The shopping brain: Math anxiety modulates brain responses to buying decisions

Page 1

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research

and educational use, including for instruction at the author’s institution and sharing with colleagues.

Other uses, including reproduction and distribution, or selling or

licensing copies, or posting to personal, institutional or third party websites are prohibited.

In most cases authors are permitted to post their version of the

article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies are encouraged to visit:

http://www.elsevier.com/copyright

Page 2

T

Wa

b

c

a

ARAA

KMCNFPD

1

hetmsFKtfbbttrg

r

3

0d

Biological Psychology 89 (2011) 201–213

Contents lists available at SciVerse ScienceDirect

Biological Psychology

journa l homepage: www.e lsev ier .com/ locate /b iopsycho

he shopping brain: Math anxiety modulates brain responses to buying decisions

illiam J. Jonesa,∗, Terry L. Childersb, Yang Jiangc

Department of Marketing, Wayne State University, Detroit, MI, USADepartment of Marketing, Iowa State University, Ames, IA, USAUniversity of Kentucky Chandler Medical Center, University of Kentucky, Lexington, KY, USA

r t i c l e i n f o

rticle history:eceived 15 September 2010ccepted 14 October 2011vailable online 24 October 2011

eywords:ath anxiety

onsumer psychologyeuromarketingN400

a b s t r a c t

Metacognitive theories propose that consumers track fluency feelings when buying, which may havebiological underpinnings. We explored this using event-related potential (ERP) measures as twenty high-math anxiety (High MA) and nineteen low-math anxiety (Low MA) consumers made buying decisionsfor promoted (e.g., 15% discount) and non-promoted products. When evaluating prices, ERP correlates ofhigher perceptual and conceptual fluency were associated with buys, however only for High MA femalesunder no promotions. In contrast, High MA females and Low MA males demonstrated greater FN400amplitude, associated with enhanced conceptual processing, to prices of buys relative to non-buys underpromotions. Concurrent late positive component (LPC) differences under no promotions suggest dis-crepant retrieval processes during price evaluations between consumer groups. When making decisions

Author's Personal Copy

3ecision-making

to buy or not, larger (smaller) P3, sensitive to outcome responses in the brain, was associated with buyingfor High MA females (Low MA females) under promotions, an effect also present for males under no pro-motions. Thus, P3 indexed decisions to buy differently between anxiety groups, but only for promoteditems among females and for no promotions among males. Our findings indicate that perceptual andconceptual processes interact with anxiety and gender to modulate brain responses during consumer

choices.

. Introduction

Buying decisions are malleable (Schwarz, 2004). Rather thanolding well defined preferences, consumers often construct pref-rences spontaneously using whatever information is available tohem (Bettman et al., 1998). Recent interest in neuroeconomic

ethods to study buying behavior underscores attempts to under-tand the processes by which consumers make these decisions.or example, using functional magnetic resonance imaging (fMRI),nutson et al. (2007) showed that excessive prices were linked

o increased insular activity and decreased activity in medial pre-rontal regions. These findings confirm the positive relationshipetween perceived price unfairness and negative affect that haseen proposed elsewhere (Xia et al., 2004), and are largely consis-ent with Bechara and Damasio’s (2005) somatic marking notionhat our brains map anticipated outcomes of purchases from inte-oceptive emotional signals prior to decision making, which then

uide choice.

The metacognitive experiences literature suggests that inte-oceptive signals guide choices as early as perceptual processes

∗ Corresponding author at: 318 Prentis, 5201 Cass Avenue, Detroit, MI 48202-930, USA. Tel.: +1 313 577 4466; fax: +1 313 577 4515.

E-mail address: [email protected] (W.J. Jones).

301-0511/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.biopsycho.2011.10.011

© 2011 Elsevier B.V. All rights reserved.

(Reber et al., 1998). So-called fluency signals, which are hedonicallymarked signals of processing ease (Winkielman et al., 2003), havebeen shown to influence later judgments in which they are non-integral. For example, a difficult-to-pronounce amusement ridename (e.g., Vaiveahtoishi) increases the perception the ride willmake you sick (Song and Schwarz, 2009 – Study 3). Behaviorally,the subparts of processing fluency, perceptual fluency and con-ceptual fluency have repeatedly been shown as separable if ofteninterrelated (e.g., Cabeza and Ohta, 1993; Lee and Labroo, 2004;Whittlesea, 1993). Perceptual fluency, defined as the ease withwhich perceptual processing occurs, has been shown to influencemetacognitive judgments directly. For example, Whittlesea et al.(1990) introduced a noise mask to target words in a repetition prim-ing task and found that when noise level was light, consequentlymaking the perceptual features of target words more readily dis-cernible, participants in an experiment were more likely to falselyjudge such items as having been previously presented.

Later work by Whittlesea (1993) extended the influence of flu-ency on familiarity-based recognition judgments to also includeconceptual fluency. In this study, Whittlesea demonstrated thatmanipulating either the perceptual or conceptual fluency of current

processing such as via repetition priming or semantically relatedprimes can induce feelings of familiarity. These feelings may ormay not be treated as diagnostic to decision making dependingon “the expectations aroused by a current judgment” (p. 1249).

Page 3

2 l Psych

Bcmqjeiac

oWvtirnaas2fibslpt(pcptes

scoteaagwictlatptosn

wflcbemeht

Author's Personal Copy

02 W.J. Jones et al. / Biologica

ecause the feeling of familiarity engendered in perceptually oronceptually facilitated processing or both is separable from actualemory retrieval, it must be meaningfully inferred as a conse-

uence of past experience for it to be treated as diagnostic toudgment (Whittlesea, 1993). Stated simply, the process by whichither, both, or neither forms of fluency are incorporated into famil-arity judgments is, as Whittlesea (1993) says, highly sophisticatednd wholly dependent on considerations of past experience andurrent processing.

In order to dissociate familiarity-based recognition from rec-llection, Whittlesea and colleagues (Whittlesea et al., 1990;hittlesea, 1993) studied items in which a target was never pre-

iously presented. Because in these trials no memory trace for thearget was ever created, this allowed Whittlesea and colleagues tosolate familiarity effects from recollection – i.e., those aspects ofecognition independent of familiarity feelings. Accounts of recog-ition memory in which familiarity and recollection are considereds distinct factors are typically labeled as dual-process accounts,nd have been investigated widely both behaviorally (for a review,ee Yonelinas, 2002) and neurally (for review, see Rugg and Curran,007). Horton et al. (1993), citing Bartlett (1932), remind us thatamiliarity and recollection need not be mutually exclusive, and cannclude additional processes such as judging likeness of a remem-ered object. We argue that consumer contexts (here prices) arecenarios which might be replete with both familiarity and recol-ection aspects. Consumers are accurate at using information aboutreviously stored prices, but they often seem to store this informa-ion as fuzzy memory traces rather than exact price representationsVanhuele and Drèze, 2002). Thus, this pathway assumes that aresented price that feels familiar, owing to perceptual and/oronceptual fluency feelings, may be judged as acceptable becauserices are not typically judged relative to exact prices. At the sameime, aspects of memory associated with recollection might also bengaged during price evaluations because consumers are adept attoring and identifying price ranges (Vanhuele and Drèze, 2002).

A second pathway by which fluency might affect buying deci-ions is via an affect-as-information heuristic. At least partially,onsumer choice is frequently related to affect. That is, consumersften buy what they like and reject what they do not like. Thoughhe present study is not concerned with preference judgmentsxplicitly, Reber et al. (1998) showed that perceptual fluency isffectively positive and selectively enhances positive (but not neg-tive) judgments. In the context of non-preference buying, feelingood might be judged as having previously determined that a priceas acceptable. In this sense, the process is identical to the famil-

arity prediction; familiarity feels good. However, for enhancedonceptual fluency, Lee and Labroo (2004 – Study 4) demonstratedhat it does not necessarily follow that enhanced conceptual fluencyeads to positive judgments in a product evaluation task even if thectual processing fluency aspect itself is experienced positively. Inheir study, Lee and Labroo used a negatively valenced conceptualrime (Lice Shampoo) and demonstrated reduced product evalua-ions of a low familiarity brand of conditioner relative to evaluationf a control product (alkaline batteries). In other words, it is pos-ible that enhanced processing fluency can also be associated withot buying.

Our study was designed to examine what differences, if any,ould be present among individuals that differ in their ability touently consider prices presented to them. We recorded electroen-ephalography (EEG) to study how neural responses, as measuredy event-related potentials (ERPs), to the onset of a price differ-ntiate buys from non-buys. Our low fluency group, those high in

ath anxiety (hereafter High MA), was contrasted with a high flu-

ncy group, those low in math anxiety (Low MA). Past researchas shown that High MA results in reduced processing fluency,hough accuracy is often left intact (Ashcraft and Kirk, 2001).

ology 89 (2011) 201–213

Anxiety-inducing stimuli in general induce disfluency by compet-ing for attention within fronto-cingulate cortices (Bishop et al.,2004). Though High MA has yet to be studied elsewhere using neu-roscience techniques, behavioral research has shown High MA tobe entirely consistent with other forms of anxiety (for a review,see Eysenck et al., 2007). Extreme avoidance of the object, event, orcontext that brings about or intensifies anxiety symptoms is a hall-mark of all anxiety disorders (Maner et al., 2007; Paulus and Stein,2010). Under threat-related processing, anxiety prone individualsalso frequently rely on compensatory strategies such as increasingeffort and deploying cognitive resources disproportionate to thetask at-hand (Eysenck et al., 2007), which may be related to a formof risk avoidance to committing errors (Paulus et al., 2003).

Yet, we note that the purpose of our research is not to explicitlydescribe relationships between emotions and specific ERP compo-nents. Indeed, these relationships may be non-specific to valenceand contingent on simultaneous appraisal processes (Hajcak et al.,2006; Grandjean and Schere, 2008). In our work, we instead drawa distinction between emotions and feelings, which we view asless specific though related to emotion (and sometimes interre-lated components of emotion), that represent ongoing monitoringof internal processes (for similar views, see Pribram, 1970; Scherer,2004). We argue that process-induced fluency feelings, as mea-sured by event-related potentials, might be utilized as sources ofinformation in the non-preference related buying task describedin this paper. Notwithstanding, we emphasize emotional effectsthrough our simultaneous focus on trait math anxiety. In theconsumer behavior literature, Raghunathan et al. (2006) havedemonstrated that anxiety leads to a risk-reduction buying men-tality. Brand et al. (2006) argue that optimal decision making underrisk is characterized by “both cognitive strategies as well as bias-ing emotional signals” (p. 1273). Thus, despite our non-predictionsrelated to emotions and specific ERP components, we would predictthat High MA renders consumers more susceptible to incorporat-ing feeling-related information into their choice making. In thissense, we expect that High MA consumers will demonstrate morebuy versus non-buy differences as measured by ERPs than Low MAconsumers.

Past ERP studies of economic transactions have shown effects onreward processing magnitude and valence (e.g., Yeung and Sanfey,2004). To the best of our knowledge no study has directly inves-tigated ERP correlates of buys versus non-buys. However, severalERP studies which have shown fluency effects guided our thinking.Curran and colleagues have amassed an impressive body of ERPstudies to support dual-process accounts of recognition memory(for a review, see Curran et al., 2006). These studies have centeredon the frontal N400 (FN400) as an index of familiarity. For example,Curran (2000) analyzed ERPs to previously presented words, sim-ilar words with different pluralities, and new words. Results for aFN400 between 300 and 500 ms varied as a function of familiaritysuch that FN400 was more negative for new words than for simi-lar and old words. A late positive component or LPC (400–800 ms),was associated with both recollection and recognition, and con-versely showed more positive amplitudes for studied words thanfor old or similar words. Work by Voss and Paller and their col-leagues (e.g., Voss and Federmeier, 2011; Voss and Paller, 2007;Voss et al., 2010) calls into question this interpretation of FN400as a singular index of familiarity. Instead, they contend that FN400might only indicate conceptual processing (Voss and Federmeier,2011; Voss et al., 2010), and therefore may be functionally indis-tinct from N400 effects (Voss and Federmeier, 2011). In this sense,FN400 effects found in prior studies (more positive for familiar

objects) really indexed reduced conceptual processing. Thus, sub-jects in these studies possibly used reduced conceptual processingsignals in making familiarity judgments. For N400 itself, there isprecedent for its meaningfulness within ERP studies on arithmetic.

Page 4

l Psych

FNcaNea

tweffpjaisbaEortnbri

ibdsc(udu

essnaoS1rprSptMet2ataew

Author's Personal Copy

W.J. Jones et al. / Biologica

or example, Niedeggen and Rösler (1999) demonstrated increased400 amplitude for correct versus incorrect solutions in a multipli-ation verification task. On the contrary, correct solutions lead topositive component at about 450 ms. Similar to Curran (2000),400 was attenuated when incorrect solutions were related torrors of a small or medium distance from the true answer withinmultiplication table (e.g., 5 × 8 = 32, the product of 4 × 8).

Perceptual priming studies have generally shown that repe-ition enhances early frontocental positive potentials in a timeindow close to the P200 (Schendan and Kutas, 2003; Woollams

t al., 2008). Recently, Paynter et al. (2009) demonstrated that arontal P200 and a frontocentral P300 index a “feeling of knowing”or arithmetic problems prior to retrieval, which they likened toerceptual fluency. Whereas familiarity attenuates FN400, the sub-

ective feeling of knowing amplifies P200. Conceptually, this raisesn interesting question. Because fluency refers to processing ease,t follows that less (not more) activation of conceptual knowledgetructures implicated in enhanced FN400 and N400 effects shoulde present (Voss and Federmeier, 2011). Kotchoubey (2006) hasrgued that functional differences underlie positive and negativeRP components such that negative ERPs may reflect mobilizationf cortical resources and positive ERPs reflect consumption of theseesources.1 Because fluency should be positively marked, it followshat disfluency should be less so, which may indicate that enhancedegative ERPs such as N400 are effortful (Holcomb, 1993) and maye associated with a subjectively more negative experience. Indeed,esearch by Garbarino and Edell (1997) indicates that when efforts required the subjective experience of negative affect occurs.

Unlike previous research on the neural consequences of buy-ng, we controlled for numeracy ability. Theoretically, High MA cane concurrent with and reciprocally interconnected to numeracyeficits (Hembree, 1990). Highly numerate individuals have beenhown as less susceptible to numerical biases and to have betteralibrated affect expectancies when making numerical judgmentsPeters et al., 2006). To test boundary conditions of buying effectsnder demanding cognitive conditions, we further included a con-ition with percentage price promotions (e.g., 15% off), a widelysed marketing tactic.

Also unlike previous research, we explicitly take into accountffects of gender. Anecdotal evidence in the popular press fromelf-styled “neuromarketers” indicates that neuromarketing con-ultancy firms have obtained data demonstrating that women areeurally more sensitive and aware of prices than men (Karlinskynd Patrick, 2011). If true, these findings echo other findingsbserved from research on gender selectivity theory (Darley andmith, 1995; Meyers-Levy, 1988; Meyers-Levy and Maheswaran,991). According to this theory, males are selective processors thately on a limited set of details whereas females are comprehensiverocessors that consider a broader variety of information cues, botheadily apparent and subtle, when making decisions (Darley andmith, 1995). The increased sensitivity of women to prices mightresent itself as a systematic tendency to encode more prices ando elaborate on prices with greater frequency (cf. Meyers-Levy and

aheswaran, 1991). Gender differences for math anxiety are wellstablished with women tending to exhibit math anxiety more sohan men (Hembree, 1990; Hopko et al., 2003; Miller and Bischel,004). However, these differences seem to have less to do withbility and more to do with social factors such as gender stereo-ypes (Eccles and Jacobs, 1986). The relationship between math

nxiety and math ability generally is unclear, though math anxi-ty concurrently seems to reduce willingness to take math courses,hich affects acquisition of math skills (Ashcraft, 2002). Of specific

1 See p. 56 of Kotchoubey (2006) for a discussion as it relates to N400.

ology 89 (2011) 201–213 203

interest to the present study, Miller and Bischel (2004) found thatmath anxiety impacts females’, but not males’, abilities to completeapplied math problems. Perhaps these effects result from males’more selective use of decision inputs, which may render them moreimmune to math anxiety effects.

In brief, our study complements fMRI work dissociating buysfrom non-buys (e.g., Knutson et al., 2007), though extends that workto a different method, examines boundary conditions related tomath anxiety, promotion formats, and gender, all while controllingfor numeracy, which is a plausible alternative explanation to mathanxiety.

2. Methods

2.1. Participants

We utilized data from thirty nine introductory business students (17 female;mean age = 21 years, SD = .84) who volunteered for extra credit and were paid $10plus a chance at up to $20 extra based on relative accurate buys from a set asidefund (range = $2–20; mean = $7.59). Due to excessive artifact, we excluded EEG andbehavioral data from one Low MA male participant. Prior to recruitment participantswere unknowingly prescreened for math anxiety from sections of undergraduatebusiness courses using the Abbreviated Math Anxiety Scale (AMAS; Hopko et al.,2003). We recruited for volunteers with high and low scores on AMAS out of asample of 709 administered students. For High MA consumers, we recruited onlythose participants deemed to have clinically significant or pathological math anx-iety scores as recommended by the scale developer, which is at or around ∼32 or1.5 SD or greater than the average sample mean (D. Hopko, personal communica-tion, October 25, 2010; cf. Hopko et al., 2003). Sample characteristics were: HighMA AMAS M = 35.29, SD = 5.60; Low MA AMAS M = 12.33, SD = 3.00. Seven Low MAfemales, 11 Low MA males, and 10 each High MA females and males were held out forfurther analysis. All participants were right-handed, English speakers with normalor corrected to normal vision, and none had neurological or psychiatric disorders.Participants gave informed consent before taking part in the study.

2.2. Stimuli and procedure

We developed a stimulus list of products and their images familiar to the popula-tion studied. A total of 50 products were used twice. Critical stimuli included productprices and a “Buy?” stimulus. To include a range of overpriced and underpriced items,we developed prices by first identifying price category midpoints from a local, pop-ular retailer. Within conditions, we applied 10–34% markups/discounts to half ofthe midpoint prices each without replacement to arrive at final price stimuli. Thisrange reflects research on price differences that consumers are adept at identifying(Hoch et al., 1994). For the price promotions, we further added promotional stimuli(7–33%), which we used twice each with random assignment. We excluded 10% asit is processed with distinct ease. While we displayed final prices under no promo-tions (e.g., $12.14), we displayed the equivalent of final prices under promotions forthe promotion condition (e.g., $13.13, 20% off). Displayed prices ranged from $.90to $198.39.

One week before the experimental session, participants completed a numeracyscale (Lipkus et al., 2001) and retook AMAS (retest r = .91, p < .001) by web survey.Retesting of AMAS was done to substantiate the initial scores, which were collectedwithin a larger prescreening study given to a college wide subject pool. Conversa-tions with participants during debriefing suggested to us that participants eitherdid little to connect the online survey to the laboratory portion of the study and/orforgot about the survey at the time of the lab session. In addition, participants werebriefly shown the product stimuli and asked to provide absolute reference pricesand assessments of their confidence in these prices via a Likert scale anchored at 1– “Not very Confident” to 5 – “Very Confident”. Due to a scheduling issue, this datawas unavailable from one High MA female.

The experimental session was run in blocks of all no promotions or all pro-motions. Block order was counterbalanced across participants taking into accountgroup and gender. For both conditions, participants completed four practice trialsto acclimate to the task (Fig. 1). Participants were instructed to evaluate the pricespresented to them and then to make choices. Because participants would not haveenough time to ascertain exact answers we asked them to estimate after-discountprices within the promotion condition. Past research has shown that consumersdo this spontaneously (Morwitz et al., 1998). We instructed participants that theyshould only buy if they felt they could not get a better deal elsewhere locally rel-ative to the screen offer. Each trial began with the name of the product presentedfor 1000 ms and then an image of the product for 2000 ms followed by the price ofthe offering presented for 4000 ms (hereafter Price). A stimulus featuring the word

Buy with a question mark (hereafter Buy) followed Price and was presented for4000 ms. For the Price and Buy stimuli presentation, we used fixed timing intervals.Fig. 1 provides a pictorial overview of the task. Past research indicates a differ-ential response to collection of reaction time latencies and/or time pressure as afunction of math anxiety (Faust et al., 1996; Tsui and Mazzocco, 2007). Though

Page 5

204 W.J. Jones et al. / Biological Psychology 89 (2011) 201–213

Fig. 1. Illustration of the experimental task. Shown is a sequence for the promotion condition. Substitute no promotion as shown in figure. Each trial began with the nameof a product followed by its image. Stimuli of interest came thereafter including an offer price (labeled as Price), then a stimulus requesting that the participant decide onwhether to buy or not (labeled Buy), and finally a response stimulus. Answer choices for the response stimulus varied pseudo-randomly whose order was unknown to thep , or 30a

pvasubPus

2

rprlfi

tre(eff4(aFC

3

3

m

Author's Personal Copy

articipant. Pseudo-randomly determined stimulus onset asynchronies of 100, 200re denoted with asterisks.

recluding the collection of reaction time latencies, our use of fixed time inter-als strengthens the conclusions we can draw by ruling out time pressure as anlternative explanation for our results. Following these two critical stimuli, we pre-ented a pseudo-randomly varying yes/no response stimulus for 500 ms, which wassed to dissociate the expected motor response from potentially contaminating theuying decision. A 1000 ms fixation of a centered cross was used between trials.articipants were asked to blink and move their eyes only when the fixation, prod-ct name, or product image appeared on the screen. Pseudo-randomly determinedtimulus onset asynchronies of 100, 200, or 300 ms separated stimuli of interest.

.3. EEG recording and ERP analysis

EEG was recorded from 32 Ag/AgCl electrodes (<5 k�) using a linked-mastoideference including two electro-oculogram electrodes and amplified with a band-ass of .1–100 Hz and sampled at 1000 Hz/channel. Off-line analysis followedemoval of eye movement and blink artifacts (>± 75 �V). Each averaging epochasted 4.2 s, including a 200 ms baseline before the two critical stimuli, was offlineltered at .1–30 Hz, and baseline-corrected.

EEG data were averaged separately for trials in which participants chose yeso buy or not to buy. We focused our analysis on the Price and Buy stimuli. As aule, we required a minimum of 10 trials per epoch. Post-artifact rejection, actualpochs contained an average of 21.71 trials (SD = 6.99) under Price and 21.55 trialsSD = 6.61) under Buy. Between-group epoch sizes were statistically identical. Badlectrodes were treated as missing data, which only reduced the number of channelsor interpolation of topographic maps. Statistically significant effects are reportedor Price at P200 (200–300), FN400 (300–500), and a late positive complex (LPC;00–700). For the Buy decision, we reported results for P200 (200–300) and P300300–500). We quantified P200 and FN400 using five anterior electrodes centeredt FCz (Fz, FC3, FCz, FC4, and Cz). P3 was quantified using FCz, C3, Cz, C4, and CPz.inally, LPC was quantified using five posterior electrodes centered at CPz (Cz, CP3,P4, CPz, and Pz).

. Results

.1. Behavioral results

Behavioral data was analyzed via constructing linear mixedodels with numeracy as a covariate (used in all analyses

0 separated stimuli of interest. A cross fixation separated trials. Significant effects

heretofore). Follow-up tests incorporated a Bonferroni adjust-ment. We analyzed reference price confidence via an analysis withMath Anxiety (MA: High vs. Low) and Gender (male vs. female) asbetween-subjects factors. A significant main effect of gender (F(1,1604) = 24.51, p < 001) and a MA group × Gender interaction effectwas noted (F(1, 1604) = 11.26, p < 001). The main effect of genderresulted because males had greater reference price confidence(M = 3.81, SEM = .03) than females (M = 3.57, SEM = .03). Follow-upanalyses to examine the interaction found that Low MA females(M = 3.48, SEM = .06) demonstrated lower levels of reference priceconfidence than High MA males (M = 3.73, SEM = .05) and Low MAmales (M = 3.89, SEM = .04), but not High MA females (M = 3.48,SEM = .06, p > .10). Low MA males demonstrated greater referenceprice confidence than High MA females (p < .001) and High MAmales at a marginally significant level (p < .10). High MA femalesand High MA males did not differ with respect to reference priceconfidence (p > .10). No effects were noted for an analysis of theaccuracy of consumers’ reference prices defined as a percentagedeviation of their stated prices relative to the actual price.

We also analyzed buys (yes vs. no) via a 2 (MA: High vs. Low) × 2(Condition: no promotion vs. promotion) × 2 (Gender: male vs.female) × 2 (median split of reference price confidence: high vs.low) linear mixed-model analysis. Significant effects were foundfor condition (F(1, 3670) = 17.58, p<.001), gender (F(1, 3670) = 5.13,p = .024), and an MA × reference price confidence interaction (F(1,3676) = 6.13, p = .013). The condition effect found that no promo-tion trials (M = .62, SEM = .01) lead to more buys than promotiontrials (M = .54, SEM = .01). For the gender effect, males (M = .60,SEM = .01) were more likely overall to buy than females (M = .56,

SEM = .01). Finally, follow-up analyses of the interaction effectshowed that High MA consumers were less likely to buy under lowconfidence (M = .55, SEM = .02) than Low MA consumers (M = .60,SEM = .01; F(1, 1242) = 4.07, p = .04). For High MA consumers, a

Page 6

l Psych

md(

3

rpcca2toIm

BLtwttweBsoosrBti

epseplcm1siwll(thfSfrmptH

3

(B

Author's Personal Copy

W.J. Jones et al. / Biologica

arginally significant trend was observed such that high confi-ence (M = .60, SEM = .01) lead to more buys than low confidenceM = .55, SEM = .02; F(1, 1835) = 3.78, p = .052).

.2. ERP analysis

Mean amplitudes and component latencies were analyzed usingepeated measures analyses of variance (ANOVAs). When appro-riate, the Greenhouse–Geisser correction was applied. Data wasollected using amplifiers by Neuroscan (Neuroscan, Inc.) and pro-essed using Neuroscan software. Given advantages of fractionalpproaches to ERP latencies over peak measures (e.g., Woodman,010), component latencies were calculated using the full dura-ion at half maximum amplitude (FDHM), which we implementedn our averaged files imported into EMSE software (Source Signalmaging, Inc.). This value reflects the difference between the half

aximum amplitude values from the component tails.Our within-subjects factor (Buys) was buys versus non-buys.

etween-subjects factors were math anxiety group (MA: High vs.ow), promotion condition (Condition: no promotion vs. promo-ion), and gender (Gender: male vs. female). As noted previously,e predicted a priori that High MA females will be most likely

o monitor changes in their physiological state and to incorporatehese changes into their buying decisions. Also noted previously,e included numeracy as a covariate in addition to subject-level

ffects. Thus, we expect significant four-way interaction effects ofuys × MA × Condition × Gender throughout. Conceptually, this isimilar to predicting a MA × Condition × Gender interaction effectf a difference wave analysis of buys versus non-buys. However,ur analyses utilize the original buy and non-buy dependent mea-ures due to known reliability issues with difference scores (for aeview, see Peter et al., 1993). Thus, we explore the presence ofuys × MA × Condition × Gender effects via planned comparisonsesting of the MA × Condition × Gender combinations across buy-ng levels.

Given the complex design of this study, and for clarity of hypoth-sis testing, we omit discussion of effects of the omnibus tests ofractical insignificance (Kirk, 1996). Grasso and Simons (2011) con-ider partial eta square values of .05 a small effect, .1 a mediumffect, and greater than .2 a large effect. Cohen (1988) describedartial eta square values of .01, .06, and .14 as small, medium, and

arge effect sizes, respectively. However, others have emphasizedaution with interpreting partial-eta square values because theeasure is directly impacted by sample size (Murray and Dosser,

987). Our approach was to report all significant results wherebyignificant planned comparisons were demonstrated for effects ofnterest. We also report all meaningfully significant effects, which

e define as those demonstrating partial-eta square values of ateast .03. In addition, we report Cohen’s d for follow-up tests calcu-ated based on means and standard deviations. As others have notedKirk, 1996), Cohen’s d provides a true effect size measure relativeo variance accounted for measures such as partial-eta square. Postoc and planned comparisons applied the Bonferroni adjustment

or the number of tests. Reported p-values reflect this adjustment.ignificant interactions with the buy versus non-buy dependentactor in the presence of the four-way interaction are explicated byeference to the planned comparisons. In order to provide infor-ation about the magnitude of buy versus non-buy differences,

lanned comparisons are reported in the form of paired t-tests with-values and significant differences derived from the full model (cf.owell, 2009).

.2.1. P200 – priceSignificant effects were noted for Buys × MA × Condition

F(1, 370) = 11.97, p < .001, �p2 = .03) as well as for the

uys × MA × Condition × Gender interaction effect (F(1,

ology 89 (2011) 201–213 205

370) = 41.93, p < .001, �p2 = .10). Significant buy versus non-

buy effects were shown for Low MA females under no promotionsas well as for High MA females under no promotions and pro-motions (see Fig. 2). For Low MA females, non-buys (7.39 �V)lead to larger P200 amplitudes than buys (5.34 �V; t(34) = 2.97,p = .04, d = .50) under no promotions. In contrast, High MA femalesdemonstrated larger P200 amplitudes to buys (7.38 �V) relativeto non-buys (5.82 �V) under no promotions (t(54) = −2.92, p = 04,d = −.30), but larger amplitudes for non-buys (7.72 �V) comparedto buys (6.12 �V) under promotions (t(54) = 4.11, p < .001, d = .44).

Past research indicates that P200 is susceptible to effects ofdegree of familiarity (Caharel et al., 2002). In other words, in first-half trials it is possible that P200 effects are due to decreasedfamiliarity with the perceptual features of the stimuli rather thanby a rapid assessment of price evaluation feasibility. Given thispossibility, we conducted additional exploratory analyses of theP200 by constructing epochs for trials within the second halfof price trials only. Results of this analysis again found a sig-nificant Buys × MA × Condition × Gender interaction effect (F(1,370) = 14.70, p < .001, �p

2 = .04). Follow-up analyses confirmedeffects for High MA and Low MA females under no promotions(High MA: Mnonbuy = 6.12 �V vs. Mbuy = 8.01 �V, t(49) = −3.69 �V,p = .005, d = −.39, Low MA: Mnonbuy = 7.13 �V vs. Mbuy = 4.47 �V;t(34) = 4.57, p < .001, d = .60). Interestingly, we also observed a sig-nificant effect for Low MA males under promotions (t(34) = 4.57,p < .001, d = .45, Mnonbuy = 6.54 �V vs. Mbuy = 4.07 �V). Under pro-motions Low MA males behaved similarly to Low MA females underno promotions, at least for second-half trials, which we accept onlytentatively. This result for Low MA males suggests that as Low MAmales become more familiar with the price evaluation task, per-ceptual processes become increasingly diagnostic for them withrespect to decision making. The absence of such effects during thistime window among High MA females for promotions (t(49) = 1.89,p > .10) likely indicates that P200 effects owe to stimulus decodingprocesses. In other words, the P200 effect might reflect the initialshock value of the promoted prices for High MA females, but noth-ing more. We did not observe P200 latency effects for Price eitherfor the main analysis or the exploratory analysis.

3.2.2. FN400 – priceAs expected, a significant Buys × MA × Condition × Gender

interaction effect emerged (F(1, 370) = 38.65, p < .001, �p2 = .10).

The interactive effect of Buys × Condition was also meaningfullysignificant (F(1, 370) = 12.03, p < .001, �p

2 = .03). Planned compar-isons demonstrated significant buy versus non-buy differencesfor Low MA males under promotions and for High MA femalesunder both promotions and no promotions (see Fig. 3). For LowMA males, buys (M = 1.33 �V) were less positive than non-buys(M = 2.75 �V, t(54) = 3.14, p = .02, d = .35) for promotions. Likewise,among High MA females, buys (M = 2.01 �V) were less positivethan non-buys (M = 3.04 �V; t(54) = 2.91, p = .04, d = .34) also forpromotions. In contrast, for High MA females buys were more posi-tive (M = 3.78 �V) than non-buys (M = .95 �V; t(54) = −7.17, p < .001,d = −.68) under no promotions.

We observed several FN400 latency effects. Effectswere noted for MA (F(1, 370) = 15.26, p < .001, �p

2 = .04),MA × Condition (F(1, 370) = 9.78, p < .002, �p

2 = .03),Buys × MA × Gender (F(1, 370) = 9.78, p < .001, �p

2 = .03), andBuys × MA × Condition × Gender (F(1, 370) = 6.06, p = .014,�p

2 = .02) factors. Overall, High MA lead to delayed latencies(M = 402 ms) compared to Low MA (M = 386 ms). Among the High

MA, no promotions were delayed (M = 411 ms) relative to pro-motions (M = 392 ms, p = .005), which was significantly differentthan Low MA under no promotions (M = 380 ms, p < .001) andpromotions (M = 389 ms, p = .001). Planned comparisons testing for

Page 7

206 W.J. Jones et al. / Biological Psychology 89 (2011) 201–213

Fig. 2. Event-related potentials (ERPs) at FCz are shown in the top two rows for females and males, respectively, for the Price stimulus. The time window corresponding tot outcomw icate tt rpretav

tapt

3

wpfLnInddn

btitMpL

Author's Personal Copy

he P200 interval is shown shaded. Buy outcomes are shown in red and non-buyaves for the shaded interval (Buy minus Non-buy). Columns from left to right ind

hen for Low MA consumers. Significant effects are denoted with asterisks. (For inteersion of the article.)

he four-way interaction revealed significant buying differencesmong High MA males under no promotions only (t(49) = −4.72,< .001, d = −.69). Non-buys elicited faster latencies (M = 394 ms)

han buys (M = 431 ms).

.2.3. LPC – priceA significant Buys × MA × Condition × Gender interaction effect

as observed (F(1, 370) = 12.97, p < .001, �p2 = .03). Planned com-

arisons revealed significant differences for Low MA males, Low MAemales, and High MA females all under no promotions (see Fig. 4).ow MA males exhibited larger buy amplitudes (M = 2.61 �V) thanon-buy amplitudes (M = 1.32 �V; t(54) = −4.08, p < 001, d = −.36).

n contrast, Low MA females demonstrated larger amplitudes toon-buys (M = 5.82 �V) than buys (M = 4.38, t(34) = 3.06, p = 03,= .56). Interestingly, and like Low MA males, High MA femalesemonstrated larger buy amplitudes (M = 5.06 �V) compared toon-buy amplitudes (M = 2.67 �V; t(54) = −4.59, p < .001, d = −.38).

No significant latency differences were observed with respect touying. However, we observed a MA × Condition × Gender interac-ion effect (F(1, 370) = 14.62, p < .001, �p

2 = .04). We explicated thisnteraction so as to gain insights into the individual difference fac-

ors affecting consumers’ price evaluation processes. Among Low

A males, no promotion decisions (M = 554 ms) peaked later thanromotion decisions (M = 444 ms, p < .001, d = .84). Interestingly,PC peaked later among Low MA males under no promotions than

es are shown in blue. Bottom two rows show topographical maps of differenceshe promotion condition then no promotion condition for High MA consumers andtion of the references to color in this figure legend, the reader is referred to the web

among High MA males under no promotions (M = 500 ms, p < .001,d = .71) and promotions (M = 506 ms, p < .001, d = .62). Finally, LPClatency was delayed for Low MA females (531 ms) relative to LowMA males (494 ms, p < .001, d = .58) under promotions. Thus, theonly significant within group difference for promotion type wasobserved among Low MA males.

3.2.4. P200 – buyFor the Buy stimulus, significant effects were obtained for

a Buys × MA × Gender interaction (F(1, 370) = 12.64, p < .001,�p

2 = .03) as well as a significant Buys × MA × Condition × Genderinteraction (F(1, 370) = 6.57, p = .01, �p

2 = .02). Planned compar-isons relegated this effect to Low MA males under promotions(t(54) = −3.40, p = .008, d = −.43). As per the price stimuli, we againre-epoched the P200 in an exploratory analysis designed to testthe veracity of the P200 finding relative to the second half of trials.However, we were unable to support this finding as a result of thisanalysis. We did not observe P200 latency effects for Buy withinthe main analysis or exploratory analysis.

3.2.5. P3 – buy

As expected, a statistically significant Buys × MA ×

Condition × Gender interaction was observed (F(1, 370) = 7.76,p = .006, �p

2 = .02). We also observed meaningfully significanteffects for Condition (F(1, 370) = 54.20, p < .001, �p

2 = .13), Gender

Page 8

W.J. Jones et al. / Biological Psychology 89 (2011) 201–213 207

Fig. 3. Event-related potentials (ERPs) at FCz are shown in the top two rows for females and males, respectively, for the Price stimulus. The time window corresponding tothe FN400 interval is shown shaded. Buy outcomes are shown in red and non-buy outcomes are shown in blue. Bottom two rows show topographical maps of differencesw icate tt asterii

(3(bpdtHadLeiembodnofntph

Author's Personal Copy

aves for the shaded interval (Buy minus Non-buy). Columns from left to right indhen for Low MA consumers. Significant effects are denoted with asterisks. Doublen this figure legend, the reader is referred to the web version of the article.)

F(1, 370) = 23.42, p < .001, �p2 = .06), a Buys × MA interaction (F(1,

70) = 41.03, p < .001, �p2 = .10), and a MA × Gender interaction

F(1, 370) = 14.92, p < .001, �p2 = .04). Condition effects were driven

y larger overall no promotion amplitudes (3.19 �V) compared toromotion amplitudes (.90 �V) (see Fig. 5). Gender effects wereriven by larger overall amplitudes among males (2.84 �V) relativeo females (1.25 �V). The MA × Gender interaction resulted fromigh MA females who demonstrated significantly smaller P300mplitudes (M = .42 �V) than High MA males (M = 3.23, p < .001,= −.65), Low MA females (M = 2.08 �V, p = .003, d = −.38), andow MA males (M = 2.46, p < .001, d = −.45), respectively. Wexplicated the Buys × MA interaction by reference to the 4-waynteraction. Among males, significant follow-up tests for buyingffects were confined to no promotion conditions. For Low MAales, non-buys (M = 3.50 �V) elicited larger amplitudes than

uys (M = 2.53 �V; t(54) = 3.88, p = .002, d = 38). We observedpposite-pattern effects for High MA males (t(49) = −2.93, p = .04,= −.44, Mnonbuy = 3.49 �V vs. Mbuy = 4.93 �V). With females, sig-ificant follow-up tests were confined to promotion conditionsnly. Interestingly, we observe the same pattern of amplitude dif-erences within females under promotions as among males under

o promotions. That is, Low MA females elicited larger P300 ampli-udes to non-buys (M = 1.59 �V) than buys (M = .04 �V; t(34) = 2.38,= .02, d = .48) under promotions. High MA females, on the otherand, demonstrated larger buy amplitudes (M = .12 �V) relative

he promotion condition then no promotion condition for High MA consumers andsks denote a significant latency effect. (For interpretation of the references to color

to non-buys (M = −1.91 �V; t(54) = −3.62, p = .008, d = −.61) underpromotions.

A gender effect was observed for latency (F(1, 370) = 17.55,p < .001, �p

2 = .05). Follow-up analyses found faster latencies amongfemales (M = 387 ms) than among males (M = 398 ms).

4. Discussion

Our emerging understanding of the neural network underly-ing buying behavior has largely emphasized activations in brainregions using fMRI, which typically does not capture neuralresponses within 1 second. We used electrophysiological mark-ers of consumers’ price evaluations and buying processes to show,for the first time, differential neural activity between buys andnon-buys within 200–300 ms. Behavioral results indicated thatfemales, particularly Low MA females, felt less confident in theirreference prices than males. Females also bought at lower ratesthan males. Despite this, math anxiety status and gender did notimpact reference price accuracy. In a similar vein, past researchhas found that men are often overconfident in the area of financialdecision-making, and therefore our study extends previous find-

ings to the area of consumer decision-making (Barber and Odean,2001). Notwithstanding, High MA consumers were less likely tobuy when their reference price confidence was low as comparedto Low MA consumers, and less likely to buy under low confidence

Page 9

208 W.J. Jones et al. / Biological Psychology 89 (2011) 201–213

Fig. 4. Event-related potentials (ERPs) at CPz are shown in the top two rows for females and males, respectively, for the Price stimulus. The time window corresponding tothe LPC interval is shown shaded. Buy outcomes are shown in red and non-buy outcomes are shown in blue. Bottom two rows show topographical maps of differences wavesfor the shaded interval (Buy minus Non-buy). Columns from left to right indicate the promotion condition then no promotion condition for High MA consumers and then forL n of tho

tb

ritiepalmccdfpd

4

tc

Author's Personal Copy

ow MA consumers. Significant effects are denoted with asterisks. (For interpretatiof the article.)

han high confidence. No promotion formats positively impacteduying likelihood for all groups.

Our ERP results demonstrated differential sensitivity to neu-al processes during buying for a group that should process pricenformation relatively less fluently, High MA consumers, comparedo a more fluent group, Low MA consumers. Fluency effects alsonteracted with promotion condition and gender. We interpret theffects obtained during Price as possibly related to differences inrocessing fluency (i.e., perceptual fluency and conceptual fluency)nd recollection which may vary due to math anxiety, promotionevel, and gender factors. Similar to others, we interpret P200 as a

arker of perceptual processing, FN400 as a neural signature foronceptual processing, and LPC as indicative of recollective pro-essing (e.g., Rugg et al., 1998). Component differences emerginguring Buy likely characterize the actual buying process itself. Dif-erences during Price, on the other hand, reflect the dynamics ofrocessing the numerical information that feed into the buyingecision. We address each of these separately below.

.1. Price effects

Differences in P200 for Price lend credence to the notionhat processing fluency is utilized as a source of information byonsumers largely automatically (Schwarz, 2004). P200 was larger

e references to color in this figure legend, the reader is referred to the web version

for buys under no promotions for High MA females, but larger fornon-buys among Low MA females also under no promotions (andfor Low MA males under promotions in our exploratory analysis).Trying to understand these effects, we look to Paynter et al. (2009)who found a similar frontocentral P200 to arithmetic problems thatbecame more positive as a function of problem familiarity. Payn-ter et al. interpreted their P200 findings as indicative of enhancedperceptual fluency, which their subjects presumably used to inferwhether they felt like they knew the answer to the arithmetic prob-lem. Accordingly, when subjects felt like they knew the answer tothe problem, they initiated a retrieval attempt. Perhaps, a similarprocess is operating within our High MA female group for no pro-motions, which we discuss further below. Opposite effects amongLow MA females are intriguing and challenge this interpretation.Voss et al. (2010) reported a frontopolar P170 that was more posi-tive to novel versus repeated visual stimuli, though they also notethe atypicality of this finding (cf. Schendan and Kutas, 2003). Unlikethat study, our experimental paradigm did not employ a percep-tual priming manipulation. So, while it is still possible that LowMA females utilized perceptual fluency signals in assessing non-

promoted prices, an alternative and more plausible explanation isthat for Low MA females P200 indexes a form of selective attention.Carretié et al. (2001) demonstrated larger P200 magnitude whenadditional attentional resources were deployed to negative versus

Page 10

W.J. Jones et al. / Biological Psychology 89 (2011) 201–213 209

Fig. 5. Event-related potentials (ERPs) at Cz are shown in the top two rows for females and males, respectively, for the Buy stimulus. The time window corresponding to theP s are st promoL n of tho

poapt

PombmgfMafcbmecuptmfl

Author's Personal Copy

3 interval is shown shaded. Buy outcomes are shown in red and non-buy outcomehe shaded interval (Buy minus Non-buy). Columns from left to right indicate theow MA consumers. Significant effects are denoted with asterisks. (For interpretatiof the article.)

ositive stimuli. In the same sense, the perceived negative valencef the price stimuli (i.e., “bad” price versus “good” price) couldccount for the P200 difference among Low MA females. Nonap-earance of similar P200 effects for other consumers might indicatehat buying decisions for them are made further downstream.

Arithmetic fact-retrieval models (e.g., Siegler, 1988) supportaynter et al.’s P200 interpretation, which we likewise attribute tour finding among High MA females under no promotions. Theseodels proffer the idea of confidence criteria that must be met

efore an individual will attempt to retrieve an answer to an arith-etic problem or resort to a backup strategy such as informed

uessing, writing out a problem, etcetera. Theoretically, High MAemales could set higher retrieval confidence criteria than their Low

A female counterparts when evaluating non-promoted prices ascompensatory mechanism given felt comfortableness with this

ormat (LeFevre et al., 1993). As noted already, this process mayritically depend on a perceptual fluency assessment. This effect isorne out in the behavioral data such that High MA consumers areore likely to buy when confident in their reference prices. Further

nhancing the plausibility of this interpretation, High MA femalesompared to Low MA females show differential FN400 responsesnder no promotions (more positive to buys). Given the dual-

rocess perspective, High MA females possibly attempt to integratehe price offered within the context of the product presented before

aking a decision to buy, which may be supported by a conceptualuency assessment. Consequently, increased amplitude under no

hown in blue. Bottom two rows show topographical maps of differences waves fortion condition then no promotion condition for High MA consumers and then fore references to color in this figure legend, the reader is referred to the web version

promotions among High MA females for non-buys might operatesimilarly to the typical N400 incongruity effect (Kutas and Hillyard,1980). The consumer compares the price to her internal referenceprice, which if rejected is reflected in an N400 mismatch effect.Reduced amplitude for buys might indicate that a price that is con-ceptually easier to process is more likely to be bought. Thus, itwould appear that High MA females utilize both perceptual flu-ency and conceptual fluency in judging whether to buy under nopromotions. No effects for FN400 noted for other consumers underno promotions might indicate that a conceptual fluency assessmentis not part of their decision making process or at least that suchassessments are not supported by conceptual fluency processes.

The resource allocation interpretation proposed by Paynter et al.(2009) additionally implies opposite decision-making styles underno promotions among High MA versus Low MA females. We specu-late that, absent promotions, High MA females may seek to confirmwhether a price is acceptable whereas Low MA females may seekto reject the acceptability of an offered price. Recent research byDutta et al. (2011) supports the notion that consumers sometimesadopt different decision styles when evaluating price informationas a function of their regulatory focus. Indeed, states such as anx-iety are well known to invoke a systematic processing bias (Clore

et al., 1994), and adoption of a confirmatory decision-making styleby High MA females may be part of this bias. If true, this leads tothe prediction that High MA females should conduct a more robustmemory search for accepted prices (i.e., buys) whereas the reverse

Page 11

2 l Psych

mLrbifstlirtmfrcceaml

powetfmHdnraTtotad

flti2ibpMtpcoabwsigwItceae

viduals are more risk-avoidant with respect to committing errorsand subsequently are more likely to activate an anterior cingu-late/medial prefrontal network during decision making even when

Author's Personal Copy

10 W.J. Jones et al. / Biologica

ight be found for Low MA females. Indeed, the amplitude data forPC under no promotions finds just that. The LPC is larger for buyselative to non-buys among High MA females under no promotions,ut larger for non-buys relative to buys for Low MA females. This

ndicates that Low MA females might carry out a more robust searchor conflicting price information and compare the outcome of thatearch to the offered price, which results in a larger non-buy ampli-ude. The decision-style explanation also provides a reason for theack of FN400 differences among Low MA females in that reject-ng a price may less critically depend on familiarity processes thately on conceptual fluency. In fact, one should expect the price noto feel conceptually familiar as the strategy goal is to retrieve the

ore familiar reference price. For clarity, we surmise that High MAemales under no promotions assess whether they can retrieve aeference price for an offer price, appraise whether the price feelsonceptually familiar to them for the product, and finally search theontents of memory in an effort to match the offer price to a refer-nce price. For Low MA females, we put forward that they initiallyssess whether they can reject an offer price before conducting aemory search to authenticate the rejection, which results in a

arger LPC to a rejected offer.In our introductory discussion of gender selectivity theory, we

ointed out that females often process comprehensively, relyingn a broader variety of information cues when making decisions,hereas males gravitate toward more selective processing. These

ffects are largely confirmed under no promotions with the excep-ion of LPC differences among Low MA males. LPC was largeror buys relative to non-buys under no promotions for Low MA

ales, an effect identical to that observed among High MA females.owever, Low MA males may experience numerical processingifferently than High MA males such that they are more metacog-itively aware of numerical processing dynamics such as priceetrieval. Under promotions, we again observe similar ERP effectsmong Low MA males and High MA females with respect to FN400.his is consistent with the view that High MA females would behe most biased toward a comprehensive processing style, perhapswing to an anxiety bias and their gender role. This is also consis-ent with the view that Low MA males are more metacognitivelyware of numerical processing dynamics, which they utilize in theirecision making.

FN400 effects also bear directly on the interplay of conceptualuency and buying. As discussed earlier, FN400 has been proposedo be generally related to familiarity (Rugg et al., 1998) or specif-cally related to conceptual processing (Voss and Paller, 2006,007). Under promotions, our results are more consistent with the

dea that FN400 is indicative of conceptual processing, at least foruys. Greater FN400 (less positive) was observed for buys underromotions for both High MA females and Low MA males. For HighA females and Low MA males, this might signify that to the extent

hat these consumers activate a semantic processing subroutine forrice promotions this is inferred as beneficial and increases pur-hasing. Thus, it would appear that FN400 may vary as a functionf shopping task demands. Possibly this reflects an initial attemptt identifying the post-promotion price via computation. Worky Zhou et al. (2006) identifies an anterior negative componentithin a similar time window (N300) that became larger when

ubjects invoked multiplication routines. Our subjects may bencorporating elements of multiplication into their decision strate-ies when they ultimately buy. If true, this finding echoes recentork on the “instrumentality heuristic” (Labroo and Kim, 2009).

n brief, the instrumentality heuristic refers to the notion that, inhe absence of a goal, effort decreases subjective evaluations. In

ontrast, when pursuing goals effort leads to enhanced subjectivevaluations. Under promotions, the consumer is presented withgoal of identifying a post-promotion price, and therefore the

ffort enacted via engaging conceptual processing subroutines

ology 89 (2011) 201–213

(e.g., related to multiplying or computation) to arrive at that pricemight be construed positively and treated as diagnostic to theirbuying decision accordingly. This is not to say that the subjectiveexperience of effort is hedonically marked. Rather, among High MAfemales who chronically experience negative affect under numer-ical processing, this may signal for them an uncommonly tolerablelevel of that state. Curiously, we also observed a latency delay forbuys compared to non-buys among Low MA males. Thus, FN400amplitude for Low MA and High MA females, as well as latencydelay for High MA males, may all be indicative of these consumersusing conceptual processing feelings, possibly experienced aseffortful, to infer the quality of the post-promotion price.

4.2. Buy effects

When making buying decisions, High MA females exhibitedincreased P3 amplitudes under promotions whereas Low MAfemales exhibited increased P3 amplitudes for non-buys also underpromotions. Under no promotions, this pattern is observed formales such that for High MA males buy amplitudes are larger thannon-buys, but for Low MA males non-buys are larger than buys.Significant differences were not present for females under no pro-motions or for males under promotions. From a selectivity theoryperspective, females might emphasize promotions at this stage dueto increased cognitive demands associated with evaluating pro-motions whereas males might emphasize no promotions as theyare more selective processors disinclined to reconsider metacogni-tive aspects of evaluating promotions. FN400 effects were observedfor males under promotions and ERPs delineate several differencesfor females under no promotions. We note the striking similari-ties in the scalp distributions of these parallel P3 findings. For HighMA participants, the scalp distributions appear to be frontocentralfor females under promotions and for males under no promo-tions. For Low MA participants, the scalp distributions appear tobe central-parietal for females under promotions and males underno promotions.

The P3 component may therefore be a P3a-like potential forHigh MA consumers (hereafter P3a), which might indicate greaterreliance on rapid motivational and emotional factors.2 For the LowMA group, the P3 is probably a P3b, indicating a qualitatively dif-ferent process. We address each of these possibilities in tandemas follows, though a bit of clarity is first required in relating ourHigh MA findings to the P3a literature. In assessing the extant lit-erature on P3a, Polich (2007) argues that “the P3a, novelty, andno-go P300 are most likely variants of the same ERP that varies inscalp topography as a function of attentional and task demands”(p. 2134). Novelty and no-go P3 relate to specific operationaliza-tions of P3 in two- and three-stimulus experimental paradigms ofthe oddball task to distractor stimuli (for a discussion, see Polichand Comerchero, 2003). As we did not employ these paradigms,we will instead emphasize the purported function and neuralunderpinnings of the entire class of P3a subcomponents, whichlargely results when a demanding stimulus commands frontalattention (Polich, 2007; Polich and Comerchero, 2003; Polich andCriado, 2006). Citing Desimone et al. (1995), Polich and Criado(2006) propose that P3a may be subtended by neural changes inanterior cingulate function when new stimuli replace the con-tents of working memory. Paulus et al. (2003) demonstrated that,compared to low trait anxious individuals, high trait anxiety indi-

2 We thank Christopher Paynter for making this suggestion and for other helpfulcomments on this paper.

Page 12

l Psych

ttc

pPieueMdfivb(bwaPfowip

rmmeaicaoo

eraslrtUwoosgPFwoPatimfi

wrair

Author's Personal Copy

W.J. Jones et al. / Biologica

he chance of committing an error is low. Fjell et al. (2007) showedhat P3a habituates more rapidly for high sensation seekers; theonverse may hold true for avoidant-seeking high anxiety types.

It is well known that P3b is associated with memory updatingrocesses (Polich and Criado, 2006). Kok (2001) puts forward that3b represents an “event categorization” decision when a stimuluss matched relative to an internal representation. Larger P3bs arelicited for improbable events whereas smaller P3bs are observednder difficulty (Donchin, 1981; Kok, 2001). P3 effects (and appar-nt locus) lend additional credence to the suggestion that HighA and Low MA consumers differ with respect to their consumer

ecision styles. Our view that High MA consumers seek to con-rm buys is consistent with the finding that P3a is larger for buysersus non-buys. High MA consumers possibly seek to confirm buysecause they are risk avoidant with respect to committing errorsPaulus et al., 2003). Already mentioned, this is consistent with theehavioral data that High MA consumers are more likely to buyhen reference price confidence is high and is supported by the

voidance-tendency literature (Paulus and Stein, 2010). Increased3a for buys thusly might index the affective salience of buyingor High MA consumers. Quite possibly, P3a reflects the allocationf additional frontal cortical resources (likely anterior cingulate)hen High MA consumers override their tendency to avoid buy-

ng decisions predicated on quantitative assessments of an item’srice.

The decision to buy by Low MA consumers may well stem from aeject decision style. For this group, the dominant buying proclivityay be to buy. This is consistent with the notion that buying is nor-ally a default mode, which does not occur when painful (Knutson

t al., 2007). In this sense, the P3 that they exhibit might representn updating of the dominant response to buy of which not buy-ng would reflect an update. Alternatively, larger P3bs to non-buysould reflect the demands of the task. Because the task is to buy,nd not to avoid buying, it may be that P3b codes the improbabilityf the offer price. A larger P3b is generated to non-buys when anffer is perceived as probably not a good buy.

Moreover, the decision style view confirms work by Polezzit al. (2010) who found that P3 amplitude is most sensitive toisk proneness in situations where one has a preference for takingrisk, which varies individually. In their study, participants who

howed a risk preference in a low-risk gambling task also showedarger P3 within that task relative to participants that showed aisk preference in a higher-risk task; the converse was true forhose with a risk preference in the high-risk task under high risk.nlike that study, the present study shows that P3 can be associatedith inter-individual variability within the same task as a function

f response type. Also interesting, Polezzi et al. found P3 effectsver both anterior and posterior regions. However, and unlike ourtudy, these authors emphasized participants P3 responses to bothains and losses. Reminiscent of others (Yeung and Sanfey, 2004),olezzi et al. found that P3 s were larger for gains versus losses.urthermore, these differences were amplified for the participantsith a risk preference within the low-risk task and significant

nly frontally (but marginally significant over the posterior). Thus,olezzi et al. (2010) demonstrated P3a sensitivity and larger P3amplitude among a lower risk-seeking group when they take risks,hough we again note the effect is topographically widespread asn the present study. To the extent our High MA consumers avoid

aking purchases, their buys might be akin to taking risks and ourndings are therefore similar to those of Polezzi et al. (2010).

In closing this discussion section, it is worth consideringhether the P3 findings we observe are associated with response-

elated processes. Verleger et al. (2005) contend that P3b may betransitory stage between perceptual processing and response

nitiation; however, P3a may be selectively driven by stimulus-elated processes. Why High MA consumers would demonstrate

ology 89 (2011) 201–213 211

stimulus-related process differences as evidenced by an underlyingP3a difference while Low MA consumers demonstrate response-related differences as evidenced mainly by an underlying P3bdifference is unclear. One possibility is that High MA consumersruminate over their decisions to buy, thereby emphasizing stim-ulus over response-related processes, perhaps owing to an erroravoidance tendency. In contrast, Low MA consumers are moreinclined to act and therefore P3 might reflect both the closure oftheir decision cycle and response preparation. On the other hand,P3 effects within High MA females seem to be at least partiallyaccounted for by an overlapping contingent negative variation(CNV) and specifically for non-buys. It is well known that frontalCNV is associated with response-related processes and is largerfor negative or threatening stimuli (Carretié et al., 2004). Perhaps,CNV indexes preparation of the rejection response for High MAfemales.

4.3. Summary of results

In summary, our results indicate that math anxiety, promo-tion format, and gender combine to influence consumer purchaseprocesses. For High MA females and Low MA males, strongerconceptual activation (FN400 effects) under promotions was asso-ciated with buying; for High MA males, a similar effect is observedsuch that FN400 latencies are delayed for buys.

The reverse effect also occurred for High MA females under nopromotions with non-buys presenting larger FN400 than buys. LPCamplitude was larger for High MA females and Low MA malesunder promotions, which we attribute to recollective processing.We have argued that these effects might be due to a bias to engagein comprehensive processing among High MA females and a greatersensitivity among Low MA males to the processing dynamics ofquantitative reasoning relative to other males. Unlike High MAfemales, Low MA females demonstrated larger LPC under no pro-motions for non-buys. We hypothesized that this may be due tothe adoption of a decision style which seeks to reject offer priceswhereas High MA females may assess prices via attempting to con-firm them. Differences between High MA and Low MA femalesemerged at least as early as P200. Based on past research, weinterpret P200 as indicative of perceptual processing. High MAfemales demonstrated larger P200 for buys and Low MA femalesdemonstrated larger P200 for non-buys, possibly reflecting the ini-tial stages of alternative buying strategies. Under promotions, weobserved identical P3 patterns such that High MA females exhib-ited larger amplitudes for buys and Low MA females exhibitedlarger amplitudes for non-buys. We observed this same patternamong males but under no promotions. Apparent anteriorizationof P3 among High MA may indicate greater reliance on emotionaland motivational factors when making buying decisions. Thereappeared to be a more posterior P3 for Low MA, which suggeststhat these consumers do more to integrate available informationwhen making decisions to buy or not buy.

5. Concluding remarks

Using fMRI, Knutson et al. (2007) investigated the neural con-sequences of buying (Knutson et al., 2007). In the present study,we explored individual differences in neural correlates of buyingwith ERP. Apart from our method being more sensitive in time,our work extended Knutson et al. (2007) by taking into accountmath anxiety, promotion format, and gender. Additionally, whileKnutson et al. utilized a preference buying task, we instructed par-

ticipants in our study to buy when they felt that the offer price wasthe best deal. In their study, Knutson et al. (2007) determined thatanterior insula and medial frontal cortex play key roles in shapingbuying behavior. Insular connections with frontal cortex, especially

Page 13

2 l Psych

at(hiCsawsfhiaaiwaa2tdvmsimcaiu

R

A

A

B

B

B

B

B

B

B

B

C

C

C

C

C

C

Author's Personal Copy

12 W.J. Jones et al. / Biologica

nterior cingulate and striatum, may be used to send signals abouthe state of the organism, which then biases how choices are madeBechara and Damasio, 2005; Xue et al., 2010). Thus, in preferenceow one literally feels about a product such as thinking its price

s too high is incorporated into a decision of whether to buy it.onsider that in response to high prices, for example, consumersometimes say that they have been gouged (i.e., price gouged),word which literally can mean to remove the eye of anotherith one’s thumb. When buying is rule based, it is less clear that

trong interoceptive signals should be present and that theseeelings should necessarily bias behavior. Mental arithmetic is,owever, widely known to induce an array of biomarkers in certain

ndividuals such as muscle tension, increased cardiac responses,nd so on (Berdina et al., 1972). Insula may play a pivotal role inssessing these biomarkers, and this response may be exaggeratedn High MA consumers (Paulus and Stein, 2006). Anterior cingulate,

hich receives information regarding the interoceptive state fromnterior insula, may mediate the process of deploying additionalttention resources as a gain-control function (Botvinick et al.,004; Paulus and Stein, 2006). Anteriorization of P3 and differenceshereof during buying may be related to interoceptive biasing ofecision-making within High MA consumers. The considerable buyersus non-buy differences that we find here for High MA femalesay result as a function of these tendencies. Among Low MA con-

umers, particularly males, interoceptive signals might be used tonfer the quality of their information processes, which themselves

ediate buying or not buying rather than an altered decision pro-ess per se. Future research utilizing our task and greater sourceccuracy techniques may aid in clarifying this relationship. Assess-ng whether P3a indexes interoceptive biasing of decision makingnder preference seems worthy of future research likewise.

eferences

shcraft, M.H., 2002. Math anxiety: personal, educational, and cognitive conse-quences. Current Directions in Psychological Science 11, 181–185.

shcraft, M.H., Kirk, E.P., 2001. The relationships among working memory, mathanxiety, and performance. Journal of Experimental Psychology: General 130,224–237.

arber, B.M., Odean, T., 2001. Boys will be boys: gender, overconfidence, and com-mon stock investment. Quarterly Journal of Economics 116, 261–292.

artlett, F.C., 1932. Remembering: A Study in Experimental and Social Psychology.Cambridge University Press, Cambridge, UK.

echara, A., Damasio, A., 2005. The somatic marker hypothesis: a neural theory ofeconomic decision. Games and Economic Behavior 52, 336–372.

erdina, N.A., Kolenko, O.L., Kotz, I.M., Kuznetzov, A.P., Rodionov, I.M., Savtchenko,A.P., Thorevsky, V.I., 1972. Increase in skeletal muscle performance during emo-tional stress in man. Circulation Research 30, 642–650.

ettman, J.R., Luce, M.F., Payne, J.W., 1998. Constructive consumer choice processes.Journal of Consumer Research 25, 187–217.

ishop, S.J., Duncan, J., Brett, M., Lawrence, A., 2004. Prefrontal cortical function andanxiety: controlling attention to threat-related stimuli. Nature Neuroscience 7,184–188.

otvinick, M., Cohen, J.D., Carter, C.S., 2004. Conflict monitoring and anterior cingu-late cortex: an update. Trends in Cognitive Sciences 8, 539–546.

rand, M., Labudda, K., Markowitsch, H.J., 2006. Neuropsychological correlatesof decision-making in ambiguous and risky situations. Neural Networks 19,1266–1276.

aharel, S., Poiroux, S., Bernard, C., Thibaut, F., Lalonde, R., Rebaï, M., 2002. ERPsassociated with familiarity and degree of familiarity during face recognition.International Journal of Neuroscience 112, 1499–1512.

abeza, R., Ohta, N., 1993. Dissociating conceptual priming, perceptual priming, andexplicit memory. European Journal of Cognitive Psychology 5, 35–53.

arretié, L., Mercado, F., Tapia, M., Hinojosa, J.A., 2001. Emotion, attention, and the‘negativity bias,’ studied through event-related potentials. International Journalof Psychophysiology 41, 75–85.

arretié, L., Mercado, F., Hinojosa, J.A., Martín-Loeches, M., Sotill, M., 2004. Valence-related vigilance biases in anxiety studied through event-related potentials.Journal of Affective Disorders 78, 119–130.

lore, G.L., Schwarz, N., Conway, M., 1994. Affective causes and consequences of

social information processing. In: Wyer, R.S., Srull, T.K. (Eds.), The Handbookof Social Cognition. , 2nd ed. Lawrence Erlbaum Associates, Hillsdale, NJ, pp.323–417.

ohen, J., 1988. Statistical Power Analysis for the Behavioral Sciences. LawrenceErlbaum Associates, Hillsdale, N.J.

ology 89 (2011) 201–213

Curran, T., 2000. Brain potentials of recollection and familiarity. Memory & Cognition28, 923–938.

Curran, T., Tepe, K.L., Piatt, C., 2006. ERP explorations of dual process recognitionmemory. In: Zimmer, H.D., Mecklinger, A., Lindenberger, U. (Eds.), Binding inHuman Memory: A Neurocognitive Approach. Oxford University Press, Oxford,pp. 467–492.

Darley, W.K., Smith, R.E., 1995. Gender differences in information-processing strate-gies: an empirical test of the selectivity model in advertising response. Journalof Advertising 24, 41–56.

Desimone, R., Miller, E., Chelazzi, L., Lueschow, A., 1995. Multiple memory systemthe visual cortex. In: Gazzaniga, M. (Ed.), The Cognitive Neurosciences. MIT Press,Cambridge, MA, pp. 475–486.

Donchin, E., 1981. Surprise! . . . surprise? Psychophysiology 18, 493–513.Dutta, S., Biswas, A., Grewal, D., 2011. Regret from post-purchase discovery of lower

market prices: do price refunds help? Journal of Marketing 75, 124–138.Eccles, J.S., Jacobs, J.E., 1986. Social forces shape math attitudes and performance.

Signs 11, 367–380.Eysenck, M.W., Derakshan, N., Santos, R., Calvo, M.G., 2007. Anxiety and cognitive

performance: attentional control theory. Emotion 7, 336–353.Faust, M.W., Ashcraft, M.H., Fleck, D.E., 1996. Mathematics anxiety effects in simple

and complex addition. Mathematical Cognition 2, 25–62.Fjell, A.M., Walhovd, K.B., Fischl, B., Reinvang, I., 2007. Cognitive function, P3a/P3b

brain potentials, and cortical thickness in aging. Human Brain Mapping 28,1098–1116.

Garbarino, E.C., Edell, J.A., 1997. Cognitive effort, affect, and choice. Journal of Con-sumer Research 24, 147–158.

Grandjean, D., Scherer, K.R., 2008. Unpacking the cognitive architecture of emotionprocesses. Emotion 8, 341–351.

Grasso, D.J., Simons, R.F., 2011. Perceived parental support predicts enhanced latepositive event-related brain potentials to parent faces. Biological Psychological86, 26–30.

Hajcak, G., Moser, J.S., Simons, R.F., 2006. Attending to affect: appraisal strategiesmodulate the electrocortical response to arousing pictures. Emotion 6, 517–522.

Hembree, R., 1990. The nature, effects, and relief of mathematics anxiety. Journal forResearch in Mathematics Education 21, 33–46.

Hoch, S.J., Drèze, X., Purk, M., 1994. EDLP, Hi-Lo, and margin arithmetic. Journal ofMarketing 58, 16–27.

Holcomb, P.J., 1993. Semantic priming and stimulus degradation: implications forthe role of the N400 in language processing. Psychophysiology 30, 47–61.

Hopko, D.R., Mahadevan, R., Bare, R.L., Hunt, M.K., 2003. The Abbreviated Math Anx-iety Rating Scale (AMAS). Construction, Validity, and Reliability Assessment 10,178–182.

Horton, D.L., Pavlick, T.J., Moulinjulian, M.W., 1993. Retrieval-based and familiarity-based recognition and the quality of Information in episodic memory. Journal ofMemory and Language 32, 39–55.

Howell, D.C., 2009. Multiple Comparisons with Repeated Measures. University ofVermont, Burlintongton, VT, Retrieved from: http://www.uvm.edu/∼dhowell/StatPages/More Stuff/RepMeasMultComp/RepMeasMultComp.html (accessed08.01.11).

Karlinsky, N., Patrick, M., 2011. Brain activity measures responses to ads, commer-cials. ABC News, Retrieved from: http://abcnews.go.com/Technology/scientists-response-ads-measuring-brainwaves/story?id=12841570 (accessed 30.04.11).

Kirk, R.E., 1996. Practical significance: a concept whose time has come. Educationaland Psychological Measurement 56, 746–759.

Knutson, B., Rick, S., Wimmer, G.E., Prelec, D., Loewenstein, G., 2007. Neural predic-tors of purchases. Neuron 53, 147–156.

Kok, A., 2001. On the utility of P3 amplitude as a measure of processing capacity.Psychophysiology 38, 557–577.

Kotchoubey, B., 2006. Event-related potentials, cognition, and behavior: a biologicalapproach. Neuroscience and Biobehavioral Reviews 30, 42–65.

Kutas, M., Hillyard, S.A., 1980. Reading senseless sentences: brain potentials reflectsemantic incongruity. Science 207, 203–205.

Labroo, A.A., Kim, S., 2009. The instrumentality heuristic: why metacogni-tive difficulty is desirable during goal pursuit. Psychological Science 20,127–134.

Lee, A.Y., Labroo, A.A., 2004. Effects of conceptual and perceptual fluency on brandevaluation. Journal of Marketing Research 41, 151–165.

LeFevre, J., Greenham, S.L., Waheed, N., 1993. The development of procedural andconceptual knowledge in computational estimation. Cognition and Instruction11, 95–132.

Lipkus, I., Samsa, G., Rimer, B., 2001. General performance on a numeracy scaleamong highly educated samples. Medical Decision Making 21, 37–44.

Maner, J.K., Richey, J.A., Cromer, K., Mallott, M., Lejuez, C.W., Joiner, T.E., Schmidt,N.B., 2007. Dispositional anxiety and risk-avoidant decision-making. Personalityand Individual Differences 42, 665–675.

Meyers-Levy, J., 1988. The influence of sex roles on judgment. Journal of ConsumerResearch 14, 522–530.

Meyers-Levy, J., Maheswaran, D., 1991. Exploring differences in males’ and females’processing strategies. Journal of Consumer Research 18, 63–70.

Miller, D., Bischel, M.I., 2004. Anxiety, working memory, gender, and math perfor-mance. Personality and Individual Differences 37, 591–606.

Morwitz, V.G., Greenleaf, E.A., Johnson, E.J., 1998. Divide and prosper: consumers’reactions to partitioned prices. Journal of Marketing Research, 453–463.

Murray, L.W., Dosser, D.A., 1987. How significant is a significant difference? Prob-lems with the measurement of magnitude of effect. Journal of CounselingPsychology 34, 68–72.

Page 14

l Psych

N

P

P

P

P

P

P

P

P

P

P

P

R

R

R

R

S

S

S

S

S

Author's Personal Copy

W.J. Jones et al. / Biologica

iedeggen, M., Rösler, F., 1999. N400 effects reflect activation spread during retrievalof arithmetic facts. Psychological Science 10, 271–276.

aulus, M.P., Rogalsky, C., Simmons, A., Feinstein, J.S., Stein, M.B., 2003. Increasedactivation in the right insula during risk-taking decision making is related toharm avoidance and neuroticism. NeuroImage 19, 1439–1448.

aulus, M.P., Stein, M.B., 2006. An insular view of anxiety. Biological Psychiatry 60,383–387.

aulus, M.P., Stein, M.B., 2010. Interoception in anxiety and depression. Brain Struc-ture and Function 214, 451–463.

aynter, C.A., Reder, L.M., Kieffaber, P.D., 2009. Knowing we know before we know:ERP correlates of initial feeling-of-knowing. Neuropsychologia 47, 796–803.

eter, J.P., Churchill, G.A., Brown, T.J., 1993. Caution in the use of difference scoresin consumer research. Journal of Consumer Research 19, 655–662.

eters, E., Västfjäll, D., Slovic, P., Mertz, C.K., Mazzocco, K., Dickert, S., 2006. Numeracyand decision making. Psychological Science 17, 407–413.

olezzi, D., Sartori, G., Rumiati, R., Vidotto, G., Daum, I., 2010. Brain correlates of riskydecision-making. Neuroimage 49, 1886–1894.

olich, J., 2007. Updating P300: an integrative theory of P3a and P3b. Clinical Neu-rophysiology 118, 2128–2148.

olich, J., Comerchero, M.D., 2003. P3a from visual stimuli: typicality, task, andtopography. Brain Topography 15, 141–152.

olich, J., Criado, J.R., 2006. Neuropsychology and neuropharmacology of P3a andP3b. International Journal of Psychophysiology 60, 172–185.

ribram, K.H., 1970. Feelings as monitors. In: Arnold, M. (Ed.), Feelings and Emotions.Academic Publisher, New York, pp. 41–53.

aghunathan, R., Pham, M.T., Corfman, K.P., 2006. Informational properties of anx-iety and sadness, and displaced coping. Journal of Consumer Research 32,596–601.

eber, R., Winkielman, P., Schwarz, N., 1998. Effects of perceptual fluency on affectivejudgments. Psychological Science 9, 45–48.

ugg, M.D., Curran, T., 2007. Event-related potentials and recognition memory.Trends in Cognitive Sciences 11, 251–257.

ugg, M.D., Mark, R.E., Walla, P., Schloerscheidt, A.M., Birch, C.S., Allan, K., 1998.Dissociation of the neural correlates of implicit and explicit memory. Nature392, 595–598.

cherer, K.R., 2004. Feelings integrate the central representation of appraisal-drivenstatement organization in emotion. In: Manstead, A.S.R., Frijda, N.H., Fischer,A.H. (Eds.), Feelings and Emotions: The Amsterdam Symposium. CambridgeUniversity Press, Cambridge, pp. 136–157.

chendan, H.E., Kutas, M., 2003. Time course of processes and representationssupporting visual object identification and memory. Journal of Cognitive Neu-roscience 15, 111–135.

chwarz, N., 2004. Meta-cognitive experiences in consumer judgment and decision

making. Journal of Consumer Psychology 14, 332–348.

iegler, R.S., 1988. Strategy choice procedures and the development of multiplicationskill. Journal of Experimental Psychology: General 117, 258–275.

ong, H., Schwarz, N., 2009. If it’s difficult to pronounce, it must be risky: fluency,familiarity, and risk perception. Psychological Science 20, 135–138.

ology 89 (2011) 201–213 213

Tsui, J.M., Mazzocco, M.M.M., 2007. Effects of math anxiety and perfectionism ontimed versus untimed math testing in mathematically gifted sixth graders.Roeper Review 29, 132–139.

Vanhuele, M., Drèze, X., 2002. Measuring the price knowledge shoppers bring to thestore. Journal of Marketing 66, 72–85.

Verleger, R., Jaskowski, P., Wascher, E., 2005. Evidence for an integrative role ofP3b in linking reaction to perception. Journal of Psychophysiology 19, 165–181.

Voss, J.L., Federmeier, K.D., 2011. FN400 potentials are functionally identical toN400 potentials and reflect semantic processing during recognition testing.Psychophysiology 48, 532–546.

Voss, J.L., Paller, K.A., 2006. Fluent conceptual processing and explicit memory forfaces are electrophysiologically distinct. Journal of Neuroscience 26, 926–933.

Voss, J.L., Paller, K.A., 2007. Neural correlates of conceptual implicit memory andtheir contamination of putative neural correlates of explicit memory. Learning& Memory 14, 259–267.

Voss, J.L., Schendan, H.E., Paller, K.A., 2010. Finding meaning in novel geometricshapes influences electrophysiological correlates of repetition and dissociatesperceptual and conceptual priming. NeuroImage 49, 2879–2889.

Whittlesea, B.W.A., 1993. Illusions of familiarity. Journal of Experimental Psychol-ogy: Learning, Memory, and Cognition 19, 1235–1253.

Whittlesea, B.W.A., Jacoby, L.L., Girard, K., 1990. Illusions of immediate memory: evi-dence of an attributional basis for feelings of familiarity and perceptual quality.Journal of Memory and Language 29, 716–732.

Winkielman, P., Schwarz, N., Fazendeiro, T., Reber, R., 2003. The hedonicmarking of processing fluency: implications for evaluative judgment. In:Musch, J., Klauer, K.C. (Eds.), The Psychology of Evaluation: Affective Pro-cesses in Cognition and Emotion. Lawrence Erlbaum Associates, Mahwah, NJ,pp. 189–217.

Woodman, G.F., 2010. A brief introduction to the use of event-related potentials instudies of perception and attention. Attention, Perception, & Psychophysics 72,2031–2046.

Woollams, A.M., Taylor, J.R., Karayanidis, F., Henson, R.N., 2008. Event-related poten-tials associated with masked priming of test cues reveal multiple potentialcontributions to recognition memory. Journal of Cognitive Neuroscience 20,1114–1129.

Xia, L., Monroe, K., Cox, J., 2004. That price is unfair! A conceptual framework ofprice fairness perceptions. Journal of Marketing 68, 1–15.

Xue, G., Lu, Z., Levin, I.P., Bechara, A., 2010. The impact of prior risk experienceson subsequent risky decision-making: the role of the insula. NeuroImage 50,709–716.

Yeung, N., Sanfey, A., 2004. Independent coding of magnitude and valence in thehuman brain. Journal of Neuroscience 24, 6258–6264.

Yonelinas, A.P., 2002. The nature of recollection and familiarity: a review of 30 yearsof research. Journal of Memory and Language 46, 441–517.

Zhou, X., Chen, C., Dong, Q., Zhang, H., Zhou, R., Zhao, H., Qiao, S., Jiang, T., Guo, Y.,2006. Event-related potentials of single-digit addition, subtraction, and multi-plication. Neuropsychologia 44, 2500–2507.