Deconstructing the seductive allure of neuroscience explanations

13 Pages·2015·0.2 MB·English

JudgmentandDecisionMaking,Vol.10,No.5,September2015,pp.429–441
Deconstructing the seductive allure of neuroscience explanations
DeenaSkolnickWeisberg∗ JordanC.V.Taylor† EmilyJ.Hopkins†
Abstract
Previousworkshowedthatpeoplefindexplanationsmoresatisfyingwhentheycontainirrelevantneuroscienceinformation.
Thecurrentstudiesinvestigatewhythiseffecthappens.InStudy1(N=322),subjectsjudgedpsychologyexplanationsthatdid
ordidnotcontainirrelevantneuroscienceinformation.Longerexplanationswerejudgedmoresatisfying,aswereexplanations
containing neuroscience information, but these two factors made independent contributions. In Study 2 (N=255), subjects
directlycomparedgoodandbadexplanations. Subjectsweregenerallysuccessfulatselectingthegoodexplanationexcept
whenthebadexplanationcontainedneuroscienceandthegoodonedidnot. Study3(N=159)testedwhetherneuroscience
jargon was necessary for the effect, or whether it would obtain with any reference to the brain. Responses to these two
conditionsdidnotdiffer.Theseresultsconfirmthatneuroscienceinformationexertsaseductiveeffectonpeople’sjudgments,
whichmayexplaintheappealofneuroscienceinformationwithinthepublicsphere.
Keywords:explanation,neuroscience,reasoning,seductiveallure.
1 Introduction
cially, the neuroscience information was irrelevant to the
logicoftheexplanationsandevenmadethegoodexplana-
Attention to neuroscience is growing within the public tionsworse,accordingtotheratingsofexperts.
sphere. Neuroscientificfindingsnowplayakeyroleinpub- When the explanations contained neuroscience informa-
licconversationsabouteconomics, marketing, andthelaw,
tion, ratings were significantly higher than when they did
among other areas (e.g., Ariely & Berns, 2010; Camerer,
not. Thiswasespeciallytrueforthebadexplanations(per-
Loewenstein & Prelec, 2005; Farah, 2012; Greene & Co-
haps because people have trouble detecting circularity in
hen, 2004; Roskies, 2002; Satel & Lilienfeld, 2013). For
arguments, see Rips, 2002). That is, non-experts judged
example, neuroscience data are often used in courtrooms
that psychological phenomena are explained better using
asevidenceofadefendant’sresponsibilityorguilt(Morse,
thelanguageofneuroscience,althoughthislanguageshould
2011;Saks,Schweitzer,Aharoni&Kiehl,2014;Schweitzer
makenodifference, assumingthatanexplanation’s quality
etal.,2011). Butitisnotentirelyclearhowmembersofthe
is drawn primarily from the strength of its logic. One re-
publicviewthesefindings. Dotheyunderstandtherolethat
cent study (Scurich & Shniderman, 2014) also found that
neuroscience information plays in explanations of people’s
subjects give higher ratings to studies that included neuro-
beliefsandbehaviors?
scienceinformation,butonlywhentheconclusionsofthese
Previous research suggests that the answer to this ques-
studiesconfirmedtheirpriorbeliefs. However,theabsence
tion is “no”. People are unduly swayed to think favor-
of a no-neuroscience control condition in this study makes
ably of psychology explanations that include references to
itdifficulttodrawfirmconclusionsaboutthegeneraleffect
neuroscience—even when such neuroscience information
ofneuroscienceinformation.
is logically irrelevant to the explanations (Weisberg, Keil,
Goodstein, Rawson & Gray, 2008). In this study, subjects Three other studies did include the appropriate controls,
read descriptions of psychological phenomena. Each phe- and both confirmed Weisberg et al.’s (2008) findings. One
nomenonwasfollowedbyoneoffourtypesofexplanation, used the same stimuli in an exact replication (Fernandez-
constructed by crossing explanation quality (good or bad) Duque, Evans, Christian & Hodges, 2015). The other
with neuroscience information (present or absent). Cru- two used different sets of stimuli in a conceptual replica-
tion (Rhodes, Rodriguez & Shah, 2014; Rhodes & Shah,
WethankMatthewBateman,MarthaFarah,DiegoFernandez-Duque, 2015), inwhichsubjectsreadamocknewsarticledescrib-
FrankKeil,andthemembersofthePennCognition&DevelopmentLab. ing psychological research; the article either did or did not
ThisresearchwasfundedbytheTempletonFoundation(VarietiesofUn-
contain irrelevant neuroscience information. Neuroscience
derstandingprojectgranttoDSW).
Copyright: ©2015. Theauthorslicensethisarticleunderthetermsof information thus exerts a seductive allure effect, whereby
theCreativeCommonsAttribution3.0License. peoplewithoutadvancedtrainingbelievethatreferencesto
∗University of Pennsylvania, Department of Psychology, 3720
brain processes improve the quality of a psychological ex-
Walnut St., Solomon Labs, Philadelphia, PA 19104. Email:
planation, even when these references are logically irrele-
[email protected].
†UniversityofPennsylvania. vant. Thiseffectcouldbethoughtofaspartofafamilyof
429
JudgmentandDecisionMaking,Vol.10,No.5,September2015 Seductiveallureofneuroscienceexplanations 430
heuristicsthatpeopleuseforjudgingthequalityofexplana- thantheexplanationsthatdidnot. Subjectsmayhavesim-
tions, which includes teleological information (Lombrozo ply rated longer explanations as better. Indeed, other work
&Carey,2006)andanintuitivesenseofsatisfaction(Trout, showed that people prefer longer explanations, even if the
2002). addedlengthdidnotaddtotheexplanation’squality(Kikas,
Onestudyclaimedthatneuroscience imagesarerespon- 2003).Study1thusbeginsourinvestigationofthiseffectby
sible for the effect (McCabe & Castel, 2008), suggesting replicatingWeisbergetal.(2008)withtheadditionofacon-
thatpeopleareseducedbythevisualappealofimagesgen- trolforthelengthoftheexplanations.
erated by fMRI scans and other neuroscientific techniques. A second possible explanation for the effect is that neu-
However, many later studies have failed to replicate this roscience information may appeal due to its authoritative
finding (Gruber & Dickerson, 2012; Hook & Farah, 2013; aesthetic: Explanations containing neuroscience informa-
Keehner, Mayberry & Fischer, 2011; Michael, Newman, tion may look as though they have come from a suitably
Vuorre,Cumming&Garry,2013;seeFarah&Hook,2013, scientific process, and so may be perceived as trustworthy
forreview). Totestdirectlywhetherbrainimagesaddvalue and therefore convincing, regardless of their content (see
to explanations that already contained neuroscience text, Sperber, 2010). We address this issue in Study 2 by ask-
Fernandez-Duque et al. (2015) presented subjects with ex- ingsubjectstodirectlycomparegoodandbadexplanations
planations that either contained no neuroscience informa- whentheydoanddonotcontainneuroscienceinformation.
tion,containedirrelevantneuroscienceinformation,orcon- Thethirdpossibilitythatweinvestigateisthatpeopleare
tainedirrelevantneuroscienceinformationandwereaccom- attracted to any kind of scientific-sounding jargon because
panied by a neuroscience image. These researchers found they believe that use of these fancy terms signals higher-
that people rated explanations with neuroscience informa- quality science. Indeed, math-based jargon has precisely
tion as better than explanations without this information, thiseffect(Eriksson,2012). WeaddressthisissueinStudy
as noted above, but images did not have any additional ef- 3 by comparing subjects’ ratings of explanations that use
fect.Further,Weisbergetal.(2008),Fernandez-Duqueetal. simplereferencestobrainprocesseswiththeirratingsofex-
(2015), Rhodes et al. (2014), and Rhodes and Shah (2015) planationsthatusemoretechnicalterms.
obtained the seductive allure effect without the use of any
pictures. These studies strongly suggest that neuroscience
imageryisnotthesourceoftheeffect. 2 Study 1
Why, then, does this effect happen? The importance of
answeringthisquestionbecomesevidentwhenweexamine Study1wasdesignedtodeterminewhethertheseductiveal-
the many ways in which neuroscience information is used lureeffectresultsfromsubjects’responsestoneuroscience
(and misused) in the public sphere. The proliferation of informationitselforfromthetendencyforexplanationscon-
headlines proclaiming that some drug or activity “literally tainingneuroscienceinformationtobelongerthanexplana-
changes your brain” illustrates both how appealing neuro- tionswithoutthisinformation. Previousworksuggeststhat
scienceinformationistothegeneralpublicandhowpoorly lengthdoesnotaccountfortheeffect: Fernandez-Duqueet
thisinformationisunderstood.Totakeaweightierexample, al.(2015)foundthatexplanationswithaddedneuroscience
attorneysmayappealtoneuroscience-basedevidenceinor- informationwereratedmorehighlythanunembellishedex-
dertoconvinceajuryofalegalfact. Butbecausethiskind planations, but explanations with added social psychology
of information is intuitively compelling even when it is ir- information were not. In addition, Rhodes et al. (2014)
relevant,suchevidencemayundulybiasthejury,potentially foundthatstimuliwithneuroscienceinformationwererated
threateningthefairnessofthejudicialsystem(seeGreene& more highly than length-matched stimuli without neuro-
Cohen, 2004; Morse, 2004). Similarly, in the field of edu- scienceinformation.Theseresultssuggestthattheseductive
cation, unsubstantiated claims about how children’s brains allureeffectcannotbeaccountedforsolelybytheexplana-
changeorfundamentaldifferencesbetweenboys’andgirls’ tions’length.
brains can lead to the implementation of educational poli- Study1continuedthisinvestigationoftheroleoflength
cies or practices that seem appealing but may not actually and addressed a potential issue with the method used in
benefitthestudents(seeBruer,1997;Goswami,2006). previous studies. Both Fernandez-Duque et al. (2015)
Learning why neuroscience information is alluring can and Rhodes et al. (2014) controlled for length by mak-
help us to develop techniques to reverse some of these ing the without-neuroscience explanations longer, so as to
trends. The current studies begin to address this issue by match the length of the with-neuroscience stimuli. But
investigatingthreefactorsthatmightcontributetotheseduc- thisadditionalinformationmayhaveaffectedhowsubjects
tiveallureeffect: length(Study1),explicitappealofneuro- rated the without-neuroscience explanations. For example,
science(Study2),andjargon(Study3). Intermsoflength, Fernandez-Duqueetal.(2015)comparedexplanationswith
the explanations in Weisberg et al. (2008) that contained superfluousinformationfromsocialscienceorhardsciences
irrelevant neuroscientific information were always longer to those with superfluous neuroscience information. How-
JudgmentandDecisionMaking,Vol.10,No.5,September2015 Seductiveallureofneuroscienceexplanations 431
ever, this information from other fields may have seemed Materials. We selected four of the 18 items presented to
less relevant to the explanations than the neuroscience in- subjects in Weisberg et al. (2008) and Fernandez-Duque et
formation,potentiallyloweringsubjects’ratings. Thusthis al. (2015) (babies’ abilities to do simple arithmetic, atten-
design does not separate the effect of length from the ef- tionalblink,genderdifferencesinspatialreasoning,anddif-
fect of different types of added information. The current ferencesbetweenseeingandimaginingobjects;seesupple-
study made the with-neuroscience explanations shorter so mentalmaterialsforfullstimulusitems). Thesewereitems
astomatchthelengthofthewithout-neuroscience stimuli. for which subjects in a pilot sample consistently judged
This more fully un-confounds the variables of length and the bad version of the without-neuroscience explanation as
neuroscienceinformation. worse than the good version of that explanation. Each of
Study 1 thus provides a more complete investigation of the four items consisted of a description of a psycholog-
the potential effect of length on the seductive allure effect, ical phenomenon and eight different explanations for that
whichwillallowustodeterminehowneuroscienceinforma- phenomenon. The good explanations are the ones that the
tionaffectspeople’sjudgments.Iftheseductiveallureeffect researchersthemselvesprovidedforthephenomenaorthat
is only due to a general tendency to judge longer explana- were provided in psychology textbooks. The bad explana-
tions as better, then it should disappear when the explana- tions were circular restatements of the phenomena with no
tions that do and do not contain neuroscience are matched mechanisticinformationthatcouldgiveareasonforthephe-
for length. But if something about neuroscience informa- nomenon. Itemsinallstudiesareinthesupplement.
tionleadstomorepositivejudgmentsofexplanations,then The explanations used in the Without Neuroscience-
the effect of neuroscience should remain regardless of the Short and the With Neuroscience-Long conditions exactly
lengthoftheexplanation. matchedthoseusedinWeisbergetal. (2008). Toconstruct
theWithoutNeuroscience-Longexplanations,weaddedsu-
perfluous wording to the existing Without Neuroscience-
2.1 Method
Shortexplanationstomakethemthesamelengthasthecor-
Subjects. We recruited subjects from two populations: responding With Neuroscience-Long explanations. Impor-
undergraduatestudentsfromthepsychologysubjectpoolat tantly,thisadditionalwordingreferredonlytopsychological
theUniversityofPennsylvaniaandworkersonMechanical constructsandnevertoothersciences,mathematics,orneu-
Turk. Because previous work on this topic has primarily roscience,andthisinformationdidnotaddanyvaluetothe
usedundergraduatesassubjects,weaddedtheMTurkwork- explanation. ToconstructtheWithNeuroscience-Shortex-
ers in order to assess the generality of the effect in a more planations,weeditedtheexistingWithNeuroscience-Long
representative population. This study included 204 under- explanations to make them the same length as the corre-
graduates (143 women, 61 men; mean age = 19.8 years, sponding Without Neuroscience-Short explanations. Re-
range = 18–50) and 177 MTurk workers (85 women, 92 gardless of length, the irrelevant neuroscience information
men;meanage=37.5years,range=19–70). Undergradu- wasidenticalacrossthegoodandbadversionsoftheexpla-
atesreceivedcoursecreditforparticipatinginthestudy,and nationsforeachphenomenon.
MTurkworkersreceived20cents.
Procedure. All subjects completed an online survey dis-
Design. Subjects were divided into 4 conditions accord- tributedonQualtrics. Ineachtrial,subjectsreadadescrip-
ingtoa2(Neuroscience: with,without)x2(Length: long, tionofapsychologicalphenomenon,whichappearediniso-
short)design. Thesewerebothbetween-subjectsvariables, lationonthescreenfor10secondsbeforetheywereallowed
soanindividual subjectsawexplanations thateitherallin- toadvancetothenextscreen. Onthesecondscreenofeach
cludedoralldidnotincludeneuroscienceinformation,and trial,thephenomenonappearedagainatthetop,followedby
theirexplanationswouldallcomefromthesamelengthcat- oneoftheeightpossibleexplanationsforthatphenomenon.
egory. There were 43 MTurk workers and 44 undergrad- Subjectswereaskedtojudgehowsatisfyingtheyfoundthis
uates in With Neuroscience-Long, 40 MTurk workers and explanationonaseven-pointscale,from–3(veryunsatisfy-
65 undergraduates in With Neuroscience-Short, 49 MTurk ing)to+3(verysatisfying),with0astheneutralmidpoint.
workers and 50 undergraduates in Without Neuroscience- Eachsubjectsawallfourstimulusitems,onepertrial,in
Long, and 45 MTurk workers and 45 undergraduates in arandomizedorder.Oneachtrial,thephenomenonwaspre-
WithoutNeuroscience-Short. Qualitywasawithin-subjects sentedalongwithoneversionoftheexplanation. Subjects’
variable; for each trial, the survey software randomly de- conditiondeterminedwhethertheysawalongorshortver-
termined whether to show the good or bad version of the sionandawith-orwithout-neuroscienceversion. Whether
explanation.1 theysawagoodorbadversionoftheexplanationwasran-
domly determined on each trial (as described above in the
1Duetotherandomization,therewere47subjectswhosaweithergood
Designsection). Attheendofthesurvey,subjectsprovided
explanationsoneverytrialorbadexplanationsoneverytrial.Theinclusion
ofthesesubjectsdidnotaffectanyanalyses,sotheywereleftinthesample. basic demographic information: age in years, gender, and
JudgmentandDecisionMaking,Vol.10,No.5,September2015 Seductiveallureofneuroscienceexplanations 432
Table 1: Study 1 mixed-effects linear regression model (∗ Figure 1: Average ratings of explanation quality in Study
p<.05). 1. Errorbarsrepresent95%confidenceintervalsaroundthe
means.
Predictor Estimate [95%CI] t
Intercept 0.24 [0.14,0.34] 5.00∗ Study 1
Item2 0.24 [0.10,0.38] 3.21∗ 3
Good Explanations
Item3 –0.61 [–0.74,–0.47] –8.82∗ Bad Explanations
Item4 0.65 [0.21,0.79] 9.41∗ 2
Length 0.12 [0.02,0.21] 2.55∗
Neuroscience 0.13 [0.03,0.22] 2.70∗ 1
Group 0.15 [0.06,0.25] 3.18∗ Rating
QNueualriotsyciencexItem2 –00..0356 [[–00..2189,,00..4140]] –80..4772∗ Average 0
NeurosciencexItem3 0.20 [0.06,0.34] 2.93∗ −1
NeurosciencexItem4 –0.24 [–0.37,–0.10] –3.45∗
GroupxItem2 0.03 [–0.11,0.17] 0.50 −2
GroupxItem3 0.16 [0.01,0.30] 2.34∗
GroupxItem4 –0.19 [–0.34,–0.05] –2.71∗ −3
Without Neuroscience With Neuroscience
QualityxItem2 0.08 [–0.08,0.20] 1.06
QualityxItem3 –0.14 [–0.28,–0.02] –1.94∗
QualityxItem4 –0.25 [–0.38,–0.12] –3.57∗
Note:Thisregressionpredictedsubjects’ratingsofthequal-
This test revealed a main effect of Quality (Figure 1),
ityoftheexplanations. TheinterceptrepresentstheWithout
wheregoodexplanations(M =0.58,SD=1.68)wererated
Neuroscience condition, short explanations, undergraduate
morehighlythanbadexplanations(M =–0.13,SD=1.80).
subjects, and bad explanations. Item is deviation coded,
TherewasalsoamaineffectofNeuroscience: Explanations
such that the coefficient for each level represents deviation
that contained neuroscience (M = 0.34, SD = 1.75) were
fromthegrandmean;Item1isthereferencelevel.
ratedmorehighlythanexplanationsthatdidnot(M =0.11,
SD = 1.80). We also found a main effect of Length: Long
levelofeducation(fortheMTurksubjects)orclassyearand explanations(M =0.34,SD=1.76)wereratedmorehighly
major(fortheundergraduates). thanshortexplanations(M=0.12,SD=1.79).Finally,there
wasamaineffectofGroup: MTurkworkers(M =0.37,SD
2.2 Results =1.73)gaveoverallhigherratingsthanundergraduates(M
=0.10,SD=1.80).
Unlike in Weisberg et al. (2008), some subjects in the cur-
rentstudyreceivedunequalnumbersofgoodandbadexpla- TheeffectsofGroup,Neuroscience,andQualityalsovar-
nations. In order to deal with this, we conducted a mixed- iedbyitem, asindicatedbythesignificantinteractions. To
effects linear regression. The model included random in- examinetheseinteractions,weconductedseparatelinearre-
terceptsbysubjectaswellasrandomslopesbysubjectfor gressions for each item examining main effects of Group,
theeffectofQuality(theonlywithin-subjectsvariable). We Neuroscience,Length,andQuality. Theresultsaresumma-
tested effects of Item, Group (MTurk or undergraduates), rized in Table 2. Although the magnitudes (and therefore
Length (long or short), Neuroscience (present or absent), significance levels) of the effects varied by item, only two
andQuality(goodorbad)andtheirinteractions2;themodel
effectswerenotinthepredicteddirections;forItem4,there
thatbestfitthedataisshowninTable1.
were non-significant negative effects of Group and Neuro-
2Preliminaryanalysesrevealedoneeffectofgender:aninteractionbe- science. TheeffectsofneuroscienceforItem2andquality
tweengenderandexplanationlength.Men’sratingsdidnotdifferforlong
for Item 4 were small and non-significant, but in the pre-
explanations(M=0.27,SD=1.75)andshortexplanations(M=0.33,SD
=1.74). However,womenratedlongexplanations(M=0.38,SD=1.76) dicteddirections.
morehighlythanshortexplanations(M=–0.03,SD=1.81). Wehaveno
explanationforthisunexpectedgenderdifference,andbecausegenderdid
notaffecttheothervariables,wedidnotconsidergenderfortheremainder ofouranalyses.
JudgmentandDecisionMaking,Vol.10,No.5,September2015 Seductiveallureofneuroscienceexplanations 433
maybeduetothefactthattheneuroscienceinformationin
Table2:Regressioncoefficientsforindividualitemanalysis
theexplanationsofthisphenomenonisentirelycontainedin
inStudy1(∗p<.05,+p<.10).
the first sentence, separate from the explanatory (or circu-
Item1 Item2 Item3 Item4 lar)informationinthesecondsentence. Thisstructuremay
havemadeiteasierforsubjectstoseethattheneuroscience
Length 0.04 0.05 0.17+ 0.19∗ informationwasnotrelevanttotheexplanation’squality.
Neuroscience 0.22∗ 0.08 0.34∗ –0.10 Overall,Study1demonstratesthattheseductiveallureef-
Group 0.14 0.18∗ 0.30∗ –0.06 fectreplicatesandisnotsolelyduetolength.Study2begins
Quality 0.69∗ 0.44∗ 0.23∗ 0.08 tomoredirectlyaddresswhytheeffecthappens. Todoso,
ratherthanaskingsubjectstoratesingleexplanationsfora
phenomenon,weaskthemtochoosewhichoftwoexplana-
tionstheyfindmostsatisfying. Eachpaircontainedagood
2.3 Discussion
explanationandabadexplanation,buteitherbothcontained
Study1wasdesignedtoreplicatetheseductiveallureeffect neuroscience, neither contained neuroscience, or only the
andtestforthecontributionofexplanationlength. Subjects bad explanation contained neuroscience. This is a some-
did indeed judge longer explanations as better than shorter whatlessecologicallyvaliddesign,sinceitisrarethatpeo-
ones overall, demonstrating a general bias towards longer plewouldneedtoevaluatemultipleexplanationsforasingle
explanations.However,thislengthpreferencedoesnotfully phenomenon. However,thisdesignallowsustotestdirectly
explaintheseductiveallureofneuroscience. Makingexpla- how neuroscience information may interfere with people’s
nationslongerdoesmakethemseembetter,butaddingneu- ability to distinguish good from bad explanations. Given
roscienceinformationdoesaswell,andthesetwomodifica- previousresults,weexpectedthatsubjectswouldgenerally
tionshadindependenteffects. Thisresultconfirmsotherre- beabletodistinguishgoodfrombadexplanationsifbothor
centstudiesthatshowthattheseductiveallureeffectobtains neither contained neuroscience. If, however, neuroscience
whenexplanationlengthiscontrolled(Fernandez-Duqueet informationhastheeffectofmaskingthepoorqualityofthe
al.,2015;Rhodesetal.,2014). badexplanations,peopleshouldbelesslikelytodistinguish
goodfrombadexplanationswhenonlythebadonecontains
There was also a strong effect of explanation quality:
neuroscience.
Good explanations were judged as better than bad expla-
nationsoverall. Thisresultdemonstratesthatpeoplearenot
generally confused about what makes certain explanations
3 Study 2
betterthanothersandareabletodistinguishbetweengood
andbadexplanations.However,asnotedabove,theaddition
ofneuroscienceinformationinterfereswiththisability. 3.1 Method
Finally,wefoundthatundergraduatesgaveoveralllower
Subjects. Thisstudyincluded130undergraduates(86fe-
ratingsthanMTurkworkers.Thisislikelynotduetotheun-
male,44male; meanage=19.5years,range=18–27)and
dergraduateshavingahigherlevelofeducation,since99%
130MTurkworkers(90female,37male,threeunreported;
oftheMTurkworkersreportedhavingatleastsomecollege
mean age = 40.6 years, range = 19–71). The undergradu-
education, and 50% reported earning an advanced degree.
ateswererecruitedfromthepsychologysubjectpoolatthe
Instead, the experience of participating in research as part
University of Pennsylvania and received course credit for
ofaclass,orofcurrentlybeingamemberofaneducational
theirparticipation. TheMTurkworkerswererecruitedfrom
community, may serve to increase overall skepticism. Re-
Amazon’s system and were paid 20 cents for their partici-
gardless,bothpopulationsshowedthesamegeneralpattern
pation. Anadditionalsevensubjects(threeMTurkworkers
ofresponsestotheexplanations(i.e.,therewerenosignifi-
andfourundergraduates)wererecruitedbutexcludedfrom
cantinteractionswithsubjectgroup).
the final analyses for failing an attention check (described
These main effects appeared in nearly the same way for
below).
allfouritems;however,thereweresomedifferencesinhow
subjects responded to the four phenomena. Notably, the
phenomenondescribingthedifferencesbetweenseeingand Design. There were three between-subjects conditions in
imaginingobjects(Item4)didnotshowaneffectofQuality thisstudy. AsinStudy1,therewerefourtrialspersubject,
orNeuroscience.Thisitemwasratedhigheroverallthanthe eachofwhichusedadifferentphenomenon(orderrandom-
others(asindicatedbythesignificantmaineffectforItem4 ized). Eachphenomenonwasaccompaniedbybothagood
in the regression), and there was little difference in ratings and a bad explanation. In the Without Neuroscience con-
between the different versions of the explanation. In addi- dition (41 MTurk workers and 43 undergraduates), neither
tion, the phenomenon describing attentional blink (Item 2) explanationcontainedanyneuroscienceinformation,andin
did not show as strong of an effect of Neuroscience. This theWithNeurosciencecondition(42MTurkworkersand45
JudgmentandDecisionMaking,Vol.10,No.5,September2015 Seductiveallureofneuroscienceexplanations 434
undergraduates), both explanations contained neuroscience
Table3: Study2mixed-effectslogisticregressionmodel(∗
information. ThecrucialconditionwastheMixedcondition
p<.05).
(47 MTurk workers and 42 undergraduates), in which the
goodexplanationdidnotcontainneuroscienceinformation
Predictor Estimate SE z
andthebadonedid,pittingqualityandneuroscienceagainst
eachother. Intercept 0.05 0.14 0.33
Item2 0.67 0.13 5.05∗
Materials. WeusedthesamefourphenomenaasinStudy Item3 –0.25 0.12 –2.08∗
1, accompanied by the short versions of the four possible Item4 –0.94 0.13 –7.44∗
explanationsforeachphenomenon: goodandbadexplana- Group 0.25 0.14 1.75
tions both with and without irrelevant neuroscience infor- WithoutNeurosciencecondition 0.63 0.21 3.08∗
mation(seesupplementalmaterials).
WithNeurosciencecondition 1.12 0.22 5.18∗
GroupxItem2 –0.20 0.13 –1.52
Procedure. Subjects completed an online survey dis-
GroupxItem3 0.19 0.12 1.53
tributedonQualtrics.Foreachtrial,theyfirstreadadescrip-
tionofapsychologicalphenomenon,whichappearediniso- GroupxItem4 0.37 0.12 3.05∗
lationonthescreenfor10secondsbeforetheywereallowed WithoutNeurosciencecond.xGroup–0.49 0.21 –2.37∗
toadvancetothenextscreen. Onthesecondscreenofeach WithNeurosciencecond.xGroup –0.53 0.21 –2.48∗
trial, the phenomenon appeared again at the top, followed
Note. TheinterceptrepresentstheMixedconditionandun-
by the prompt, “Please choose which explanation you find
dergraduatesubjects. Itemisdeviationcoded,suchthatthe
more satisfying.” Subjects always saw one good explana-
coefficientforeachlevelrepresentsdeviationfromthegrand
tionandonebadexplanationaswellasthechoice“bothare
mean;Item1isthereferencelevel.
equal.” The “equal” option always appeared in the center,
withtheleft/rightpositionofthegoodandbadexplanations
randomized across trials. After making their choice, sub-
jectswereaskedtoexplainwhytheyhadmadethatchoice of gender, so this variable is not considered further. The
inone ortwosentences. Therewerefour suchtrialsinthe modelthatbestfitthedataispresentedinTable3.
experiment, eachinvolvingadifferentphenomenon andits The primary result from these analyses is the significant
accompanyingexplanations. main effect of condition: Subjects found it more difficult
Afterthesecondofthefourtrials,subjectsengagedinan todeterminewhichwasthegoodexplanationintheMixed
attention check. Following methods recommended in Op- condition where only the bad explanation contained neu-
penheimer,Meyvis&Davidenko(2009),thischeckingtrial roscience (Figure 2). Specifically, subjects were signifi-
presentedanotherphenomenonandtwoexplanationssothat cantly more likely to select the good explanation in either
itlookedsuperficiallyliketheotherfourtrials.Forthistrial, theWithNeuroscience(72.0%oftrials)orWithoutNeuro-
weincluded instructionsattheendofthephenomenon de- science(63.5%oftrials)conditionsthanintheMixedcon-
scriptiontellingsubjectstochoosethe“equal”optionandto dition(51.7%oftrials).
write that they had done so as their justification. As noted
There was also a significant Group x Condition interac-
above, we excluded seven subjects who failed this check
tion. Follow-up regressions conducted on each group sep-
by selecting a different option. At the end of the survey,
arately showed that the MTurk workers were significantly
subjects provided the same demographic information as in
morelikelytoselectthegoodexplanationintheWithNeu-
Study1: age,gender,andeducationlevel.
roscienceconditionascomparedtotheMixedcondition(β
=0.58,p<.05),buttherewasnosignificantdifferencebe-
3.2 Results tween the Without Neuroscience and Mixed conditions (β
= 0.15, p = .59). For undergraduate subjects, subjects se-
To analyze the data, we conducted a mixed-effects logistic lected the good option significantly more often in both the
regressionpredictingwhethersubjectsselectedthegoodex- With Neuroscience (β = 1.68, p < .001) and Without Neu-
planationoneachtrial,consideringthisresponseascorrect roscience(β =1.14,p<.001)conditionsthanintheMixed
and the other two responses (selecting the bad explanation condition. Thus,althoughthemaineffectofconditionwas
orthe“bothareequal”option)asincorrect. Themodelin- significant in the whole sample, it was driven more by the
cluded random intercepts by subject. We tested effects of undergraduatesubjectsthanbytheMTurkworkers. Finally,
Item,Group(MTurkworkersorundergraduates),Condition although there were significant differences between items,
(WithNeuroscience,WithoutNeuroscienceorMixed),and asinStudy1,therewasnotasignificantItemxConditionor
possible interactions. Preliminary analyses found no effect ItemxConditionxGroupinteraction. Therefore,theeffect
JudgmentandDecisionMaking,Vol.10,No.5,September2015 Seductiveallureofneuroscienceexplanations 435
frontallobeswithoutexplainingwhatisactuallyhappening
Figure 2: Average number of trials on which subjects se-
isbullshit.”
lectedthegoodexplanationinStudy2. Errorbarsrepresent
Eachsubjectwasgivenascore(outof4)forthenumber
95%confidenceintervalsaroundthemeans. Thedottedline
ofpositivebrain-basedjustificationstheygave. A2(Group:
representschanceperformancesinceselectingthegoodex-
MTurkworkers,undergraduates)x2(Condition:WithNeu-
planationwasoneofthreepossibleresponsesoneachtrial.
roscience,Mixed)ANOVArevealedonlyasignificantmain
effectofGroup, F(1,166) =7.08, p<.01, η2 =.04: Under-
Study 2
graduates (M = 0.62, SD = 0.84) were overall more likely
4 than MTurk workers (M = 0.31, SD = 0.66) to refer to the
MTurk Workers
Undergraduates brainasaddingvaluetoanexplanation. Therewasneither
was selected 3 aCosnigdnitiifiocnanintteerffaecctitoonf. Condition nor a significant Group x
which good explanation 2 3Wof.3haepnsaDysckihseocdluotgosiscciaholonpohseenboemtweneeonn,gsouobdjeacntsdsbealedcetexdpltahneagtiooonds
Average # of trials on 1 eawpxloshps.iilctTahinvhaeestlusiyeob.njreeTcastsausklbtetseenincnodtgonegmdfierottmhoreesrrs,auatbtethjiesetcfhsyteesi’ngrgreoasootuidnnltgstehsxdefperlmomamnaoajnotSisrottiurtnyadstyeomf1t,othriraine-t
subjects understand the difference between good and bad
explanations.
0
Mixed Without Neuroscience With Neuroscience The one exception to this conclusion is the Mixed con-
dition, in which the bad explanation contained irrelevant
neuroscienceinformationandthegoodexplanationdidnot.
Here, subjects were seduced by the presence of neuro-
science information, which made them less likely to pre-
ofConditionandtheConditionxGroupinteractionwerenot
fer the good explanations than in the other conditions. Al-
significantlydifferentacrossitems.
though the main effect of condition was significant in the
To gain further insight into subjects’ responses, for the full sample, it was driven primarily by the undergraduates.
two conditions that used explanations containing neuro- Indeed, the undergraduates’ justifications were more likely
science language (With Neuroscience and Mixed), we per- to mention neuroscience in a positive light. This suggests
formedatextsearchonsubjects’justificationsforwordsre- thatthepresenceofneuroscienceinformationplayedakey
latedtoneuralprocessesgenerally(“brain”,“lobe,”,“scan”, role in convincing these subjects that the bad explanations
“neur*”)andforthespecificneurosciencetermsusedinthe were satisfactory. As the justifications quoted above illus-
explanationsthemselves(“premotor,”“cortex”);58%ofjus- trate,someundergraduatesappearedtorelyonthepresence
tifications inthesetwoconditions contained atleastone of ofneuroscienceasaheuristictojudgethequalityofanex-
these terms. A research assistant, who was blind to condi- planation.
tion,group,andstudyhypotheses,furthercodedthesejusti- Itisnotentirelyclearwhyundergraduateswouldbemore
ficationsforwhethersubjectsreferredtothebrainasadding attracted to explanations containing neuroscience informa-
valuetoanexplanation. Forexample,“thisgivesabiologi- tionthanMTurkworkers,althoughcurrentlylearningabout
calexplanationandusesbrainpartstoexplain”,“brainscans psychological and neuroscientific phenomena might have
andtimingseemmoreaccurate”,and“it’smoreindepthand swayedtheundergraduatestolendmoreweighttothepres-
seems more factual because it is talking about brain parts enceofneuroscience. Onepossibilityisthat,sincetheycur-
and stuff.” Of justifications that referenced the brain, 84% rentlyarelearningaboutthefunctionsofthebrain,reading
didsointhispositiveway.Theremainingjustificationssug- aboutspecificphenomenainwhichthebrainappearstoplay
gested a good grasp of the irrelevance of this information, a causal role leads them to judge explanations with neuro-
suchastheseundergraduates: “Idonotwanttohearabout scienceexplanationsmorefavorably.Insupportofthisargu-
what the brain is doing. I am interested in WHY the phe- ment, students who were currently taking a course on neu-
nomenon occurs in a more general sense” and “Saying be- roscience(Weisbergetal.,2008,Study2)showedastronger
cause of frontal lobe areas is not a sufficient explanation, attractiontoexplanationscontainingirrelevantneuroscience
but just states where the processing is occurring.” Or, as information than students recruited from the introductory
oneMTurkworkerputit,“Talkingmumbojumboaboutthe psychologypool(Weisbergetal.,2008,Study1).
JudgmentandDecisionMaking,Vol.10,No.5,September2015 Seductiveallureofneuroscienceexplanations 436
Finally, subjects’ performance in this study can start to gon,thentheNeurosciencePlusJargonexplanationsshould
explain one of the item effects seen in Study 1. Specifi- be rated more highly than the other two, which should not
cally, in Study 1, the seeing/imagining item (Item 4) was differ. Finally, there might be an additive effect of jargon
judged similarly regardless of quality or presence of neu- andneurosciencelanguage,inwhichcasetheNeuroscience
roscience. Here in Study 2, about twice as many subjects Plus Jargon explanations would be rated more highly than
chosethe“bothareequal”optionforthisitemthanforthe theSimpleNeuroscienceexplanations,whichwouldinturn
other three, indicating that they could not see a difference beratedmorehighlythantheWithoutNeuroscienceexpla-
between the good and bad explanations for this item. The nations.
differencebetweenthetwoversionswasveryslight,chang-
ingonly“usesthesameprocess”to“resultsinthesamear-
4.1 Method
rayofresponses”.Infact,anumberofparticipantsexplicitly
statedintheirjustificationsthatthetwoexplanationsseemed Subjects. Thefinalsampleforthisstudyincluded88un-
thesame(e.g., “Bothexplanationssoundliketheyaresay- dergraduates (63 female, 25 male; mean age = 19.5 years,
ingthesamething”,“Bothexplanationsaresimilarandsay range=18–22)and82MTurkworkers(42female,38male,
thesamethingjustinaslightlydifferentway.”) twounreported;meanage=35.0years,range=19–67). As
inpreviousstudies,theundergraduateswererecruitedfrom
thepsychologysubjectpoolattheUniversityofPennsylva-
4 Study 3 nia and received course credit for their participation. The
MTurk workers were recruited from Amazon’s system and
Havingdeterminedthatlengthdoesnotunderlietheseduc- werepaid20centsfortheirparticipation. Anadditional50
tive allure effect, and that neuroscience information is ef- subjects completed the survey but were excluded from the
fective at disguising bad explanations, Study 3 tested an- final analyses for failing an attention check (described be-
otherpossiblereasonthatneuroscienceexplanationsareap- low; 22MTurkworkersand28undergraduates). Although
pealingtosubjects,namelythatthisinformationtendstoin- more subjects failedthe attention check here than inStudy
cludetechnicaljargon. Ifsubjectsareattractedtoscientific- 2, thedesignofthisstudywasdifferentandmayhavepre-
soundingterms,thenneuroscienceinformationperseisnot sentedalessengagingtaskthanStudy2,andthesenumbers
seductive;subjects’responsescanbeinfluencedbythepres- areinlinewithotherstudiesthatincludedsimilarattention
enceofanyjargon. However,technicallanguageandrefer- checks(seeOppenheimeretal.,2009).
encestothebrainwereconfoundedintheexplanationsused
thusfar,preventingusfromdeterminingwhetheranyrefer- Design. Thisstudyuseda2(Group: MTurk,undergradu-
ence to the brain would be sufficient or whether technical ate)x2(Neuroscience:SimpleNeuroscience,Neuroscience
jargonisnecessary. Study3constructedalternativeversions Plus Jargon) x 2 (Quality: good, bad) design. Group and
oftheseexplanationsinordertoteaseoutwhichtypeofin- Neuroscience were between-subjects variables and Quality
formationisresponsiblefortheseductiveallureeffect. was a within-subjects variable.3 Subjects were assigned to
This study’s design mirrored that of Study 1, in which either the Neuroscience Plus Jargon condition (42 MTurk
subjectsreaddescriptionsofpsychologicalphenomenaone workersand44undergraduates)ortheSimpleNeuroscience
at a time and then rated one explanation of each phe- condition(40MTurkworkersand44undergraduates). Data
nomenon. InStudy3,theseexplanationscamefromoneof from the 45 MTurk workers and 45 undergraduates in the
two sets: Simple Neuroscience, in which the explanations Without Neuroscience-Short condition from Study 1 were
referred to brain scans and neural processes but in simple alsousedhereforcomparison.
languagewithoutreferencetospecificbrainareas,andNeu-
roscience Plus Jargon, in which the explanations included Materials. Toconstructthestimuli,weusedthesamefour
technical terms to refer to the type of brain scan used and psychologicalphenomenaasinStudies1and2,andmodi-
theindividualareasofthebrain. Inbothcases, thesestim- fiedtheWithNeuroscience-Shortexplanationstofitthenew
uliwereconstructedbymodifyingtheShortversionsofthe conditions(seesupplementalmaterials).Explanationsinthe
stimuliusedinStudy1,andtheShort-WithoutNeuroscience SimpleNeuroscienceconditionremovedreferencestospe-
conditionfromthatstudyservesasacontrolconditionhere. cific brain-scanning techniques or brain areas and replaced
Thissetofstimuliallowsustotestamongthreehypothe- themwithgenericterms(“brainscans”,“visualarea”). Ex-
ses. If neuroscience information alone is responsible for planations in the Neuroscience Plus Jargon condition en-
theseductiveallureeffect,weshouldexpectsimilarratings hanced existing references to include as much specific jar-
for the Neuroscience Plus Jargon and the Simple Neuro-
science explanations, both of which should be rated more 3AsinStudy1,therandomizationalgorithmledto11subjectsreceiving
eithergoodexplanationsoneverytrialorbadexplanationsoneverytrial.
highlythantheWithoutNeuroscienceexplanations. Ifneu-
Theinclusionofthesesubjectsdidnotaffectanyanalyses,sotheywere
roscienceinformationappealsbecauseitcontainsfancyjar- leftinthesample.
JudgmentandDecisionMaking,Vol.10,No.5,September2015 Seductiveallureofneuroscienceexplanations 437
gon as possible (“fMRI scans”, “parietal lobe”). Each ex-
Table4: Study3mixed-effectslinearregressionmodel
planation had a good and a bad version, and this modified
(∗p<.05).
informationwasexactlythesameinbothversions.
Predictor Estimate [95%CI] t
Procedure. All subjects filled out an online survey on
Intercept 0.02 [–0.19,0.24] 0.17
Qualtrics. As in Study 1, for each of the four trials, sub-
jectsfirstreadadescriptionofoneofthefourpsychological Item2 0.33 [0.06,0.62] 2.19∗
phenomena, which appeared on the screen for 10 seconds Item3 –0.91 [–1.24,–0.62] –5.97∗
beforetheywereallowedtoadvance. Onthesecondscreen, Item4 0.98 [0.69,1.25] 6.44∗
this phenomenon description appeared again, followed by Neuroscience 0.27 [0.05,0.51] 2.44∗
one of the four possible explanations of the phenomenon,
Group 0.15 [0.05,0.27] 2.64∗
accordingtothesubject’sassignedcondition;whetherthey
saw the good or bad version of the explanation was ran- Quality 0.57 [0.35,0.78] 5.39∗
domly determined on each trial (as described above in the NeurosciencexItem2 –0.37 [–0.68,-0.07] –2.42∗
Designsection). Subjectswereaskedtoratehowsatisfying NeurosciencexItem3 –0.04 [–0.36,0.27] –0.23
theyfoundtheexplanationona–3(veryunsatisfying)to+3
NeurosciencexItem4 –0.05 [–0.38,0.28] –0.35
(verysatisfying)scale. Theywerethenaskedtojustifytheir
QualityxItem2 0.12 [–0.18,0.44] 0.82
ratinginoneortwosentences.
QualityxItem3 –0.39 [–0.67,–0.07] –2.62∗
As in Study 2, after the first two trials, we included an
attentioncheck. Thisattentioncheckpresentedanotherde- QualityxItem4 –0.57 [–0.85,–0.25] –3.83∗
scription of a psychological phenomenon and an explana-
Note:Thisregressionpredictedsubjects’ratingsofthequal-
tion for it in exactly the same way as the other trials, ex-
ityoftheexplanations.Theinterceptrepresentstheexplana-
cept that the last sentence of the explanation told subjects
tionsthatdidnotcontainneuroscience,undergraduatesub-
toselect3onthescale. Asnotedabove, 50subjectsfailed
jects, and bad explanations. Item is deviation coded, such
thisattentioncheck(byfailingtoselect3and/orbydemon-
thatthecoefficientforeachlevelrepresentsdeviationfrom
stratingalackofattentivenessintheirjustificationsforthis
thegrandmean;Item1isthereferencelevel.
item)andarenotincludedinouranalyses. Attheendofthe
survey, subjects responded to the same basic demographic
questionsasinStudies1and2,reportingtheirage,gender,
absent),andQuality(goodorbad)andtheirinteractions;the
andhighestlevelofeducation.
modelthatbestfitthedataisshowninTable4.
The analysis revealed significant main effects of Group,
4.2 Results Quality, and Neuroscience (Figure 3). MTurk subjects (M
= 0.33, SD = 1.90) gave higher overall ratings than under-
In order to have a control condition with which to com-
graduatesubjects(M=0.03,SD=1.89),andsubjectsrated
pare the current subjects’ responses, the analyses for this
goodexplanations(M =0.52,SD=1.86)morehighlythan
studyadditionallyincludetheresponsesfromthesubjectsin
bad explanations (M = –0.19, SD = 1.88), replicating the
theWithoutNeuroscience-ShortconditionfromStudy1(45
results of Study 1. The significant main effect of Neuro-
MTurkworkersand45undergraduates). Preliminaryanaly-
science indicates that explanations were rated more highly
ses revealed no effects of gender, so it was not included in
inthetwoconditionsthatusedneurosciencelanguage(M=
ouranalyses.
0.29, SD = 1.95) than in the Without Neuroscience condi-
As in Study 1, we conducted a mixed-effects linear re-
tion(M=–0.03,SD=1.79). Therewasnosignificanteffect
gression analysis. The model included random intercepts
ofJargon, meaningthattheNeurosciencePlusJargoncon-
by subject as well as random slopes by subject for the
dition(M =0.29,SD=1.91)wasnotsignificantlydifferent
effect of Quality (the only within-subjects variable). We
fromtheothertwocombined(M=0.12,SD=1.89).
created two dummy variables to examine the effects of
As in Study 1, the effects of Neuroscience and Quality
neuroscience and jargon; the Neuroscience variable coded
alsovariedbyitem,asindicatedbysignificantinteractions.
whether neuroscience information was present (the Simple
Toexaminetheseinteractions,weconductedseparatelinear
Neuroscience and Neuroscience Plus Jargon conditions) or
regressionsforeachitemexaminingmaineffectsofGroup,
absent(theWithoutNeurosciencecondition). Similarly,the
Neuroscience, Length, and Quality. The results are sum-
Jargonvariablecodedwhetherjargonwaspresent(theNeu-
rosciencePlusJargoncondition)orabsent(theSimpleNeu- marized in Table 5. Although the magnitudes (and there-
roscience and Without Neuroscience conditions). The re- fore significance levels) of the effects varied by item, only
gressiontestedeffectsofItem,Group(MTurkorundergrad- two effects were not in the predicted directions; consistent
uates),Neuroscience(presentorabsent),Jargon(presentor with Study 1, Item 2 had a negative, non-significant effect
JudgmentandDecisionMaking,Vol.10,No.5,September2015 Seductiveallureofneuroscienceexplanations 438
Figure3: AverageratingsofexplanationqualityinStudy3, Table5:Regressioncoefficientsforindividualitemanalysis
including the Without Neuroscience condition from Study inStudy3(∗p<.05).
1. Errorbarsrepresent95%confidenceintervalsaroundthe
Item1 Item2 Item3 Item4
means.
Neuroscience 0.27∗ –0.09 0.24∗ 0.22∗
Study 3
Quality 0.55∗ 0.69∗ 0.17 –0.01
3 Group 0.04 0.32∗ 0.18 0.08
Good Explanations
Bad Explanations
2
4.3 Discussion
1
Explanations with neuroscience information, whether pre-
Rating sentedassimplyaspossiblewithoutjargonorwithreference
Average 0 tiosfsypinegcitfihcanneeuxrpallatneacthionniqsuwesithanodutbnreauinroasrceiaesn,cweeirnefomrmoraetisoant-.
However, there was no difference between the two neuro-
−1
scienceconditions. Thissuggeststhatanyreferencetoneu-
roscience is sufficient to cause the effect, and that adding
−2 technicaljargondoesnotincreasesubjects’ratings.
Additionally, subjects judged good explanations more
−3 highlythanbadones,andundergraduatesubjectsgaveover-
Without Neuroscience Simple Neuroscience Neuroscience Plus Jargon all lower ratings than MTurk workers. These effects repli-
cate the findings of Study 1 and suggest two additional
conclusions. First, people can generally discriminate good
frombadexplanations. Second,participatinginresearchas
partofone’seducationalexperienceseemstomakesubjects
ofNeuroscience,andItem4hadanegative,non-significant
moreskepticaloverall,butdoesnoteliminatetheseductive
effectofQuality.
allureeffect.
Toanalyzesubjects’justifications,wesearchedforallref- These main effects were significant overall, but varied
erencestoneuroscience,asinStudy2;24%ofalljustifica- somewhatbyitem. AsinStudy1,theattentionalblinkitem
tionsreferencedthebrain,and58%ofthosedidsoinapos- (Item2)didnotshowaneffectofneuroscienceinformation
andtheseeing/imaginingitem(Item4)didnotshowanef-
itivemanner. A2(Group: MTurkworkers,undergraduates)
fect of quality. In the case of the former, as noted above,
x 2 (Condition: Simple Neuroscience, Neuroscience Plus
the neuroscience information was contained in a separate
Jargon)ANOVArevealednosignificanteffectsonsubjects’
sentenceratherthanbeingdirectlylinkedtotheexplanatory
averagenumberofpositivebrain-basedjustifications.
information. This may have made it easier for subjects to
However, subjects’ justifications provide further insight realize that this information was irrelevant. In the case of
into one of the item effects. Many of the justifications for latter, as in Studies 1 and 2, subjects seemed generally un-
the attentional blink item (Item 2), which did not show abletotellthedifferencebetweenthegoodandbadversions
a significant neuroscience effect, mentioned that the neu- ofthisitem.
roscience information seemed unconnected with the phe-
nomenon: e.g., “The explanation mentions the frontal lobe
5 General discussion
areas but does not really say how the areas relate to atten-
tional blink”, and “The explanation does not explain how
The seductive allure effect of neuroscience, first observed
frontalareasarerelatedtothetemporalrelationshipbetween
by Weisberg et al. (2008), occurs when subjects judge that
the two houses.” This item effect was consistent with the
explanations for psychological phenomena (especially bad
findings from Study 1, and justifications such as this sup-
ones) that contain irrelevant neuroscience information are
port our suggestion that putting the information about the
betterthanexplanationsthatdonot.Thecurrentstudiespro-
frontallobeinaseparatesentencemayhavemadeiteasier videnewinsightintowhythiseffecthappens.
for subjects to separate this information from the body of First, although the original stimuli used to demon-
theexplanation,explainingthelessereffectofneuroscience strate this effect confounded neuroscience information
forthisitem. with length, our Study 1 and independent replications by