indeed evolve based on plausible interaction

strategies between agents (Axelrod & Dion, 1988; Axelrod & Hamilton, 1981; Nowak, 2006).

Importantly, strategies that allow for cooperation to evolve and that are successful with respect to

the long-term outcomes for all agents (e.g. “tit-for-tat” and variants thereof, Nowak, 2006)

include two general behavioral tendencies: A willingness to cooperate initially (i.e., a cooperative

action) and a willingness to reinstate cooperation even after defection (i.e., a cooperative

reaction). It is this very distinction that recently formed part of the theoretical basis (Ashton &

Lee, 2001) for what is currently one of the most prominent models of basic personality structure,

the HEXACO model (Ashton & Lee, 2007, 2008a; Ashton, Lee, & De Vries, 2014).

In lexical studies across a broad set of languages, Ashton, Lee, and colleagues

consistently recovered a six-factor structure of trait adjectives (Ashton et al., 2004; Lee &

Ashton, 2008), giving rise to their corresponding six-factor personality model (Honesty-Humility,

Emotionality, eXtraversion, Agreeableness, Conscientiousness, and Openness, thus HEXACO).

Therein, they explicitly distinguish between basic tendencies of (active) fairness versus

exploitation – subsumed under the Honesty-Humility (HH) factor – and (reactive) forgiveness

versus retaliation – subsumed under the Agreeableness (AG) factor – as complementary aspects

of reciprocal altruism. Specifically, Honesty-Humility characterizes individuals “cooperating

with others even when one might exploit them without suffering retaliation” (Ashton & Lee,

2007, p. 156), whereas Agreeableness refers to those “cooperating with others even when one

might be suffering exploitation by them” (Ashton & Lee, 2007, p. 156). This distinction between
Running Head: HEXACO Agreeableness 4

active versus reactive cooperativeness (Hilbig, Zettler, Leist, & Heydasch, 2013; Zhao & Smillie,

2015) exactly mirrors the two main ingredients of strategies underlying the evolution of

cooperation and arguably constitutes the primary difference between the HEXACO model and its

closest predecessor, the Big Five approach (McCrae & Costa, 2008).

Given both the theoretical importance of the HH-AG distinction and the counterargument

that the two are merely aspects of one single broad (Agreeableness) factor as specified in the Big

Five (DeYoung, 2010; McCrae & Costa, 2008; van Kampen, 2012), it is of primary importance

to scrutinize the empirical evidence concerning this distinction. Ashton and Lee’s argument of

the cross-language emergence of six factors and thus, by implication, a distinction between HH

and AG was disputed in some subsequent lexical studies (De Raad, Barelds, Levert, et al., 2010),

though not in others (De Raad et al., 2014). Both this inconclusive picture and especially the

corresponding debate on the number of to-be-distinguished basic traits (Ashton & Lee, 2010; De

Raad, Barelds, Mlačić, et al., 2010) are, in our view, more telling about the inconclusiveness of

lexical approaches for the question at hand than about either of the positions taken in the debate.

Rather, strong evidence for the proposed distinction would require a pattern of dissociations to

the effect that either of the two factors can be exclusively linked to some criteria that the other

cannot account for. In particular, we concur with Zhao and Smillie (2015) that use of “robustly

established behavioral paradigms” will allow for testing “core postulates […] by examining the

theoretical division between honesty-humility and agreeableness through their ‘double

dissociation’”, ultimately representing “a major shift in trait psychology, from mere description

to explanatory models” (p. 294).

On the one hand, there is now ample evidence that HH accounts for diverse criteria that

the HEXACO variant of AG cannot predict. These include crime, delinquency, and

counterproductive work behavior (Dunlop, Morrison, Koenig, & Silcox, 2012; van Gelder & de
Running Head: HEXACO Agreeableness 5

Vries, 2013; Zettler & Hilbig, 2010), dishonesty and cheating (Hershfield, Cohen, & Thompson,

2012; Hilbig & Zettler, 2015; Thielmann, Hilbig, Zettler, & Moshagen, in press), as well as

sexual harassment, sexual “quid pro quos”, and infidelity (Ashton & Lee, 2008b; Hilbig,

Moshagen, & Zettler, 2015; Lee, Gizzarone, & Ashton, 2003) and thus cover a wide range of

immoral, exploitative behaviors that are linked to low HH. Most importantly, HH was repeatedly

shown to positively predict actively fair (versus exploitative) behavior in fully incentivized

economic paradigms and allocation decisions (e.g., Baumert, Schlösser, & Schmitt, 2014; Hilbig,

Glöckner, & Zettler, 2014; Hilbig, Thielmann, Hepp, Klein, & Zettler, 2015; Thielmann et al., in

press), whereas HEXACO AG was largely unrelated to said behavior (see also Ackermann, Fleiß,

& Murphy, 2016; Hilbig et al., 2013). Thus, in summary, empirical evidence strongly supports

the unique role of HH (as opposed to AG) for capturing fairness versus exploitation – thereby

supporting the proposed pattern of dissociation for active cooperativeness.

On the other hand, the evidence for a unique association between HEXACO AG and

forgiveness versus retaliation is notably less convincing. First off, studies investigating self-

reports of AG-related criteria such as a reciprocity scale (Perugini, Gallucci, Presaghi, &

Ercolani, 2003), a revenge planning scale (Lee & Ashton, 2012), several forgiveness scales

(Romero, Villar, & López-Romero, 2015; Shepherd & Belicki, 2008), or a vengeance scale

(Sheppard & Boon, 2012) indeed found medium to large effects for AG (typical |r| between .30

and .70). However, all of these criteria were also substantially linked to HH (typical |r| between

.20 and .40), typically in the same direction as AG. Thus, although the effects sizes tend to be

larger for AG than for HH, the findings do not corroborate a conclusive pattern of dissociation. A

somewhat more encouraging picture evolved from studies based on hypothetical economic

games, especially the ultimatum game (Güth, Schmittberger, & Schwarze, 1982) in which a

responder can reject unfair offers made by a proposer (for details of the game, see below): It was
Running Head: HEXACO Agreeableness 6

repeatedly found that AG predicts responders’ ultimatum game decisions – that is, individuals

low in AG are more likely to reject offers and thus to retaliate – whereas HH typically does not

(Hilbig et al., 2013; Thielmann & Hilbig, 2014; Thielmann, Hilbig, & Niedtfeld, 2014).

However, the effect sizes for AG were notably smaller than in the self-report studies summarized

above (typical |r| between .15 and .20). Also, a recent study using fully incentivized economic

games again found that whereas only HH predicted positive reciprocity, both AG and HH

predicted negative reciprocity and thus the tendency to retaliate (Ackermann et al., 2016). Hence,

in summary, there is only limited evidence for the unique role of AG (over and above HH) in

explaining forgiveness versus retaliation and thus the implied pattern of dissociation for reactive

cooperation.1

Discouraging though this picture may appear at first sight, we maintain that it is, at least

in part, due to methodological aspects of previous studies. In particular, the only behavioral

measure of forgiveness versus retaliation used so far, the ultimatum game, does not offer a

sufficiently direct measure of said tendency. In the ultimatum game, one player (the proposer)

makes an offer to the other (the responder) about how to split an endowment between the two; the

responder can either accept or reject the offer. If she accepts, the endowment is split as proposed,

whereas if she rejects the entire endowment is lost and neither player receives anything (for

variants, see Suleiman, 1996). Thus, rejection of an offer conflates retaliation with the

willingness to forgo gains because retaliation is costly (Brethel-Haurwitz, Stoycos, Cardinale,

Huebner, & Marsh, 2016). Problematically, whereas the intention to retaliate should be a mark of

low AG, the willingness to forgo gains could actually be a matter of high HH. Consequently, the

behavior in question – rejecting unfair offers in the ultimatum game – may necessitate a trait

1
It should be noted explicitly that the empirical picture is even less in favor of the claim that HH and AG should be
subsumed under Big Five Agreeableness. Unlike HH, Big Five Agreeableness has neither been linked consistently to
fairness versus exploitation in economic games (for an overview see Hilbig, Thielmann, et al., 2015), nor has it been
more consistently linked to forgiveness versus retaliation than HEXACO AG (Zhao & Smillie, 2015).
Running Head: HEXACO Agreeableness 7

pattern (low AG, but high HH) which, almost by definition, is unlikely to ever be strong as “there

are few if any behaviors or traits that depend on the contrast between the two reciprocally

altruistic tendencies represented by H[H] and A[G]” (Ashton et al., 2014, p. 146, emphasis

original). Stated simply, it is thinkable that AG could only strongly predict ultimatum rejections

if it were generally negatively associated with HH which makes little sense theoretically and has,

to best of our knowledge, never once been found empirically.

The conflation of different behavioral tendencies or motives within the same economic

game is neither uncommon (Thielmann, Böhm, & Hilbig, 2015) nor inherently undesirable. For

example, it has been a seminal insight that humans are indeed willing to retaliate even at some

cost (Fehr & Gächter, 2000, 2002). However, for the reasons sketched above, costly retaliation is

unsuitable for the specific purpose of testing the unique role of AG within the HEXACO

framework. The latter will require a behavioral task in which retaliation is not – or, at least

arguably much less – conflated with one’s willingness to forego gains. In the following, we will

present a corresponding paradigm which we will name the “Uncostly Retaliation Game” (URG)

and subsequently use it in two experiments which test the ability of AG to (substantially and

uniquely) predict retaliation in the face of exploitation.

The Uncostly Retaliation Game

In general terms, the URG is a variant of the ultimatum game, based on the idea of

decoupling retaliation from costs (Anderson & Putterman, 2006). In particular, it implements a

second player who can retaliate against a first player’s allocation decision at no personal cost

(Houser & Xiao, 2010; Leibbrandt & López-Pérez, 2014; note, however, that their paradigms

involved fixed but non-zero costs for retaliation). Thus, the URG is a sequential, two-stage

extensive-form game, though with incomplete information for the first player (see below). The
Running Head: HEXACO Agreeableness 8

game structure is depicted in Figure 1: Player 1 divides a resource of, say, 10 monetary units

(MU) between herself and player 2, selecting either a fair distribution (i.e. the equal split of 5, 5)

or an unfair distribution with higher payoffs for herself (7.5, 2.5).

Next, player 2 reacts to this allocation, choosing whether and to what extent to retaliate by

reducing the payoffs of player 1. Importantly, this choice has no consequences for the payoffs of

player 2, that is, she neither gains through retaliation nor is the latter costly in any way. The

specific retaliation options of player 2 depend on the move of player 1 2: In case of a fair

distribution decision by player 1, player 2 can either leave player 1’s payoffs unaltered or reduce

them by 2.5 MU, leading to final payoffs of (5, 5) or (2.5, 5), respectively. Note that reducing

player 1’s payoffs despite a fair distribution decision can be considered “antisocial punishment”

(Herrmann, Thöni, & Gächter, 2008) and should actually not be predicted by AG which explicitly

refers to situations in which one is suffering exploitation. By contrast, AG should (negatively)

account for retaliation in case of an unfair move by player 1: In this case, player 2 has four

options for reducing player 1’s payoffs, namely by 0, 2.5, 5, or 7.5 MU resulting in final payoffs

of (7.5, 2.5), (5, 2.5), (2.5, 2.5), and (0, 2.5), respectively. Thus, player 2 can react in a forgiving

manner (leaving player 1’s payoffs unaltered) or retaliate with varying degrees of severity (in the

extreme, reducing the payoffs of player 1 to zero). The extent to which player 2 reduces the

payoffs of player 1 in this situation is consequently a measure of forgiveness versus retaliation in

the face of prior exploitation; importantly, since none of the options incurs any costs for player 2

(her payoffs are fully determined by the allocation decision of player 1), the game arguably

provides a more direct measure of retaliation as compared to the ultimatum game and related

game paradigms.

2
This dependency will later be relaxed in Experiment 2 in which we extended the game to yield entirely symmetric
retaliation options for player 2, independent of the move made by player 1.
Running Head: HEXACO Agreeableness 9

Note that, in more game theoretic terms, player 2 behavior signals her preferences for

different relative payoffs. In other words, player 2 has the power and opportunity to change how

much player 1 ultimately gains relative to herself. The essential aspect of the paradigm is thus

how these preferences depend on the behavior of player 1. If preferences reflect a more negative

(or less positive) relative payoff for player 1 if and only if player 1 behaved unfairly, this is

interpreted as an instance of retaliation. If, in turn, player 2 is willing to leave the relative payoffs

unaltered even though player 1 chose an unfair distribution, this is considered an act of

forgiveness. In essence, the URG thus measures one’s preferences for relative outcomes

conditional on another’s fairness while ensuring that minimizing player 1’s payoff (and thus

maximizing one’s own in relative terms) is not costly.

For clarification, consider the following examples: Assume player 1 decides to split the

10 MU in an unfair manner, thus assigning 7.5 MU to herself and 2.5 MU to player 2. If player 2

decides to leave the allocation unaltered, player 1 finally receives 7.5 MU whereas player 2

receives 2.5 MU. In refraining to reduce player 1’s payoff, player 2’s behavior hence arguably

mirrors a forgiving action. If player 2, by contrast, decides to reduce player 2’s payoff by, say, 5

MU, both players finally end up with a 2.5 MU payoff (i.e., player 1: 7.5 MU – 5 MU reduction

by player 2; player 2: 2.5 MU, as assigned by player 1). In this case, player 2’s behavior is hence

interpretable in terms of retaliation which, in this particular example, restores equality between

players.

In summary, the hypothesis that AG negatively predicts forgiveness versus retaliation in

response to being exploited can be tested in terms of an interaction between the move of player 1

and individual AG scores: If and only if player 1 acts unfairly, participants lower in AG should

be more likely to retaliate, that is, more strongly reduce player 1’s final payoff. This is in line

with the general recommendation to test trait effects via specific interactions (Appelt, Milch,
Running Head: HEXACO Agreeableness 10

Handgraaf, & Weber, 2011) in experimental setups. Correspondingly, the behavior of player 1

merely serves as the experimental manipulation, rather than being of interest in and of itself. With

this in mind, we implemented the game with asymmetrically incomplete information: Player 1

was neither informed that the game involves two stages, nor about the specific choice options of

player 2. Player 1 was thus unaware that an unfair distribution might later result in retaliation.

This approach seemed prudent to ensure that a sufficient proportion of unfair decisions would

actually be made by participants in the role of player 1 (given that knowledge of player 2’s

retaliation options would certainly shift player 1’s behavior towards strategic fairness, cf. Fehr &

Gächter, 2000; Hilbig, Zettler, & Heydasch, 2012). The advantage of this procedure is that the

experimentally manipulated variable is less likely to be severely skewed. However, as a

drawback, this renders the choice made by player 1 ultimately uninterpretable which is why we

exclusively focus on player 2’s behavior in what follows.

EXPERIMENT 1

Methods

Measures

To assess the HEXACO personality dimensions – most prominently AG and HH – we

used the German 60-item version of the HEXACO Personality Inventory-Revised (Ashton &

Lee, 2009). The HEXACO-60 contains 10 items for each of the six HEXACO factors. All items

are responded to on a five-point Likert-type scale ranging from 1 = strongly disagree to 5 =

strongly agree. The German version of the HEXACO-60 has been shown to provide a reliable

(internal consistencies: .74 ≤ α ≤ .83; test-retest reliability: .72 ≤ rtt ≤ .90) and valid measurement

of the six HEXACO dimensions (Moshagen, Hilbig, & Zettler, 2014; Thielmann et al., in press).
Running Head: HEXACO Agreeableness 11

To test the unique role of AG in predicting retaliation in the face of unfairness or

exploitation, respectively, we used the URG as described above. Participants in the role of player

1 received an endowment of 10€ (approx. US$13.15 at the time of data collection) which they

were asked to distribute between themselves and another unknown participant (simply called “the

other”) by choosing one of the described allocation options. Once participants in the role of

player 1 made their decision, their respective player 2 was informed about the corresponding

distribution and instructed that she could now react to player 1’s decision by choosing one of the

described final payoff distributions. The full verbatim instructions, materials, and data can be

found at https://osf.io/ntmhv/.

Procedure

Data collection involved two separate measurement occasions. In the first part of the

experiment, participants completed an online survey assessing demographic information and the

HEXACO-60. In the second part (at least 24 hours later), participants were invited to the lab to

play the URG. In each lab session, there were at least four participants to preserve anonymity.

Upon arrival, participants were seated in front of a computer and asked to provide informed

consent. The experiment was run using zTree (Fischbacher, 2007). In the URG, participants were

first randomly assigned to either the role of player 1 or of player 2 and matched with another

unknown participant in the room (in the opposite role).3 After receiving detailed instructions

about the rules of the URG (see above), participants indicated their choices as either player 1 or

3
In sessions with an uneven number of participants, one of the experimenters participated in the URG,
following a standardized protocol (role of player 1: unfair share; role of player 2: no change in payoff
distribution following a fair allocation, reduction of payoff by 5.00€ following an unfair allocation).
Participants were fully aware of this procedure (but not of the specific protocol the experimenter would
follow) in advance. Of course, data analyses involve only “real” player 2 decisions made by actual
participants.
Running Head: HEXACO Agreeableness 12

player 2. At the end of the game, they were informed about their final payoffs. Following some

unrelated tasks, participants finally received their payments (M = 4.20€ for the URG) as well as

information about the purpose of the experiment.

Participants

Sample size considerations were based on our main hypothesis of an interaction between

AG and player 1 behavior in predicting retaliation decisions in a multiple regression analysis (i.e.,

increase in R² beyond the main effects). A corresponding a priori power analysis using G*Power

(Faul, Erdfelder, Buchner, & Lang, 2009) yielded a required sample size of N = 81 (in the role of

player 2 and thus N = 162 in total) to detect a small- to medium-sized effect (f² = .10)4 of an

additional (third) predictor in a linear multiple regression analysis with satisfactory power (1-β =

.80). Correspondingly, we recruited N = 172 participants from a local participant pool, of whom

N = 156 fulfilled the criteria for inclusion in the data analysis (i.e., completion of all tasks).

About two thirds (62.2%) of participants were female, and they were aged between 18 to 46 years

(M = 21.89, SD = 4.43). In the URG, n = 76 acted in the role of player 1 whereas n = 80 acted in

the role of player 2.

Results and Discussion

Table 1 summarizes the descriptive statistics, internal consistencies of the trait scales, and

inter-correlations between the focal variables assessed in Experiment 1. The overall pattern of

decisions is depicted in Figure 2 (left panel). As intended, about half of player 1 participants (n =

37) chose a fair split of the endowment (5€ for both players) whereas the other half (n = 39)
4
Prior evidence based on the ultimatum game indeed suggests a small- to medium-sized effect of AG on
retaliation behavior (see above; Hilbig et al., 2013; Thielmann & Hilbig, 2014; Thielmann, et al., 2014).
Given that the URG arguably provides a more direct measure of retaliation decisions, we considered the
effect sizes from the ultimatum game to be lower bound effect size estimates.
Running Head: HEXACO Agreeableness 13

realized an unfair split (7.50€ for themselves and 2.50€ for player 2). In turn, 40% of player 2

participants (n = 32) decided to retaliate, that is, to reduce player 1’s payoffs. Importantly,

retaliation was almost exclusively limited to unfair distribution decisions by player 1, covering

90.6% of retaliating choices (n = 29). In reacting to an unfair distribution, most participants opted

for mild retaliation, reducing player 1’s payoffs by 2.50€ (n = 13) or 5.00€ (n = 12). Only 4

individuals retaliated to the maximum extent, that is, reducing player 1’s payoffs by 7.50€.

Following a fair move of player 1, only three participants decided to reduce player 1’s payoffs by

2.50€ (“antisocial punishment”). Confirming that the game produced the intended pattern of

behavior, a Chi-square test of independence showed that the extent of retaliation was strongly

dependent on player 1’s prior (fair versus unfair) behavior, χ²(df = 3) = 28.49, p < .001, implying

a large effect when considering the odds ratio of retaliation versus no retaliation in reaction to fair

versus unfair player 1 behavior, OR = 0.05, 95% CI [0.01, 0.18].

To investigate the influence of AG and HH on player 2 behavior, we based all

corresponding analyses on the maximum-likelihood estimator as implemented in Mplus (version

7.3, Muthén & Muthén, 2012) which provides Satorra-Bentler adjusted standard errors and test

statistics (Satorra & Bentler, 2001; MLM in MPlus) and thus accounts for non-normality in the

data (as observed for the URG). In line with the hypotheses, AG showed a negative link to

forgiveness versus retaliation in reaction to unfair behavior by player 1 (r = -.27, one-tailed5 p =

.029; see Table 1) whereas HH did not (r = -.14, p = .438). Correspondingly, in a multiple

regression analysis, the effect of AG on player 2 behavior in response to exploitation remained

stable when including HH as additional predictor, β = -.25, 95% CI [-.53, .03], p = .042 (one-

tailed) – with HH providing no unique contribution, β = -.05, 95% CI [-.37, .27], p = .770. In

5
Note that the directional nature of the hypothesis relating AG to retaliation decisions (see above) warrants and,
strictly speaking, even necessitates one-tailed significance testing in this specific case. For all non-directional
tests/hypotheses, we relied on two-tailed tests, so as to being more conservative.
Running Head: HEXACO Agreeableness 14

reaction to fair behavior, in turn, AG had no predictive power for player 2 behavior (r = .02, p =

.735). Interestingly, HH was negatively linked to retaliation decisions in the face of fair behavior

by player 1, that is, antisocial punishment (r = -.33, p = .003).6 Overall, as implied by the pattern

of results (and supporting the main hypothesis), a multiple regression analysis including AG, HH

(both centered on the sample mean), player 1 behavior (unfair vs. fair, coded 0 vs. 1), and the

interactions between each trait dimension and player 1 behavior revealed a significant interaction

between AG and player 1 behavior in predicting the extent to which player 2 participants

retaliated, β = .19, 95% CI [.003, .38], p = .047. In other words, low levels of AG were (only)

related to forgiveness versus retaliation behavior in response to being exploited by player 1 (see

Figure 3). In addition, the regression only revealed a main effect of player 1 behavior on

retaliation decisions, β = -.59, 95% CI [-.72, -.46], p < .001, but no reliable main effect of AG (β

= -.26, 95% CI [.-.56, .03], p = .083) or of HH (β = -.06, 95% CI [-.42, .31], p = .769) and no

interaction between HH and player 1 behavior, β = -.04, 95% CI [-.31, .23], p = .787.

In sum, the results corroborate the hypothesis that HEXACO AG is uniquely associated

with forgiveness versus retaliation and thus support the theoretical distinction between AG and

HH. Specifically, AG (but not HH) predicted whether and to what extent individuals retaliated in

the face of being exploited. Interestingly, HH (but not AG) was a significant predictor of

retaliation decisions in the face of fair player 1 behavior, that is, of antisocial punishment which

is in line with prior research pointing to the strong (negative) link between HH and antisocial

tendencies such as psychopathy (Lee & Ashton, 2014; Lee et al., 2013). The URG itself revealed

a highly plausible pattern of behavior in that retaliation strongly depended on the fairness of

player 1’s move. As such, our findings using the URG are generally compatible with typical

6
Given that the base-rate for choosing antisocial punishment was very low (only n = 3 participants opted for this
behavior), we considered it informative to additionally regard the HH levels of these individuals. Notably, the mean
z-standardized HH score of participants opting for antisocial punishment was = -1.27, thus indicating that these
individuals indeed scored at the lower end of the HH continuum.
Running Head: HEXACO Agreeableness 15

findings from costly punishment, showing that the severity of punishment increases with larger

deviations of another’s behavior from the fairness norm (Fehr & Gächter, 2000, 2002).

However, a potential limitation of the URG as used in Experiment 1 may be seen in the

asymmetry in choice options for player 2 as a function of player 1’s prior behavior. That is, as

detailed above, in case of a fair move by player 1, player 2 could only choose between two

options: reducing player 1’s payoffs by either nothing or 2.50€. By contrast, in case of an unfair

move by player 1, player 2 could choose among four options, including three different levels of

retaliation (i.e., reducing player 1’s payoffs by 2.50€, 5€, or even 7.50€) versus forgiveness (i.e.,

leaving player 1’s payoffs unaltered). In Experiment 2, we therefore extended the game to a fully

symmetrical version in which player 2 could choose among the same four options, irrespective of

player 1’s prior move.

As a second potential caveat, one may argue that player 2 decisions in the URG are, at least

in part, attributable to general, stable other-regarding preferences rather than retaliation intentions

in reaction to another’s exploitation. That is, reducing player 1’s (absolute and thus relative)

payoffs may be driven by inequality aversion (i.e., an intention to restore equal payoffs) rather

than or in addition to actual retaliation. Previous research indeed implies that costly punishment

is driven by both a desire for revenge and a desire for equality (Bone & Raihani, 2015;

Leibbrandt & López-Pérez, 2012). Likewise, associations between individual levels of Social

Value Orientation (SVO; e.g. Murphy & Ackermann, 2014; Van Lange, 1999) – denoting “the

weights people assign to their own and others’ outcomes in situations of interdependence”

(Balliet, Parks, & Joireman, 2009, p. 533) – and rejection decisions in the ultimatum game

suggest that a pro-social SVO is related to a lower willingness to reject offers (Baumert et al.,

2014; Karagonlar & Kuhlman, 2013). Correspondingly, in the URG – particularly in response to

exploitation – individuals’ choices might not only be driven by retaliation intentions, but also by
Running Head: HEXACO Agreeableness 16

a desire to reduce inequality (which is realizable via retaliation) versus a desire to maximize joint

outcomes (which is realizable via forgiveness) or, more generally speaking, by stable social

preferences. To rule out this alternative and, more importantly, to critically test the role of AG

once this alternative is controlled for, we additionally sought measures of social preferences in

Experiment 2.

EXPERIMENT 2

Methods

Measures

As in Experiment 1, we again relied on the German version of the HEXACO-60 (Ashton

& Lee, 2009; Moshagen, Hilbig, & Zettler, 2014) to measure the six HEXACO dimensions. In

addition, we used the 15-item SVO Slider Measure (Murphy, Ackermann, & Handgraaf, 2011) to

assess individual’s SVO (six primary SVO Slider items) as well as their level of inequality

aversion versus joint gain maximization (nine secondary SVO Slider items). Each SVO Slider

item reflects a choice between nine joint distributions of outcomes for oneself and another

person. Depending on the specific item, for example, an individual faces a conflict between

maximizing the personal versus collective interest, or between maximizing versus minimizing

outcome (in)equality. Outcomes are presented in points, ranging between 15 and 100.

Individuals’ allocation decisions in the six primary SVO items were integrated into the so-called

SVO angle which provides a continuous, one-dimensional measure of SVO, with higher values

indicating higher pro-sociality. Based on individuals’ allocation decisions in the nine secondary

SVO items, we calculated the inequality aversion (IA) index which provides a continuous

measure of individuals’ preference for inequality aversion (IA index = 0) versus joint gain

maximization (IA index = 1). However, note that the IA index is only readily interpretable for
Running Head: HEXACO Agreeableness 17

individuals indicating a pro-social inclination in the primary items (for details, see Ackermann &

Murphy, 2012). Thus, all analyses involving the IA index were exclusively based on individuals

with a pro-social SVO.

To measure forgiveness versus retaliation, we again relied on the URG. However, as

sketched above, we implemented a fully symmetric decision structure (see grey lines in Figure 1)

in Experiment 2. In particular, we extended the URG as used in Experiment 1 such that player 2

had equivalent choice options, irrespective of player 1’s prior (fair vs. unfair) behavior. We hence

implemented two additional choice options for player 2 in reaction to a fair move by player 1:

Player 2 could not only leave the fair split or reduce player 1’s payoffs by 2.50€, but she could

also reduce player 1’s payoffs by 5.00€ or 7.50€. In the extreme, player 2 could thus realize a

negative payoff for player 1, essentially extending the “antisocial punishment” options.

Procedure

As in Experiment 1, we separated the personality assessment from the behavioral tasks.

That is, participants first completed an online survey including demographic information and the

HEXACO-60. At least one week later, participants were invited to the lab to participate in an

experiment including the SVO Slider Measure and the (extended) URG. Again, each session

comprised at least four participants to preserve anonymity. Both SVO and URG were run using

the psynteract software for interactive experiments (Henninger, Kieslich, & Hilbig, in press) in

OpenSesame (Mathôt, Schreij, & Theeuwes, 2012).7

In the first part of the lab experiment, participants completed the SVO Slider Measure

with the 15 SVO items presented in random order (see Murphy et al., 2011). Participants received
7
In case of an uneven number of participants, we used the “ghost mode” as implemented in the psynteract
software in which the excess participant (who could initially not be assigned to a unique partner) receives
the same actual input as another participant in the same role (see Henninger, Kieslich, & Hilbig, in press,
for further details).
Running Head: HEXACO Agreeableness 18

the information that, for each allocation decision, they were randomly assigned to a new,

unknown partner in the room. Participants were aware that two randomly selected SVO trials

(one as an allocator and one as a recipient) were actually paid out at the end of the experiment

(with a conversion rate of 100 points = 1€, i.e. approx. US$1.09 at the time of data collection).

Following the SVO, participants completed the extended URG which was implemented in the

same manner as the original URG in Experiment 1 (i.e., using random assignment to role and

partner and providing initial endowments of 10€ to player 1). We explicitly informed participants

in the role of player 2 that player 1 was unaware of player 2’s opportunity to react to player 1’s

distribution decision, thus increasing transparency. Subsequent to the URG, participants received

information about their payoffs across SVO and URG (M = 6.21€) which was finally paid out

following another unrelated task. Participants also received information about the purpose of the

experiment.

Participants

A sample of N = 104 members of a local participant pool was invited to take part in the

experiment. The majority of participants (81.7%) were female and participants’ age ranged from

18 to 43 years (M = 22.81, SD = 3.99). In the URG, n = 52 were assigned to either role. Note that,

in general, we aimed at analyzing the main hypotheses using the entire sample from Experiments

1 and 2 in an overall analysis (see below). Thus, we considered it sufficient to recruit a slightly

smaller sample in Experiment 2.

Results and Discussion

Table 2 summarizes the descriptive statistics, internal consistencies, and inter-correlations

between the focal variables assessed in Experiment 2, and Figure 2 (right panel) depicts the
Running Head: HEXACO Agreeableness 19

decision pattern in the URG. Compared to Experiment 1, a slightly smaller proportion of player 1

participants chose to realize an unfair split of the endowment (n = 19) which, in consequence,

also led to lower levels of retaliation overall (n = 12). Strikingly, retaliation exclusively occurred

following an unfair move by player 1, again mostly in the form of mild retaliation (only one

participant retaliated to the maximum extent, reducing player 1’s payoffs by 7.50€). Vice versa,

in response to a fair move by player 1, no-one decided to reduce player 1’s payoffs, thus also

implying that the additional retaliation options of the extended URG were discarded entirely. A

Chi-square test of independence hence indicated different distributions of retaliating choices

following fair versus unfair player 1 behavior, χ²(df = 3) = 27.10, p < .001. Correspondingly,

comparing player 2’s reactions to player 1’s unfair behavior in the original URG (Experiment 1)

and the extended URG (Experiment 2) yielded no differences in the distribution of choices across

experiments, χ²(df = 3) = 1.64, p = .651, BF10 = 0.428 (see Figure 2).

To again test the unique role of AG (as opposed to HH) in predicting behavior in response

to being exploited (as player 2) in the URG, we relied on the same analytic approach as in

Experiment 1 (using the MLM estimator as implemented in Mplus). Corroborating our previous

findings, we again found a negative correlation between AG and the willingness to retaliate in

response to an unfair move by player 1 which was notably larger in size as compared to

Experiment 1 (r = -.53, one-tailed p = .002; see Table 2). For HH, by contrast, again no

comparable link with retaliation decisions in response to an unfair move by player 1 was apparent

(r = -.15, p = .518). Correspondingly, the relation between AG and retaliation in response to

unfair player 1 behavior remained virtually unaltered when simultaneously including HH as

8
The Bayes Factor (BF) is the ratio of the posterior odds of the alternative hypothesis and those of the null
hypothesis, given the data and the statistical model. BF10 < 1 thus reflects evidence in favor of the null hypothesis,
whereas BF10 > 1 reflects evidence in favor of the alternative hypothesis. The BF was calculated using the
BayesFactor package (Morey, Rouder, & Jamil, 2015) in R. In particular, we relied on the “poisson” sampling plan
as implemented in the contingencyTableBF function, given that there was no random assignment of participants to
experimental conditions (i.e., fair vs. unfair player 1 behavior) and the sample size in either condition was not fixed.
Running Head: HEXACO Agreeableness 20

predictor in a multiple regression analysis, β = -.53, 95% CI [-.88, -.17], one-tailed p = .002 –

with HH itself yielding no meaningful effect, β = -.14, 95% CI [-.55, .28], p = .527. In response

to a fair move by player 1, by contrast, the zero-order correlations between AG as well as HH and

player 2 decisions were essentially zero given that all participants responded in the same way

(i.e., showing no retaliation or antisocial punishment, respectively, at all). Thus, basically

reflecting this pattern of results, a multiple regression analysis predicting player 2 behavior by

AG and HH, player 1 behavior, and the two interaction terms (i.e., AG x player 1 behavior and

HH x player 1 behavior; cf. Experiment 1) indicated that the link between AG and the willingness

to retaliate was stronger in reaction to an unfair distribution decision by player 1 than in reaction

to a fair distribution decision, as implied by a significant interaction between AG and player 1

behavior, β = .72, 95% CI [.20, 1.23], p = .006 (see Figure 4). Additionally, both player 1

behavior (β = -.68, 95% CI [-.87, -.49], p < .001) and AG (β = -.82, 95% CI [-1.42, -.23], p =

.006) showed significant main effects, whereas no effect was apparent for HH (β = -.24, 95% CI

[-.95, .48], p = .520) or its interaction with player 1 behavior (β = .20, 95% CI [-.42, .83], p =

.520). Overall, Experiment 2 using the extended URG with symmetric choice options for player 2

hence largely replicated the results from Experiment 1 with asymmetric choice options.

As sketched above, another aim of Experiment 2 was to test the potential effect of

individuals’ stable social preferences (i.e., SVO and inequality aversion/joint gain maximization)

on forgiveness versus retaliation in the URG, particularly in case of unequal payoffs for both

players (i.e., following an unfair move by player 1). However, for both SVO and the IA index, we

found no effect on retaliation decisions in reaction to an unfair distribution decision (r = -.02, p =

.871 for SVO and r = .08, p = .811 for IA; see Table 2).9 In other words, neither individuals’

general social preferences nor pro-socials’ preference for inequality aversion versus joint gain

9
Correspondingly, adding either SVO or IA as predictors in a multiple regression analysis did not alter the
results in any meaningful way.
Running Head: HEXACO Agreeableness 21

maximization accounted for individuals’ willingness to retaliate in the face of exploitation. These

findings, in turn, strongly suggest that the URG indeed serves as a relatively direct measure of

retaliation, unperturbed by considerations for one’s own costs or one’s general social preferences

and inequality aversion versus concerns for joint gain maximization. In sum, Experiment 2 hence

further supported the hypothesis that retaliation intentions are the primary driver of player 2

decisions in the URG and that HEXACO AG (but not HH) is a substantial and unique predictor

of corresponding behavior.

Overall Analysis

As mentioned above, we considered it informative and valuable in terms of increasing

statistical power to finally run an overall analysis of our data from Experiments 1 and 2. That is,

we again tested our main hypothesis regarding the unique power of AG (as opposed to HH) to

account for the willingness to retaliate, specifically in response to being exploited by another.

Note that the precondition for such an overall analysis on the two versions of the URG was met

given that we found no differences in the distributions of forgiving versus retaliating choices

across experiments (see above and Figure 2). In this overall analysis, we also relied on Bayes

Factors as an indicator of the strength of evidence in favor of the alternative versus null

hypothesis. Specifically, we used Bayesian correlation tests (Wetzels & Wagenmakers, 2012),

calculating BF10 as the ratio of the alternative to the null hypothesis (see Footnote 8 for further

information).

As summarized in Table 3 – and corroborating our results from above – this overall

analysis again revealed the expected link between AG and retaliation decisions in the face of

exploitation (r = -.33, one-tailed p = .002) whereas the corresponding link for HH did not reach a

conventional level of statistical significance (r = -.17, p = .235) and was substantially smaller. In
Running Head: HEXACO Agreeableness 22

terms of Bayes Factors, however, evidence was in favor of the alternative hypothesis for AG (H1

being more than three times as likely as the null hypothesis, BF10 = 3.29) whereas for HH,

evidence was in favor of the null hypothesis (being more than four times as likely as the

alternative, BF10 = 0.23). Correspondingly, the relation between AG and retaliation behavior

following an unfair treatment remained virtually unaltered when including HH as an additional

predictor in a multiple regression analysis, β = -.31, 95% CI [-.54, -.08], one-tailed p = .005. For

HH, there was no evidence of a unique contribution, β = -.08, 95% CI [-.32, .17], p = .549. In

reaction to a fair distribution decision by player 1, in contrast, AG did not account for significant

variance in player 2 behavior (r = -.03, p = .305) – whereas HH did (r = -.29, p = .002). In line

with this pattern of results, a multiple regression analysis on the overall player 2 data (N = 132)

involving AG, HH (both centered on sample mean), player 1 behavior (dummy coded; unfair = 0

vs. fair = 1), and the two-way interactions between AG and player 1 behavior as well as HH and

player 1 behavior as predictors of player 2 behavior confirmed the hypothesized interaction

between AG and player 1 behavior implying a stronger (negative) effect of AG in reaction to

unfair compared to fair behavior, β = .28, 95% CI [.08, .49], p = .008. In addition, both player 1

behavior, β = -.62, 95% CI [-.72, -.52], p < .001, and AG, β = -.37, 95% CI [-.64, -.09], p = .009,

yielded significant main effects in the regression model. HH, in turn, was again unrelated to

retaliation decisions, β = -.09, 95% CI [-.38, .20], p = .547, as was its interaction with player 1

behavior, β = .02, 95% CI [-.22, .25], p = .895. In sum, the overall analysis hence fully

corroborated our results from the single experiments10, once again demonstrating the unique role

of AG for retaliation behavior in response to being exploited.

10
We replicated all analyses including the specific experiment (1 versus 2) as a dummy-coded variable, both in terms
of a main effect as well as in interaction with the primary effect under investigation. In all but one analysis, the
corresponding interaction with this dummy-variable failed to detect a meaningful effect. That is, only for the
regression of player 2 behavior in reaction to a fair distribution decision by player 1 on HH, the analysis revealed a
significantly stronger effect of HH in Experiment 1 – which is unsurprising given that in Experiment 2 there was
zero variance in player 2 behavior following a fair move by player 1 (i.e., complete absence of antisocial
Running Head: HEXACO Agreeableness 23

GENERAL DISCUSSION

To a noteworthy extent, the evolution of cooperation rests on both the tendency to actively

cooperate and the tendency to reinstate cooperation in reaction to exploitation. This very

distinction constitutes part of the theoretical basis of the HEXACO model of personality structure

(Ashton & Lee, 2001, 2007) which attributes the tendency for fairness versus exploitation (active

cooperation) to the Honesty-Humility (HH) factor and the tendency for forgiveness versus

retaliation (reactive cooperation) to the Agreeableness (AG) factor (Ashton et al., 2014; Hilbig et

al., 2013; Zhao & Smillie, 2015). However, whereas recent research has provided abundant

evidence for the strong and unique role of HH for active cooperation, the empirical picture

concerning the role of AG for reactive cooperation is less clear. Specifically, several self-report

criteria mirroring forgiveness versus retaliation appear to be predicted by AG, but also by HH,

albeit to a lesser extent (e.g., Shepherd & Belicki, 2008; Sheppard & Boon, 2012). Behavioral

measures from the ultimatum game, in turn, indeed corroborated the role of AG; however, the

effect sizes of AG were relatively small (e.g., Hilbig et al., 2013; Thielmann et al., 2014) and

effects of HH were also observed at times (Ackermann et al., 2016).

Herein, we have argued that methodological aspects of prior investigations can account for

these findings. Specifically, the only behavioral measure of forgiveness versus retaliation used so

far, the ultimatum game, conflates the tendency to retaliate with individual’s payoff concerns

because retaliation is costly. To overcome this limitation, we introduced an alternative behavioral

paradigm (similar to those put forward by Houser & Xiao, 2010; Leibbrandt & López-Pérez,

2014) which was designed to provide a measure of uncostly retaliation (thus called “Uncostly

punishment). For the remaining analyses (as summarized above) no interaction between the primary effect under
investigation and the dummy-variable coding Experiment 1 versus Experiment 2 emerged.
Running Head: HEXACO Agreeableness 24

Retaliation Game”; URG). In this game, a first mover (player 1) decides between a fair and an

unfair distribution of money between herself and a second mover (player 2) who can retaliate by

reducing player 1’s payoffs to different degrees. Importantly, retaliation has no consequences for

player 2’s payoffs, but exclusively reduces player 1’s payoffs. In two fully incentivized

experiments involving real players, we used this game to test the main hypothesis that AG should

predict player 2’s retaliation decisions in response to exploitation (i.e., an unfair distribution

made by player 1) – over and above HH.

Across the two experiments, results mirrored the hypothesized pattern: AG was negatively

linked to the extent to which player 2 participants retaliated in the face of an exploitative decision

by player 1 (and only then), amounting to a medium-sized effect in the overall analysis. This link

was unique in that HH, by contrast, was not associated with retaliation in response to exploitation

and only AG explained unique variance in a multiple regression including both factors. The

findings thus corroborate the HH-AG-distinction and complete the pattern of dissociations found

in prior work by additionally showing that AG is a unique and substantial predictor of reactive

cooperation whereas HH is not.

Interestingly, the game further revealed a second, complementary dissociation, namely that

retaliation in response to a fair distribution ("antisocial punishment", Herrmann et al., 2008) was

negatively linked to HH (but not AG). This is perfectly in line with prior research on the

(negative) link between HH and antisocial tendencies such as psychopathy (Lee & Ashton, 2014;

Lee et al., 2013). Note, however, that there was very limited variance in participants’ reactions to

a fair distribution and thus the findings should be treated with some caution. Indeed, the fact that

retaliation decisions were strongly dependent on the fairness of player 1’s distribution decisions

(i.e., that retaliation was almost exclusively limited to unfair distributions) fortifies the

interpretation of the game as a relatively “pure” measure of retaliation: Participants’ preferred a

Running Head: HEXACO Agreeableness 25

more negative absolute and relative payoff for player 1 if and only if player 1 behaved unfairly –

corroborating the interpretation of URG behavior as reflecting retaliation versus forgiveness. In

addition, findings from the second experiment confirmed that retaliation decisions were also

unrelated to stable social preferences in terms of Social Value Orientation (Van Lange, 1999)

and, more specifically, inequality aversion (Murphy et al., 2011), thus ruling out that these

decisions are primarily driven by individuals’ concerns for payoff equality or joint gain

maximization rather than retaliation.

Admittedly, the associations obtained between the personality scales and behavior in the

URG were only small to moderate in size. However, it must also be noted that the URG only

provides a single observation of directly consequential (forgiving versus retaliatory) behavior.

Common method variance or shared response tendencies (such as socially desirable responding)

with personality scales are thus reduced to a minimum. Also, the game is highly specific and

entirely one-shot for each player and thus lacks the type of reliability that is sought in how

personality scales are typically constructed. Finally, the personality scales used are designed to

cover a broad range of dispositional tendencies which include, but are certainly not limited to, the

specific type of behavior tapped by the URG. In light of these limitations, we maintain that small-

to medium-sized (but no larger) effects are to be expected. Indeed, similar investigations linking

personality scales to behavior in economic games have rarely reported effects beyond |r| = .30

(see Zhao & Smillie, 2015).

Also, it should be openly acknowledged that the samples sizes in the present studies were

only modest in size – especially due to the game structure itself in which only a subset of

responses are actually relevant to the key hypotheses (those with unfair allocations by player 1).

Though this can be considered a drawback of the game, it is inherent to any kind of paradigm that

targets actual unfair behavior rather than relying on hypothetical behavior or, worse yet,
Running Head: HEXACO Agreeableness 26

deceiving participants (Ortmann & Hertwig, 2002). Although the overall analysis can be

considered to yield acceptable power (and indeed provides more conclusive test statistics than the

single studies), future replications are certainly to be encouraged.

In summary, we have demonstrated how a methodological approach rooted in behavioral

economics can greatly profit personality research (Zhao & Smillie, 2015) – if attention is paid to

the specific motives underlying behavior in different game paradigms (Thielmann et al., 2015).

On the substantive side, our findings provide further evidence for the distinction between traits

driving active cooperation (fairness versus exploitation) and reactive cooperation (forgiveness

versus retaliation) as made in the HEXACO model of personality. More generally, this shows

how well-specified and theoretically-grounded models of personality structure can help explain

individual differences in actual interpersonal behavior.

Running Head: HEXACO Agreeableness 27

REFERENCES

Ackermann, K. A., Fleiß, J., & Murphy, R. O. (2016). Reciprocity as an individual difference. Journal of

Conflict Resolution, 60(2), 340-367.

Ackermann, K. A., & Murphy, R. O. (2012). Tutorial on how to evaluate the svo slider measure’s

secondary items. Retrieved from http://vlab.ethz.ch/svo/SVO_Slider/SVO_Slider_scoring.html.

Anderson, C. M., & Putterman, L. (2006). Do non-strategic sanctions obey the law of demand? The

demand for punishment in the voluntary contribution mechanism. Games and Economic Behavior,

54(1), 1-24.

Appelt, K. C., Milch, K. F., Handgraaf, M. J. J., & Weber, E. U. (2011). The Decision Making Individual

Differences Inventory and guidelines for the study of individual differences in judgment and

decision-making research. Judgment and Decision Making, 6(3), 252-262.

Ashton, M. C., & Lee, K. (2001). A theoretical basis for the major dimensions of personality. European

Journal of Personality, 15(5), 327-353.

Ashton, M. C., & Lee, K. (2007). Empirical, theoretical, and practical advantages of the HEXACO model

of personality structure. Personality and Social Psychology Review, 11(2), 150-166.

Ashton, M. C., & Lee, K. (2008a). The HEXACO model of personality structure and the importance of

the H factor. Social and Personality Psychology Compass, 2(5), 1952-1962.

Ashton, M. C., & Lee, K. (2008b). The prediction of Honesty-Humility-related criteria by the HEXACO

and Five-Factor Models of personality. Journal of Research in Personality, 42(5), 1216-1228.

Ashton, M. C., & Lee, K. (2009). The HEXACO-60: A short measure of the major dimensions of

personality. Journal of Personality Assessment, 91(4), 340-345.

Ashton, M. C., & Lee, K. (2010). On the cross-language replicability of personality factors. Journal of

Research in Personality, 44(4), 436-441.

Ashton, M. C., Lee, K., & De Vries, R. E. (2014). The HEXACO Honesty-Humility, Agreeableness, and

Emotionality Factors: A Review of Research and Theory. Personality and Social Psychology

Review, 18, 139-152.

Running Head: HEXACO Agreeableness 28

Ashton, M. C., Lee, K., Perugini, M., Szarota, P., de Vries, R. E., Di Blas, L., et al. (2004). A six-factor

structure of personality-descriptive adjectives: Solutions from psycholexical studies in seven

languages. Journal of Personality and Social Psychology, 86(2), 356-366.

Axelrod, R., & Dion, D. (1988). The further evolution of cooperation. Science, 242(4884), 1385-1390.

Axelrod, R., & Hamilton, W. D. (1981). The evolution of cooperation. Science, 211(4489), 1390-1396.

Balliet, D., Parks, C., & Joireman, J. (2009). Social Value Orientation and cooperation in social dilemmas:

A Meta-Analysis. Group Processes Intergroup Relations, 12(4), 533-547.

Baumert, A., Schlösser, T., & Schmitt, M. (2014). Economic games - Performance-based assessment of

altruism and fairness. European Journal of Psychological Assessment, 30, 178-192.

Bone, J. E., & Raihani, N. J. (2015). Human punishment is motivated by both a desire for revenge and a

desire for equality. Evolution and Human Behavior, 36(4), 323-330.

Bowles, S., & Gintis, H. (2011). A cooperative species: Human reciprocity and its evolution: Princeton

University Press.

Brethel-Haurwitz, K. M., Stoycos, S. A., Cardinale, E. M., Huebner, B., & Marsh, A. A. (2016). Is costly

punishment altruistic? Exploring rejection of unfair offers in the Ultimatum Game in real-world

altruists. Scientific reports, 6.

De Raad, B., Barelds, D. P., Levert, E., Ostendorf, F., Mlačić, B., Di Blas, L., et al. (2010). Only Three

Factors of Personality Description Are Fully Replicable Across Languages: A Comparison of 14

Trait Taxonomies. Journal of Personality and Social Psychology, 98(1), 160-173.

De Raad, B., Barelds, D. P., Mlačić, B., Church, A. T., Katigbak, M. S., Ostendorf, F., et al. (2010). Only

three personality factors are fully replicable across languages: Reply to Ashton and Lee. Journal

of Research in Personality, 44(4), 442-445.

De Raad, B., Barelds, D. P., Timmerman, M. E., De Roover, K., Mlačić, B., & Church, A. T. (2014).

Towards a pan‐cultural personality structure: Input from 11 psycholexical studies. European

Journal of Personality, 28(5), 497-510.

DeYoung, C. G. (2010). Toward a theory of the Big Five. Psychological Inquiry, 21(1), 26-33.
Running Head: HEXACO Agreeableness 29

Dunlop, P. D., Morrison, D. L., Koenig, J., & Silcox, B. (2012). Comparing the Eysenck and HEXACO

models of personality in the prediction of adult delinquency. European Journal of Personality,

26(3), 194-202.

Faul, F., Erdfelder, E., Buchner, A., & Lang, A.-G. (2009). Statistical power analyses using G*Power 3.1:

Tests for correlation and regression analyses. Behavior Research Methods, 41, 1149-1160.

Fehr, E., & Gächter, S. (2000). Cooperation and punishment in public goods experiments. American

Economic Review, 90(4), 980-994.

Fehr, E., & Gächter, S. (2002). Altruistic punishment in humans. Nature, 415(6868), 137-140.

Fischbacher, U. (2007). z-Tree: Zurich toolbox for ready-made economic experiments. Experimental

economics, 10(2), 171-178.

Güth, W., Schmittberger, R., & Schwarze, B. (1982). An experimental analysis of ultimatum bargaining.

Journal of Economic Behavior & Organization, 3(4), 367-388.

Henninger, F., Kieslich, P. J., & Hilbig, B. E. (in press). Psynteract: A flexible, cross-plattform, open

framework for interactive experiments. Behavior Research Methods.

Herrmann, B., Thöni, C., & Gächter, S. (2008). Antisocial punishment across societies. Science,

319(5868), 1362-1367.

Hershfield, H. E., Cohen, T. R., & Thompson, L. (2012). Short horizons and tempting situations: Lack of

continuity to our future selves leads to unethical decision making and behavior. Organizational

Behavior and Human Decision Processes, 117(2), 298-310.

Hilbig, B. E., Glöckner, A., & Zettler, I. (2014). Personality and pro-social behavior: Linking basic traits

and social value orientations. Journal of Personality and Social Psychology, 107, 529-539.

Hilbig, B. E., Moshagen, M., & Zettler, I. (2015). Truth will out: Linking personality, morality, and

honesty through indirect questioning. Social Psychological and Personality Science, 6, 140-147.

Hilbig, B. E., Thielmann, I., Hepp, J., Klein, S. A., & Zettler, I. (2015). From personality to altruistic

behavior (and back): Evidence from a double-blind dictator game. Journal of Research in

Personality, 55, 46-50.

Running Head: HEXACO Agreeableness 30

Hilbig, B. E., & Zettler, I. (2015). When the cat’s away, some mice will play: A basic trait account of

dishonest behavior. Journal of Research in Personality, 57, 72-88.

Hilbig, B. E., Zettler, I., & Heydasch, T. (2012). Personality, punishment, and public-goods: Strategic

shifts towards cooperation as a matter of dispositional Honesty-Humility. European Journal of

Personality, 26, 245-254.

Hilbig, B. E., Zettler, I., Leist, F., & Heydasch, T. (2013). It takes two: Honesty-Humility and

Agreeableness differentially predict active versus reactive cooperation. Personality and Individual

Differences, 54, 598-603.

Houser, D., & Xiao, E. (2010). Inequality-seeking punishment. Economics Letters, 109(1), 20-23.

Karagonlar, G., & Kuhlman, D. M. (2013). The role of social value orientation in response to an unfair

offer in the ultimatum game. Organizational Behavior and Human Decision Processes, 120(2),

228-239.

Lee, K., & Ashton, M. C. (2008). The HEXACO personality factors in the indigenous personality lexicons

of English and 11 other languages. Journal of Personality, 76(5), 1001-1053.

Lee, K., & Ashton, M. C. (2012). Getting mad and getting even: Agreeableness and Honesty-Humility as

predictors of revenge intentions. Personality and Individual Differences, 52(5), 596-600.

Lee, K., & Ashton, M. C. (2014). The dark triad, the big five, and the HEXACO model. Personality and

Individual Differences, 67, 2-5.

Lee, K., Ashton, M. C., Wiltshire, J., Bourdage, J. S., Visser, B. A., & Gallucci, A. (2013). Sex, power,

and money: Prediction from the dark triad and honesty–humility. European Journal of

Personality, 27, 169-184.

Lee, K., Gizzarone, M., & Ashton, M. C. (2003). Personality and the Likelihood to Sexually Harass. Sex

Roles, 49(1-2), 59-69.

Leibbrandt, A., & López-Pérez, R. (2012). An exploration of third and second party punishment in ten

simple games. Journal of Economic Behavior & Organization, 84(3), 753-766.

Running Head: HEXACO Agreeableness 31

Leibbrandt, A., & López-Pérez, R. (2014). Different carrots and different sticks: do we reward and punish

differently than we approve and disapprove? Theory and decision, 76(1), 95-118.

Mathôt, S., Schreij, D., & Theeuwes, J. (2012). OpenSesame: An open-source, graphical experiment

builder for the social sciences. Behavior research methods, 44(2), 314-324.

McCrae, R. R., & Costa, P. T. (2008). The five-factor theory of personality. In Handbook of personality

psychology: Theory and research (3rd ed., pp. 159-181). New York, NY: Guilford Press.

Morey, R. D., Rouder, J. N., & Jamil, T. (2015). BayesFactor. R package (Version 0.9.12-2). Retrieved

from https://cran.r-project.org/web/packages/BayesFactor/index.html.

Moshagen, M., Hilbig, B. E., & Zettler, I. (2014). Faktorenstruktur, psychometrische Eigenschaften und

Messinvarianz der deutschsprachigen Version des 60-Item HEXACO Persönlichkeitsinventars

[Factor structure, psychometric properties, and measurement invariance of the German-language

version of the 60-item HEXACO personality inventory]. Diagnostica, 60, 86-97.

Murphy, R. O., & Ackermann, K. A. (2014). Social value orientation: Theoretical and measurement issues

in the study of social preferences. Personality and Social Psychology Review, 18, 13-41.

Murphy, R. O., Ackermann, K. A., & Handgraaf, M. J. J. (2011). Measuring social value orientation.

Judgment and Decision Making, 6(8), 771-781.

Muthén, L. K., & Muthén, B. O. (2012). Mplus user’s guide (7th ed). Los Angeles, CA: Muthén &

Muthén.

Nowak, M. A. (2006). Five rules for the evolution of cooperation. Science, 314, 1560-1563.

Ortmann, A., & Hertwig, R. (2002). The costs of deception: Evidence from Psychology. Experimental

Economics, 5, 111-131.

Perugini, M., Gallucci, M., Presaghi, F., & Ercolani, A. P. (2003). The Personal Norm of Reciprocity.

European Journal of Personality, 17(4), 251-283.

Romero, E., Villar, P., & López-Romero, L. (2015). Assessing six factors in Spain: Validation of the

HEXACO-100 in relation to the Five Factor Model and other conceptually relevant criteria.

Personality and Individual Differences, 76, 75-81.

Running Head: HEXACO Agreeableness 32

Satorra, A., & Bentler, P. M. (2001). A scaled difference chi-square test statistic for moment structure

analysis. Psychometrika, 66(4), 507-514.

Shepherd, S., & Belicki, K. (2008). Trait forgiveness and traitedness within the HEXACO model of

personality. Personality and Individual Differences, 45(5), 389-394.

Sheppard, K. E., & Boon, S. D. (2012). Predicting appraisals of romantic revenge: The roles of honesty–

humility, agreeableness, and vengefulness. Personality and Individual Differences, 52(2), 128-

132.

Suleiman, R. (1996). Expectations and fairness in a modified Ultimatum game. Journal of Economic

Psychology, 17(5), 531-554.

Thielmann, I., Böhm, R., & Hilbig, B. E. (2015). Different games for different motives: comment on

Haesevoets, Folmer, and Van Hiel (2015). European Journal of Personality, 29(4), 506-508.

Thielmann, I., & Hilbig, B. E. (2014). Trust in me, trust in you: A social projection account of the link

between personality, cooperativeness, and trustworthiness expectations. Journal of Research in

Personality, 50, 61-65.

Thielmann, I., Hilbig, B. E., & Niedtfeld, I. (2014). Willing to give but not to forgive: Borderline

personality features and cooperative behavior. Journal of Personality Disorders, 28, 778–795.

Thielmann, I., Hilbig, B. E., Zettler, I., & Moshagen, M. (in press). On measuring the sixth basic

personality dimension: A comparison between HEXACO Honesty-Humility and Big Six Honesty-

Propriety. Assessment.

van Gelder, J.-L., & de Vries, R. E. (2013). Rational Misbehavior? Evaluating an Integrated Dual-Process

Model of Criminal Decision Making. Journal of Quantitative Criminology, 30(1), 1-27.

van Kampen, D. (2012). The 5-Dimensional Personality Test (5DPT): Relationships with two lexically

based instruments and the validation of the absorption scale. Journal of personality assessment,

94(1), 92-101.

Van Lange, P. A. M. (1999). The pursuit of joint outcomes and equality in outcomes: An integrative

model of social value orientation. Journal of Personality and Social Psychology, 77(2), 337-349.
Running Head: HEXACO Agreeableness 33

Wetzels, R., & Wagenmakers, E.-J. (2012). A default Bayesian hypothesis test for correlations and partial

correlations. Psychonomic Bulletin & Review, 19(6), 1057-1064.

Zettler, I., & Hilbig, B. E. (2010). Honesty-Humility and a person-situation-interaction at work. European

Journal of Personality, 24(7), 569–582.

Zhao, K., & Smillie, L. D. (2015). The Role of Interpersonal Traits in Social Decision Making Exploring

Sources of Behavioral Heterogeneity in Economic Games. Personality and Social Psychology

Review, 19, 277-302

Running Head: HEXACO Agreeableness 34

FIGURES

Figure 1. Uncostly Retaliation Game with payoffs for player 1 (underlined) and player 2 (italics).
Game structure in Experiment 1 depicted in black, extended game structure in Experiment 2
depicted in grey.
Running Head: HEXACO Agreeableness 35

Figure 2. Distribution of retaliation decisions (player 2) in the Uncostly Retaliation Game as a

function of player 1 behavior (fair vs. unfair) in Experiment 1 (left panel) and Experiment 2 (right
panel).
Running Head: HEXACO Agreeableness 36

Figure 3. Predicted extent of retaliation (player 2) in the Uncostly Retaliation Game as a function
of HEXACO AG (centered on sample mean) and conditional on player 1 behavior (fair vs.
unfair). Shaded areas mark the 95% confidence bands around the prediction (Experiment 1).
Running Head: HEXACO Agreeableness 37

Figure 4. Predicted extent of retaliation (player 2) in the Uncostly Retaliation Game as a function
of HEXACO AG (centered on sample mean) and conditional on player 1 behavior (fair vs.
unfair). Shaded areas mark the 95% confidence bands around the prediction (Experiment 2).
Running Head: HEXACO Agreeableness 38

TABLES

Table 1
Means, Standard Deviations (in Parentheses), and Inter-Correlations (95% Confidence
Intervals in Brackets) between Behavior in the Uncostly Retaliation Game and Personality
(Honesty-Humility and Agreeableness) in Experiment 1, with Internal Consistency
Reliabilities (Cronbach’s α) in the Diagonal (in Parentheses).
Scale M (SD) Correlations

HH AG

HH 1-5 3.25 (0.64) (.76)

AG 1-5 3.12 (0.50) .18* [.02, .33] (.68)

URG player 2 0-7.5 1.62 (2.29) -.06 [-.33, .21] -.07 [-.29, .16]

given fair player 1 (n = 36) 0-2.5 0.21 (0.70) -.33** [-.54, -.11] .02 [-.10, .14]

given unfair player 1 (n = 44) 0-7.5 2.78 (2.48) -.14 [-.48, .21] -.27*a [-.54, .01]

Note. All inter-correlations reported for the URG are based on the maximum-likelihood
estimator with Satorra-Bentler adjusted standard errors and test statistics.
HH = Honesty-Humility, AG = Agreeableness, URG = Uncostly Retaliation Game.
a
: one-sided testing;
** p < .01, * p < .05.
Running Head: HEXACO Agreeableness 39

Table 2
Means, Standard Deviations (in Parentheses), and Inter-Correlations (95% Confidence Intervals in Brackets) between Behavior in the Uncostly
Retaliation Game and Personality (Honesty-Humility, Agreeableness, and Social Value Orientation) in Experiment 2, with Internal Consistency
Reliabilities (Cronbach’s α) in the Diagonal (in Parentheses).
Scale M (SD) Correlations

HH AG SVO IA

HH 1-5 3.59 (0.56) (.70)

AG 1-5 3.29 (0.53) .23* [.04, .41] (.73)

SVO +/- 180 30.13 (10.57) .05 [-.15, .24] .10 [-.09, .29] – –

IAa 0-1 0.33 (0.23) -.10 [-.31, .11] -.03 [-.24, .18] -.10 [-.30, .12] –

URG player 2 0-7.5 0.82 (1.69) .11 [-.14, .35] -.20 [-.43, .02] .06 [-.16, .29] -.04 [-.42, .33]

given fair player 1 (n = 33)b 0-7.5 0.00 (0.00) – – – –

given unfair player 1 (n = 19) 0-7.5 2.24 (2.19) -.15 [-.61, .31] -.53**c [-.87, -.19] -.02 [-.29, .25] .08 [-.59, .75]

Note. All inter-correlations reported for the URG are based on the maximum-likelihood estimator with Satorra-Bentler adjusted standard errors and
test statistics.
HH = Honesty-Humility, AG = Agreeableness, SVO = Social Value Orientation (SVO primary items), IA = Inequality Aversion Index (SVO
secondary items); URG = Uncostly Retaliation Game.
a
: Analyses based on subsample of individuals with a pro-social SVO (n = 18).
b
: There was no variance in player 2 behavior in response to a fair move by player 1.
c
: One-sided testing.
* p < .05, ** p < .01.
Running Head: HEXACO Agreeableness 40

Table 3
Inter-Correlations (95% Confidence Intervals) and Bayes Factors (for the alternative over the
null hypothesis) between Behavior in the Uncostly Retaliation Game and Personality
(Honesty-Humility and Agreeableness) in the Overall Analysis across Experiments 1 and 2.
HH AG

r 95% CI BF10 r 95% CI BF10

URG player 2 -.06 [-.27, .14] 0.09 -.13 [-.30, .03] 0.22

given fair player 1 (n = 69) -.29** [-.47, -.11] 1.62 -.03 [-.10, .03] 0.10

given unfair player 1 (n = 63) -.17 [-.44, .11] 0.23 -.33**a [-.56, -.11] 3.29

Note. All inter-correlations reported for the URG are based on the maximum-likelihood
estimator with Satorra-Bentler adjusted standard errors and test statistics.
Bayes Factors (BF10) are based on the Bayesian correlation test introduced by Wetzels and
Wagenmakers (2012).
HH = Honesty-Humility, AG = Agreeableness, URG = Uncostly Retaliation Game.
a
: one-sided testing;
** p < .01.
Running Head: HEXACO Agreeableness 41

AUTHOR NOTE

The work reported herein was supported by a grant from the German Research Foundation to

the first author [HI 1600/6-1].

Running Head: HEXACO Agreeableness 42

Highlights
The evolution of cooperation requires fairness and forgiveness

The HEXACO thus differentiates Honesty-Humility and Agreeableness (AG)

We test the AG-forgiveness-link in a novel paradigm, the uncostly retaliation game

AG (uniquely) predicts forgiveness vs. retaliation in response to exploitation

Results also confirm the usefulness of the new paradigm for studying forgiveness