the advantages of being unpredictable: how emotional inconsistency extracts concessions in...

Journal of Experimental Social Psychology 49 (2013) 498–508

Contents lists available at SciVerse ScienceDirect

Journal of Experimental Social Psychology

j ourna l homepage: www.e lsev ie r .com/ locate / jesp

The advantages of being unpredictable: How emotional inconsistency extractsconcessions in negotiation☆

Marwan Sinaceur a, Hajo Adam b, Gerben A. Van Kleef c, Adam D. Galinsky d,⁎a INSEAD, Franceb Rice University, United Statesc University of Amsterdam, Netherlandsd Columbia University, United States

H I G H L I G H T S

► Emotional inconsistency and unpredictability make recipients comply in negotiation.► Emotional inconsistency induces recipients to concede more than express anger.► This effect occurs because recipients feel less control.► Emotional inconsistency was manipulated by alternating between emotions.► The results speak to research on emotions and to theories on unpredictability.

☆ This research was generously funded by the INSEAD2520-609) through an INSEAD R&D Committee Grant that⁎ Corresponding author.

E-mail address: [email protected]

0022-1031/$ – see front matter © 2013 Elsevier Inc. Allhttp://dx.doi.org/10.1016/j.jesp.2013.01.007

a b s t r a c t
a r t i c l e i n f o
Article history:Received 13 July 2012Revised 31 December 2012Available online 25 January 2013

Keywords:EmotionInconsistencyChangeUnpredictabilityFeeling control over outcomeNegotiation

Integrating recent work on emotional communication with social science theories on unpredictability, weinvestigated whether communicating emotional inconsistency and unpredictability would affect recipients'concession-making in negotiation. We hypothesized that emotional inconsistency and unpredictabilitywould increase recipients' concessions by making recipients feel less control over the outcome. In Experiment1, dyads negotiated face-to-face after one negotiator within each dyad expressed either anger or emotionalinconsistency by alternating between anger and happiness. In Experiment 2, participants received angryand/or happy messages from a simulated negotiation opponent. In Experiment 3, participants read a scenarioabout a negotiator who expressed either anger or emotional inconsistency by alternating between anger anddisappointment. In all three experiments, emotional inconsistency induced recipients tomake greater concessionscompared to expressing a consistent emotion. Further, in all three experiments, the effect of emotionalinconsistency was mediated by recipients' feeling less control. These findings qualify previous research onanger in negotiation and demonstrate the importance of feelings of control for negotiation outcomes.

© 2013 Elsevier Inc. All rights reserved.

Introduction

In dealingwith foreign powers, Queen Elizabeth I exhibited “baffling”displays of emotional inconsistency. During her negotiations, sheshowed “endless short-term shifts that every historian of Elizabethanforeign policy has charted… Yet, it is worth noticing how often thesenegotiations succeeded in achieving the desired objectives” (Loades,2006, pp. 307–308). According to this historical account, communicatingemotional inconsistency and unpredictability helped Elizabeth I securegreater concessions fromopponents in conflict andnegotiation. Interest-ingly, other leaders practiced this strategy of communicating emotionalinconsistency, believing this would make recipients concede in conflict

Alumni Fund (INSEAD Grantwas awarded to the first author.

(A.D. Galinsky).

rights reserved.

and negotiation, among them Soviet Premier Nikita Khrushchev andFrench President Charles de Gaulle (Gaddis, 2005). Negotiators, theseleaders believed, “must project an aura of mystery” through their emo-tional reactions to obtain compliance (Gaddis, 2005, p. 298). Appearinginconsistent and unpredictable has even become part of contemporaryU.S. military strategy: a report from the U.S. Strategic Command (1995,p. 7) suggests that it is deemed as essential to deterrence. Indeed, bothconflict theorists (Schelling, 1960) and negotiation practitioners(Koren & Goodman, 1992) have proposed that appearing inconsistentand unpredictable yields compliance in competitive interactions. Yet,despite the abundant anecdotal evidence and theorizing, there is littleempirical evidence for the relationship between communicating emo-tional inconsistency and others' compliance.

Across three experiments, we investigated whether emotionalinconsistency and unpredictability would affect recipients' concessionsin negotiation. We also investigated a mechanism behind this effect,testing whether the recipients' feeling less control over the outcome

http://dx.doi.org/10.1016/j.jesp.2013.01.007

mailto:[email protected]

http://dx.doi.org/10.1016/j.jesp.2013.01.007

http://www.sciencedirect.com/science/journal/00221031

499M. Sinaceur et al. / Journal of Experimental Social Psychology 49 (2013) 498–508

would mediate their making concessions to emotionally inconsistentnegotiators.

Emotional communication in negotiation

Negotiation often involves the communication of emotions(Elfenbein, 2007; Van Kleef, De Dreu, & Manstead, 2010). These com-municated emotions typically bear important social consequences(Frijda, 1986; Morris & Keltner, 2000) as they offer crucial informationabout intentions and what behavior to expect from the expresser(Keltner & Haidt, 1999). Indeed, people are quite adept at noticingthe discrete emotions communicated by others in social interactions(Frijda, 1986). Thus, negotiators draw inferences from a counterpart'semotional communication as they try to decipher and predict thecounterpart's behavior (Van Kleef et al., 2010); for example, an angryexpresser is perceived as tough (Sinaceur & Tiedens, 2006). Supportingthe theory that emotions have a social function in conflict and negotia-tion (Morris & Keltner, 2000), recent research has documented thatpeople can strategically communicate emotions such as anger to elicitconcessions from recipients in negotiation (Kopelman, Rosette, &Thompson, 2006; Lelieveld, Van Dijk, Van Beest, Steinel, & Van Kleef,2011; Lelieveld, Van Dijk, Van Beest, & Van Kleef, 2012; Overbeck,Neale, & Govan, 2010; Sinaceur & Tiedens, 2006; Steinel, Van Kleef, &Harinck, 2008; Van Kleef, De Dreu, & Manstead, 2004, Van Kleef et al.,2010; Van Kleef & Côté, 2007).

Emotional inconsistency

To date negotiation research has predominantly explored emotionsthat were communicated consistently throughout the negotiation. Onegeneralfinding that has emerged is that anger communication is a highlyeffective strategy to extract concessions from recipients (Sinaceur &Tiedens, 2006; Van Kleef et al., 2004, 2010).

One exception to the focus of negotiation research on communi-cating consistent emotions is a qualitative study suggesting that con-trasting positive and negative emotions elicit compliance (Rafaeli &Sutton, 1991). However, the field setting of that studymakes it difficultto separate emotions from moves, such as the offers being made. Also,emotions were broadly defined (e.g., “positive emotions” includedapproval, respect, and empathy). Although not in the domain of emo-tions, experimental work also suggests that contrasting positive (coop-erative) and negative (competitive) stances across encounters, as in thegood-cop/bad-cop strategy, elicit compliance (Hilty & Carnevale, 1993;seeHarford & Solomon, 1967). However, in this earlierwork, argumentswere not independent of offers, making it difficult to infer whether thecontrast effectwas due to the offers beingmade or the arguments beingexpressed. And, the arguments expressed involved little or no emotion.The current research further departs from this prior work by explor-ing inconsistency rather than contrast. By “inconsistency”, we meanoscillating or fluctuating between different psychological states over thecourse of a single encounter (Barrett, 2009; Frijda, 1986). Accordingly,we emphasize a mechanism of a different nature – feelings of control –that departs from perceptual contrast where prior behavior serves asa reference point against which later behavior is judged (Hilty &Carnevale, 1993).

Emotional inconsistency represents an important, thoughunderstudied, phenomenon. The predominant focus of negotiationresearch on communicating consistent emotions in a given situation(Van Kleef et al., 2010) precludes more intricate configurations ofemotion, such as emotional inconsistency or fluctuation over thecourse of a single encounter (Frijda, 1986). Indeed, inconsistency inemotions over time may often be typical (Barrett, 2009; Kuppens,Oravecz, & Tuerlinckx, 2010; Larsen, Augustine, & Prizmic, 2009;Scherer, 2009), particularly so in the domain of conflict and negotiation(Morris & Keltner, 2000; Rothman &Wiesenfeld, 2007). Emotion theo-rists have noted that people often oscillate between opposite emotions

over time (Larsen et al., 2009). In particular, emotional communicationsmay briskly change from positive to negative and vice-versa, especiallywhen new information (such as offers in a negotiation context) isbrought in (Frijda, 1986). For example, expressions of aggressive emo-tions such as anger may turn into expressions of contentment such ashappiness over the course of a single exchange and vice-versa.

Emotion theorists have even argued that emotions are essentiallycharacterized by their shifting and dynamic nature (Russell, 2009)and that “our lives are characterized by affective ups and downs, changesand fluctuations”, rather than by emotional consistency (Kuppens et al.,2010, p. 1042). Some have even proposed that emotions have a signalingfunction because they are inconstant (Frijda, 1986; Scherer, 2009). Draw-ing from these arguments, we propose to move beyond the focus ofnegotiation research on the consistent communication of emotions.We suggest that negotiators' alternating between communicating differ-ent emotions over the course of a single exchange (such as alternatingbetween anger and happiness) can affect concessions from recipients.

Inconsistency and unpredictability in competitive interactions

In conflict and negotiation, one primary social effect that emotionalinconsistency could have is conveying a sense of unpredictability toopponents. In general, conflict theorists have argued that strategiesthat project inconsistency essentially create a sense of unpredictabilityand surprise in opponents (Ellsberg, 1959; Schelling, 1960). Importantly,inconsistency reduces an opponent's ability to gain intelligence aboutone's past strategy, thus preventing them from deducing tell-taleregularities (Schelling, 1960). In negotiation, emotional inconsistencycan make it hard for opponents to anticipate one's future behavior(Ellsberg, 1959; Frank, 1988).

Conflict theorists have long theorized that being seen as inconsistentand unpredictable will thus increase recipients' compliance in competi-tive interactions (Frank, 1988; Schelling, 1960). This assertion resonateswith arguments across the social sciences. For example, sociologistsspeculate that recipients yield more before actors who appear inconsis-tent and unpredictable in their reactions (Crozier & Friedberg, 1980).Clinical psychologists have observed how unexpected, surprisingmoves can make recipients open to change (Watzlawick, Weakland, &Fisch, 1974). Extending these arguments, cognitive scientists theorizethat unpredictable, non-contingent behavior increases outcomes incompetitive contexts. “Irreducible uncertainty,” they propose, cannotbe learned by competitors, so organisms should behave with someinconsistency before competition (Glimcher, 2003).

Despite these arguments, empirical research on the effects ofinconsistency and unpredictability in competitive interactions has beenscant. Although several lines of argument suggest that a negotiator'semotional inconsistency and unpredictability might increase recipients'concessions, the current research is the first to directly test thisrelationship.

Feeling little control over outcomes

One reason why emotional inconsistency and concomitantunpredictability could have positive effects on concessions, we propose,is that it is likely to make recipients feel little control over the outcomeof the negotiation. By “control”, we mean the belief that one possessesoneself a response that can causally influence outcomes, for example,that one's own actions will determine outcomes (e.g., Litt, 1988;Stevens, Bavetta, & Gist, 1993). Social psychological research suggestsindeed that when people face inconsistency and unpredictability, theyoften feel a lack of control, which is a potent force that drives their re-actions (Kay, Shepherd, Blatz, Chua, & Galinsky, 2010; Kay, Whitson,Gaucher, & Galinsky, 2009; Menon, Sheldon, & Galinsky, under review;Sullivan, Landau, & Rothschild, 2010; Waytz et al., 2010; Whitson &Galinsky, 2008). For example, when people face inconsistency andunpredictability they over-attribute causal influence to opponents

500 M. Sinaceur et al. / Journal of Experimental Social Psychology 49 (2013) 498–508

(Sullivan et al., 2010). In interpersonal interactions, surprising movesfrom an opponent make people over-attribute control to that opponent(Menon et al., under review). That is, peoplewho face inconsistency andunpredictability are likely to feel that they themselves have little causalinfluence over outcomes, and that their opponents have great causalinfluence over outcomes (Menon et al., under review; Sullivan et al.,2010). It seems plausible that this relationship between facing inconsis-tency and feeling that one has little causal influence over outcomeswould extend to the domain of negotiation.

In a negotiation, feeling little control over outcomes involves thebelief that one can do little to alter the negotiated outcome, whichtranslates into lesser demands and therefore greater concession-making to an opponent (Stevens et al., 1993). In general, feeling littlecontrol makes people more likely to be subject to the influence ofothers and adopt others' positions (e.g., Eagly & Chaiken, 1993; DeVries, De Dreu, Gordijn, & Schuurman, 1996; Morris, Sim, & Girotto,1995). AsMorris et al. (1995, p. 215) put it, feeling little causal influenceover an outcome makes oneself less likely “to act so as to influence”the outcome. This relationship between feeling little control andconcession-making is consistent with the notion that, when one feelsthat one's behavior is externally motivated, one's concessions appearless psychologically costly to the self (Pruitt, 1981). Drawing fromthis prior research, we investigate whether the inconsistency andunpredictability in a negotiator's emotions might positively affect therecipients' concessions by inducing lesser feelings of control.

Overview

Integrating recent work on emotional communication (Van Kleef etal., 2010) with social science theories on unpredictability (Schelling,1960), we investigated whether communicating emotional inconsis-tency would affect recipients' concession-making in negotiation. InExperiment 1, emotional inconsistency was manipulated in actual, face-to-face interactions. Dyads negotiated after one negotiator within eachdyad was advised to express either anger or emotional inconsistency(by alternating emotional expressions of opposite valence, i.e., angerand happiness). In Experiment 2, emotional inconsistency was manipu-lated using a more controlled, computer-based paradigm. Specifically, asimulated opponent alternated opposite emotions (anger and happi-ness) over a sequence of rounds in a computer-mediated negotiation.In Experiment 3, emotional inconsistencywasmanipulated using a con-trolled scenario. Participants read a dialogue in which a negotiatorexpressed either anger or emotional inconsistency by alternatingbetween anger and disappointment. In Experiment 1, we controlledfor recipients' perceptions about the expresser's demands statistically,and in Experiments 2–3 we held the actual offers (actual concessions)constant. All three experiments further tested our argument by examin-ing whether recipients' feeling little control wouldmediate the effect ofemotional inconsistency on concessions.

Experiment 1

In Experiment 1, we used a face-to-face negotiation paradigm. Totest the effectiveness of emotional inconsistency as a strategy, we com-pared it against the consistent expression of anger in order to allowcomparison with prior research and to provide a conservative test ofour hypothesis (e.g., Lelieveld et al., 2011, 2012; Steinel et al., 2008;Van Kleef & De Dreu, 2010). Indeed, anger is highly effective to extractconcessions in negotiation (for example, significantly more so thanno emotion, Van Kleef et al., 2010). We manipulated emotional incon-sistency by alternating emotional expressions of opposite valence,i.e., anger and happiness. We did so for two reasons. First, emotiontheorists have noted that alternating in emotional valence (such asbetween anger and happiness) represents a common, typical case ofemotional inconsistency (Frijda, 1986; Larsen et al., 2009), particularlyin conflict and negotiation (Frijda, 1986; Morris & Keltner, 2000;

Rothman & Wiesenfeld, 2007). Second, the hypothesis that a mixtureof anger and happiness (i.e., less anger overall) could lead to more con-cessions than anger may not be straightforwardly derived from priornegotiation research since anger has been shown to elicit significantlymore concessions than does happiness (Van Kleef et al., 2010). Finally,we further tested our argument by examining whether recipients' feel-ing little controlwouldmediate the effect of emotional inconsistency onrecipients' concessions.

Method

ParticipantsSixtyMaster inManagement students (50.0% female) participated in

the study. Participants were randomly assigned into 30 same-genderdyads.

Experimental designDyads were randomly assigned to one of two experimental condi-

tions: Emotional Inconsistency or Anger. In each dyad, one participantwas randomly assigned to being the expresser and the other to beingthe recipient.

MaterialsNegotiation exercise. Participants role-played a two-party negoti-

ation exercise (see Sinaceur, 2010; Sinaceur, Van Kleef, Neale,Adam, & Haag, 2011). Participants were randomly assigned to eitherone of two roles in negotiating a new business venture. This negoti-ation was between two firms who were considering setting up aventure to develop, manufacture and market a new product. Bothprior research (Sinaceur, 2010; Sinaceur et al., 2011) and pre-testing showed that role did not affect negotiation outcomes; indeed,it did not, so this factor is not discussed further. Participants kept allmaterials confidential.

The negotiation was about the terms of the prospective businessventure and included four issues: profit sharing, manufacturing facili-ties, R&D investment, and sales coverage. Payoff schedules for eachrole specified the point values that participants could reach on eachissue. Two issues were distributive (i.e., the parties' preferences werein opposition): profit sharing and sales coverage. And two issues wereintegrative (i.e., parties cared about the issues differently, which madetrading-off possible to increase joint gain): manufacturing facilitiesand R&D investment. Across the experimental conditions, all partici-pants were told their goal was to maximize their own points.

Manipulation of the expression of emotional inconsistencyTo manipulate the expression of emotional inconsistency, we drew

from successful manipulations of emotional expressions from priorresearch (Kopelman et al., 2006; Sinaceur & Tiedens, 2006; Sinaceuret al., 2011; see Overbeck et al., 2010). In each dyad, one participantwas randomly assigned to being the expresser. Along with their payoffinformation, these participants were given a set of negotiation recom-mendations that was adapted from prior research (Kopelman et al.,2006; Sinaceur & Tiedens, 2006; Sinaceur et al., 2011). In order to createconsistency across the two emotion conditions, participants in both theanger and emotional inconsistency conditionswere advised to use theiremotions to obtain concessions and negotiate successfully. However,the two emotion conditions varied in terms of what the recommenda-tion was.

In the anger expression condition (N=16) participants were advisedto express anger throughout the negotiation. They were given recom-mendations about expressing anger consistently, such as using angrysentences (e.g., “This negotiation really makes me angry”, “You'rebeginning to get on my nerves”, “I'm very angry now”). These angrysentences were taken from previous research on anger in negotiation(Sinaceur & Tiedens, 2006; Sinaceur et al., 2011; Van Kleef & Côté,2007; Van Kleef et al., 2004) and had been pilot tested as expressing


anger reliably in the context of negotiation (Sinaceur et al., 2011; VanKleef et al., 2004). In the emotional inconsistency expression condition(N=14) participants were advised to express emotional inconsistencyby alternating between expressing anger and happiness throughout thenegotiation. Theywere given recommendations about expressing angerand happiness alternately, such as alternately using angry sentences(similar to the anger condition) and happy sentences (e.g., “This nego-tiation reallymakesme happy”, “I'm happywe're negotiating together”,“I'm very happy now”). Recipients were given no emotion instructions.

ProcedureParticipants had 25 min to read their instructions and prepare for

the negotiation. Then, they were given 25 min to negotiate face-to-face in dyads. After negotiating, participants indicated the terms ofthe agreement they reached, if any.

Measures

Value claiming. Value claiming was the percentage of the total pointsthat one party got for himself/herself (e.g., Sinaceur & Tiedens, 2006).Recipients' lower value claimed indicates greater concessions bythem.

Following the negotiation, participants answered items on 7-pointscales (1= little, 7=very much).

Lack of control. The recipients' feeling little controlwas our hypothesizedmediator for the effect of emotional inconsistency. It was measured byasking participants how much control they felt they had over thenegotiation.

Emotional inconsistency manipulation check. Participants also ratedhow consistent the counterpart was during the negotiation.

Alternative explanations.We also wanted to rule out several importantalternative explanations due to the nature of our manipulation,namely that, across conditions, recipients might have (a) perceivedexpressers as differentially accepting the recipients' offers, (b) perceivedexpressers as differentially demanding, (c) perceived expressers asdifferentially losing their mind while negotiating, or (d) put differentialamount of thinking into their making offers. Thus, we also measuredhow much the counterpart's behavior was indicative of his/her accep-tance of the offers made by participants, how demanding the counter-part was, how much the counterpart seemed to have lost his/hermind, and how much thinking participants had put into their makingoffers.

Note. A Bootstrapping procedure showed that themediator) was significant. Numbers represent stanparentheses represent simultaneous standardized r* p < .05, ** p < .01, *** p < .001

EmotionalInconsistency vs.

Anger

Feeling Lack Control.37*

.57*** (.39*

Fig. 1. Experiment 1: Mediation of emotional

Results

Manipulation checkRecipients facing emotionally inconsistent expressers rated the

expresser as less consistent (M=3.21, SD=.97) than did recipientsfacing angry expressers (M=4.19, SD=1.22, t(29)=2.39, pb .03,d=.17).

Value claimingRecipients facing emotionally inconsistent expressers claimed less

value (M=48.10%, SD=2.63%) than did recipients facing angryexpressers (M=50.94%, SD=1.55%, t(29)=3.67, pb .001, d=.32).Thus, recipients facing emotionally inconsistent expressersmade greaterconcessions.

Value creationWe also conducted an exploratory analysis on value creation.

Value creation was measured by the dyad's total points. A one-wayemotion expression ANOVA showed no effect (p>.90).

Lack of controlRecipients facing emotionally inconsistent expressers felt less

control (M=4.29, SD=.91) than did recipients facing angry expressers(M=4.94, SD=.77, t(29)=2.12, p=.04, d=.14).

Alternative explanationsEmotional inconsistency did not affect how much the expressers'

behavior was indicative of their acceptance of the offers madeby recipients (p>.27), how demanding expressers were perceived(p>.88), how much expressers seemed to have lost their mind(p>.89), or how much thinking recipients put themselves into theirmaking offers (p>.87). Therefore these could not mediate the effectof emotional inconsistency. Importantly, the mediation results belowheld whether controlling for these or not.

Mediation by lack of controlNext, we examined whether recipients' feeling little control would

mediate the effect of emotional inconsistency on concessions. Fig. 1presents the results of the regression analyses. A bootstrapping pro-cedure using 5,000 resamples demonstrated that the indirect path(i.e., the path through the mediator) was significant (95% CI Low=.026; 95% CI High=.567). Overall, the recipients' feeling little controlpartiallymediated the effect of emotional inconsistency on concessions.

Experiment 2

We conducted a second experiment to further document theeffect of emotional inconsistency and unpredictability on recipients'

indirect effect (i.e., the path through thedardized regression coefficients. Numbers inegression coefficients.

of .62*** (.48**)

)Concessions

inconsistency by feeling lack of control.

Table 1Experiment 2: Emotional messages used to manipulate anger versus happiness.

Angry reaction Happy reaction

In round 2 I am very angry with your firstoffer! This negotiation startsvery badly.

I am very happy with your firstoffer! This negotiation starts verywell.

In round 3 I've received your second offer.In fact, this second offer gets muchon my nerves! I am not happy atall. It makes me irritated.

I've received your second offer.In fact, this second offer pleasesme much! I am very happy.It makes me feel good.

In round 4 Is this your third offer? This is notserious!!! I'm fed up with this.I am very angry with thisnegotiation.

Is this your third offer?This is really cool!!! I'mdelighted about this. I am veryhappy with this negotiation.

In round 5 I just saw your fourth offer. Itreally makes me feel angry!It's really not serious!!!It pisses me off.

I just saw your fourth offer.It really makes me feel happy!It's super cool!!! I am very happy.

Note. The actual offers (actual concessions) made by the computer were held constantacross conditions. Specifically, participants received amessage about the offer beingmadeafter receiving the emotionalmessage. Participantswere told that only the offermessage –but not the emotional message – indicated what the computer would actually do next.Thus, the order was as follows: participants received an emotional message, then amessage containing the actual offer, and then responded to the latter by makingthemselves an offer.


concessions. Experiment 2 offered a more controlled test of our argu-ment as it used a paradigm in which the arguments and offers wereheld constant. We manipulated emotional inconsistency by alternatingemotional communications of opposite valence (anger and happiness)made by the opponent over a sequence of rounds in a computer-mediated negotiation. Also, we included the last move's emotion as acounterbalancing factor to insure that our results would be indepen-dent of the valence or order of alternating, thus having a fully-crosseddesign. Experiment 2 again examined whether recipients' feeling littlecontrol would mediate the effect of emotional inconsistency onconcessions.

Method

Experimental design and participantsOne-hundred and fifty-two university students (51.3% female)were

randomly assigned to a 2 (Emotional Inconsistency: constant emotionsvs. alternating emotions)×2 (Last Move: anger vs. happiness) factorial,between-subjects design.

ProcedureParticipants were randomly assigned to the experimental condi-

tions. For each lab session, four to six participants were seated in sepa-rate cubicles equippedwith computers. Participants were told that theywould engage in a computer-mediated negotiationwith another partic-ipant. Actually, the other participant's behavior was simulated by thecomputer, which allowed us to obtain a high level of experimental con-trol. Specifically, we manipulated the opponent's emotional statementswhile holding constant the opponent's offers and arguments.

The negotiation task was adapted from Van Kleef et al. (2004; e.g.,Van Kleef & Côté, 2007). Participants played the role of a seller ofmobile phones who negotiated with a buyer (i.e., the computer) in anoffer and counteroffer sequence. The negotiation involved three issues:price, warranty period, and number of models. For each issue therewere nine possible agreement positions, which yielded different pointsfor participants: position 1 yielded the maximum points; position 9yielded the minimum points. Each issue had an equal weight. Partici-pants were incentivized to maximize their own points through thepossibility of winning MP3 players.

Participants were told that the negotiation could last up to fiverounds of offer and counteroffer exchanges between them and thecounterpart. The computer always followed the same pre-determinedpattern in making offers throughout the rounds (e.g., Van Kleef et al.,2004; Van Kleef & Côté, 2007). Specifically, over five negotiation roundsthe computer offered the following levels of agreement for price, war-ranty, and number of models, respectively: 8–7–8 (round 1), 8–7–7(round 2), 8–6–7 (round 3), 7–6–7 (round 4), and 7–6–6 (round 5).Prior research has shown that these specific offers have face validityand are seen as intermediate in cooperativeness and competitiveness(De Dreu & Van Lange, 1995). After each simulated counterpart's offer,participants responded by making an offer themselves over the fivenegotiation rounds. Hence, an offer and counteroffer sequence ensued,with participants themselves going from making an initial offer tomaking a final offer.

ManipulationsParticipants were told that they had been randomly selected to

receive the counterpart's reactions in addition to (but separately from)receiving his/her offers. They received a message from the allegedbuyer disclosing his/her emotional reactions in rounds 2, 3, 4, and 5.These emotional messages contained either angry or happy statementsthat were adapted from prior research (Sinaceur & Tiedens, 2006; VanKleef et al., 2004). Table 1 presents these emotional messages. As canbe seen, emotions were directed at the offer made by participants inthe prior round (independent of that offer) to make the counterpart'salternating in emotional valence more understandable and believable.

This way to manipulate emotions as reactions to prior offers drawsfrom past research (e.g., Lelieveld et al., 2011, 2012; Sinaceur et al.,2011; Steinel et al., 2008; Pietroni, Van Kleef, De Dreu, & Pagliaro,2008; Van Kleef et al., 2004, 2010).

It is important to note that the actual offers (i.e., actual decisions)being made by the computer were held constant across conditions.Specifically, participants received a message containing the offerafter receiving an emotional message. And, they received the messagecontaining the offer before making their own decision. Overall, theorder was as follows: participants received an emotional message,then a message containing an offer (held constant across conditions),and then responded to the latter by making an offer themselves.There was always a 15-second interval between when participantsreceived the emotional message (the manipulation) and when partic-ipants received the message containing the offer (held constant). Theoffer message always read: “I am going to offer X–Y–Z” with X–Y–Zbeing the specific offer. This offer message remained visible whileparticipants were given time to enter the terms of their offer. Further,the instructions made clear to the participants that the emotionalmessages were “current thoughts and reactions to the participants'prior offers” rather than “intentions to act.” Thus, the emotional mes-sages provided information about “the opponent's process of thinking”rather than about what the opponent would do next. Participants weretold that only the messages containing an offer – but not the emotionalmessages – were informative of the computer's decision-making andacceptance or rejection of the participants' offers. In short, the instruc-tions made clear that the emotions were reactions to participants'past behavior, not indicative of the opponent's future behavior(e.g., Lelieveld et al., 2011, 2012; Van Kleef et al., 2010).

Opponent's last move. Participants were reminded before round 5 thatit was the last round. In the angry (happy) last move condition, par-ticipants received an angry (happy) message before receiving a finaloffer message and making their final offer.

Opponent's emotional inconsistency. In the consistent emotion condition,participants received emotional messages that were of the same emo-tional valence in all rounds (thus creating angry–angry–angry–angryor happy–happy–happy–happy sequences, depending on the last movevalence). In the inconsistent emotion condition, participants receivedemotional messages that alternated in emotional valence every round


(thus yielding happy–angry–happy–angry or angry–happy–angry–happy sequences, depending on the last move valence).

Dependent measures

Concessions. The final offer made by participants was the critical out-come (e.g., Adam, Shirako, & Maddux, 2010; Sinaceur et al., 2011; VanKleef & Côté, 2007). It was obtained by adding together the levels ofagreement that participants finally proposed for the three negotiationissues. A higher score indicated that participants made larger conces-sions. The possible range was 3–19, with 3 being the minimum conces-sion and 19 being the final offer by the computer.

After the negotiation, participants answered items measuring theirperceptions on 9-point scales (1= little, 9=very much).

Last move manipulation checks. Participants rated how much theircounterpart expressed anger and happiness at the end of the negotia-tion, respectively. They also ratedhowmuch their counterpart expressedanger and happiness overall throughout the negotiation, respectively.These latter measures were used as control variables for our last movecheck analyses.

Emotional inconsistency manipulation check. There were two checks forthe emotional inconsistency manipulation. Inconsistency was the aver-age of four items (e.g., how constant the counterpart was throughoutthe negotiation [reverse-scored], how changing the counterpart wasthroughout the negotiation; α=.86). In addition, we wanted to verifythat the recipients' sense of unpredictability was augmented by ourmanipulation as intended. Sense of unpredictability was the averageof four items (e.g., how easy participants had found it to predicttheir counterpart's reactions [reverse-scored], how much partici-pants had believed the counterpart's behavior to be unpredictable;α=.68).

Lack of control. The recipients' feeling lack of control was the hy-pothesized mediator of the effect of emotional inconsistency. Itwas the average of four items (e.g., how much participants hadfelt in control during the negotiation [reverse-scored], how muchparticipants had felt they lacked control over the negotiating situa-tion; α=.63).

Alternative explanations.We also wanted to rule out several importantalternative explanations for the effect of emotional inconsistency.First, it is possible that, compared to consistent emotion, emotionalinconsistency might have induced recipients to perceive differentiallythe informational value of the emotional messages with regard to theacceptability of the offers; that is, the emotional messages might beperceived as communicating more information about the acceptabilityof the offers. Further, emotional inconsistency might have inducedrecipients to attach different importance to the emotional messages orto put different amount of thinking into understanding these. Finally,emotional inconsistency might have induced recipients to perceivethe counterpart as having lost his/her mind or simply as not caringabout his/her decisions. To rule out all these alternative explanations,we measured (a) how much the emotional messages communicatedinformation about the acceptability of the offers through four items(e.g., how much the counterpart's emotional messages were indicativeof his/her actual satisfaction with the offers made by participants, howmuch the counterpart's emotional messages were indicative of thedecisions s/he would be really making; α=.77); (b) howmuch impor-tance participants attached to the emotional messages through twoitems (e.g., howmuch importance participants attached to the counter-part's emotional messages; α=.76); (c) how much thinking partici-pants put into understanding the emotional messages through twoitems (e.g., how much thought participants gave to understand thecounterpart's emotional messages; α=.76); (d) how much the

counterpart seemed to have lost his/her mind through four items(how irrational, odd, fanciful, and unbalanced the counterpart was;α=.75); and (e) how much the counterpart seemed to care abouthis/her decisions through four items (e.g., how much the counterpartseemed to think through his/her decisions, how calculating the coun-terpart was; α=.66).

Results

Manipulation checksLast move manipulation checks. The anger last move check was sub-

mitted to a 2 (Emotional Inconsistency: constant emotions vs. alter-nating emotions)×2 (Last Move: anger vs. happiness) ANCOVA. Itcontrolled for anger expressed overall throughout the negotiation toensure that results would not merely reflect perceptions of emotionsexpressed prior to the last move. Participants in the angry lastmove con-dition thought that the counterpart expressed more anger at the end(M=6.99, SD=1.96) than did participants in the happy last movecondition (M=4.31, SD=3.18; F(1, 150)=17.18, pb .001, η2=.11).There were no other effects.

Similarly, the happiness last move check was submitted to a 2(Emotional Inconsistency)×2 (Last Move) ANCOVA, which controlledfor happiness expressed overall throughout the negotiation. Partici-pants in the angry last move condition thought that the counterpartexpressed less happiness at the end (M=3.40, SD=2.35) than didparticipants in the happy last move condition (M=5.47, SD=3.07;F(1, 151)=12.91, pb .001, η2=.08). There were no other effects.

Emotional inconsistency manipulation check. Inconsistency was sub-mitted to a 2 (Emotional Inconsistency)×2 (Last Move) ANOVA. Par-ticipants in the emotional inconsistency condition thought that thecounterpart was more inconstant (M=5.21, SD=2.14) than did par-ticipants in the emotional consistency condition (M=2.65, SD=1.64;F(1, 151)=66.86, pb .001, η2=.31). There were no other effects.

Sense of unpredictabilityParticipants' measured sense of unpredictability was submitted to

a 2 (Emotional Inconsistency)×2 (Last Move) ANOVA. The ANOVAshowed amain effect of emotional inconsistency, such that participantsexperienced a greater sense of unpredictability with an emotionallyinconsistent counterpart (M=5.25, SD=1.69) thanwith an emotionallyconsistent one (M=3.88, SD=1.66), F(1, 151)=25.93, pb .001, η2=.15. There was also a main effect of the last move emotion but inthe opposite direction, i.e., participants experienced a lower sense ofunpredictability when the counterpart's last move was angry (M=4.21, SD=1.74) than happy (M=4.92, SD=1.82), F(1, 151)=7.00,pb .01, η2=.05. Importantly, there was no interaction.

Concession-makingFig. 2 displays means for concessions. Concessions were submitted

to a 2 (Emotional Inconsistency: constant emotions vs. alternatingemotions)×2 (Last Move: anger vs. happiness) ANCOVA. It controlledfor participants' initial offers (at round 1, i.e., before any manipulation)since prior research showed that outcomes are strongly driven by initialoffers (Galinsky & Mussweiler, 2001). Replicating prior work, theANCOVA showed a main effect of the last move emotion, such that par-ticipantsmademore concessionswhen the counterpart's lastmovewasangry (M=15.04, SD=2.82) than happy (M=13.31, SD=3.50), F(1,151)=13.40, pb .001, η2=.08. Above and beyond that, the ANCOVAalso showed a main effect of emotional inconsistency, such that partic-ipants made more concessions to an emotionally inconsistent counter-part (M=14.81, SD=2.88) than to an emotionally consistent one (M=13.54, SD=3.52), F(1, 151)=7.29, pb .01, η2=.05. Importantly, therewas no interaction between the two factors (p>.61), which indicatesthat the effect of emotional inconsistency was independent of the effectof the last move.

16EMOTIONALCONSISTENCY

EMOTIONALINCONSISTENCY

15

Tot

al c

once

ssio

ns (

last

off

er m

ade)

14

13

12HAPPY LAST MOVE ANGRY LAST MOVE

Note. A higher score indicates larger concessions. ANCOVA showed two main effects of thelast move emotion (p < .001) and of emotional inconsistency (p < .01), respectively.Importantly, there was no interaction between the two factors (p > .61).

Fig. 2. Experiment 2: Total of concessions made by participants.


Lack of controlParticipants' feeling lack of control was submitted to a 2 (Emotional

Inconsistency)×2 (Last Move) ANOVA. The ANOVA showed a maineffect of emotional inconsistency, such that participants felt a lack ofcontrol to a greater extentwith an emotionally inconsistent counterpart(M=3.31, SD=1.49) than with an emotionally consistent one (M=2.59, SD=1.42), F(1, 151)=9.23, pb .005,η2=.06. The ANOVA showedno other effects.

Alternative explanations2×2 ANOVAs showed that emotional inconsistency did not affect

howmuch the emotional messages communicated information aboutthe acceptability of the offers (p>.31), how much importance partic-ipants attached to the emotional messages (p>.33), how much think-ing participants put into understanding the emotional messages (p>.24), how much the counterpart seemed to have lost his/her mind(p>.77), or how much the counterpart seemed to care about his/her

Note. Regressions that include emotional inconsistfactor and their interaction term (thus forming orthcontrolled for in all regressions that are conductedstandardized regression coefficients. Numbers in pregression coefficients.^ p < .10, * p < .05, ** p < .01, *** p < .001

Emotional Inconsistency

0 = No1 = Yes

Feeling LackControl.24**

.20** (.14^


decisions (p>.78). The 2×2 ANOVAs showed no effects on these vari-ables, but a main effect of the last move on information about theacceptability of the offers such that an angry last move communicatedmore information about the acceptability of the offers than did ahappy last move (pb .001). Therefore these variables could not mediatethe effect of emotional inconsistency. Importantly, the mediationresults below held whether controlling for these or not.

Mediation by lack of controlNext, we examined whether recipients' feeling little control would

mediate the effect of emotional inconsistency on concession-making,controlling for participants' initial offers. Fig. 3 presents the resultsof the regression analyses. A bootstrapping procedure using 5,000resamples demonstrated that the path through the mediator was sig-nificant (95% CI Low=.042; 95% CI High=.278). Thus, recipients'feeling little control mediated the effect of emotional inconsistencyon concession-making.

ency as a factor also include the last moveogonal factors). Participants’ initial offers are on concessions. Numbers representarentheses represent simultaneous standardized

of .29*** (.24***)

)Concessions



Experiment 3

We conducted a third experiment to further document the effect ofemotional inconsistency and unpredictability on recipients' conces-sions. First, Experiment 3 used a scenario-based paradigm to furtherensure that emotionswere conveyed in amanner thatwas independentof the acceptability of the demands and the arguments being made.In this scenario, wemanipulated emotions as the emotional tone associ-ated with the opponent's statements, while keeping entirely constantthe substance of all the opponent's statements. Thus, in Experiment 3,the manipulation of emotions was perfectly orthogonal to all theopponent's statements.

We also wanted to examine whether our effects might extend toemotional inconsistency within valence. Whereas in Experiments 1–2we had manipulated emotional inconsistency through alternations ofemotional valence, in Experiment 3wewanted tomanipulate emotionalinconsistency through alternations of emotions within the same neg-ative domain. We focused on anger and disappointment, which areboth negative emotions, for three reasons. First, these emotions arecommon in negotiation (Lelieveld et al., 2011, 2012; Van Dijk &Zeelenberg, 2002). Second, both are reactions to undesirable out-comes; hence, both convey similar levels of dissatisfaction withanother's last proposal (Frijda, 1986; Lelieveld et al., 2011, 2012;Van Dijk & Zeelenberg, 2002). Third, anger and disappointment notonly share a negative valence but also possess similar levels of inten-sity and appropriateness (Lelieveld et al., 2011, 2012). Because angerhas been the focus of most prior research on emotions in negotiation,and because prior research suggests that anger is highly effective toextract concessions in negotiation (for example, significantly more sothan no emotion, Van Kleef et al., 2010), we again compared emotionalinconsistency (varying between anger and disappointment) to anger.

Again, we tested several important alternative explanations forthe emotional inconsistency effect. In particular, it is possible thatan opponent's emotional inconsistency might lead participants totake the task less seriously as they could think that an emotionallyinconsistent opponent is kidding around. Therefore, we wanted toestablish that emotional inconsistency is taken at face value and notseen as comical or unrealistic.

Method

Experimental design and participantsNinety university students and professionals (41.1% female) were

randomly assigned to one of two experimental conditions: EmotionalInconsistency or Consistent Anger. Emotional inconsistency wasmanipulated through alternating between two negative emotions:anger and disappointment.

ProcedureParticipants read a negotiation scenario drawn from prior research

(Adam et al., 2010; Sinaceur & Tiedens, 2006). This procedure provideduswith a high level of experimental control. Importantly, prior researchhas demonstrated that it yields results similar to those of other proce-dures (e.g., Adam et al., 2010; Sinaceur, 2010; Sinaceur & Tiedens,2006; Van Kleef & Côté, 2007). Specifically, all participants were askedto imagine that they were playing the role of a negotiator who wasfinalizing a deal about the sale of technical equipment (see Appendix Afor full text). The scenario informed participants that they were negoti-ating the terms of the warranty, which involved three issues: liabilitycoverage; time needed for possible repair; and a discount for spareparts. The negotiation dialogue was comprised of eleven statementsmade by the participants' character and their counterpart. The onlyaspect of the scenario that varied across conditions was the emotionaltone of the counterpart's last four statements.

Emotional inconsistency manipulation. The last four statements madeby the counterpart in the dialogue were presented either with thephrase “Counterpart, in an angry way:” or “Counterpart, in a disap-pointed way:” displayed right before the statement. In the consistentanger condition (N=46), all four statements were preceded with theangry prefix (thus yielding an angry–angry–angry–angry sequence).In the emotional inconsistency condition (N=44), the four statementswere alternatively preceded with the disappointed prefix and theangry prefix (thus yielding a disappointed–angry–disappointed–angrysequence). It is important to note that the dialogue itself did not vary.Literally two words differed across conditions. Thus, the emotionswere conveyed in a manner that was independent of all the demandsand arguments being made.

Dependent variable

Participants were instructed to consider what terms they wouldagree to for the warranty. The liability component was measured asfollows: “What liability coverage would you grant? Please indicateany number between 0.1 million and 15 million. 0.1 million is thesmallest concession possible, 15 million is the biggest concessionpossible.” The repair time component was measured as follows: “Whatrepair time would you grant? Please indicate any number between36 hours and 12 hours. 36 hours is the smallest concession possible,12 hours is the biggest concession possible.” The discount componentwas measured as follows: “What discount for spare parts would yougrant? Please indicate any number between 10% and 60%. 10% is thesmallest concession possible, 60% is the biggest concession possible.”

A total concessions variable was created. For liability and discount,a higher score indicated that participants made a bigger concession onthose issues. However, for repair time, a lower score indicated thatparticipants made a bigger concession on that issue. Thus, the repairtime component was reversed. Since each issue was on a differentscale, we standardized them, and then we added them together tocreate a total concessions score.

MeasuresAfter deciding on what terms they would agree to, participants

answered items on 7-point scales (1= little, 7=much).

Manipulation checks. There were two checks for the emotional inconsis-tency manipulation. Consistency was measured through two items (howemotionally consistent the counterpart was during the negotiation, howconstant the counterpart was throughout the negotiation; α=.73). Wealso wanted to verify that the recipients' sense of unpredictability wasaugmented by our manipulation as intended. Sense of unpredictabilitywas measured through two items (how easy participants had foundit to predict the counterpart's reactions, how much participants hadthought they could predict the counterpart's reactions; α=.74).

Lack of control. The recipients' feeling lack of control was the hypoth-esized mediator of the effect of emotional inconsistency. It was theaverage of four items (α=.90). To ensure the robustness of our mea-sure, in this study two items were about control over the negotiationprocess (e.g., how much control participants had felt over how thenegotiation evolved) and two items were about control over thenegotiation outcome (e.g., how much control participants had feltover the outcome). A confirmatory factor analysis verified that allitems loadedon one factor (it yielded only one factorwith an eigenvaluegreater than 1, i.e., 3.08, which explained 76.92% of the total variance;also, all items loaded positively on that factor with all loadings l's>.78).

Alternative explanations. We also wanted to rule out potential alterna-tive explanations for the effect of emotional inconsistency. First, wemeasured how much the counterpart's emotional reactions weretaken at face value through four items (how much the counterpart's


emotional reactions made sense, were believed to be consequential,conveyed important information, and were meaningful; α=.81). Wealso measured how much the counterpart's emotions seemed realisticthrough four items (e.g., howmuch participants had found the counter-part's emotions to be genuine; α=.66) and comical through two items(e.g., how much participants had found the counterpart's emotions tobe comical; α=.96).

Results

Manipulation checksParticipants in the emotional inconsistency condition thought that

the counterpart was less consistent (M=3.98, SD=1.52) than didparticipants in the consistent anger condition (M=4.68, SD=1.38;F(1, 89)=5.35, pb .03, η2=.06).

Participants in the emotional inconsistency condition also thoughtthat the counterpart's reactions were less predictable (M=3.91, SD=1.60) than did participants in the consistent anger condition (M=4.65, SD=1.23; F(1, 89)=6.14, pb .02, η2=.07).

ConcessionsA one-way ANOVA was conducted on the total amount of conces-

sions granted. Participants in the emotional inconsistency conditiongrantedmore concessions (M=+.58, SD=2.34) than in the consistentanger condition (M=− .56, SD=1.72; F(1, 89)=6.95, pb .01, η2=.07).

Lack of controlParticipants in the emotional inconsistency condition felt less

control (M=3.81, SD=1.42) than in the consistent anger condition(M=4.43, SD=1.23; F(1, 88)=4.89, pb .03, η2=.05).

Alternative explanationsEmotional inconsistency did not affect how much the counterpart's

emotional reactions were taken at face value (p>.95), how much thecounterpart's emotions seemed realistic (p>.84), or how much thecounterpart's emotions seemed comical (p>.76). Therefore thesecould not mediate the effect of emotional inconsistency. Also, t-testsagainst the mid-point of the scale indicated that, in both conditions,the counterpart's emotional reactions were taken at face value, seemedrealistic, and did not seem comical.

Mediation by lack of controlNext, we examined whether participants' feeling little control

would mediate the effect of emotional inconsistency on concessionsgranted. Fig. 4 presents the results of the regression analyses. Abootstrapping procedure using 5,000 resamples demonstrated thatthe path through the mediator was significant (95% CI Low=.010;95% CI High=.426). Thus, participants' feeling little control mediated

Note. Numbers represent standardized regression csimultaneous standardized regression coefficients.^ p < .10, * p < .05, ** p < .01, *** p < .001

EmotionalInconsistency

0 = No1 = Yes

Feeling Lack Control.23*

.27** (.19^)


the effect of emotional inconsistency on concessions granted. Theseresults replicate and extend those of Experiments 1–2.

General discussion

Three experiments investigated whether communicating emotionalinconsistency would secure concessions in negotiation. Experiment 1examined actual, face-to-face interactions. Experiments 2–3 usedmore controlled, computer- and scenario-based paradigms. In all threeexperiments, emotional inconsistency induced recipients to makegreater concessions compared to emotional consistency. Further, allthree experiments showed that this effect of emotional inconsistencywas mediated by recipients' feeling less control.

Emotion in negotiation

Most research on emotion in conflict and negotiation (e.g., Van Kleefet al., 2010) has focused on emotions communicated consistentlythroughout the interaction. We extend this work by moving beyondthat focus and examining the effects of emotional inconsistencythroughout the negotiation (i.e., over time) rather than emotional con-sistency. Our results have important implications because inconsistencyin emotions may often be typical (Barrett, 2009; Kuppens et al., 2010;Scherer, 2009), particularly in conflict and negotiation (Morris &Keltner, 2000; Rothman & Wiesenfeld, 2007). By documenting theeffectiveness of emotional inconsistency as an instrument of influence,our results support the theory that inconsistency in emotions has, initself, a signaling function (Frijda, 1986; Scherer, 2009). Emotions “havemeaning because they change” (Kuppens et al., 2010, p. 1042).

Inconsistency and instigating a lack of control in competitive interactions

The current experiments also speak to social theories onunpredictability and compliance. Social scientists (e.g., Schelling,1960) have long theorized that creating unpredictability throughalternating strategies increases compliance in competitive interactions.This assertion has been quite influential, both in the realms of theory(Glimcher, 2003) and practice (U.S. Strategic Command, 1995). Yet,no prior work had empirically demonstrated this long held assertion.The current research is the first to experimentally show that inconsis-tency and concomitant unpredictability positively affect concessions.

Our work also contributes to theorizing on the benefits of inconsis-tency and unpredictability in competitive interactions by documentingamechanismbehind this effect. Specifically, it resonateswith the notionthat feeling little control importantly drives people's reactions in theface of inconsistency and unpredictability (Kay et al., 2010; Menon etal., under review; Sullivan et al., 2010). Indeed, feeling little controlover outcomesmakes peoplemore susceptible to the influence of others(e.g., De Vries et al., 1996; Morris et al., 1995). And, in negotiation, this

oefficients. Numbers in parentheses represent

of.35*** (.31**)

Concessions



translates into greater concession-making to others (Stevens et al.,1993). Yet, we extend this prior research by showing that lack of controlcan be deliberately created through communicating emotions.

Limitations and future research

The current research is the first to demonstrate that emotionalinconsistency and unpredictability can positively affect concessions.Yet, this basic effect could bemoderated by a number of factors. For ex-ample, it is possible that alternating between two – or more – emotionsother than the ones we manipulated would yield different results. Thecurrent experiments focused on anger, happiness, and disappointmentbecause these are common emotions in negotiation (Van Kleef et al.,2010). However, expressing less common emotions might not be as ef-fective or perceived as normal. In addition, the emotional inconsistencyeffect may be moderated by negotiators' power because power canimpact effects of emotions in negotiation (e.g., Overbeck et al., 2010).In particular, in situationswhere the recipient could exit the negotiationbecause s/he has an attractive alternative to the negotiation, emotionalinconsistency might not lead to more concession-making. Indeed, hav-ing an attractive alternative makes recipients generally less susceptibleto a counterpart's emotions (e.g., Van Kleef et al., 2010). Finally, thecurrent experiments focused on single-shot negotiations. But, in subse-quent interactions, recipients might elect to avoid inconsistent oppo-nents. Thus, the effect of emotional inconsistency may be moderatedby important factors, such as the prototypicality of emotions, recipients'power, and time horizon.

Conclusion

Prior research on why people comply in conflict and negotiationtypically emphasized the role of toughness (e.g., De Dreu, 1995; VanKleef & De Dreu, 2010). Thus, it highlighted the benefits of anger, atough emotional strategy (Van Kleef et al., 2010).

The current experiments show that emotional inconsistency andunpredictability are another path to eliciting concessions. Our resultssupport the theory (e.g., Frijda, 1986; Russell, 2009; Scherer, 2009)that there is a signaling value in inconstant emotions. The cur-rent research is also the first to empirically demonstrate theinfluential prediction (Schelling, 1960) that inconsistency andconcomitant unpredictability positively affect concessions incompetitive interactions.

Appendix A. (Experiment 3)

Imagine that you work for a company that designs, manufactures,and sells technical equipment. You have a final negotiation meetingwith one of your clients today.

Your counterpart and you have agreed on the majority of theissues relating to the deal, but there is still a warranty to be negotiated.You donot see any point in including awarranty in the deal because it iswell-known in the industry that your products are highly reliable. Still,your counterpart keeps insisting on it.

Your counterpart wants a warranty and wants the warranty toinclude agreements about: (a) the amount of liability coverage in caseof a breakdown; (b) the time needed for repairing the equipment, incase of a breakdown; (c) and a discount applicable to the purchase ofspare parts.

Your company has provided you information about what kind ofdeal would be acceptable. In general, your company wants (1) thelowest amount of liability coverage, (2) the most amount of timeavailable for fixing repairs, and (3) the lowest discount for spareparts. At the same time, your company set maximum conditions foryou. Your company will accept (1) at most 15 million dollars liability,(2) at least 12 h needed for repairing any damaged equipment, and(3) at most a 60% discount for spare parts. Again, this corresponds

to the maximum conditions you can afford so that the profit on thedeal remains positive for your company, but would not be consideredideal.

Here is how the negotiation proceeds:

Counterpart: “How about a warranty for the material? If there areproblems, wewant you to give us damages, at least 5 million dollarsin damages.”You: “Our reputation for the quality of this technical equipment isalmost unmet in the industry… The technology we developed hasbeen well-tried and is highly safe.”Counterpart: “Good. Then you won't mind formally including thethree warranty items in the contractual agreement.”You: “We have never encountered any failure problem with thisequipment. It is highly reliable. You can check our clients' referencesas well as the quality tests.”Counterpart, in a disappointed way (in an angry way): “But this isimportant for us. We cannot afford to have your equipmentdecreasing or stopping the production in our factory even for aday. We cannot play with the production rates.”You: “We have never failed our clients. We care a lot about therelationships with our clients. Our maintenance engineers areexperts in their field. And our aftersales service is very efficientand reacts instantly. We would repair the equipment within36 hours.”Counterpart, in an angry way (in an angry way): “We can't put theproduction at that much risk. A 20-hours repair time is the most Ican agree to. I also need at least a 50% discount when purchasingthe spare parts. So I want those in addition to the 5 million dollarsin damages I mentioned earlier.”You: “But what you're asking for now was not in the originalagreement.”Counterpart, in a disappointed way (in an angry way): “You talkedabout quality and services, and now you don't want to put them inthe agreement…”

You: “We cannot be held responsible for every failure that wouldoccur. For example, if there's a power cut…”

Counterpart, in an angry way (in an angry way): “What do youmean? Who is responsible then?”

Note. Emotion manipulation appears in the last four statementsmade by the counterpart. In the emotional inconsistency conditionthe prefix alternated between “Counterpart, in a disappointed way”and “Counterpart, in an angry way” and in the consistent anger condi-tion the prefix always was “Counterpart, in an angry way.”

References

Adam, H., Shirako, A., & Maddux, W. (2010). Cultural variance in the interpersonaleffects of anger in negotiation. Psychological Science, 21, 882–889.

Barrett, L. (2009). Variety is the spice of life: A psychological construction approach tounderstanding variability in emotion. Cognition and Emotion, 23, 1284–1306.

Crozier, M., & Friedberg, E. (1980). Actors and systems: The politics of collective action (A.Goldhammer, Trans.). Chicago: The University of Chicago (Original work published1977).

De Dreu, C. K. W. (1995). Coercive power and concession making in bilateral negotiation.Journal of Conflict Resolution, 39, 646–670.

De Dreu, C. K. W., & Van Lange, P. A. M. (1995). The impact of social value orientationson negotiator cognition and behavior. Personality and Social Psychology Bulletin, 21,1178–1188.

De Vries, N. K., De Dreu, C. K. W., Gordijn, E., & Schuurman, M. (1996). Majority and mi-nority influence: A dual-role interpretation. In W. Stroebe, & M. Hewstone (Eds.),European review of social psychology, Vol. 7. (pp. 145–172). Chichester, UK: Wiley.

Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. Texas: Harcourt.Elfenbein, H. A. (2007). Emotion in organizations: A review in stages. In A. Brief, & J.

Walsh (Eds.), Academy of Management Annals, Vol. 1. (pp. 315–386)Amsterdam:Elsevier.

Ellsberg, D. (1959). The political uses of madness. Paper presented at the Department ofGovernment, Harvard University.


Frank, R. (1988). Passions within reason: The strategic role of the emotions. New York:Norton.

Frijda, N. (1986). The emotions. Cambridge, England: Cambridge University Press.Gaddis, J. (2005). Strategies of containment: A critical appraisal of American national

security policy during the Cold War. New York: Oxford University Press.Galinsky, A. D., & Mussweiler, T. (2001). First offers as anchors: The role of perspective-

taking and negotiator focus. Journal of Personality and Social Psychology, 81, 657–669.Glimcher, P. (2003). Decisions, uncertainty, and the brain: The science of neuroeconomics.

MA: MIT Press.Harford, T., & Solomon, L. (1967). “Reformed sinner” and “lapsed saint” strategies in

the prisoner's dilemma game. Journal of Conflict Resolution, 11, 104–109.Hilty, J. A., & Carnevale, P. J. (1993). Black-hat/white-hat strategy in bilateral negotiation.

Organizational Behavior and Human Decision Processes, 55, 444–469.Kay, A. C., Shepherd, S., Blatz, C. W., Chua, S. N., & Galinsky, A. D. (2010). For God (or)

country: The hydraulic relation between government instability and belief inreligious sources of control. Journal of Personality and Social Psychology, 99, 725–739.

Kay, A. C., Whitson, J. A., Gaucher, D., & Galinsky, A. D. (2009). Compensatory control:Achievingorder through themind, our institutions, and theheavens. Current Directionsin Psychological Science, 18, 264–268.

Keltner, D., & Haidt, J. (1999). Social functions of emotions at four levels of analysis.Cognition and Emotion, 13, 505–521.

Kopelman, S., Rosette, A., & Thompson, L. (2006). The three faces of eve: Strategicdisplays of positive negative and neutral emotions in negotiation. OrganizationalBehavior and Human Decision Processes, 99, 81–101.

Koren, L., & Goodman, P. (1992). The haggler's handbook: One hour to negotiating power.WW Norton & Company.

Kuppens, P., Oravecz, Z., & Tuerlinckx, F. (2010). Feelings change: Accounting forindividual differences in the temporal dynamics of affect. Journal of Personalityand Social Psychology, 99, 1042–1060.

Larsen, R. J., Augustine, A., & Prizmic, Z. (2009). A process approach to emotion andpersonality: Using time as a facet of data. Cognition and Emotion, 23, 1407–1426.

Lelieveld, G. -J., VanDijk, E., VanBeest, I., Steinel,W., &VanKleef, G. A. (2011). Disappointedin you, angry about your offer: Distinct negative emotions induce concessions viadifferent mechanisms. Journal of Experimental Social Psychology, 47, 635–641.

Lelieveld, G. J., Van Dijk, E., Van Beest, I., & Van Kleef, G. A. (2012). Why anger and dis-appointment affect other's bargaining behavior differently: The moderating role ofpower and the mediating role of reciprocal and complementary emotions. Person-ality and Social Psychology Bulletin, 38, 1209–1221.

Litt, M. D. (1988). Cognitive mediators of stressful experience: Self-efficacy and per-ceived control. Cognitive Therapy and Research, 12, 241–260.

Loades, D. (2006). Elizabeth I: A life. : Hambledon Press.Menon, T., Sheldon, O., & Galinsky, A. D. (under review). Agents of unluckiness: Feeling

jinxed and contaminated by contact with friendly enemies.Morris,M.W., & Keltner,D. (2000).Howemotionswork: Ananalysis of the social functions

of emotional expression in negotiation. Research in Organizational Behavior, 22, 1–50.Morris, M. W., Sim, D. L. H., & Girotto, V. (1995). Time of decision, ethical obligation,

and causal illusion: Temporal cues and social heuristics in the prisoner's dilemma.In R. M. Kramer, & D. M. Messick (Eds.), Negotiation as a social process. (pp.209–239). Thousand Oaks, CA: Sage.

Overbeck, J. R., Neale, M. A., & Govan, C. L. (2010). I feel, therefore you act: Intrapersonaland interpersonal effects of emotion on negotiation as a function of social power.Organizational Behavior and Human Decision Processes, 112, 126–139.

Pietroni, D., Van Kleef, G. A., De Dreu, C. K., & Pagliaro, S. (2008). Emotions as strategicinformation: Effects of other's emotional expressions on fixed-pie perception,

demands, and integrative behavior in negotiation. Journal of Experimental SocialPsychology, 44, 1444–1454.

Pruitt, D. G. (1981). Negotiation behavior. New York: Academic Press.Rafaeli, A., & Sutton, R. I. (1991). Emotional contrast strategies as a means of social

influence: Lessons from criminal interrogators and bill collectors. Academy ofManagement Journal, 34, 749–775.

Rothman, N. B., & Wiesenfeld, B. M. (2007). The social consequences of expressingemotional ambivalence in groups and teams. In E. Mannix, M. A. Neale, & C.Anderson (Eds.), Research on managing groups and teams: Affect and groups, Vol.10. (pp. 275–308)Amsterdam: Elsevier.

Russell, J. A. (2009). Emotion, core affect, and psychological construction. Cognition andEmotion, 23, 1259–1283.

Schelling, T. C. (1960). The strategy of conflict. Cambridge, MA: Harvard University Press.Scherer, K. R. (2009). The dynamic architecture of emotion: Evidence for the component

process model. Cognition and Emotion, 23, 1307–1351.Sinaceur, M. (2010). Suspending judgment to create value: Suspicion and trust in

negotiation. Journal of Experimental Social Psychology, 46, 543–550.Sinaceur, M., & Tiedens, L. Z. (2006). Get mad and get more than even: When and why

anger expression is effective in negotiation. Journal of Experimental Social Psychology,42, 314–322.

Sinaceur, M., Van Kleef, G. A., Neale, M. A., Adam, H., & Haag, C. (2011). Hot or cold: Iscommunicating anger or communicating threats more effective in negotiation?Journal of Applied Psychology, 96, 1018–1032.

Steinel, W., Van Kleef, G. A., & Harinck, F. (2008). Are you talking to me?! Separatingthe people from the problem when expressing emotions in negotiation. Journalof Experimental Social Psychology, 44, 362–369.

Stevens, C. K., Bavetta, A., & Gist, M. (1993). Gender differences in the acquisition ofsalary negotiation skills: The role of goals, self-efficacy, and perceived control. Journalof Applied Psychology, 78, 723–735.

Sullivan, D., Landau, M. J., & Rothschild, Z. (2010). An existential function ofenemyship: Evidence that people attribute influence to personal and politicalenemies to compensate for threats to control. Journal of Personality and SocialPsychology, 98, 434–449.

U.S. Strategic Command (1995). Essentials of post-cold war deterrence. Available fromhttp://nautilus.org/archives/nukestrat/USA/Advisory/essentials95.txt

Van Dijk, W. W., & Zeelenberg, M. (2002). Investigating the appraisal patterns of regretand disappointment. Motivation and Emotion, 26, 321–331.

Van Kleef, G. A., & Côté, S. (2007). Expressing anger in conflict: When it helps and whenit hurts. Journal of Applied Psychology, 92, 1557–1569.

Van Kleef, G. A., & De Dreu, C. K. W. (2010). Longer-term consequences of angerexpression in negotiation: Retaliation or spillover? Journal of Experimental SocialPsychology, 46, 753–760.

Van Kleef, G. A., De Dreu, C. K. W., & Manstead, A. (2004). The interpersonal effects of angerand happiness in negotiation. Journal of Personality and Social Psychology, 86, 57–76.

Van Kleef, G. A., De Dreu, C. K. W., & Manstead, A. S. R. (2010). An interpersonalapproach to emotion in social decision making: The emotions as social informationmodel. Advances in Experimental Social Psychology, 42, 45–96.

Watzlawick, P., Weakland, J., & Fisch, R. (1974). Change: Principles of problem formulationand problem resolution. New York: Norton.

Waytz, A., Morewedge, C., Epley, N., Monteleone, G., Gao, J. -H., & Cacioppo, J. (2010).Making sense by making sentient: Effectance motivation increases anthropomor-phism. Journal of Personality and Social Psychology, 99, 410–435.

Whitson, J., & Galinsky, A. D. (2008). Lacking control increases illusory pattern perception.Science, 322, 115–117.

http://nautilus.org/archives/nukestrat/USA/Advisory/essentials95.txt

the advantages of being unpredictable: how emotional inconsistency extracts concessions in...

Documents