Full Terms & Conditions of access and use can be found athttp://www.tandfonline.com/action/journalInformation?journalCode=hpli20

Psychological InquiryAn International Journal for the Advancement of Psychological Theory

ISSN: 1047-840X (Print) 1532-7965 (Online) Journal homepage: http://www.tandfonline.com/loi/hpli20

On the Prioritized Processing of Threat in a DualImplicit Process Model of Evaluation

David S. March, Lowell Gaertner & Michael A. Olson

To cite this article: David S. March, Lowell Gaertner & Michael A. Olson (2018) On the PrioritizedProcessing of Threat in a Dual Implicit Process Model of Evaluation, Psychological Inquiry, 29:1,1-13

To link to this article: https://doi.org/10.1080/1047840X.2018.1435680

Published online: 21 Mar 2018.

Submit your article to this journal

View related articles

View Crossmark data

ORIGINAL ARTICLE

On the Prioritized Processing of Threat in a Dual Implicit Process Model of Evaluation

David S. March, Lowell Gaertner, and Michael A. Olson

Department of Psychology, University of Tennessee, Knoxville, Tennessee

ABSTRACTDual-process models of attitudes distinguish between implicit and explicit processes in which the valence(i.e., positivity or negativity) of a stimulus influences judgments and behavior toward the stimulus.Developing parallel to the dual-process literature has been a threat detection literature suggesting thatthe mind is preferentially attuned to threats to immediate bodily harm. That literature reveals earlyprivileged responses (e.g., shorter latency of detection, stronger reflexive reactions, and faster andstronger physiological responses) to threatening stimuli relative to negative, neutral, and positive stimuli.By integrating those literatures, we develop the dual implicit process model that postulates twofunctionally distinct and serially linked automatic processes in which an implicit threat process precedes(and potentially influences) an implicit valence process (positive vs. negative), which precedes (andpotentially influences) explicit processes. In addition to explicating the nature of the model, we examineinsights it offers various research areas, and conclude by identifying open questions regarding the model.

Lions and tigers, and bears. Oh my! – Dorothy, The Wizard of Oz

Imagine two experiences: (a) You were once terrorized by aknife-wielding funhouse clown and now automatically recoilwhenever you see one; (b) you have a strong automatic dislikeof mimes. Existing dual-process models of evaluation (e.g.,Fazio & Olson, 2014; Gawronski & Bodenhausen, 2006) wouldconceptualize both as implicit negative evaluations. We arguethat the two reactions are the product of distinct evaluative pro-cesses, both of which can be activated quickly and withoutintention yet have unique causes and consequences, and likelydifferent analogues in the brain. We propose, in particular, thatyour reaction to the clown is the result of an implicit threatevaluation that is functionally distinguishable from an implicitnegative evaluation responsible for your reaction to the mime.In what follows, we review existing dual-process models of eval-uation and a threat detection literature that developed parallelto (but in isolation of) the former. We then consider limitationsof dual-process approaches given the threat literature, andintroduce the dual implicit process model (DIPM), which inte-grates key aspects of both literatures. Finally, we examineinsights that the DIPM offers various research areas, and weconclude with open questions about the model.

Dual-Process Models of Evaluation

Dual-process models propose that evaluation occurs throughboth automatic (i.e., implicit, spontaneous) and controlled (i.e.,explicit, deliberate) processes. Automaticity implies that aprocess functions with at least one of the four characteristicsof being unintentional, effortless, uncontrollable, or beyondawareness (Bargh, 1994; Greenwald & Banaji, 1995). In terms

of latency, automatic processes are relatively fast and controlledprocesses are relatively slow. Such a dual-process frameworkhas been applied quite broadly to account for a variety ofpsychological phenomena including attitude–behaviorcorrespondence (Fazio, 1990), attribution (Trope, 1986), deci-sion making (Sloman, 1996), impression formation (Brewer,1988; Fiske & Neuberg, 1990), persuasion (Chaiken, 1980; Petty& Cacioppo, 1986), and prejudice (Devine, 1989; Gaertner &Dovidio, 1986; Pettigrew & Meertens, 1995). A number ofdual-process models of evaluation have been developed thatdiffer in regard to operating principles (i.e., mechanisms bywhich stimuli are processed) and operating conditions (i.e., cir-cumstances in which the mechanisms function). We describethe current state of several dual-process models with a focus onthe aspects of each that are relevant to our argument that threatprocessing is distinct from other forms of automatic evaluativeprocessing.

The Motivation and Opportunity as Determinants Model

The Motivation and Opportunity as DEterminants (MODE)model (Fazio, 1990; Fazio & Olson, 2014) proposes that atti-tudes can be expressed in judgments and behavior throughspontaneous and/or deliberative processes. From the perspec-tive of the MODE model, the starting point for evaluativeresponding is whether an attitude of sufficient strength is acti-vated automatically upon perception of an attitude object.Spontaneous processing occurs automatically, without inten-tion, and can influence downstream judgments and behavior.Deliberative processing occurs downstream, is effortful, andmay attempt to steer judgments and behaviors away from theimplications of spontaneously activated evaluations given

CONTACT David S. March dmarch@vols.utk.edu Department of Psychology, University of Tennessee, Knoxville, TN 37996, USA.

Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/hpli.© 2018 Taylor & Francis Group, LLC

PSYCHOLOGICAL INQUIRY2018, VOL. 29, NO. 1, 1–13https://doi.org/10.1080/1047840X.2018.1435680

ample motivation and opportunity to do so. But, if the sponta-neously activated evaluation is sufficiently strong, it can biasinitial stages of processing that have downstream consequenceson deliberative perceptions, judgments, behaviors—in terms ofguiding attention toward the object (Roskos-Ewoldsen & Fazio,1992) and construal of it (Smith, Fazio, & Cejka, 1996). Motiva-tion and opportunity together determine whether the expressedjudgment or behavior mainly reflects spontaneous and/ordeliberate processing. Motivation refers to the desire to engagein deliberative processing, whether in a general pursuit of accu-racy, or in pursuit of a specific conclusion (e.g., to avoid preju-dice). Opportunity refers to the capacity in a given situation toengage in deliberative processing; some situations, like thoserequiring fast reactions, limit the capacity to deliberatively con-trol responses. Similarly, fatigue, intoxication, and other factorsthat limit cognitive capacity also limit opportunity. In theabsence of motivation or opportunity, the spontaneous evalua-tion of the object in question is the main determinant of judg-ments and behavior. With both motivation and opportunitypresent, evaluative judgments and behaviors are more likely toreflect deliberately held motives rather than automatically acti-vated attitudes.

For instance, imagine a chocolate lover encountering achocolate bar. A strong positive preexisting attitude towardchocolate would lead one to automatically categorizea chocolate bar as something delicious to consume. If thechocolate lover, however, were on a diet (motivated toavoid empty calories) and fully rested and sober (able tocontemplate calories), the chocolate lover would likelyabstain. On the other hand, if the dieting chocolate loverwere hungry, intoxicated, or stressed, (i.e., a compromisedcapacity to think), the spontaneously positive evaluationwould likely lead to consumption of the chocolate.

For the present purposes, it is important to note that theMODE model considers automaticity of the attitude along acontinuum defined only by an object-evaluation association,that is, the association in memory between an object and itssummary evaluation. It does not consider other factors thatmight determine spontaneity of activation. Regardless of thenature of the object, attitudes characterized by stronger object-evaluation associations are more likely to have such propertiesof automaticity as spontaneous activation of the evaluation inresponse to perceiving the object, attitude-consistent construalof the object, and spontaneous approach/avoidance behaviortoward the object.

The Associative-Propositional Evaluation Model

The associative-propositional evaluation (APE) model(Gawronski & Bodenhausen, 2006; 2007) is focused on operat-ing principles (i.e., characteristics of the mechanism operating)in contrast to the MODE model’s focus on operating conditions(i.e., when a mechanism is operating). The APE model arguesthat evaluation manifests through associative and propositionalprocesses, with the former generally being automatic and thelatter being controlled (Gawronski & Bodenhausen, 2009,2014). The associative process links objects and events throughcontiguity (i.e., environmental or spatiotemporal co-occur-rence; e.g., evaluative conditioning; Jones, Olson, & Fazio,

2010). The activation of associative processes occurs via featuresimilarity (e.g., “that looks like chocolate”) and is experiencedas automatic affect. Propositional reasoning is more effortfuland includes truth/falsehood tags by weighing the validity ofthe activated associations using both preexisting informationand contextual cues.

Because associative and propositional processes glean infor-mation from different sources, they may not always agree.Whatever “momentarily considered” propositions are activatedin response to the associative input determines whether thoseassociations are accepted or rejected. Such acceptance orrejection depends on principles of cognitive consistency. Forexample, if the proposition is congruent with the activatedassociation, the associative evaluation will likely influenceevaluative judgments. But, if propositional reasoning divergesfrom the associative evaluation, the association is rejected as abasis for evaluative judgments (and possibly behavior). Becausepropositional (i.e., correction) processing occurs last, it is theoperation most vulnerable to constraints on attention, time, orcognitive resources (analogous to the “opportunity” factor ofthe MODE model).

Continuing the chocolate example, chocolate is often por-trayed as delicious, and repeated co-occurrence of chocolateand delicious would yield an associatively formed positive eval-uation of chocolate. Yet a chocolate-loving dieter’s positiveautomatic associative response to chocolate is inconsistent withthe momentarily considered propositional belief that chocolateis high in calories. Similar to MODE model logic, whetherpropositional reasoning overrules the associative link betweenchocolate and delicious is influenced by operating conditions.If one has both motivation (e.g., to lose weight) and opportu-nity, the propositional concern with weight gain (e.g., “I believethat eating that delicious chocolate would be inconsistent withmy weight loss goals”) would override the positive associationsactivated in response to the chocolate. Lacking either (e.g., ifstarving or intoxicated) would likely lead to judgments reflect-ing the initial association.

Like the MODE model, the APE model does not make dis-tinctions about the nature of the evaluative associations thatproduce automatic evaluative responses. All that matters aretheir associative strength.

The Quad Model

The quad model distinguishes between four processes ofevaluation that involve the automatic versus controlled dis-tinction: activation, guessing, discrimination, and overcom-ing bias (Conrey, Sherman, Gawronski, Hugenberg, &Groom, 2005). The activation (AC) process activates exist-ing evaluative associations in response to objects encoun-tered in the environment. The guessing (G) process drivesresponses in the absence of other information, often result-ing in response biases (i.e., a default response tendency in agiven context, such as to respond with approach ratherthan avoidance). The discrimination (D) process usesexplicit information from previous experience or the envi-ronment to determine possible outcomes, and the overcom-ing bias (OB) process monitors for the appropriate responseand attempts to suppress automatically activated evaluative

2 D. S. MARCH ET AL.

associations or response tendencies. These processes are notpurely automatic or controlled. Although AC and G usuallyfunction automatically, G can function more deliberatively.Conversely, although D and OB usually function delibera-tively, they can acquire automatic features.

For instance, imagine that a chocolate lover receiveswhat appears to be chocolate but it is actually the terrible-tasting chocolate substitute carob. A preexisting positiveautomatic evaluation of chocolate is activated (AC) becausethe chocolate lover’s default response toward anythingresembling chocolate is to assume (G) that it is indeedchocolate. But, having been tricked into eating carob in therecent past, the perceiver is motivated to overcome the ini-tial bias (OB) to ensure that they discriminate (D) what thecandy bar actually is, delicious chocolate or terrible carob,before eating it.

Like the preceding models, the QUAD model is agnosticregarding what sorts of objects might become associated withautomatic evaluation. It does not make distinctions about thenature of the associations within the AC and G processes.

Other Relevant Models

Dual-attitudes models (e.g., Wilson, Lindsey, & Schooler, 2000)propose that attitudes can form both implicitly and explicitlytoward the same object and coexist as distinct mental entities.This implies the possibility of evaluative dissociation, with theimplicit attitude and explicit attitudes having opposite valence.

Dual-system models locate automatic and controlled pro-cesses, or what these models refer to as reflexive versusreflective (Lieberman, Gaunt, Gilbert, & Trope, 2002) orintuitive versus rule based (Sloman, 1996, 2014), in distinctmental systems with unique neural substrates. As withdual-attitudes models, these dual-systems approaches sug-gest that the automatic and controlled processes can yielddistinct attitudes of opposite valence (McConnell & Rydell,2014). Downstream reflective (rule-based) processing mayinhibit the reflexive (intuitive) system similarly to theMODE model.

Resuming the chocolate example, upon an initial enjoyableexposure to chocolate the automatic/intuitive/reflexive systemmay form a positive attitude toward chocolate. After learningabout the high caloric content of chocolate, the controlled/rule-based/reflective system forms a second negative evalua-tion. Dual-attitudes and dual-systems models agree that thepositive implicit attitude and negative explicit attitude nowcoexist, but dual-attitudes models assume one system isresponsible for both evaluations, whereas dual-systems modelsplace each evaluation within the confines of a unique system.Whether the automatic positive evaluation of chocolate is con-strained by the controlled system is again dependent on theMODE model’s operating conditions (i.e., motivation andopportunity).

In summary, a number of dual-process models of evaluationhave been proposed. They differ in detail but share the com-mon distinction between automatic and controlled evaluativeprocessing. And they do not make functional distinctionsbetween different kinds of automatic or implicit evaluative

processing. As we discuss in subsequent sections, this is a limi-tation that the literature on threat processing brings to light.

Threat Processing

Threat is a broadly employed concept in psychological research,such as threat to identity (Steele, 1997), ingroup status andresources (Stephan & Stephan, 2000), and self-esteem (Bau-meister, Smart, & Boden, 1996). Anchored in an evolutionaryframework, the threat- (i.e., fear-) processing literature hasfocused exclusively on threats to immediate bodily harm, whichis our focus here. Ancestors who reacted more quickly whenconfronted with such threats were more likely to survive thanwere their slower conspecifics (Blanchette, 2006), which islikely why humans overestimate the threat relevance of stimuliin ambiguous situations (Haselton & Buss, 2000; Nesse, 2005).A psychology of threat perception ostensibly evolved as anadaptive mechanism for the rapid avoidance of physical danger(e.g., Schaller & Neuberg, 2012).

€Ohman has made the case for such an adaption in the formof a hypothetical “fear module” responsible for processing andinitiating reaction to threat (€Ohman & Mineka, 2001, 2003).Such processing ostensibly utilizes neural circuitry that evolvedprior to the cortices, enabling it to function semi-independentof (i.e., in parallel to) non-threat-relevant mental processes(Calder, Lawrence, & Young, 2001). Consistent with such apossibility is what LeDoux (1996, 2012) referred to as the “lowroad,” a subcortical pathway to the amygdala capable of detect-ing threat and activating associated responses without explicitprocessing (in contrast to the “high road,” which is slower butprovides more processed, i.e., cortical, information). Such a lowroad allows for rapid responses after only superficial processingof the stimuli, whereas the high road undertakes more compre-hensive processing before instigating a response (Morris,€Ohman, & Dolan, 1999).

Although the amygdala is involved in the general processingof affective and motivationally relevant information (includingnovel and extremely positive stimuli; Cunningham & Brosch,2012), it is particularly attuned to the initial processing ofthreatening information (Campeau et al., 1991; Cunningham,Packer, Kesek, & Van Bavel, 2009; Lewis, Critchley, Rotshtein,& Dolan, 2007; Pascoe & Kapp, 1985). The amygdala may thusinitiate responses to stimuli and activate associated processesbefore neocortical structures have received, interpreted, andresponded to the same information. Neuroanatomical researchof the human brain supports such a “low road” capable ofdetecting and evaluating emotionally salient information andinitiating responses without explicit awareness (Garrido,Barnes, Sahani, & Dolan, 2012; Garvert, Friston, Dolan, & Gar-rido, 2014; Whalen et al., 2004; cf. Pessoa & Adolphs, 2010).

As we review, research is consistent with the possibility thathumans inherited an ability to preferentially process threats toimmediate survival—an effect that has been dubbed the threat-superiority effect (Blanchette, 2006; E. Fox & Damjanovic,2006). Stimuli such as lions, snakes, sharks, and weapons posean imminent threat to survival and necessitate quick detectionand avoidance. Clearly, such threatening stimuli are negative.But not all negative stimuli are threatening. Important to note,the existing research indicates that the mind preferentially

THREAT IN A DUAL IMPLICIT PROCESS MODEL 3

processes threatening stimuli, not simply negative stimuli—thisis a point to which we return. Furthermore, the preferentialresponse to threatening stimuli relative to negative, neutral,and positive stimuli occurs in terms of very early responses,such as shorter latency of detection, stronger reflexive reactions,and faster and stronger physiological responses, all of whichunderlie the presumed adaptive value of quickly detecting andreacting to survival threats.

Consistent with the possibility that threatening stimuli arepreferentially processed is research indicating that people morequickly detect threatening stimuli (e.g., snake, spider, gun,knife) than innocuous stimuli (e.g., flowers, mushrooms;Blanchette, 2006; Brosch & Sharma, 2005; E. Fox, Griggs, &Mouchlianitis, 2007; €Ohman, Flykt, & Esteves, 2001) and arequicker to detect angry faces than happy or sad faces (Foxet al., 2000; €Ohman et al., 2001; €Ohman, Lundqvist, & Esteves,2001). Eye-tracking research has shown that initial attention ismore frequently drawn to threatening images than to positiveor neutral images (Hermans, Vansteenwegen, & Eelen, 1999;Rinck & Becker, 2006). Research using continuous flash sup-pression (which prevents awareness of stimuli for multiple sec-onds) suggests that misattribution of affect occurs for angry butnot happy faces (Almeida, Pajtas, Mahon, Nakayama, & Cara-mazza, 2013) and threat-relevant faces emerge from suppres-sion and into consciousness more quickly than do neutral orhappy faces (Yang, Zald, & Blake, 2007).

Similarly, research suggests that both supraliminally andsubliminally presented threatening stimuli elicit unique physio-logical responses (Knight, Waters, & Bandettini, 2009; Morris,€Ohman, & Dolan, 1998; Whalen et al., 1998). For example, sub-liminally presented angry faces increase skin conductanceresponses while happy faces do not (Esteves, Dimberg, &€Ohman, 1994), subliminally presented fearful faces increaseamygdala activity relative to happy faces (Whalen et al., 1998),and classical conditioning occurs to subliminal angry but nothappy faces (Esteves, Parra, Dimberg, & €Ohman, 1994). Also,participants conditioned to supraliminal angry faces continueto show evidence of conditioning (i.e., increased skin conduc-tance) after the aversive shock is removed (i.e., during extinc-tion), but show no lingering conditioning to happy or neutralfaces (Dimberg & €Ohman, 1996). Supraliminally presenteddirectly threatening stimuli elicit stronger and faster amygdalaresponses than do other types of negative stimuli (Kveragaet al., 2014) and supraliminally presented threatening stimulielicit earlier cortical activity than do other types of negative andpositive stimuli (Costa et al., 2014). The amygdala also showsmore activity to subliminally presented fearful than neutralfaces independent of fusiform facial area activity, suggestingthat such activation is independent of attention (Vuilleumier,Armony, Driver, & Dolan, 2001)

Although nondiagnostic of a threat versus negative differ-ence, similar patterns are reported in studies that collapsedacross threatening stimuli (e.g., guns) and nonthreatening-neg-ative stimuli (e.g., attributes such as rude). People, for example,are better able to identify the presence and categorize thevalence of subliminally presented threatening/negative wordsthan neutral or positive words (Dijksterhuis & Aarts, 2003).Subliminally presented threatening/negative but not positivestimuli potentiate startle-eyeblinks relative to neutral stimuli

(Reagh & Knight, 2013) and event-related brain potentials arelarger during evaluative categorization of supraliminally pre-sented threatening/negative stimuli than positive or neutralstimuli (Ito, Larsen, Smith, & Cacioppo, 1998). When attentionis distracted from the emotional content of stimuli, amygdalaactivity is heightened in response to threatening/negative stim-uli relative to positive and neutral stimuli (Straube, Pohlack,Mentzel, & Miltner, 2008).

Our own research provides direct evidence that threateningstimuli, not simply negative stimuli, elicit preferential responses(March, Gaertner, & Olson, 2017). To distinguish threateningstimuli and nonthreatening-negative stimuli, we pilot testedimages to obtain four stimulus categories: threatening, non-threatening-negative, positive, and neutral. The nonthreaten-ing-negative category consisted of stimuli that are unpleasantbut do not pose an immediate survival threat (e.g., rotten teeth,excrement, sick or dead animals, maggots). Important to note,both the threatening and nonthreatening-negative stimuli wereevaluated as negative but only the former were evaluated ashighly threatening. All four categories were equivalent in redvalue and luminance. We subsequently employed those stimuliin three paradigms that assess very early responses: visualsearch, eye-tracking, and startle-eyeblink. Consistent with theargument that threat is preferentially processed, participants(a) were faster to detect a threatening than nonthreatening-neg-ative image when each was embedded among positive or neu-tral images; (b) oriented their initial gaze more frequentlytoward threatening than nonthreatening-negative, positive, orneutral images; and (c) evidenced larger startle-eyeblinks tothreatening than to nonthreatening-negative, positive, or neu-tral images.

Hence the threat-processing literature yields robust evidenceof a threat-superiority effect that manifests as faster and stron-ger perceptual, physiological, and behavioral reactions to stim-uli that pose an immediate survival threat. Important to note,this preferential processing occurs in regard to early reactionsand does not imply that it occurs at a longer or more protractedtime course. Indeed, the functional utility of the effect isostensibly rooted in the survival value of avoiding imminentdanger. Notably, this early preferential processing of threat car-ries an interesting and important implication for dual-processmodels of evaluation.

Limitation of Dual Process

Dual-process models are valuable for understanding the differ-ences between and interplay of implicit versus explicit evalua-tion and have established those processes as being serial innature. Implicit processing is associative, intuitive, and reflex-ive, and as such it is fast and efficient. Explicit processing iscontrolled, propositional, deliberative, and reflective, and assuch it is slow, effortful, and downstream of implicit processing.In terms of dual-process models, the fast and early responses ofthe threat-superiority effect constitute an implicit response.Existing dual-process models, however, cannot explain oraccount for threat superiority. From the perspective of thesemodels, the automaticity of an implicit response is thought tooccur equally for positively and negatively evaluated stimuli.The MODE model, for example, conceptualizes automaticity of

4 D. S. MARCH ET AL.

evaluative responses solely in terms of the strength of theobject-evaluation association and does not qualify that tenantin terms of valence or threat potential. Similarly, the APEmodel discusses activated associations between objects andaffect without specifying further the nature of thoseassociations. In short, these models do not articulate a speed orstrength difference between evaluations of different valences.The threat literature, on the other hand, indicates a qualitativedistinction in implicit processing that existing dual-processmodels do not capture. Threatening stimuli are evaluated nega-tively, as are nonthreatening-negative stimuli, but threateningstimuli systematically elicit faster and stronger responses thando other negative or positive stimuli. Here the mind is queuednot just to valence but particularly to threat.

When juxtaposed with dual-process models, the threatliterature implies the possibility of two serially linked implicit pro-cesses. One implicit process, evaluative threat processing, isattuned to survival threats such that objects are evaluated implic-itly as to whether they pose an immediate survival threat. Theother implicit processes, evaluative valence processing, is attunedto valence such that objects are evaluated implicitly as to whetherthey are good or bad. In what follows, we integrate those processesinto a DIPM and demonstrate the conceptual value of distinguish-ing implicit threat processing and implicit valence processing.

The Dual Implicit Process Model

The DIPM postulates two functionally distinct and seriallylinked automatic evaluative processes: the first implicit process(i1) is solely oriented toward threats to bodily harm. This pro-cess precedes and potentially influences the subsequent implicitprocess (i2) that encompasses the full evaluative continuum(negative to positive). As we define it, i2 aligns well withnotions of automatic and implicit evaluative responses in thedual-process models we reviewed earlier. Both i1 and i2 precedeand potentially influence subsequent explicit (“e”) processing(see Figure 1). Based on findings that threatening stimuli arepreferentially processed via the low-road (e.g., LeDoux, 1996),the DIPM purports that i1 processing occurs first to activate

threat responses, given proper input. The slower operatinghigh-road processes nonthreatening stimuli via i2 to providemore detailed evaluative responses. Given that i2 and e func-tioning are consistent with extant dual-process models, wefocus primarily on explicating i1. The following sections detailthe inputs to i1, outputs of i1, and influences on the magnitudesof the outputs, and then puts it all together in the full temporalsequence of evaluative processing.

Inputs to i1

€Ohman (€Ohman et al., 2001; €Ohman & Mineka, 2001,2003) proposed the idea of threat sensitivity in terms of aneural “fear module,” which implies the possibility thatthreat evaluation is queued uniquely to the phylogeneticstimuli (e.g., snakes, spiders) on which it evolved. To befunctional, however, a threat evaluation process shouldlearn and incorporate new threats (Gould & Vrba, 1982).Indeed, fear in humans can be learned both directlythrough firsthand experience and indirectly through sociallearning, with the amygdala playing a central role in bothinstances (Olsson & Phelps, 2007). Consistent with the pos-sibility of a flexible system, threat-superiority research indi-cates that ontogenetic threats (e.g., modern weapons) andphylogenetic threats (e.g., animals, fire) have the same func-tional effect of being preferentially processed (Blanchette,2006; Brosch & Sharma, 2005; E. Fox et al., 2007). In ourown research (March et al., 2017), for example, participants(a) were faster to detect both ontogenetic and phylogeneticthreats than negative stimuli; (b) oriented their initial gazemore frequently to both ontogenetic and phylogeneticthreats than to negative, positive, or neutral stimuli; and (c)evidenced larger startle-eyeblinks to both ontogenetic andphylogenetic threats than to negative, positive, and neutralstimuli. Hence, i1 is supported by a flexible system thatlearns and incorporates new objects as potential threats.

Outputs of i1

Implicit process i1 yields two interrelated outcomes: (a) imme-diate physiological reactions to protect/prepare the body fromthreat and (b) organized downstream information processingqueued to the threat. Both outcomes function to protect againstharm and ensure safety (e.g., via vigilance, active avoidancemechanisms; Bolles & Fanselow, 1980; Pratto & John, 1991;Wentura, Rothermund, & Bak, 2000).

Physiological outputs. The physiological outputs of i1 arethose typically associated with responding to threat (LeDoux,2014), including protective reflexes (e.g., freezing, flailing,defensive fighting), autonomic arousal (e.g., pupil dilation,heart rate increase, sweating), and neurological activity (e.g.,amygdala activation, secretion of epinephrine by the adrenalglands). Neurological actions prepare the body to respond tothreat by instigating overt behavioral movements and auto-nomic arousal meant to aid the body’s response to threat (L€ow,Weymar, & Hamm, 2015). Once initiated, such responses inev-itably run to completion with little opportunity for interruption(i.e., a fixed action pattern; Lorenz, 1965).

Figure 1. Diagram of the dual implicit process model (DIPM) showing the progres-sion from the first implicit process (i1) threat processing into the second implicitprocess (i2) evaluative and controlled/explicit processing and accompanyingdescriptive table.

THREAT IN A DUAL IMPLICIT PROCESS MODEL 5

Organized downstream processing. Activated i1 organizesdownstream information processing for a vigilant focus on thethreatening stimulus (e.g., Pratto & John, 1991). Such organiza-tion allocates attentional resources toward gathering furtherinformation about the threat and is enabled via neuronal path-ways between the amygdala, supplementary structures, andcortices as explicated by LeDoux (1996), €Ohman (2005) andothers (e.g., A. S. Fox, Oler, Tromp, Fudge, & Kalin, 2015).This allows for the possibility that i1 can negatively bias theevaluative process of i2 and e in a manner consistent with theactivated threat. Imagine, for example, three people: Donna dis-likes cats, Lisa likes cats, and Nancy is neutral toward cats. Ifeach were to encounter a cat in the absence of i1 activation,Donna’s automatic i2 evaluation would be negative, Lisa’swould be positive, and Nancy’s would be neutral. In contrast, ifeach were to encounter a hissing cat with exposed fangs andarched back, the now-activated i1 would negatively bias i2 forall three persons (relative to their own prior reaction): Donnawould have a faster and more negative evaluation, Lisa wouldhave a slower and less positive evaluation, and Nancy wouldhave a relatively fast and negative evaluation. Hence, i1 activa-tion influences the evaluative process of i2 and alters it fromwhat it would have been absent i1 activation.

There is also the possibility that i1 directly affects e (i.e.,independent of i2), whereby the activated threat is made appar-ent to e which then processes it directly. For example, (neutral)Nancy’s i1 response to the hissing cat promotes active attentionto the cat with deliberate avoidance (e.g., Nancy might say toherself, “I should avoid that nasty cat”). Alternatively, i1 mayindirectly affect e via i2. For example, Nancy’s i1 response tothe hissing cat promotes a negative implicit evaluation of thecat (i.e., i2), which in turn could promote a negative explicitevaluation of the cat and deliberate avoidance.

Magnitude of Outcomes Produced by i1

The magnitude of i1’s outcomes is determined by the threat’sperceived imminence and potency. Threat imminence refers tothe proximal distance of the threat in time and space, and, allelse being equal, the magnitude of the i1 response increaseswith imminence (Fanselow, 1994; Kveraga et al., 2014; L€ow,Lang, Smith, & Bradley, 2008; L€ow et al., 2015; Mobbs et al.,2007). For example, a threat at a greater distance would likelyelicit less sympathetic arousal and attentional resources thanwould a threat at a closer distance (L€ow et al., 2008). Threatpotency refers to the presumed capacity of the threat to inflictbodily harm (e.g., Lundqvist, Esteves, & €Ohman, 1999). That is,some threats (e.g., a roaring lion) are perceived to have greaterpotency to harm than other threats (e.g., hissing cat) and, allelse being equal, the magnitude of the i1 response increaseswith potency (Blanchard, Hynd, Minke, Tiffanie, & Blanchard,2001).

As depicted in Figure 2, the resulting magnitude of the i1response is conditioned simultaneously on imminence andpotency. When the threat is of high imminence and potency,the magnitude of the i1 response is maximized. For example,coming face-to-face with a bear in the woods would activate i1and yield strong physiological responses (e.g., release of epi-nephrine, elevated heartrate) and highly organized downstream

cognitive processing of the bear to maximize a protectivebehavioral response (fight or flight). In contrast, viewing a pho-tograph of a bear (or encountering the bear from a distance),for which perceived potency remains high but perceived immi-nence is reduced, would activate i1 with a weaker response(physiology, downstream cognition, behavior) than in the face-to-face encounter.

Finally, an interesting possibility to consider is the case of anambiguous threat, such as a creaking floorboard at midnight ora creepy-crawling movement across one’s back. Properties oferror management (Haselton & Buss, 2000) would lead to theoverperception of threat imminence and potency, which wouldyield a stronger i1 response than would have occurred withaccurate perception of the creaking as due to the wind ratherthan an intruder or the creepy crawling as due to a tree branchrather than a spider.

Putting It All Together

Evaluation absent i1 threat activation follows the evaluativeprocess explicated by current dual process models. Forinstance, imagine standing in a park when in the distance, younotice your friend approaching. At this point, an evaluation ofyour friend is activated. i2 activates a (presumably) summarypositive association, and likewise explicitly you are happy to seeyour friend. This chain of evaluative events does not include ani1 activation, and hence no i1 influence. The conditions laidout by current dual process models (e.g., MODE, APE) stipu-late that given the motivation and opportunity to control one’sresponses, the explicit positive attitude toward your friend isthat likely to be expressed. Alternatively, when a threat is pres-ent, the DIPM proposes an alternative process from that pro-posed by current dual process models. For example, imaginewaiting in that same park, but now instead of approachingfrom the front, your friend sneaks up behind you and scaresyou. Before consciously (e.g., e) identifying the threat, i1 is acti-vated, involving immediate behavioral and physiologicalresponses preparing you to respond to the threat but also a neg-ative biasing effect downstream. The i2 evaluation is now lesspositive than it would have been absent an i1 event, as is yourexplicit evaluation.

Figure 2. Relationship between threat potency and imminence in determiningmagnitude of the first implicit process (i1) response.

6 D. S. MARCH ET AL.

It may appear to the reader that we are merely tacking onthreat processing to dual-process models, and in a sense we are.However, as we argue next, i1 is likely to be active in a numberof domains, including those often considered within the con-text of attitudes research (e.g., prejudice), as well as in domainsinto which attitudes researchers are typically more reticent toventure (e.g., phobia, intimate partner violence), with implica-tions for evaluative responding across them all. Indeed, webelieve an important strength of the DIPM is its ability to inte-grate literatures with little cross-talk, highlighting similaritiesin evaluative processing across them, with implications forbehavioral responses as well as the potential for change.

Implications of the DIPM

By distinguishing implicit threat processing and implicitvalence processing, the DIPM provides a fuller understandingof evaluative processing and offers unique insights to variousfields of study. The following sections examine some of thoseinsights to prejudice, phobia, and intimate partner violence.

Implications for Prejudice

By distinguishing an implicit threat process from an implicitvalence process, the DIPM suggests that implicit biases towardsocial groups can be functionally distinguished in regard to threatversus automatic positive/negative evaluation. Black-danger stereo-types, for example, uniquely draw attention to Black faces whereasnon-threat-related but negative Black stereotypes do not (Donders,Correll, & Wittenbrink, 2008). Similarly, White participants showstronger amygdala activation to subliminally presented Black thanWhite faces (Cunningham et al., 2004). In these examples, threatcues ostensibly drive processing in a manner different than domerely negative (i.e., nonthreatening stereotypes) or positive (own-race face forWhite participants).

Perhaps even more interesting is the possibility that ingroupperception can yield both ingroup favoritism and outgroupfavoritism among groups associated with threat or aggression.Black participants, for example, show a shooter bias againstBlacks (Kahn & Davies, 2011) but also evidence ingroup favor-itism on other implicit measures (e.g., evaluative priming:Fazio, Jackson, Dunton, & Williams, 1995; Implicit AssociationTest: Olson, Crawford, & Devlin, 2009). Similarly, Middle East-ern participants are more likely to shoot Middle Eastern targetswearing traditional Saudi headgear than less traditional baseballhats (Schofield, Deckman, Garris, DeWall, & Denson, 2015).There has been much debate as to whether members of particu-lar marginalized groups, despite their low status, evince ingroupfavoritism or reflect society’s negative views of their group inthe form of outgroup favoritism (e.g., Jost, Gaucher, & Stern,2015; Olson et al., 2009). The DIPM offers a possible reconcilia-tion of those views by suggesting that members of marginalizedgroups—particularly those associated with aggression—mightsimultaneously show both automatic outgroup favoritism (viai1 threat responses to own-group members) and automaticingroup favoritism (via i2 positive responses to own-groupmembers).

We also see implications for the DIPM in terms of prejudicemeasurement. Specifically, the DIPM’s implications of separate

i1 and i2 responses in the domain of prejudice provide a frame-work for future research to uncover the extent to which a givenmeasure of prejudice assesses threat-related responses (as weearlier implied the shooter-bias might; Correll, Park, Judd, &Wittenbrink, 2002) versus i2-related evaluative responses (asthe Implicit Association Test and priming measures thatemploy only valence-relevant attribute items might; Fazio &Olson, 2003). This leads to the possibility of better predictionof discriminatory behavior (Guglielmi, 1999). As threatresponses and “mere” approach/avoidance responses likelylook different behaviorally, the i1/i2 distinction provides afiner-grained analysis of the sorts of behaviors that differentmeasures of prejudice are likely to predict. For example, a mea-sure more attuned to threat responses might better predict fear-ful emotions and avoidance of the target, whereas a measuremore attuned to valence might better predict anger or disgustand either approach or other information-seeking behaviors.

Finally, in line with contemporary theories of the functionalbases of prejudice (e.g., Schaller & Neuberg, 2012; Smith,1993), the DIPM suggests a functional distinction betweenmore physical-threat-based and more evaluative-based prejudi-ces (perhaps, e.g., in terms of violations of traditional values).Different interventions might be developed to address the spe-cific underlying bases of the prejudice based on this distinction.However, as discussed next, threat responses may be more diffi-cult to unlearn, suggesting that threat-based prejudices may bemore difficult to reduce.

Implications for Phobia

Phobias are the result of a dysfunctional predictive relationshipthat occurs when the level of threat associated with a stimulusbecomes catastrophic (Grillon, Pellowski, Merikangas, & Davis,1997). Some treatments appear able to reduce the magnitude offear responses, often extinguishing them entirely (Hermans,Craske, Mineka, & Lovibond, 2006). Yet, a return of fear (e.g.,renewal, reinstatement, spontaneous recovery) often occurs whenthe ostensibly extinguished fear response suddenly returns. TheDIPM suggests that phobia treatments appear to target i2 and e,with less influence on i1. In particular, the spontaneous return offear suggests that the phobic target remains associated with threatvia i1 despite becoming more favorably evaluated via i2 and e(Hermans et al., 2006). The temporary reduction of fear responsesis likely a highly controlled process through explicit behaviormod-ification supported by more favorable i2 and e evaluations thatattempt to suppress i1’s outputs.

The regulating ability of i2 and e is suggested by researchshowing that the more favorable peoples’ automatic andexplicit evaluations became of the fear inducing stimulus (i.e.,the effectiveness of treatment), the better was their ability tocontrol behavioral manifestations of fear (Huijding & de Jong,2007; Vasey, Harbaugh, Buffington, Jones, & Fazio, 2012).Further emphasizing that what is being effected is the regula-tion of fear, and not the unlearning of the association betweenthe target and threat, is research showing that spider phobiatreatment had a positive impact on evaluations of valence(good/bad) but had no impact on threat (danger/safety)evaluation (Teachman & Woody, 2003). Similarly, after under-going positive training that biases responses to emotional

THREAT IN A DUAL IMPLICIT PROCESS MODEL 7

situations, participants evaluations of emotional situationsbecame more positive, but such training had no impact on theirfearful responses to threatening stimuli (Teachman & Addison,2008). In lieu of therapy-induced change to i1’s evaluation,reduction of fear response may reflect an increased ability to (atleast temporarily) regulate fear responses. Return of fearimplies that i1 maintains at least some threat association to thephobic object even when phobic responses can be successfullysuppressed.

This research serves to raise an important issue aboutwhether it is possible to unlearn threat associations overextended time. Imagine, for example, a novice snake-handlerwith a deadly cobra. On the first few encounters, the novicewould likely experience a racing heart, sweaty palms, and pal-pable fear. Over time, however, absent fear-reinforcing events,the practiced snake-handler would likely experience less activa-tion of i1. Of interest, snakes and spiders elicit negative implicitevaluations from snake neophytes and spider neophytes andpositive implicit evaluations from snake experts and spiderexperts (Ellwart, Rinck, & Becker, 2006; Purkis & Lipp, 2007).Such a positive (i2) evaluation (and presumably positiveexplicit evaluation) might contribute to the unlearning of an i1input over time. Yet there remains an open question aboutwhether any evaluative association can be unlearned (Ebbing-haus, 1885/1964; Petty, Tormala, Bri~nol, & Jarvis, 2006), and itwould certainly be a survival benefit for threat associations tobe particularly inflexible to change. Indeed, recall researchshowing that conditioned angry faces evoke fear responses longafter the paired shock is removed but that happy and neutralfaces ceased evoking fear responses almost immediately (Dim-berg & €Ohman, 1996). This implies that even if it is possible tounlearn an association, those tied to survival value may lingermuch longer.

The amygdala is vital to conditioned fear (i.e., phobias;LaBar, Gatenby, Gore, LeDoux, & Phelps, 1998), meaning thatsuccessfully unlearning phobias may involve altering the amyg-dala’s ingrained threat associations. Therefore, phobia treat-ments aimed at making attitudes toward the object morepositive (via i2 and e) may affect i1 only indirectly as a result ofchanges in i2 and e evaluations. But important to note, changesin i2 and e are not diagnostic of a reduction in the i1 phobicevaluation. Phobias may be less susceptible to certain types ofcognitive therapy. Phobia treatments that alter the i1 evaluationmay therefore be more effective changing the long-termoutlook.

Implications for Intimate Partner Violence

Victims of intimate partner violence often stay in abusive rela-tionships for extended periods, and even after leaving oftenreturn to the abusive partner (Anderson et al., 2003; Schutte,Malouff, & Doyle, 1988). Victims often convey positive feelingsand attitudes toward their abuser, frequently even denying theoccurrence of abuse, and going so far as to defend their abuser(Wallace, 2007). Several theories have been put forth that focuson victim characteristics and logistical factors behind stay/leavedecisions (e.g., Rusbult & Martz, 1995; Strube, 1988). Theincongruence of victims’ positive self-reported attitudes andthe objective negativity of abuse is often explained though

traumatic bonding theories (i.e., Stockholm Syndrome) as asort of cognitive dissonance. Yet much is still unknown aboutthe processes that underlie victims’ stay/leave decisions (Bell &Naugle, 2005; Strube, 1988). In addition to the many factorsalready identified as predictors of stay/leave behavior, theDIPM would suggest that victims can hold positive i2 evalua-tions and concurrently associate their abuser with i1 threat.Acts of warmth or kindness intermittent with acts of abuseessentially create opposing i1 and i2/e evaluations. Decisions tostay may be partially explained by the abuse victim’s strongpositive i2 and e evaluation, effectively controlling/overcomingtheir i1 fear of their partner.

Indirect evidence comes from research showing that theamount of self-reported emotional attachment, feelings of love,loyalty, and the level of commitment toward the abusive part-ner all predict likelihood of remaining in an abusive relation-ship (Hayes & Jeffries, 2013). That is, the level of positive i2and explicit evaluation toward the abuser predicted victim stay/leave behaviors. Of interest, research has shown that increasedemotional abuse (ostensibly affecting i2) is more predictive of avictim’s decision to leave than is increased physical abuse(ostensibly affecting i1; Gortner, Berns, Jacobson, & Gottman,1997). This indicates that perhaps a negative change in i2 or eis the determining factor leading to a change in behavior. Thatis, positive i2 and e evaluations allowed the victim to controlhis or her i1 fear, but a weakened i2 or e may lessen that controland allow an increase in protective leave behaviors. More directevidence for i1/i2/e roles in stay/leave decisions awaits themore precise measurement of implicit constructs called for bythe DIPM.

Open Questions

There remain a number of open conceptual questions regardingthe DIPM that necessitate further discussion. We examine sev-eral such issues in the current section.

Summation of Successive i1 Events

Existing research suggests the possibility that the downstreamoutput of i1 is stronger when there are successive i1 eventsoccurring in short duration than when there is only a single i1event. That is, perhaps successive i1 events have a summative

Figure 3. Magnitude of the first implicit process (i1) response across time.

8 D. S. MARCH ET AL.

effect such that preceding i1 activation strengthens subsequenti1 activation (see Figure 3). Indirect evidence of this summativeeffect comes from the potentiating effect of ambient darkness(vs. ambient light) on both the startle-reflex (Grillon et al.,1997) and White participants’ expression of threat-relevantprejudice against Black targets (Schaller, Park, & Mueller,2003). In each instance, ambient darkness can be consideredthe initial i1 trigger, which strengthened (relevant to ambientlight) the subsequent startle-response to an unexpected noiseblast (second i1) and activation of threat-relevant (e.g., hostile)but not threat-irrelevant (e.g., lazy) stereotype ratings of Blacktargets (second i1; assuming Blacks are perceived as threaten-ing; see the Implication for Prejudice section). Stated otherwise,participants experiencing two threats (darkness and noise-blastor images of Black persons) evidenced larger threat responsesthan did participants experiencing a single threat response(light and noise-blast or images of Black persons). This suggeststhe summative nature of successive i1 activation.

Dual Implicit Processing in a Simple versus Complex World

In their now classic publication, Murphy and Zajonc (1993)demonstrated that both angry and smiling faces presented at4 ms yielded valance-consistent affect misattribution to a sub-sequent supraliminally presented neutral stimulus. Some read-ers might construe this finding as being inconsistent with theDIPM: If smiling faces trigger positive affect misattribution atonly a 4-ms exposure, how could there be room for earlierprocessing of a threatening stimulus? In our view, this poses noproblem for the DIPM. Murphy and Zajonc created a simpleworld in which participants were presented on any given trialwith a single stimulus (angry or smiling face) without competi-tion from other stimuli. In a complex world involving multipleco-occurring stimuli, the DIPM suggests that threatening stim-uli are processed preferentially, and, in the absence of threaten-ing stimuli, nonthreatening stimuli are processed directly viai2. So, Murphy and Zajonc is quite consistent with the DIPM.

Is i1 Less Susceptible Than i2 to Downregulation?

There is evidence that explicit goals (e) can affect the activationof implicit valence (i2). For example, Stewart and Payne (2008)demonstrated that an implementation intention to activatecounterstereotypic thoughts in response to a Black face reducedthe automatic activation of prejudice. Thus, e appears capableof downregulating i2. However, consistent with our precedingdiscussion of the phobia research, i1 may be less susceptible todownregulation. The imbalanced connections between theamygdala and the cortices (i.e., more afferent than efferent)implies that during the initial processing of a threatening stim-ulus, i1 informs i2 and e more than the reverse. This suggeststhat it might be difficult for any higher order processes (i.e., e)to impact i1. For example, explicit knowledge that you arewatching a horror movie does not stifle the threat responsewhen the axe murderer jumps out of the closet.

On the other hand, perhaps i2 and e have some potential todownregulate i1. The model posits that when encountering athreat, i1 activates and evokes the cascade of outcomes describedpreviously. But i2 and e process the same stimulus, and their

evaluations might “turn down” the i1 downstream biasing effecton information processing despite the continued physiologicalarousal. Imagine walking through the woods and hearing rus-tling sounds behind you. Initially, i1 evokes behavioral anddownstream outcomes that direct automatic attention to therustling. Upon further deliberate inspection, the rustling is real-ized to be the pattering of a baby rabbit. The identification of therabbit triggers an automatic positive (i2) evaluation. Althoughyour heart is still pounding and your palms are sweating, per-haps i1 no longer negatively biases information processing.

Indirect support of such downregulation comes from fMRI evi-dence of differential amygdala activity to Black versus White faces(with the assumption that Black faces constitute a threat to manyWhite participants). Although supraliminal presentation yieldedequivalent fMRI amygdala activation to Black and White faces(Cunningham et al., 2004; Hart et al., 2000), (a) amygdala habitua-tion was slower (i.e., activity remained heightened longer) to Blackthan White faces (Hart et al., 2000) and (b) subliminal presenta-tion yielded stronger amygdala activation to Black than Whitefaces (Cunningham et al., 2004). These results imply that the i1response to Black faces dominated at very short (i.e., subliminal)exposures that granted no opportunity for downregulation by i2or e. But, when images were presented supraliminally, though i1evinced a downstream influence (i.e., slower habituation), i2 and edownregulation of i1 was possible.

Cunningham et al. (2004) suggested that increased activityin the dorsolateral prefrontal cortex and anterior cingulate cor-tex (areas implicated in the control of regulatory responses)congruent with decreased amygdala activity reflect the suppres-sion of automatic amygdala activation when threats are con-sciously perceived. We supplement this interpretation bysuggesting that these results reflect the modulation of i1 overthe time course of processing. Interpreting these results is lim-ited by the poor temporal resolution of fMRI, which builds asummary image over multiple seconds. Were this same experi-ment able to take millisecond-precise measurements of amyg-dalar activity, it may have revealed that initial activity to Blackfaces was equal across sub- and supraliminal conditions andthat only over time (i.e., after i2 and explicit processing) wasamygdala activity dampened.

Individual Differences in i1 Processing

LeDoux (2015) argued that fear and anxiety are two manifesta-tions of the same phenomena—threat. The only difference isthat fear is a reaction to a specific stimulus, whereas anxietyoccurs toward something more diffuse. Research is starting toilluminate how individual differences may predispose certainpeople toward both fear and anxiety. If so, this may imply thatsome people have heightened i1 activity. A specific serotonintransporter gene polymorphism (5-HTTLPR) has been associ-ated with heightened sensitivity and reactivity to threat (Cheon,Livingston, Hong, & Chiao, 2014). People with one copy of theshort allele display greater amygdala activity toward threaten-ing stimuli (Munaf�o, Brown, & Hariri, 2008), which manifestsas heightened anxiety, vigilance, and fear conditioning (Canli &Lesch, 2007). In addition, individuals with posttraumatic stressdisorder show biased attention and quicker reaction to threat-ening stimuli (Block & Liberzon, 2016).

THREAT IN A DUAL IMPLICIT PROCESS MODEL 9

Indirect evidence that individual differences in i1responses lead to heightened fear-related outcomes comesfrom research showing that chronic anxiety may facilitatethreat associations, potentiate experiences of threat, andpredispose people to overperceive threat. For example,White, Ratcliff, Vasey, and McKoon (2010) showed thathigh- versus low-anxious individuals have a processingadvantage for threatening words on a lexical decision task.Pitman and Orr (1986) aversively conditioned high- andlow-anxious individuals to angry and neutral faces andfound that angry faces resisted extinction only among high-anxious participants, indicating that anxiety played a role inbuffering extinction. Bishop, Duncan, and Lawrence (2004)found that among low-anxious participants, amygdala activ-ity was reduced when instructed to attend to a house thana simultaneously appearing threatening face but that high-anxious participants showed greater overall amygdala activ-ity and no differences when attending to the house versusthe threatening face.

Cheon et al. (2014) proposed that those with an S-allele“may appraise or react to cues of outgroup threat differ-ently” such that the 5-HTTLPR “genotype may predisposeindividuals to experience more negative intergroup contactor perceptions of the environment, which may ultimatelyshape intergroup bias” (p. 1269). Perhaps individual differ-ences in (amygdala-mediated) i1 activation are likely to pre-dispose certain persons to develop specific phobias andexperience more chronic anxiety but may also make themmore likely to endorse certain group stereotype, particularlythose associated with threat.

Conclusion

We propose that automatic evaluation is made up of (at least)two distinct processes. First is an initial implicit processing ofwhether a stimulus poses a survival threat (i1). This process issensitive to threats engrained from the evolutionary past, suchas snakes and heights, as well as idiosyncratically learnedthreats unique to individual experience and social history, suchas guns and social groups. Second is a subsequent implicit proc-essing (i2) of the full evaluative spectrum (i.e., positive and neg-ative), which reflects extant dual-process models of evaluation.The DIPM’s integration of early threat-processing into implicitevaluation more fully captures the dynamics that that underliesocial cognition.

References

Almeida, J., Pajtas, P. E., Mahon, B. Z., Nakayama, K., & Caramazza, A.(2013). Affect of the unconscious: Visually suppressed angry facesmodulate our decisions. Cognitive, Affective, & Behavioral Neurosci-ence, 13, 94–101.

Anderson, M. A., Gillig, P. M., Sitaker, M., McClosky, K., Malloy, K., &Grigsby, N. (2003). “Why doesn’t she just leave?”: A descriptive studyof victim reported impediments to her safety. Journal of Family Vio-lence, 18, 151–155.

Bargh, J. A. (1994). The Four Horsemen of automaticity: Awareness, effi-ciency, intention, and control in social cognition. In R. S. Wyer, Jr. &T. K. Srull (Eds.), Handbook of social cognition, (pp. 1–40). Hillsdale,NJ: Erlbaum.

Baum, L. F. (1900). The Wizard of Oz. Chicago, IL. George M. HillCompany.

Baumeister, R. F., Smart, L., & Boden, J. M. (1996). Relation of threatenedegotism to violence and aggression: The dark side of high self-esteem.Psychological Review, 103, 5–33.

Bell, K., & Naugle, A. (2005). Understanding stay/leave decisions in violentrelationships: A behavior analytic approach. Behavior and Social Issues,14, 21–45.

Bishop, S. J., Duncan, J., & Lawrence, A. D. (2004). State anxiety modula-tion of the amygdala response to unattended threat-related stimuli.Journal of Neuroscience, 24, 10364–10368.

Blanchard, D. C., Hynd, A. L., Minke, K. A., Tiffanie, M., & Blanchard, R. J.(2001). Human defensive behaviors to threat scenarios show parallelsto fear- and anxiety-related defensive patterns of non-human mam-mals. Neuroscience and Biobehavioral Reviews, 25, 761–770.

Blanchette, I. (2006). Snakes, spiders, guns, and syringes: How specific areevolutionary constraints on the detection of threatening stimuli? TheQuarterly Journal of Experimental Psychology, 59, 1484–1504.

Block, S. R., & Liberzon, I. (2016). Attentional processes in posttraumaticstress disorder and the associated changes in neural functioning. Exper-imental Neurology, 284, 153–167.

Bolles, R. C., & Fanselow, M. S. (1980). PDR-a multi-level model of fearand pain. Behavioral and Brain Sciences, 3, 315–323.

Brewer, M. B. (1988). A dual process model of impression formation. In T.K. Srull & R. S. Wyer (Eds.), Advances in social cognition, (pp. 1–36).Hillsdale, NJ: Erlbaum

Brosch, T., & Sharma, D. (2005). The role of fear-relevant stimuli in visualsearch: A comparison of phylogenetic and ontogenetic stimuli. Emo-tion, 5, 360–364.

Calder, A. J., Lawrence, A. D., & Young, A. W. (2001). Neuropsychology offear and loathing. Nature Reviews Neuroscience, 2, 352–363.

Campeau, S., Hayward, M. D., Hope, B. T., Rosen, J. B., Nestler, E. J., &Davis, M. (1991). Induction of the c-fos proto-oncogene in rat amyg-dala during unconditioned and conditioned fear. Brain Research, 565,349–352.

Canli, T., & Lesch, K. P. (2007). Long story short: The serotonin trans-porter in emotion regulation and social cognition. Nature Neuroscience,10, 1103–1109.

Chaiken, S. (1980). Heuristic versus systematic processing and the use ofsource versus message cues in persuasion. Journal of Personality &Social Psychology, 39, 752–766.

Cheon, B. K., Livingston, R. W., Hong, Y. Y., & Chiao, J. Y. (2014). Gene xenvironment interaction on intergroup bias: The role of 5-HTTLPRand perceived outgroup threat. Social Cognitive and Affective Neurosci-ence, 9, 1268–1275.

Conrey, F. R., Sherman, J. W., Gawronski, B., Hugenberg, K., & Groom, C.J. (2005). Separating multiple processes in implicit social cognition:The quad model of implicit task performance. Journal of Personalityand Social Psychology, 4, 469–487.

Correll, J., Park, B., Judd, C. M., & Wittenbrink, B. (2002). The policeofficer’s dilemma: Using ethnicity to disambiguate potentiallythreatening individuals. Journal of Personality and Social Psychol-ogy, 6, 1314–1329.

Costa, T., Cauda, F., Crini, M., Tatu, M.-K., Celeghin, A., de Gelder, B., &Tamietto, M. (2014). Temporal and spatial neural dynamics in the per-ception of basic emotions from complex scenes. Social Cognitive andAffective Neuroscience, 9, 1690–1703.

Cunningham, W. A., & Brosch, T. (2012). Motivational salience: Amygdalatuning from traits, needs, values, and goals. Current Directions in Psy-chological Science, 21, 54–59.

Cunningham, W. A., Johnson, M. K., Raye, C. L., Gatenby, J. C., Gore, J. C.,& Banaji, M. R. (2004). Separable neural components in the processingof black and white faces. Psychological Science, 15, 806–813.

Cunningham, W. A., Packer, D. J., Kesek, A., & Van Bavel, J. J. (2009).Implicit measurement of attitudes: A physiological approach. In R. E.Petty, R. H. Fazio, & P. Bri~nol (Eds.), Insights from the new implicitmeasures (pp. 485–512). New York, NY: Psychology Press.

Cunningham, W. A., Preacher, K. J., & Banaji, M. R. (2001). Implicit atti-tude measures: Consistency, stability, and convergent validity. Psycho-logical Science, 12, 163–170.

10 D. S. MARCH ET AL.

DeSteno, D., Dasgupta, N., Bartlett, M. Y., & Cajdric, A. (2004). Prejudicefrom thin air: The effect of emotion on automatic intergroup attitudes.Psychological Science, 15, 319–324.

Devine, P. G. (1989). Stereotypes and prejudice: Their automatic and con-trolled components. Journal of Personality Social Psychology, 56, 5–18.

Dijksterhuis, A., & Aarts, H. (2003). On wildebeests and humans: The pref-erential detection of negative stimuli. Psychological Science, 1, 14–18.

Dimberg, U., & €Ohman, A. (1996). Behold the wrath: Psychophysiologicalresponses to facial stimuli.Motivation and Emotion, 20, 149–182.

Donders, N. C., Correll, J., & Wittenbrink, B. (2008). Danger stereotypespredict racially biased attentional allocation. Journal of ExperimentalSocial Psychology, 44, 1328–1333.

Ebbinghaus, H. (1885/1964). Memory: A contribution to experimental psy-chology. New York, NY: Dover

Ellwart, T., Rinck, M., & Becker, E. S. (2006). From fear to love: Individualdifferences in implicit spider associations. Emotion, 1, 18–27.

Esteves, F., Dimberg, U., & €Ohman, A. (1994). Automatically elicited fear:Conditioned skin conductance responses to masked facial expressions.Cognition & Emotion, 5, 393–413.

Esteves, F., Parra, C., Dimberg, U., & €Ohman, A. (1994). Nonconsciousassociative learning: Pavlovian conditioning of skin conductanceresponses to masked fear-relevant facial stimuli. Psychophysiology, 31,375–385.

Fanselow, M. S. (1994). Neural organization of the defensive behavior sys-tem responsible for fear. Psychonomic Bulletin & Review, 1, 429–438.

Fazio, R. H. (1990). Multiple processes by which attitudes guide behavior:The MODE model as an integrative framework. Advances in Experi-mental Social Psychology, 23, 75–109.

Fazio, R. H., Jackson, J. R., Dunton, B. C., & Williams, C. J. (1995). Vari-ability in automatic activation as an unobtrusive measure of racial atti-tudes: A bona fide pipeline? Journal of Personality and SocialPsychology, 69, 1013–1027.

Fazio, R. H., & Olson, M. A. (2003). Implicit measures in social cognitionresearch: Their meaning and use. Annual Review of Psychology, 1, 297–327.

Fazio, R. H., & Olson, M. A. (2014).The MODE model: Attitude-BehaviorProcesses as a Function of Motivation and Opportunity. In J. W., Sher-man, B. Gawronski, & Y. Trope (Eds.), Dual-process theories of thesocial mind, pp. 155–171. New York, NY: Guilford Press

Field, A. P. (2006). Is conditioning a useful framework for understandingthe development and treatment of phobias? Clinical Psychology Review,26, 857–875.

Fiske, S. T., & Neuberg, S. L. (1990). A continuum of impression forma-tion, from category- based to individuating processes: Influences ofinformation and motivation on attention and interpretation. In M. P.Zanna (Ed.), Advances in experimental social psychology (Vol. 23, pp.1–74). New York, NY: Academic Press.

Fox, A. S., Oler, J. A., Tromp, D. P., Fudge, J. L., & Kalin, N. H. (2015).Extending the amygdala in theories of threat processing. Trends inNeurosciences, 38, 319–329.

Fox, E., & Damjanovic, L. (2006). The eyes are sufficient to produce athreat superiority effect. Emotion, 6, 534–539.

Fox, E., Griggs, L., & Mouchlianitis, E. (2007). The detection of fear-rele-vant stimuli: Are guns noticed as quickly as snakes? Emotion, 7, 691–696.

Fox, E., Lester, V., Russo, R., Bowles, R. J., Pichler, A., & Dutton, K. (2000).Facial expressions of emotion: Are angry faces detected moreefficiently? Cognition & emotion, 14, 61–92.

Gaertner, S. L., & Dovidio, J. F. (1986). The aversive form of racism. In J. F.Dovidio & S. L.Gaertner (Eds.), Prejudice, discrimination, and racism(pp. 61–89). Orlando, FL: Academic Press.

Garrido, M. I., Barnes, G. R., Sahani, M., & Dolan, R. J. (2012). Functionalevidence for a dual route to amygdala. Current Biology, 22, 129–134.

Garvert, M. M., Friston, K. J., Dolan, R. J., & Garrido, M. I. (2014). Subcor-tical amygdala pathways enable rapid face processing. Neuroimage,102, 309–316.

Gawronski, B., & Bodenhausen, G. V. (2006). Associative and proposi-tional processes in evaluation: An integrative review of implicit andexplicit attitude change. Psychological Bulletin, 5, 692–731.

Gawronski, B., & Bodenhausen, G. V. (2007). Unraveling the processesunderlying evaluation: Attitudes from the perspective of the APEmodel. Social Cognition, 5, 687–717.

Gawronski, B., & Bodenhausen, G. V. (2009). Operating principlesversus operating conditions in the distinction between associativeand propositional processes. Behavioral and Brain Sciences, 32,207–208.

Gawronski, B., & Bodenhausen, G. V. (2014). The associative-proposi-tional evaluation model: Operating principles and operating conditionsof evaluation. In J. W. Sherman, B. Gawronski, & Y. Trope (Eds.),Dual-process theories of the social mind (pp. 188–203). New York, NY:Guilford Press

Gortner, E., Berns, S. B., Jacobson, N. S., & Gottman, J. M. (1997). Whenwomen leave violent relationships: Dispelling clinical myths. Psycho-therapy: Theory, Research, Practice, Training, 4, 343–352.

Gould, S. J., & Vrba, E. S. (1982). Exaptation—A missing term in the sci-ence of form. Paleobiology, 8, 4–15.

Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: Atti-tudes, self–esteem, and stereotypes. Psychological Review, 102, 4–27.

Grillon, C., Pellowski, M., Merikangas, K. R., & Davis, M. (1997). Darknessfacilitates the acoustic startle reflex in humans. Biological Psychiatry,42, 453–460.

Guglielmi, R. S. (1999). Psychophysiological assessment of prejudice: Pastresearch, current status, and future direction. Personality and SocialPsychology Review, 3, 123–157.

Hart, A. J., Whalen, P. J., Shin, L. M., McInerney, S. C., Fischer, H., &Rauch, S. L. (2000). Differential response in the human amygdala toracial outgroup vs. ingroup face stimuli. Neuroreport, 11, 2351–2354.

Haselton, M. G., & Buss, D. M. (2000). Error management theory: A newperspective on biases in cross-sex mind reading. Journal of Personalityand Social Psychology, 78, 81–91.

Hayes, S., & Jeffries, S. (2013). Why do they keep going back? Exploringwomen’s discursive experiences of intimate partner abuse. Interna-tional Journal of Criminology and Sociology, 2, 57–71.

Hermans, D., Craske, M. G., Mineka, S., & Lovibond, P. F. (2006).Extinction in human fear conditioning. Biological Psychiatry, 60,361–368.

Hermans, D., Vansteenwegen, D., & Eelen, P. (1999). Eye movement regis-tration as a continuous index of attention deployment: Data from agroup of spider anxious students. Cognition & Emotion, 13, 419–434.

Huijding, J., & de Jong, P. J. (2007). Beyond fear and disgust: The role of(automatic) contamination-related associations in spider phobia. Jour-nal of Behavior Therapy and Experimental Psychiatry, 38, 200–211.

Ito, T. A., Larsen, J. T., Smith, N. K., & Cacioppo, J. T. (1998). Negativeinformation weighs more heavily on the brain: The negativity bias inevaluative categorizations. Journal of Personality and Social Psychology,75, 887–900.

Jones, C. R., Olson, M. A., & Fazio, R. H. (2010). Chapter five—Evaluativeconditioning: The “How” question. Advances in Experimental SocialPsychology, 43, 205–255.

Jost, J. T., Gaucher, D., & Stern, C. (2015). “The world isn’t fair”: A systemjustification perspective on social stratification and inequality. In J.Dovidio & J. Simpson (Eds.), APA handbook of personality and socialpsychology (pp. 317–340). Washington, DC: American PsychologicalAssociation.

Kahn, K. B., & Davies, P. G. (2011). Differentially dangerous? Pheno-typic racial stereotypicality increases implicit bias among ingroupand outgroup members. Group Processes & Intergroup Relations,14, 569–580.

Knight, D. C., Waters, N., & Bandettini, P. A. (2009). Neural substrates ofexplicit andimplicit fear memory. Neuroimage, 45, 208–214.

Kveraga, K., Boshyan, J., Adams, R. B., Mote, J., Betz, N., Ward, N.,… Bar-rett, L. F. (2014). If it bleeds, it leads: Separating threat from mere nega-tivity. Social Cognitive and Affective Neuroscience, 10, 28–35.

LaBar, K. S., Gatenby, J. C., Gore, J. C., LeDoux, J. E., & Phelps, E. A.(1998). Human amygdala activation during conditioned fear acquisi-tion and extinction: a mixed-trial fMRI study. Neuron, 20, 937–945.

LeDoux, J. E. (1996). The emotional brain: The mysterious underpinnings ofemotional life. New York, NY: Simon and Schuster.

THREAT IN A DUAL IMPLICIT PROCESS MODEL 11

LeDoux, J. E. (2012). Rethinking the emotional brain. Neuron, 73, 653–676.LeDoux, J. E. (2014). Coming to terms with fear. Proceedings of the

National Academy of Sciences, 111, 2871–2878.LeDoux, J. E. (2015). Anxious: Using the brain to understand and treat fear

and anxiety. New York, NY: Penguin.Lewis, P. A., Critchley, H. D., Rotshtein, P., & Dolan, R. J. (2007). Neural

correlates of processing valence and arousal in affective words. CerebralCortex, 17, 742–748.

Lieberman, M. D., Gaunt, R., Gilbert, D. T., & Trope, Y. (2002). Reflexionand reflection: A social cognitive neuroscience approach to attribu-tional inference. Advances in Experimental Social Psychology, 34, 199–249.

Lorenz, K. (1965). Evolution and modification of behaviour. Chicago, IL:University of Chicago Press.

L€ow, A., Lang, P. J., Smith, J. C., & Bradley, M. M. (2008). Both predatorand prey: Emotional arousal in threat and reward. PsychologicalScience, 19, 865–873.

L€ow, A., Weymar, M., & Hamm, A. O. (2015). When threat is near, get outof here dynamics of defensive behavior during freezing and activeavoidance. Psychological Science, 26, 1706–1716.

Lundqvist, D., Esteves, F., & €Ohman, A. (1999). The face of wrath:Critical features for conveying facial threat. Cognition & Emotion,13, 691–711.

March, D. S., Gaertner, L., & Olson, M. A. (2017). In harm’s way: On pref-erential response to threatening stimuli. Personality and Social Psychol-ogy Bulletin, 43, 1519–1529. doi:10.1177/0146167217722558

March, D. S., & Graham, R. (2015). Exploring implicit ingroup and outgroupbias toward Hispanics.Group Processes & Intergroup Relations, 18, 89–103.

McConnell, A. R., & Rydell, R. J. (2014). The systems of evaluation model:A dual-systems approach to attitudes. In J. W. Sherman, B. Gawronski,& Y. Trope (Eds.), Dual process theories of the social mind (pp. 204–217). New York, NY: Guilford

Mobbs, D., Petrovic, P., Marchant, J. L., Hassabis, D., Weiskopf, N., Sey-mour, B., … Frith, C. D. (2007). When fear is near: Threat imminenceelicits prefrontal-periaqueductal gray shifts in humans. Science, 317,1079–1083.

Morris, J. S., €Ohman, A., & Dolan, R. J. (1998). Conscious and unconsciousemotional learning in the human amygdala. Nature, 393, 467–470.

Morris, J. S., €Ohman, A., & Dolan, R. J. (1999). A subcortical pathway tothe right amygdala mediating “unseen” fear. Proceedings of theNational Academy of Sciences, 96, 1680–1685.

Munaf�o, M. R., Brown, S. M., & Hariri, A. R. (2008). Serotonin transporter(5-HTTLPR) genotype and amygdala activation: A meta-analysis. Bio-logical Psychiatry, 63, 852–857.

Murphy, S., & Zajonc, R. (1993). Affect, cognition, and awareness: Affec-tive priming with optimal and suboptimal stimulus exposures. Journalof Personality and Social Psychology, 5, 723–739.

Nesse, R. M. (2005). Natural selection and the regulation of defenses: A sig-nal detection analysis of the smoke detector principle. Evolution andHuman Behavior, 26, 88–105.

€Ohman, A. (2005). The role of the amygdala in human fear: Automaticdetection of threat. Psychoneuroendocrinology, 30, 953–958.

€Ohman, A., Flykt, A., & Esteves, F. (2001). Emotion drives attention:Detecting the snake in the grass. Journal of Experimental Psychology:General, 3, 466–478.

€Ohman, A., Lundqvist, D., & Esteves, F. (2001). The face in the crowdrevisited: A threat advantage with schematic stimuli. Journal of Person-ality and Social Psychology, 3, 381–396.

€Ohman, A., & Mineka, S. (2001). Fears, phobias, and preparedness:Toward an evolved module of fear and fear learning. PsychologicalReview, 108, 483–522.

€Ohman, A., & Mineka, S. (2003). The malicious serpent: Snakes as a proto-typical stimulus for an evolved module of fear. Current Directions inPsychological Science, 12, 5–9.

Olson, M. A., Crawford, M. T., & Devlin, W. (2009). Evidence for theunderestimation of implicit in-group favoritism among low-statusgroups. Journal of Experimental Social Psychology, 45, 1111–1116.

Olson, M. A., & Fazio, R. H. (2009). Implicit and explicit measures of atti-tudes: The perspective of the MODE model. In R. E. Petty, R. H. Fazio,

& P. Bri~nol (Eds.), Attitudes: Insights from the new implicit measures(pp. 19–63). New York, NY: Psychology Press.

Olsson, A., & Phelps, E. A. (2007). Social learning of fear. Nature Neurosci-ence, 10, 1095–1102.

Osgood, C. E., Suci, G. J., & Tannenbaum, P. H. (1957). The measurementof meaning. Urbana: University of Illinois Press.

Pascoe, J. P., & Kapp, B. S. (1985). Electrophysiological characteristics ofamygdaloid central nucleus neurons during Pavlovian fear condition-ing in the rabbit. Behavioural Brain Research, 16, 117–133.

Pessoa, L., & Adolphs, R. (2010). Emotion processing and the amygdala:From a ‘low road’ to ‘many roads’ of evaluating biological significance.Nature Reviews Neuroscience, 11, 773–783.

Pettigrew, T. F., & Meertens, R. W. (1995). Subtle and blatant prejudice inWestern Europe. European Journal of Social Psychology, 25, 57–75.

Petty, R. E., & Cacioppo, J. T. (1986). The Elaboration likelihood model ofpersuasion. In L. Berkowitz (Ed.), Advances in experimental socialpsychology (pp. 123–205). New York, NY: Academic Press.

Petty, R. E., Tormala, Z. L., Bri~nol, P., & Jarvis, W. B. G. (2006). Implicitambivalence from attitude change: An exploration of the PAST model.Journal of Personality and Social Psychology, 90, 21–41.

Pitman, R. K., & Orr, S. P. (1986). Test of the conditioning model of neuro-sis: Differential aversive conditioning of angry and neutral facialexpressions in anxiety disorder patients. Journal of AbnormalPsychology, 95, 208–213.

Pratto, F., & John, O. (1991). Automatic vigilance: The attention-grabbingpower of negative social information. Journal of Personality and SocialPsychology, 61, 380–391.

Purkis, H. M., & Lipp, O. V. (2007). Automatic attention does not equalautomatic fear: Preferential attention without implicit valence. Emo-tion, 7, 314–323.

Reagh, Z., & Knight, D. (2013). Negative, but not positive emotionalimages modulate the startle response independent of conscious aware-ness. Emotion, 13, 782–791.

Rinck, M., & Becker, E. S. (2006). Spider fearful individuals attend tothreat, then quickly avoid it: Evidence from eye movements. Journal ofAbnormal Psychology, 115, 231–238.

Roskos-Ewoldsen, D. R., & Fazio, R. H. (1992). On the orienting value ofattitudes: Attitude accessibility as a determinant of an object’s attrac-tion of visual attention. Journal of Personality and Social Psychology,63, 198–211.

Rudman, L. A., & Kilianski, S. E. (2000). Implicit and explicit attitudestoward female authority. Personality and Social Psychology Bulletin, 26,1315–1328.

Rusbult, C. E., & Martz, J. M. (1995). Remaining in an abusive relationship:An investment model analysis of nonvoluntary dependence. Personal-ity & Social Psychology Bulletin, 6, 558–571.

Rydell, R. J., & McConnell, A. R. (2006). Understanding implicit andexplicit attitude change: A systems of reasoning analysis. Journal of Per-sonality and Social Psychology, 91, 995–1008.

Schaller, M., & Neuberg, S. L. (2012). Danger, disease, and the nature ofprejudice(s). In J. Olson & M. P. Zanna (Eds.), Advances in experimen-tal social psychology (pp. 1–55). Burlington, VT: Academic Press.

Schaller, M., Park, J. H., & Mueller, A. (2003). Fear of the dark: Interactiveeffects of beliefs about danger and ambient darkness on ethnic stereo-types. Personality and Social Psychology Bulletin, 29, 637–649.

Schofield, T. P., Deckman, T., Garris, C. P., DeWall, C. N., & Denson, T. F.(2015). Evidence of ingroup bias on the shooter task in a Saudi sample.SAGE Open, 5, 1–6.

Schutte, N. S., Malouff, J. M., & Doyle, J. S. (1988). The relationshipbetween characteristics of the victim, persuasive techniques of the bat-terer, and returning to a battering relationship. Journal of SocialPsychology, 5, 605–610.

Sloman, S. A. (1996). The empirical case for two systems of reasoning.Psychological Bulletin, 119, 3–22.

Sloman, S. A. (2014). Two systems of reasoning, an update. In J. Sherman,B. Gawronski, & Y. Trope (Eds.), Dual-process theories of the socialmind. New York, NY: Guilford Press.

Smith, E. R. (1993). Social identity and social emotions: Toward new con-ceptualizations ofprejudice. In D. M. Mackie & D. L. Hamilton (Eds.),

12 D. S. MARCH ET AL.

Affect, cognition, and stereotyping (pp. 297–316). New York, NY:Academic Press

Smith, E. R., & DeCoster, J. (2000). Dual-process models in social andcognitive psychology: Conceptual integration and links to underly-ing memory systems. Personality and Social Psychology Review, 4,108–131.

Smith, E. R., Fazio, R. H., & Cejka, M. A. (1996). Accessible attitudes influ-ence categorization of multiply categorizable objects. Journal ofPersonality and Social Psychology, 5, 888–898.

Steele, C. (1997). A threat in the air: How stereotypes shape intellectualidentity andperformance. American Psychologist, 52, 613–629.

Stephan, W. G., & Stephan, C. W. (2000). An integrated threat theory ofprejudice. In S. Oskamp (Ed.), Reducing prejudice and discrimination(pp. 23–46). Hillsdale, NJ: Erlbaum.

Stewart, B. D., & Payne, B. K. (2008). Bringing automatic stereotypingunder control: Implementation intentions as efficient means ofthought control. Personality and Social Psychology Bulletin, 10,1332–1345.

Straube, T., Pohlack, S., Mentzel, H. J., & Miltner, W. H. (2008). Differen-tial amygdala activation to negative and positive emotional picturesduring an indirect task. Behavioural Brain Research, 191, 285–288.

Strube, M. J. (1988). The decision to leave an abusive relationship: empiri-cal evidence and theoretical issues. Psychological bulletin, 104, 236–250.

Teachman, B. A., & Addison, L. M. (2008). Training non-threatening interpre-tations in spider fear.Cognitive Therapy and Research, 32, 448–459.

Teachman, B. A., & Woody, S. R. (2003). Automatic processing in spiderphobia: Implicit fear associations over the course of treatment. Journalof Abnormal Psychology, 1, 100–109.

Trope, Y. (1986). Identification and inferential processes in dispositionalattribution. Psychological Review, 93, 239–257.

Vasey, M. W., Harbaugh, C. N., Buffington, A. G., Jones, C. R., & Fazio, R.H. (2012). Predicting return of fear following exposure therapy with an

implicit measure of attitudes. Behaviour Research and Therapy, 50,767–774.

Vuilleumier, P., Armony, J. L., Driver, J., & Dolan, R. J. (2001). Effects ofattention and emotion on faces processing in the human brain: Anevent-related fMRI study. Neuron, 30, 829–841.

Wallace, P. (2007). How can she still love him? Domestic violence and theStockholm Syndrome. Community Practitioner, 80, 32–34.

Wentura, D., Rothermund, K., & Bak, P. (2000). Automatic vigilance: Theattention-grabbing power of approach- and avoidance-related socialinformation. Journal of Personality and Social Psychology, 78, 1024–1037.

Whalen, P. J., Kagan, J., Cook, R. G., Davis, F. C., Kim, H., Polis, S., … &Johnstone, T. (2004). Human amygdala responsivity to masked fearfuleye whites. Science, 306, 2061.

Whalen, P. J., Rauch, S. L., Etcoff, N. L., McInerney, S. C., Lee, M. B., &Jenike, M. A. (1998). Masked presentations of emotional facial expres-sions modulate amygdala activity without explicit knowledge. TheJournal of Neuroscience, 18, 411–418.

White, C. N., Ratcliff, R., Vasey, M. W., & McKoon, G. (2010). Anxietyenhances threat processing without competition among multipleinputs: A diffusion model analysis. Emotion, 10, 662–677.

Wilson, T. D., Lindsey, S., & Schooler, T. Y. (2000). A model of dual atti-tudes. Psychological Review, 107, 101–126.

Wittenbrink, B., Judd, C. M., & Park, B. (2001). Spontaneous prejudice incontext: Cariability in automatically activated attitudes. Journal ofPersonality and Social Psychology, 81, 815.

Yang, E., Zald, D. H., & Blake, R. (2007). Fearful expressions gain preferen-tial access to awareness during continuous flash suppression. Emotion,7, 882–886.

Zanna, M. P., & Rempel, J. K. (1988). Attitudes: A new look at an old con-cept. In D. Bar-Tal & W. Kruglanski (Eds.), The social psychology ofknowledge (pp. 215–334). Cambridge, England: Cambridge UniversityPress.

THREAT IN A DUAL IMPLICIT PROCESS MODEL 13