Saturday, April 17, 2021

Many negativity biases, such as negativity dominance, might not be "biases" after all but reflect people's correct encoding of evaluative ecologies

Explaining Negativity Dominance without Processing Bias. Christian Unkelbach, Alex Koch, Hans Alves. Trends in Cognitive Sciences, April 16 2021. https://doi.org/10.1016/j.tics.2021.04.005

Abstract: In a recent study, Shin and Niv explain both negativity and positivity biases in social evaluations as a function of the  diversity and low frequency of events. We discuss why negative information is indeed more diverse and less frequent, and highlight the  implications beyond social evaluations. 

Keywords: valence asymmetriespositivity biasnegativity biasessocial evaluationscognitive-ecological models


Check also Why Good Is More Alike Than Bad: Processing Implications. Hans Alves, AlexKoch and Christian Unkelbach. Trends in Cognitive Sciences, 2017. http://dx. doi. org/10. 1016/j. tics. 2016. 12. 006


Why PositiveInformationIsMoreAlikeThanNegativeInformation


 the  questionremains ‘why’ this asymmetry exists and whe there it is a feature of the  cognitive

system or afeature of the  ecology (see OutstandingQuestions). the first positionfollowsfrom

 the  affective or motivationalpotential of evaluativei nformation (Box 1). Accordingly, confrontation with a negative stimulus elicitsnegativeaffect, whichtriggersdeeperprocessing, resultingina more differentiated mental representation. However, wearguethatpositiveinformation'shigher similarity isatrueproperty of the information ecology, independent of affectiveandmotivational influences. Our idea follows the  tradition of researchers like Brunswik [13], Lewin [14], and Garner [40], whoemphasized the  importance of the informationecology for psychological

processes. 


Our explanationforpositiveinformation's higher similarity builds on the well-documented

assumption thatvalenceisafunction of attribute extremity. Aristotle [41] already recognized

that desirablequalitiesaremodestqualitiesthat are framed by both the excess and defect. That is, a

positive rangeislocatedtoward the middle of agivenattributedimensionandissurroundedby

two negativerangestoward the twoends of the dimension. the reby, positivityisnon-extreme. 

This isapparentat the mostbasiclevel: Humanlifeispossibleonlywithinasinglerange of 

temperature, oxygenconcentration, solarradiation, andsoon. Formostphysicalandchemical

dimensions thatarerelevanttohumanlife the reisa ‘too little’ and a ‘too much’. While humans

can survivewithina ‘good’ temperature range, the ycanbothfreezeandburn. the same

principle appliestointernalbiologicalstatesasprominentlyexpressedin the concept of 

homeostasis [42, 43]. Asdiscussedearlier, the perceptualsystemfollows the sameprinciple, 

as the desirablerange of prototypicalityissurroundedbyvariousdeviationsfrom the same. 


Importantly, the range principle is also ubiquitous in psychological domains. GrantandSchwartz

[44] showed thatfor virtuallyalldimensions of humanattributes, the positiveordesirablerangeis

non-extreme. Even on attribute dimensions that seemingly have one positive and onenegative

pole, the positive range reaches inflection points at which its effects turn negative. Agreeable-

ness turns into conformity, conscientiousness into perfectionism, and courage into recklessness. Consequently, desirable personality profiles are thosethat are non-extreme, which is why

 the  correlationbetweenitemmeans of personalitytestsanditemdesirabilitytypicallyexceeds

r = 0. 80 [45]. Recentresearchhasshownthat the rangeprinciplealsounderlies the mental

representation of socialgroups [46, 47]. Thatis, likeablesocialgroupsarethosethatarenon-

extreme regarding their agency and their beliefs, while non-likeable groups are those that are

extreme on these dimensions.


Of course, somequalitiesmightbelinearlyrelatedtovalence. For instance, the amount of poison

in one'sblood, or the amount of traumatic experiences one has had. However, those are

exceptions to the rule that attribute dimensions typically host one (non-extreme) range and two

(extreme) negativeranges, constituting an inverted u-shaped relation between attribute value

and valence. Further more, the reverse pattern seems evenmoreunlikely andmaybe even never

occurs; that is, attribute dimensionsthat host one(non-extreme)negativerangeandtwo

(extreme) positive ranges (but see Outstanding Questions). 


Assuming that attribute dimensions typically host one positive range framed by two negative

ranges, it follows that positive information must be on average morealikethannegative

information. the possible maximum distance between the two negativeranges on a given

attribute dimension always exceeds the distance within the positiverange. While twopositive

stimuli necessarily have to lay within the same range, two negative stimuli can lay in two different

ranges on a given attribute dimensionthat are highly distant and therefore differentfrom each

other. For example, while two attractive men must display a height that lays within the same

desirable range, two unattractive men can either be too short or too tall, and thereby highly

different.


Figure 1 (Key Figure) illustrates the rangeprincipleinatwo-dimensional attribute space in which

proximity equals similarityin accordance with a geometric model of similarity [9]. the single

positive space emerges in the center (white square), surrounded byfourdistinctnegativespaces

(dark graysquares) andfourambivalentspaces(lightgraysquares). If one would randomly

sample pairs of positive and negative stimuliandlocate the min the attributespace, the positive

stimuli willbeonaveragelocatedclosertoge the rthan the negativestimuli. 


Preponderance of NegativeConcepts


Beyond explaining the similarityasymmetry, the rangeprincipleimpliesalargernumber of 

negative states, evenonasingleattributedimension(cf. Figure 1). Evidenceforthisimplication

comes from research showing that language includes more negative than positive concepts. For example, the  majority of wordsthatcanbeusedtodescribeapersonarenegative, which has

been shown for the  English and Germanlanguages [48–50]. the sameistrueforemotion-

related wordsingeneral, as the  ‘working emotionvocabulary’ in EnglishandSpanishwasfound

to includemorenegative(50%)thanpositive(30%)words [51, 52]. Ananalysis of English ‘verbs’

also revealedapreponderance of negativeoverpositivewords [53]. Itisuncleartowhatextent

 the  preponderance of negativewordsappliestolanguagesingeneralaswearenotaware of 

research investigating the number of positiveandnegativewordsinlanguageso the rthan

English, Spanish, andGerman. 


 the  impliedlargernumber of negativestatesisals of oundinhumans’ emotional response

repertoire. While differentresearchershaveproposeddifferent ‘basic emotions’, almostall

describe moredistinctnegativethanpositiveemotions [54]. Forexample, earlyconceptualiza-

tions byWilliamJames [55] included fear, grief, rage, andlove. Later, EkmanandFriesen [56, 57]

prominently identified anger, disgust, fear, sadness, andjoyasbasicemotions, andPanksepp

[58] described the psychobiologicalsystems of fear, rage, andpanicandanappetitiveexpec-

tancy system. Howpreponderance of negativeemotionsfollowsfrom the rangeprinciplecanbe

illustrated usingappraisal the ories of emotions [59]. Accordingly, positiveemotionsresultfrom

goal-congruent appraisals, while negativeemotionsresultfromappraisals of goalincongruence

[60]. Again, while goalcongruenceconstitutesasinglecondition, the rearemanydifferentways

for conditionstobegoalincongruent. Fromthisperspective, the manifoldness of the negative

emotional repertoiremirrors the greatdiversity of goal-incongruentconditions. 


In sum, the differentialsimilarity of positiveinformationandnegativeinformationmayfollowfrom

 the  proposedrangeprinciple. We believe this explanation is plausible and parsimonious, without

denying that the re might be o the r factorscontributingtothisasymmetryinsimilarity(see

Outstanding Questions). Besidesbeinganintriguingphenomenonbyitself,positiveinforma-

tion's highsimilaritymayserveasanexplanatoryconstructfordifferencesin the processing of 

positive andnegativeinformation, whichweaddressin the following. 


No comments:

Post a Comment