Fearful Temperament and the Risk for Child and Adolescent Anxiety: The Role of Attention Biases and Effortful Control

Abstract

Fearful temperament represents one of the most robust predictors of child and adolescent anxiety; however, not all children with fearful temperament unvaryingly develop anxiety. Diverse processes resulting from the interplay between automatic processing (i.e., attention bias) and controlled processing (i.e., effortful control) drive the trajectories toward more adaptive or maladaptive directions. In this review, we examine the associations between fearful temperament, attention bias, and anxiety, as well as the moderating effect of effortful control. Based on the reviewed literature, we propose a two-mechanism developmental model of attention bias that underlies the association between fearful temperament and anxiety. We propose that the sub-components of effortful control (i.e., attentional control and inhibitory control) play different roles depending on individuals’ temperaments, initial automatic biases, and goal priorities. Our model may help resolve some of the mixed findings and conflicts in the current literature. It may also advance our knowledge regarding the cognitive mechanisms linking fearful temperament and anxiety, as well as facilitate the continuing efforts in identifying and intervening with children who are at risk. Finally, we conclude the review with a discussion on the existing limitations and then propose questions for future research.

Anxiety disorders represent one of the most commonly occurring mental health problems in early development, with prevalence estimates ranging from 6.1% to 24% in childhood and 18.6% to 31.9% in adolescence (Briggs-Gowan , Horwitz, Schwab-Stone, Leventhal, & Leaf, 2000; Kroes et al., 2001; Merikangas et al., 2010). Anxiety disorders are associated with significant impairments in current and future functioning across different domains (Rapee, Schniering, & Hudson, 2009); thus, examining the risk factors leading to a trajectory of anxiety has gained substantial attention in the past few years.

We view fearful temperament as the foundation for potential anxiety disorders. Temperament is defined as individual differences in emotional reactivity and the self-regulation of that reactivity (Rothbart & Ahadi, 1994). Negative emotional reactivity (e.g., neuroticism, fear, anger) is involved in the etiology of child psychopathology (Muris & Ollendick, 2005; Nigg, 2006). Various types of fearful reactive temperament, such as shyness, behavioral inhibition, and fear, have been demonstrated as major early risk factors that directly and indirectly promote the formation of anxiety (Degnan & Fox, 2007). Shyness is often conceptualized as wariness/anxiety in the face of social novelty and perceived social evaluation (Buss, 1985). Behavioral inhibition is characterized by negative reactivity to novelty and social reticence (Degnan & Fox, 2007). Fear typically refers to the extent of negative affect, such as worry or uneasiness, in face of potentially threatening or painful situations (Rothbart, Ahadi, Hershey, & Fisher, 2001). Buss and colleagues (2011; Buss et al., 2013) highlight the importance of examining the amount of fear that is displayed in different contexts and the degree of the match between extent of fear and level of threat in the environment. This is because children identified with fearful temperament may not form a homogenous group. Buss introduced the term dysregulated fear to describe the phenomenon of showing a high level of fear in low threatening environments. Children with dysregulated fear can be regarded as having an extremely high level of fearful temperament or they may represent a specific profile of fearful temperament (Buss, 2011; Buss et al., 2013). Although behavioral inhibition, shyness, fear, and dysregulated fear all describe a withdrawal-related style of fearful reactivity when confronted with novel stimuli, the distinctions between different subtypes of fearful temperament may be critical for noting underlying mechanisms and potential outcomes. For instance, the potential mechanisms that link shyness to anxiety may be different from the mechanisms associating dysregulated fear and anxiety (Morales, Pérez-Edgar, & Buss, 2015).

Overall, it is well established that the fearful temperament trait precedes and maintains anxiety problems across childhood and adolescence (Degnan, Almas, & Fox, 2010). For example, concurrent ratings of shyness significantly correlate with social anxiety symptoms in late childhood (10-12 years old) and in adolescence (14-16 years old; Tsui, Lahat, & Schmidt, 2017). Dysregulated fear at age 2 significantly predicts social wariness and anxiety at age 6 (Buss et al., 2013). Early maternal-report of child behavioral inhibition increases the likelihood of the child having social anxiety disorders by 3.79 times and also predicts both child and parent ratings of ongoing social anxiety symptoms by mid adolescence (Chronis-Tuscano et al., 2009). It is important to note, however, that not all children with fearful temperament develop anxiety disorders. In line with the developmental psychopathology perspective, fearful temperament does not unvaryingly and linearly lead to anxiety (Cicchetti & Rogosch, 2002; Ollendick & Hirshfeld-Becker, 2002), as noted in multiple studies. For example, Rapee (2014) reported that although behavioral inhibition at age 4 is a significant predictor of social anxiety disorder at age 15, only 37% of inhibited children were diagnosed with social anxiety disorder in adolescence.

The modest association between fearful temperament and anxiety (Degnan & Fox, 2007) stresses the importance of examining the potential mechanisms underlying the relation between temperament and anxiety as well as elucidating the factors that could alter such mechanisms, potentially directing the trajectories away from the maladaptive track. Researchers have reported that exogenous factors, such as parenting behaviors and peer relationships, perform as independent predictors of anxiety or mediators and moderators in the association between fearful temperament and anxiety (Affrunti, Geronimi, & Woodruff-Borden, 2014; Kiel & Buss, 2009; Vasey & Dadds, 2001). Researchers have also reported that endogenous factors, such as certain cognitive distortions and cognitive control mechanisms, perform as predictors of later anxiety or mediators and moderators in the link between fearful temperament and anxiety (Lonigan & Vasey, 2009; Pérez-Edgar et al., 2010). In this review, we focus on endogenous mechanisms of anxiety at the level of cognitive factors: threat-related attention bias (AB) and effortful control (EC). AB represents one of the most extensively studied cognitive distortions associated with anxiety. The findings regarding the role of AB in anxiety are mixed, however, and thus theoretical reviews are needed to summarize what we know currently to guide further research. When researchers examine the effect of AB on the association between fearful temperament and anxiety, most also emphasize the effect of effortful or strategic control ability, the hallmark of EC (Derryberry & Reed, 2002; Fox, Henderson, Marshall, Nichols, & Ghera 2005; Henderson & Wilson, 2017; Lonigan, Vasey, Phillips, & Hazen, 2004). Taking into account EC better captures the complex underlying cognitive mechanisms of anxiety.

AB appears to be involved in the onset and maintenance of anxiety for both children and adolescents (e.g., Bar-Haim, Lamy, Pergamin, Bakermans-Kranenburg, & van IJzendoorn, 2007; Hadwin, Garner, & Perez-Olivas, 2006). In Lau and Water’s (2017) recent annual review, they suggest AB as a proximal mechanism that mediates the effects of other distal (e.g., parenting behaviors, peer relationships) and predispositional ( e.g. genetic, temperamental) risk factors on anxiety. AB refers to a tendency to have one’s attention initially drawn by threat (AB toward threat), or a tendency to avoid threat (AB away from threat; Bar-Haim et al., 2007). In this context, threat is defined as socially or non-socially threatening stimuli. Of note, the perception of threat is subjective and thus the threat source or the saliency of a threatening item might be different to different people. In studies focused on fearful temperament, AB, and anxiety, the common types of stimuli include human angry faces, threat words (e.g., murder), and non-socially threatening images (e.g., threatening animals), with the most typical stimuli being angry faces. The strategy of using angry faces as threat stimuli may due to their emotional salience, as well as for the purpose of replication. A variety of reaction-based tasks has been used to measure AB, such as the dot-probe task and the spatial cueing task. During AB tasks, threatening stimuli and neutral stimuli are presented and participants are assessed on whether they are faster in identifying and processing targets (e.g., the direction of arrows) associated with threatening stimuli or neutral stimuli. If it takes participants less time to process the targets that appear in the location previously occupied by threat stimuli than by neutral stimuli, then they are considered to exhibit AB toward threat. If the contrary, then participants show AB away from threat.

It is important to note that AB measured with different stimuli presentation durations in tasks may have different underlying neural mechanisms. To be specific, short durations (e.g., 500 ms or less) typically indicate initial automatic AB. If we regard cognitive processing as a continuum that extends from automatic to regulatory, initial AB falls closer toward the automatic direction and is tied to the orienting attention system, which is a stimulus-driven system that prioritizes perceptual information for further processing (Morales, Fu, Pérez-Edgar, 2016). Orienting functions involve the superior parietal lobe, superior colliculus, temporal parietal junction, frontal eye fields, and pulvinar (Petersen & Posner, 2012). These structures are interconnected with the amygdala (Vuilleumier, Richardson, Armony, Driver, & Dolan, 2004), which plays a significant role in threat detection and in mediating AB (Cisler & Koster, 2010; Monk et al., 2008). Longer durations (e.g., 1250 ms or more) usually indicate sustained AB that can be regulated or affected by both bottom-up and top-down systems, including the prefrontal cortex underlying EC (Fu & Pérez-Edgar, 2019; Lonigan et al., 2004).

EC refers to the regulatory aspect of temperament (Kochanska, Murray, & Harlan, 2000) and specifically focuses on the voluntary processes associated with modulating reactive or dominant responses, planning, and detecting an error (Rothbart & Rueda, 2005). The development of EC relies on the maturation of the executive attention system including the anterior cingulate cortex, anterior insula, basal ganglia, and the dorsolateral prefrontal cortex (Rothbart, Sheese, & Posner, 2007). EC shows the greatest improvement across toddlerhood and childhood and then becomes stable and reaches adult-level pattern during adolescence (Gerardi-Caulton, 2000; Luna, Garver, Urban, Lazar, & Sweeney, 2004; Simonds, Kieras, Rueda, & Rothbart, 2007). EC contains several components, including attentional control (AC), inhibitory control (IC), and activational control (Rothbart et al., 2001), with the first two having been particularly associated with fearful temperament and anxiety. AC refers to the flexible control of attention (i.e., focusing and shifting) as needed. IC is the suppression of a dominant response to instead perform a subdominant one. Activational control indicates the ability to initiate behaviors when not motivated.

The role of EC in the vulnerability of psychological disorders has been well recognized (e.g., Calkins & Fox, 2002; Muris & Ollendick, 2005). Children with high EC are better able to control their behaviors, attention, and emotions, therefore being less prone to psychopathology (Eisenberg et al., 2009). EC is often studied as a moderator in the association between reactive aspects of temperament (e.g., shyness, fear) and socio-emotional problems and it is the combination of high negative reactive temperament and low EC that increases the risk of psychopathology (Muris, 2006; Oldehinkel, Hartman, Ferdinand, Verhulst, & Ormel, 2007; Verstraeten, Vasey, Raes, & Bijttebier, 2009). Interestingly, more recent studies have revealed that different components of EC, particularly AC and IC, seem to have distinct effects on the continuity of fearful temperament and the relation between fearful temperament and anxiety. For example, in a study that examined the associations between different components of EC and shyness trajectories across childhood, results indicated that high attention shifting (a component of AC) was related to lower trajectories of shyness over time, whereas high IC was related to higher trajectories of shyness over time (Eggum-Wilkens, Reichenberg, Eisenberg, & Spinrad, 2016). In addition, high levels of attention shifting buffered the risk for anxiety symptoms in toddlers with high levels of BI, whereas high levels of inhibitory control increased this risk for anxiety symptoms (White, McDermott, Degnan, Henderson, & Fox, 2011). Therefore, AC and IC may differently modify the association between fearful temperament and anxiety across early childhood. IC particularly appears to increase the risk of developing anxiety for children with high level fearful temperament by leading to a rigid over-controlled system and AC lowers the risks by facilitating the flexible shift of attention (White et al., 2011; but see Crockenberg & Leerkes, 2006; Liu, Calkins, & Bell, 2018). In light of these findings, AC and IC need to be treated separately in any discussion of how they moderate the association between fearful temperament and anxiety through AB.

The major goal of our review is to examine how fearful temperament, AB, and EC perform together across development in the prediction of anxiety. In service of this objective, we review influential theories and models in the field and highlight a number of updated studies that expand our understanding of these processes. We do not intend this paper to serve as a comprehensive review but rather a selection of up-to-date illustrative evidence from developmental and clinical areas in terms of the associations among fearful temperament, AB, EC, and anxiety. Based on current theories/models and recent studies, we propose a model to further guide the study of AB and anxiety (see Figure 1).

Figure 1.

A conceptual two-mechanism developmental model linking early fearful temperament to later anxiety. Biologically determined fearful temperament during early life leads to atypical automatic AB toward (Figure 1A) or away from threat (Figure 1B) depending on levels of fearful temperament. Figure 1A and figure 1B are mirror images of each other. Across development, the continuation of fearful temperament interacts with AB, determining its severity and longitudinal duration. As the prefrontal cortex continuously develops, EC comes into play and exhibits an increasingly stronger effect during late childhood and adolescence by moderating the association between initial AB and its persistence. Specifically, IC prolongs the initial avoidance of threat for children with high fearful temperament, thus increasing the risk of anxiety. The role of IC in children with medium fearful temperament may or may not be detrimental. AC prevents the maintenance of vigilant bias for children with medium fearful temperament and thus decreases the risk of anxiety. AC may prolong or override the initial avoidant bias for children with high fearful temperament depending on the goal priorities. The association between AB and anxiety can be bi-directional. The direct link between fearful temperament and anxiety as well as the moderating effect of EC on the fearful temperament to anxiety process are not overtly demonstrated in this model in order to emphasize the main points of our review.

Note: FT = fearful temperament; DS = dysregulated fear; IC = inhibitory control; AC = attentional control.

Note. The levels of effect for AC and IC across development are indicated by thickness of arrows, with less effect in early development and greater impact over time.

We begin with a review of current influential models focused on fearful temperament, AB, EC, and anxiety. Then we review the up-to-date empirical evidence of the associations among fearful temperament, AB, and anxiety, and how EC is involved in these associations by focusing on the different sub-components of EC, which we will detail as IC and AC. We continue by proposing our own model as a summary of what is currently known about the cognitive processes linking early fearful temperament and later anxiety outcomes. Finally, we conclude with a summary of the limitations of current work and questions to guide future research.

A Brief Summary of Previous Theoretical Models of Fearful Temperament, AB, EC, and Anxiety

Integrative model of reactive temperament, regulatory temperament, attention biases, and anxiety (Lonigan et al., 2004)

In a review paper focusing on dysregulation of negative affect and development of anxiety disorders, Lonigan and colleagues (2004) combined research evidence from affect, temperament, and attention areas, and proposed an integrative moderated mediation model. According to the model, reactive temperament, partly mediated by an automatic AB toward threat, causes elevated levels of anxiety symptoms. The risk for anxiety can be moderated by EC, however. In other words, although reactive temperament is associated with an automatic AB toward threat, the persistence of AB is moderated by EC, specifically under circumstances that allow for the deliberate control of attention. Particularly, low EC prevents automatic AB toward threat from being overcome, and the perseverance of such AB increases the risk for anxiety. Lonigan and colleagues argued that to test the moderating effect of EC, stimuli should be presented long enough to permit the intentional control of attention (i.e., 1250-1500 ms). Also in this model, EC directly moderates the association between reactive temperament and anxiety disorders, based on the rationale that AB is only one pathway through which reactive temperament links to anxiety. EC in the Lonigan model is regarded as a broad construct representing general effortful regulation ability.

Lonigan’s model (2004) represents the first attempt to demonstrate dynamic associations among fearful temperament, AB, EC, and anxiety. It sheds light on the research in the field by providing new theoretical insights and measurement guides. Nevertheless, there are limitations that may impair its usefulness. First, EC is considered as a unitary factor in the model. Given AC and IC may be different in terms of their functions in the association between fearful temperament and anxiety (e.g., White et al., 2011), it is crucial to regard AC and IC separately in any discussion of how they affect the risk and resilience processes of children with fearful temperament. Second, reactive temperament contains a variety of subtypes, such as fear, shyness, anger/frustration, sadness and so on. These factors are different in their associations with AB and anxiety (Zeman, Shipman, & Suveg, 2002). Grouping them together into a general negative reactive temperament factor may confound and obscure the important information regarding how each individual type of reactive temperament is involved in the process.

Efficiency of goal-directed attention modal (Henderson & Wilson, 2017)

To examine the underlying mechanisms by which EC affects the association between fearful temperament and anxiety, Henderson and Wilson (2017) recently proposed a model focused on the different roles of AC and IC. Specifically, Henderson and Wilson suggest that high IC prolongs initial, automatic AB toward threat. Because the combination of IC and fearful temperament creates an inflexible over-controlled system, fearful children with high IC are more likely to get stuck on threatening stimuli. In comparison, greater AC helps to direct attention from threat to non-threatening stimuli or back to the goal-directed behaviors, if applicable. Thus, high AC prevents the persistence of automatic AB toward threat.

The Henderson and Wilson model is unique in treating IC and AC separately and exploring their potentially distinct effects on either maintaining or alleviating the adverse influence of AB. In addition, the model focuses on a more specific temperamental profile, which is behavioral inhibition, rather than a general reactive temperament or negative affectivity. Although the model does not particularly specify the difference between an automatic AB and the persistence of it, it assumes that IC and AC cannot affect the initial automatic AB, which is considered as a characteristic of children with fearful temperament. Instead, IC and AC influence the behavioral inhibition and anxiety link by extending or shortening, respectively, the duration of reactive AB. This is similar to the argument in the integrative model by Lonigan and colleagues (2004): EC moderates the association between automatic AB and its perseverance.

Conceptual cascade model of AB, executive attention, and socio-emotional functioning (Morales, Fu, & Pérez-Edgar, 2016)

To shed light on the nuanced associations among AB, EC, and socio-emotional functioning across development, Morales and colleagues (2016) proposed a hybrid cascade model explaining how individual differences in AB emerge and how AB interacts with the development of attention systems to affect children’s adjustment. Instead of talking about EC itself, Morales focused on the underlying neural system of EC, which is the executive function system (Rueda, Posner, & Rothbart, 2005). Specifically, there are early emerging individual differences in AB that are determined by intrinsic and extrinsic factors, mainly driven by reactive attention systems (i.e., alerting and orienting). With the maturation of the regulatory executive attention system, the majority of children are able to modulate these early reactive biases. For those children, early-emerging AB supports adaptive patterns of vigilance in the face of threat and avoidance and helps individuals better adjust to complex environments (LoBue, Rakison, & DeLoache, 2010). This normal pattern of AB toward threat represents an appropriate level of alertness that characterizes typical early development. It is important to note, however, that a subgroup of children may show a heightened pattern of AB that can be caused by various individual risk factors, such as fearful temperament. Such atypical threat detection can also reinforce the development of psychopathology throughout childhood and adolescence.

The model by Morales and colleagues (2016) highlights the effects of developmental processes associated with maturation and experience on the formation and continuation of AB and points out that the factors involved in this model (e.g., executive attention, AB) can act as both mediators and moderators. In addition, the model argues that AB is a domain-general mechanism that is not only related to anxiety but also to other behavioral and emotional concerns, such as externalizing problems. The model also highlights the bi-directional association between AB and socio-emotional functioning.

Cognitive-motivational framework of processes underlying anxiety and AB (Mogg & Bradley, 2018)

Mogg and Bradley (2018) proposed a cognitive-motivational framework of processes underlying anxiety and AB in their recent review. The model suggests that AB and anxiety not only emerge from but also exacerbate the imbalance between salience-driven and goal-directed models of processing. Specifically, high anxious individuals tend to have over-reactivity of salience-driven processing (e.g., automatic threat salience evaluation, salience-driven switching, and salience-driven orienting) and ineffective goal-directed cognitive control (e.g., goal-directed IC, goal-directed switching, and goal-directed orienting), which further contribute to their AB and anxiety symptoms. Therefore, the associations between AB and anxiety are probably reciprocal. In addition, the model highlights that AB can be manifested in different forms: AB toward threat may increase detection and awareness of threat, therefore increasing distress. Maintaining attention on threat and inability to orient away from threat may lead to more anxious worries and thoughts. Threat avoidance may reduce immediate anxious feelings, but be harmful in the long term by impeding threat appraisal and fear habituation. The multiple cognitive mechanisms underlying AB and the different subtypes of AB encourage the development of interventions that include procedures targeting different cognitive processes (e.g., flexible shifting, goal rehearsal, threat exposure).

Summary of theoretical models

To summarize (see Table 1), with the exception of the cognitive motivational framework by Mogg and Bradley (2018), the above mentioned models discuss the roles of temperament, AB, and EC in the formation of anxiety symptoms. Mogg and Bradley focus on the cognitive mechanisms of AB and do not explicitly mention the effect of temperament. Morales and colleagues’ (2016) conceptual cascade model highlights the developmental effects in the process: constantly developing EC disrupts the persistence of atypical AB toward threat caused by early fearful temperament over the course of development (Morales et al., 2016). The other three models focus on short-term mechanisms regarding how EC affects the formation of atypical AB (e.g., overcoming the initial tendency, recovering the balance between systems).

Table 1.

Summary of Theoretical Models Regarding Fearful Temperament, AB, EC, Anxiety

Models	Main point	Temperament	Patterns and roles of AB	Forms and roles of EC	Methods for testing
Lonigan, Vasey, Phillips, & Hazen, 2004	Reactive temperament, partly mediated by an automatic AB toward threat, causes elevated levels of anxiety symptoms. Low EC prevents automatic AB toward threat from being overcome, and the perseverance of such AB increases the risk for anxiety.	Negative affectivity	Initial AB toward threat and sustained AB toward threat; mediator	A unitary factor; moderator	dot-probe; eye-tracking
Henderson & Wilson, 2017	High IC prolongs initial, automatic AB toward threat for behaviorally inhibited children and thus increase the risk of anxiety. High AC helps to direct attention from threat to non-threatening stimuli or back to the goal-directed behaviors and therefore prevents the persistence of automatic AB toward threat and reduce anxiety symptoms.	Behavioral inhibition	Attention maintenance on threat and inefficiency of return to goal-directed attention; outcome/mediator	IC and AC; moderator	dot-probe; ERP; eye-tracking
Morales, Fu, & Pérez-Edgar, 2016	The early-emerged individual differences in AB, determined by genetic individual factors and shaped by the development of attention system and contextual experience, lead to different socio-emotional adjustment outcomes.	General individual factors (e.g., fearful temperament)	Reactive AB toward threat (alerting; orienting); Moderator/mediator	Executive attention; moderator and mediator	dot-probe; ERP; eye-tracking
Mogg & Bradley, 2018	AB and anxiety are the causes and outcomes of the imbalance between salience driven and goal-directed models of processing. Interventions should include procedures targeting on different cognitive processes.	N/A	Initial orienting toward threat, maintaining attention on threat, orienting away from threat, threat distractor interference; predictor/outcome	Goal-directed cognitive control; predictor and outcome	dot-probe; emotional Stroop; eye-tracking; ERP

In regards to the patterns of AB, most models simply assume that AB is manifested as an orienting toward threat. The cognitive-motivational framework, however, suggests that different types of AB may be involved in the vulnerability of anxiety and that it is important to distinguish them for the purpose of treatment efficacy (Mogg & Bradley, 2018). The roles of AB are inconsistent across models depending on their main focus. For example, the integrative model regards AB as a mediator that links temperament and anxiety (Lonigan et al., 2004). In the conceptual cascade model, although AB was illustrated as a statistical mediator in the model, it could also act as a moderator (Morales et al., 2016). The efficiency of goal-directed attention model did not overtly specify the role of AB, but indicated that it is an outcome of the interactive effect of behavioral inhibition and EC and is involved in multiple socioemotional risks (Henderson & Wilson, 2017).

In regards to the forms and roles of EC, most models considered EC as a unitary construct. The exception is the efficiency of goal-directed attention modal, which argued that different components of EC may have different effects on the AB and anxiety process (Henderson & Wilson, 2017). The role of EC is generally treated as a moderator but may also act as an independent predictor or mediator (Morales et al., 2016).

Building the framework for a two-mechanism developmental model

Each of the models reviewed above has limitations. For example, three of the models assume that all individuals with fearful temperament show automatic AB toward threat, ignoring the potential presence of other types of atypical AB. In addition, all but one overlook the fact that AB may change over time and the effect of EC may be different at different developmental periods. Developmental effects are essential to our understanding of the formation and maintenance of anxiety. We need a developmental model with an explicit discussion of how temperament and cognitive factors work together in predicting anxiety symptoms. In the literature review that follows, we build the case for a model that integrates all the valued points of the four models and reflects the updated empirical findings. In the course of this review we emphasize three points. First, we demonstrate the potential differential effects of AC and IC and how they affect the risk and resilience processes of temperamentally fearful children (Henderson & Wilson, 2017). Second, we take into account the possibility that different types of AB may underlie the development of anxiety symptoms (Mogg & Bradley, 2018). Third, we explain the short-term mechanisms of how AB and EC are involved in the onset of anxiety, as well as the long-term developmental effect. With all three of these components included in one model, we can have a more comprehensive and fine-grained understanding of the development of anxiety and the cognitive mechanisms underlying it.

In the next section, we review empirical evidence regarding the effect of fearful temperament, AB, and EC (i.e., IC and AC) on the development of child and adolescent anxiety symptoms in developmental and clinical fields. We did our literature search by following the guidelines of writing non-systematic narrative reviews (Ferrari, 2015). The electronic search included three databases, PubMed, PsycINFO, and Google Scholar. Our search terms were “fearful temperament”, “attention bias”, “effortful control”, “anxiety”, “child”, and “adolescent”. The inclusion criteria were: empirical research articles, using experimental tasks to measure AB (e.g., dot-probe, emotional Stroop), and the participants were children or adolescents. The exclusion criteria were: not applying widely-used tasks to measure AB (e.g., attention to peer feedback), participants have other symptoms in addition to anxiety (e.g., PTSD, depression, chronic pain), and having a focus on other irrelevant constructs (e.g., sleep patterns, maternal care, body satisfaction). The retrieved articles were assessed on their suitability of the methods of testing the research questions, quality of the main results, and potential contributions to the field (Ferrari, 2015) before being included and synthetized. The references used in the articles retrieved from the keyword search were also searched, assessed, and reviewed.

Unless otherwise specified, all reviewed empirical studies measured AB with orienting-based tasks (e.g., dot-probe) that use reaction time as an index. A few of studies used other behavioral tasks (e.g., emotional Stroop) or eye-tracking to test AB (e.g., Gorline & Teachman, 2015; Shechner et al., 2017). Limitations of behavioral measures of AB (i.e., reaction time) will be discussed at the end of our review in the measurement challenges and problems section. In addition, the majority of the studies used self-reports to test EC, such as Early Adolescent Temperament Questionnaire (e.g., Susa, Benga, Pitica, & Miclea, 2014). The questionnaires generally have acceptable reliability and have been widely used in measuring child and adolescent EC. Limitations of using self-reports, however, should be considered; we discuss this in the measurement challenges and problems section as well.

Early Fearful Temperament and Threat-related Attention Biases

As one of the most powerful individual risk factors, fearful temperament is associated with the formation of extreme threat detection and AB. For example, in a study that examined AB in a group of children (4-7 years old) who tended to develop anxiety due to high temperamental shyness, results indicated that children in the high-shy group showed greater AB toward angry faces compared with children in the low-shy group (LoBue & Pérez-Edgar, 2014). Similarly, results from another longitudinal study demonstrated that fearful temperament measured during toddlerhood and early childhood was associated with AB toward threat during adolescence. Specifically, children who showed a higher level of fearful temperament tended to develop greater AB toward threat (Pérez-Edgar et al., 2010).

Interestingly, findings are not always the same regarding the direction of whether AB is toward or away from threat. Children characterized by dysregulated fear at two years of age displayed greater AB away from threat at age six. Children without dysregulated fear, however, showed no significant bias (Morales et al., 2015). Similarly, adolescents with a combination of high level of behavioral inhibition and larger N2 brain response to stimuli tended to direct their attention away from threat (Thai, Taber-Thomas, & Pérez-Edgar, 2016). The N2 component of the Event Related Potential (ERP) is elicited when individuals need to suppress a dominant or high probability response in favor of a subdominant or low probability response. Therefore, different profiles or levels of fearful temperament may cause different patterns of AB, which indicates potential different mechanisms underlying anxiety. In particular, a high level of fearful temperament, such as dysregulated fear or an extreme over-controlled system caused by the combination of fearful temperament and IC, may lead to anxiety by triggering threat avoidance. Children with dysregulated fear lack the ability to regulate emotional reactivity in a way that facilitates appropriate interactions with the context; thus, they may develop a rigid pattern of avoidant behaviors in any new environment, even those with low threat (e.g., meeting new peers; Buss, 2011). Children with medium level of fearful temperament, on the other hand, may possess a general tendency to orient toward and maintain attention on threat, which increases risk of anxiety. It is of great importance to study the implications of the two patterns of AB in the formation of anxiety for the purpose of understanding the cognitive mechanisms associated with these biases and eventually providing targeted and customized interventions. In the next section we give a detailed discussion of the association between AB and anxiety during childhood and adolescence, with a particular emphasis on the different patterns of AB: toward threat and away from threat.

Attentional Biases and the Development of Anxiety during Childhood and Adolescence

Cognitive models of anxiety suggest that AB plays a causal and maintaining role in anxiety disorders (Mogg & Bradley, 1998; Williams, Watts, MacLeod, & Mathews, 1988). Of note, researchers focused on AB in children and adolescence have produced mixed findings in terms of whether the biases are toward or away from threat.

Attentional Biases Toward Threat

AB toward threat may cause an over processing of the negative features of a stimulus, underlying the vulnerability of anxiety (e.g., Telzer et al., 2008). For example, children (8- 12 years old) with diagnoses of severe anxiety showed AB toward angry faces relative to neutral faces that were presented for 500 ms; a similar AB was not exhibited by children with less anxiety or children with no anxiety (Waters, Henry, Mogg, Bradley, & Pine, 2010). Similarly, clinically anxious children and adolescents (7-18 years old) demonstrated a greater AB toward angry faces with a 500 ms exposure time compared with non-anxious children. Moreover, this threat bias in anxious children applied to all the types of anxiety disorders examined in the study (i.e., generalized anxiety, social phobia, separation anxiety; Roy et al., 2008). Similar findings linking anxiety level with AB toward threat are reported in studies focused on community samples of children. For example, AB toward threat measured with a 500 ms stimuli presentation duration was positively associated with self-reported trait anxiety in both children and adolescents (Telzer et al., 2008). In a study investigating the duration of threat exposure, children (9-12 years old) who showed AB toward angry faces at exposure durations of both 500 ms and 1250 ms had higher levels of anxiety (Waters, Kokkoris, Mogg, Bradley, & Pine, 2010).

Evidence supporting that AB toward threat may be causally implicated in the development of anxiety comes from intervention studies. Attention bias modification (ABM) was developed to improve the current treatment of anxiety. In ABM, the visual probe task is modified so that probes requiring a response appear more frequently at the positions previously occupied by neutral versus threat stimuli, thus facilitating a consistent AB away from threat over hundreds of trials for individuals whose tendency is to have an AB toward threat (Bar-Haim, 2010). ABM has been reported as a potentially promising treatment for anxiety (e.g., Chang et al., 2018). For example, in one report children with an on-going anxiety disorder showed reduced the symptoms of anxiety after ABM, compared with children in the control conditions (Eldar et al., 2012). Findings have been replicated with adolescents; specifically the ABM group experienced a significantly greater decrease in anxiety symptoms than adolescents in the control group (Riemann, Kuckertz, Rozenman, Weersing, & Amir, 2013). In addition, ABM was reported to reduce children’s state anxiety in response to a stressor task (Bar-Haim, Morag, & Glickman, 2011) and to also change neural activity. For example, in a group of 9- to 12-year-old children with fearful temperament, ABM significantly decreased amygdala and insula activation and increased activation in the ventrolateral prefrontal cortex compared with the control group (Liu, Taber-Thomas, Fu, & Pérez-Edgar, 2018).

It is worth noting that the efficacy of ABM in treating anxiety has been challenged in recent reports. In one study, there were no effects of ABM on clinically anxious adolescents’ AB and social anxiety symptoms, and thus no significant differences between treatment group and control group (Ollendick et al., 2018). Moreover, in another study focusing on adolescents, ABM did not reduce social anxiety symptoms (Fitzgerald, Rawdon, & Dooley, 2016). In addition, the efficacy of ABM may vary by different type of anxiety disorders. For example, in one study ABM mitigated children’s separation anxiety but not social anxiety (Liu et al., 2018).

Poor outcomes of ABM may be caused by the difficult protocol (Mogg & Bradley, 2018). For example, persistent distraction is tedious and tiring, which prevents lasting compliance with treatment requirements (Amir, Kuckertz, & Strege, 2016). The context of treatment might also play a role, in that findings from a recent meta-analysis study revealed that the effect of ABM was significant when it was delivered in the clinic, but not at home (Linetzky, Pergamin-Hight, Pine, & Bar-Haim, 2015). Disappointing outcomes can also be explained by problems in target processes (Mogg & Bradley, 2018). It is possible that children who are not benefitting from ABM did not show pre-training AB. This inference is supported by the finding that the effects of ABM on eliminating AB were only significant for the participants who showed pre-training bias. When including participants without pre-training bias, the effects did not emerge (O’Toole & Dennis, 2012). There also may be risk factors other than AB causing the development of anxiety (e.g., negative parenting; Ollendick & Grills, 2016).

It is important to note that for some individuals, anxiety arises from a different pattern of AB, which is AB away from threat rather than AB toward threat. In these cases, training attention away from threat will unsurprisingly not be helpful in eliminating anxiety or may even exacerbate the severity of symptoms. Actually, there has been evidence showing that training participants to direct attention toward threat (i.e., the opposite of ABM) causes a greater decrease in social anxiety, compared with the conditions designed to elicit attention toward positive cues and neutral cues (Boettcher et al., 2013). These findings not only raise questions about ABM, with its focus on attention toward threat, as an effective treatment for all individuals with anxiety, but also highlight the potential for examining different AB patterns that might represent distinct patterns in the development of anxiety.

Attentional Biases Away from Threat

Researchers have also reported links between AB away from threat and anxiety. For example, children and adolescents (9-17 years old) with generalized anxiety disorder show greater AB away from angry faces that were presented for 500 ms compared with control children (Monk et al., 2006). Similarly, anxious children (8-9 years old) are more likely to avoid threat with a 1000 ms exposure duration, in comparison with non-anxious children (Brown et al., 2013). Results from an eye-tracking study showed that an avoidance of negative stimuli is associated with anxiety for anxious adolescents (Shechner et al., 2017). A study of non-clinical children (8-11 years old) showed that avoidance of threat (i.e., angry and fearful faces; 1000 ms) was positively associated with social anxiety symptoms (Stirling, Eley, & Clark, 2006). Moreover, in a study that examined non-anxious adolescents’ performance on the emotional Go/No Go task, adolescents who were faster to disengage from threat faces and respond to probes showed a higher level of trait anxiety (Cohen Kadosh, Heathcote, & Lau, 2014). The idea of attentional avoidance of threat serving to maintain anxiety is incorporated into the theory of fear induction (Foa & Kozak, 1986). According to the theory, avoidance of threat delays elaborate processing and habituation to anxiety-provoking stimuli, and thereby is responsible for the onset and maintaining of anxiety.

Although for a subgroup of children avoidance is the dominant AB in response to threat and is the main vulnerable risk of anxiety, these children have to detect the existence of the threat first before we can examine their AB away from threat. According to the vigilance-avoidance hypothesis, anxious individuals showed initial vigilance toward threat followed by avoidance of it, and what maintains anxiety is the maladaptive threat avoidance (Mogg & Bradley, 1998). Threats can also be detected pre-attentively via shifts in covert attention (i.e., shifts in attention without accompanying eye movements). For example, children with high anxiety sensitivity demonstrated covert AB toward social threat words that were masked (14ms) and were out of their consciousness in the dot-probe task (Hunt, Keogh, & French, 2007).

Whether following initial overt or covert detection, automatic maladaptive avoidance of threat can occur. In other words, avoidance does not always represent a strategic attention shifting from threat (Booth, 2014). In support of this idea, results from an eye-tracking study indicated that compared with non-clinical control group, anxious adolescents (age 12-17 years) directed their first fixation away from negative facial emotions (anger, disgust, fear, sadness), which occurred too quickly to be controlled by the top-down regulatory system (Gamble & Rapee, 2009). With a group of children (age 10-13 years), the positive association between fearful temperament and threat avoidance was significant only when children had low attentional control (Helzer, Connor-Smith, & Reed, 2009). In this case, avoidance seems not to be facilitated by strategic attentional processes but an automatic process. The role of attentional control is to interrupt the automatic avoidance. Thus, automatic avoidance is different from avoidance that results from the strategic disengagement from threat (Lonigan et al., 2004).

Patterns of Attention Biases, Types and Levels of Anxiety, and Treatment Response

With accumulating evidence that the risk for developing anxiety may have its origins in different patterns of AB (i.e., AB toward threat or AB away from threat), researchers have begun to distinguish between the two AB patterns with respect to correlations with later anxiety and with treatment effects. As a result, there are reports that AB toward or away from threat is associated with different types of anxiety disorders. For example, in a community-based large-scale study with 2046 children (6-12 years old), those without diagnosis and those with distress-related disorders (e.g., generalized anxiety disorder) had internalizing symptoms associated with AB toward threat. In contrast, children with fear-related disorders (e.g., specific phobia, social anxiety disorder, separation anxiety disorder) had internalizing symptoms associated with AB away from threat (Salum et al., 2013). The findings were replicated in another study. Compared to children (age 5-13 years) with no psychiatric disorder, children with distress-related disorders showed a vigilance toward threat, whereas children with fear-related disorders showed an avoidance away from threat (Waters, Bradley, & Mogg, 2014; but see Abend et al., 2018).

In addition, different patterns of AB may underlie different levels of anxiety. Although evidence suggests that threat avoidance predicts a higher level of anxiety for both clinical and non-clinical children (Stirling et al., 2006), avoidance is associated with the severity of anxiety disorders only for anxious children, whereas for non-anxious children, threat avoidance is correlated with a lower level of anxiety (Vervoort et al., 2011). The results suggest that AB away from threat represents a maladaptive avoidance for anxious children and an adaptive disengagement from threat for non-anxious children who show automatic AB toward threat. These findings raise the possibility that AB away from threat mainly characterizes children with an anxiety disorder, whereas AB toward threat represents the primary risk factor for non-clinical children who overall have a lower level of anxiety compared with anxious children.

In addition, there is evidence that individuals who show vigilant biases (AB toward) or avoidant biases (AB away from) have different responses to cognitive behavioral therapy (CBT). Anxious children with a pre-treatment bias toward threat presented greater declines in anxiety symptom severity and were less likely to meet the diagnostic criteria for anxiety disorders after receiving CBT, compared with anxious children who showed a pre-treatment avoidant bias (Waters, Mogg, & Bradley, 2012). This might be because the key to the success of exposure-based treatment is helping individuals become fully aware of and process the threatening stimuli. Thus, individuals with AB toward threat before treatment exhibit better treatment compliance relative to pre-treatment avoiders of threat (Waters et al., 2012). There are other studies, however, reporting that clinically anxious children who show AB away from severe threat profit more from child-centered CBT (Legerstee et al., 2009; Legerstee et al., 2010). The benefit may be because during the exposure-based child-focused CBT, children had to direct their attention toward threat. These exposure interventions might be particularly beneficial for children who show a strategy to avoid severe threat. The redirection of their attention toward threat during child-focused CBT may facilitate habituation of threat and reduce their anxious symptoms (Legerstee et al., 2009).

Overall, these findings suggest a complex association between AB and anxiety. It appears that both AB toward threat and AB away from threat confer risk for anxiety. That there are two distinct patterns of AB to explain the development of anxiety is in line with the results from a recent study. Fearful children with a stable pattern of AB across tasks, whether it is biases toward or away from threat, tend to have higher levels of anxiety than children who have non-stable patterns of biases. Consistent vigilance or avoidance may create sustained enhanced or reduced exposure to threat respectively, with each having distinctive implications for anxiety (Morales, Taber-Thomas, & Pérez-Edgar, 2017). Specifically, for children who show a persistent vigilance toward threat, it is the unstoppable maladaptive attention and thinking on threat that cause anxiety, whereas for youths who demonstrate a sustained avoidance, the implication of anxiety comes from the limited evaluation/appraisal of threat and delayed habituation.

It is important to distinguish between the two types of automatic AB and the persistence of those patterns within individuals. This knowledge can help improve our fine-grained understanding of the mechanisms of anxiety, as well as the appropriateness and potential efficacy of treatment.

Fearful Temperament, Attention Biases, and Anxiety

Of note, although early fearful temperament has been recognized as an important determinant or contributor to atypical AB (Morales et al., 2016), which is involved in the etiology of anxiety symptoms (Mogg & Bradley, 2016), some empirical studies failed to find direct associations between AB and fearful temperament or between AB and anxiety; instead, they report an interactive effect of fearful temperament and AB in the prediction of anxiety. For example, behavioral inhibition during toddlerhood did not directly predict AB toward threat at age 5 but interacted with AB in predicting social withdrawal at age 5 (Pérez-Edgar et al., 2011). In addition, AB at ages 5 and 7 years was not associated with behavioral inhibition in toddlerhood or anxiety at 7 years old. Instead, there was a significant interactive effect of behavioral inhibition in toddlerhood and AB at 7 years old in predicting anxiety at age 7 (White et al., 2017). This empirical evidence demonstrates the role of AB as a moderator rather than a mediator in the association between fearful temperament and anxiety.

It is possible that early fearful temperament sets the foundation for atypical AB. Over the course of development, the continuation of fearful temperament interacts with AB, determining its severity and longitudinal duration in the contribution to anxiety (Field & Lester, 2010). Therefore, more long-term systematic studies, beginning in infancy, are needed to test the complex associations among fearful temperament, AB, and anxiety across development. In addition, the null findings partly may be due to the mediating effect of AB being constrained by other factors, and thus only being established when individuals meet specific criteria or develop under certain conditions (Eisenberg et al., 2009; Susa et al., 2014). When these moderators are taken into account, a moderated mediation model may better capture the nuanced associations among factors in predicting anxiety symptoms. A relevant factor that may alter the relation between fearful temperament and AB, as well as the link between AB and anxiety, is effortful control (EC), which is the regulatory aspect of temperament (Eisenberg et al., 2009). In the next section, we will include EC as a potential moderator in the discussion of the associations among fearful temperament, AB, and anxiety.

Effortful Control and Individual Differences in the Impact of Attention Biases

As noted, empirical evidence from developmental and clinical fields suggests that fearful temperament can lead to atypical AB that is either biased toward or away from threat (LoBue & Pérez-Edgar, 2014; Pérez-Edgar et al., 2010), which further provokes the development of anxiety (Mogg & Bradley, 1998). Thus, AB may perform as a potential mediator that links early fearful temperament and anxiety. Nevertheless, not all children with fearful temperament unvaryingly develop atypical sustained AB and anxiety. There is evidence to suggest that including EC as a moderator in the examination of AB can help explain why some children do and others do not develop atypical AB and anxiety.

The Role of Attentional Control

AC has been demonstrated as a moderator in the associations among fearful temperament, AB, and anxiety. For example, in a study of children (9 to 18 years old) selected from a large screening sample for having high or low negative affectivity (NA) and EC, results indicated that EC moderated the association between NA and AB. Specifically, children showed greater AB toward threat when they had high NA and low EC. For those with high EC, however, NA was not significantly associated with AB (Lonigan & Vasey, 2009). It is worth noting that EC in this study was mainly AC measured with the self-report scales that emphasize the flexible control of attention. Moreover, AB was measured via the dot-probe task with a 1250 ms stimuli exposure duration, which is thought to reflect the persistent AB that can be affected by EC (Lonigan et al., 2004). There was no indication of automatic AB in this study. Similarly, in another community sample of children (9 to 14 years old), those with a high level of self-reported temperamental fear showed greater AB toward treat. The association between fear and AB was significant, however, only when children had low AC, but not when they had medium or high AC (Susa et al., 2014). Interestingly, AB in this study was tested with stimuli presented for only 500 ms, which is usually regarded as a duration that reflects more automatic AB (Vervoort et al., 2011).

In another study that examined the association between neuroticism, which is similar to fearful temperament, and AB in children (9 to 11 years old), AB was measured with eye-tracking and indicated by the latency to initiate an eye movement to the target in the presence of angry, happy, and neutral faces at different positions (i.e., central, parafoveal, or peripheral). Results indicated that children with higher levels of self-report neuroticism tended to show greater sustained vigilance toward threat (i.e., longer latency in orienting to target when the distractors were angry faces compared with neutral and happy faces). Moreover, when distractors were presented in the central locations (i.e., the most difficult position to be avoided), children’s AC moderated the relation of neuroticism to AB. Those with high AC were quicker at disengaging attention from threat faces and orienting to the target (Pavlou, Benson, & Hadwin, 2016).

Of note, empirical evidence suggests the possibility that a subgroup of youth with fearful temperament may show automatic avoidance in response to threat rather than biases toward it, and to these children it is the AB away from threat that causes anxiety (e.g., Monk et al., 2006; Morales et al., 2015). When we consider the different forms of automatic AB (i.e., toward or away from threat) that underlie anxiety pathology, the associations among AB, AC, and anxiety become much more complex. For children with AB toward threat as the dominant pattern of atypical AB, AC functions as a protective factor contributing to the resilient process and lowering the risk of having anxiety. For those who predominantly show AB away from threat, however, the role of AC may not be always beneficial. For example, results from a recent study indicated that AC moderated the association between automatic AB and social anxiety symptoms. That is, AB was not associated with social anxiety if children had low AC. When children displayed high AC, greater AB away from threat was correlated with a higher level of social anxiety. AC was measured via a behavioral task during which participants had to quickly shifting attention to identify targets (Grafton, Visu-Petra, Marcuş, Liebregts, & MacLeod, 2016). Similar results have been found in another study that examined the interactive effect of AB and AC on anxiety among adolescents. Specifically, AB was not related to dispositional anxiety for adolescents with low AC. For adolescents with high AC, however, AB away from threat was positively correlated with the level of anxiety (Ho, Yeung, & Mak, 2017).

Actually, level of AC is not always harmful for individuals who tend to avoid threat. There is evidence indicating that when individuals are required to achieve a specific task-related goal (i.e., identify the direction of the target arrow), AC moderates the association between fearful temperament and a consciously perceived AB that is measured with a 1250 ms stimuli exposure duration. Specifically, for children with high AC, fearful temperament is not associated with AB, whereas for those with low AC, high fearful temperament predicts greater AB away from threat. The results demonstrate that AC could disrupt the continuation of the maladaptive attentional avoidance of threat (Helzer et al., 2009).

It seems that the role of AC in individuals who show automatic threat avoidance is difficult to predict and may depend on the main goal of the individual, whether it is a task-related or emotion-regulation related goal (Mogg & Bradley, 2016). If the primary goal is to perform well in a task that requires allocation of attention to threat (e.g., target replaces angry faces in the dot-probe task), high AC may help to break the sustained avoidance of the threat cue. If the main goal is to regulate distress and discomfort, then AC may facilitate emotion regulation by keeping attention away from threat. In contrast, the role of AC in individuals who exhibit automatic AB toward threat is more consistent because the two goals are compatible: regardless of whether the motivational state is to achieve a task-goal or to regulate negative emotion, AC facilities disengaging attention from threat.

The Role of Inhibitory Control

Studies examining the function of IC in the context of fearful temperament, AB, and anxiety are limited, therefore we extend the inclusion criteria on this particular topic by reviewing adult research as well. The lack of empirical work may be because AC is thought to be more relevant to AB in the visual domain compared with IC, which is generally considered a characteristic related to behaviors or actions (Bari & Robbins, 2013). In a recent study that focused on the neural correlates of children’s behavioral inhibition, AB, and social anxiety, results indicated that the amplitude of N2 evoked by face displays in the dot-probe task moderated the association between behavioral inhibition and social anxiety. As a reminder from earlier in our review, N2 is an ERP component that is elicited when individuals need to suppress a dominant or high probability response in favor of a subdominant or low probability one, thus representing a physiological marker of IC (Lahat et al., 2014). Specifically, for children who showed larger N2, higher behavioral inhibition predicted AB away from threat (Thai et al., 2016). Researchers suggested that the results might help to explain the finding that enhanced N2 exacerbates the association between behavioral inhibition and anxiety (Henderson, 2010).

In another study that examined the moderating effect of IC between AB and social anxiety in an unselected group of undergraduate students, IC was measured using the classic color-word Stroop task and AB was tested via the emotional Stroop task. Results indicated that greater threat avoidance (i.e., shorter latencies in response to social-threat words in the emotional Stroop task compared with neutral words) predicted an enhanced level of social anxiety among those with higher IC on the color-word Stroop. In other words, IC was linked with increased risk of social anxiety for those who showed threat avoidance (Gorline & Teachman, 2015).

The outcome that IC magnifies the association between fearful temperament and threat avoidance seems inconsistent with what Henderson and Wilson (2017) suggested at first glance, which is IC prolongs sustained vigilance toward threat. The conflict can be resolved if one takes into account the two possible patterns of AB. IC combines with fearful temperament leading to a rigid over-controlled system that can amplify the original reactive tendency, whether it is AB toward or away from threat. When fearful children show AB toward threat as the primary reactive biases, IC can prolong that vigilance. When they predominantly demonstrate avoidance instead, IC will maintain that avoidance.

It is worth noting that the effect of IC may depend on the severity of fearful temperament. For example, in the Gorline & Teachman (2015) study, greater vigilance to threat was associated with higher anxiety, but only among individuals with low IC. Participants in the study had low levels of fearful temperament; thus, IC may represent an appropriate level of self-regulation instead of an over-controlled system. As such, IC may have little or even protective effect (Rhoades, Greenberg, & Domitrovich, 2009). Nevertheless, for children with greater fearful temperament, IC is likely to inflexibly maintain the initial automatic vigilance or avoidance and increase the risk of anxiety.

Overall, the associations among fearful temperament, AB, EC, and anxiety are complex. Therefore, it is of great significance to integrate evidence from different lines of research and explore the diverse potential patterns of the interactions with multiple perspectives and explanations taken into account.

Two-mechanism Developmental Framework of Attention Biases and Effortful Control

After reviewing relevant theoretical models and up-to-date empirical findings in developmental and clinical areas regarding associations among fearful temperament, AB, EC, and anxiety in children and adolescents, we propose a two-mechanism developmental framework. Our goal is to propose a possible trajectory to anxiety focused on the interplay among fearful temperament, AB and the EC components of AC and IC as the mechanisms in the course of development (see Figure 1).

We specify different pathways depending on the initial level of fearful temperament. In our two-mechanism developmental framework, medium level of fearful temperament elicits an automatic and sustained vigilance toward threat, which is involved in the onset and maintenance of medium level of anxiety or anxious symptoms in later development. The associations among fearful temperament, AB, and anxiety, however, are not linear and are moderated by children’s EC. Specifically, AC can prevent the initial automatic AB toward threat from developing into longer vigilance, thus lowering the risk of anxiety. Regarding the role of IC, on one hand, IC may cause a maladaptive sustained vigilance by adding extra regulation to the already existing reactive inhibition (i.e., fearful temperament). On the other hand, IC may not be detrimental given that the level of reactive inhibition is not extremely high and therefore the overall regulation system is not excessively inflexible (see Figure 1A).

Children who have an extreme high level of fearful temperament (e.g., dysregulated fear), on the contrary, may show a reactive tendency to avoid threat rather than orient toward it. To these children, sustained threat avoidance represents the dominant atypical AB pattern that underlies the development of severe or clinical anxiety. The rigid control system influenced by the combination of IC and high fearful temperament prolongs the initial avoidance. The effects of AC in this process are less clear, in that children with higher AC are likely to show persistent avoidance, or they may not demonstrate preserved avoidance depending on their primary goal (see Figure 1B).

Over the course of development, the continuation of fearful temperament and the continuously developed EC may moderate the longitudinal persistence of either sustained vigilance or avoidance, which affects the development of anxiety. It is noteworthy that the level of fearful temperament may change across development as a function of a series of internal and external factors (e.g., Eggum-Wilkens et al., 2016; Grady, Karraker, & Metzger, 2012); thus, the main pattern of AB underlying anxiety may change as well.

The moderating effects of AC and IC become increasingly stronger as the prefrontal cortex develops, with the most remarkable influence occurring during late childhood and adolescence (Rothbart & Rueda, 2005). For example, EC did not affect the association between negative affect and AB as well as the link between negative affect and social withdrawal for children who were 4-7 years old but exerted an effect on those relations among children between 9 to 18 years old (Cole, Zapp, Fettig, & Pérez-Edgar, et al., 2016; Lonigan & Vasey, 2009). Interestingly, although AB to threat is not directly associated with EC, high AB to reward, which may perform as a marker of externalizing problems, has been negatively associated with EC (Cole et al, 2016, Morales et al., 2019). Future research may want to examine if AB toward or away threat is directly associated with EC concurrently and longitudinally. Of note, AB represents only one possible mechanism that links fearful temperament and anxiety. It is possible that EC can directly moderate the association between fearful temperament and anxiety (Lonigan et al., 2004).

Our model needs to be interpreted with caution given that it is not derived from a systematic review. Nevertheless, the model may inspire important reflections on the complex cognitive mechanisms underlying anxiety. Our review and resulting model contribute to the discussion of developmental processes linking fearful temperament and anxiety in three different ways. First, we emphasize the two different patterns of automatic AB (i.e., toward or away from threat) that underlie the association between fearful temperament and anxiety. Children and adolescents with atypical AB do not form a homogenous group. The different patterns of atypical AB may have resulted in different types or levels of anxiety symptoms, thus affecting the effectiveness of treatment. For this reason, the two patterns of atypical AB need to be differentiated in fundamental research and in research on applied treatments.

Second, we provide a resolution of the conflicting arguments regarding the nature of AB away from threat. According to some research, avoidance of threat requires regulatory effort that helps disengaging attention from threat (Derryberry & Reed, 2002; Lonigan et al., 2004). However, it is also suggested that avoidance of threat can occur as a reactive tendency (Booth, 2014). It seems that these conflicting findings can be resolved by taking into account the two possible patterns of AB. That is, avoidance is demonstrated in one of the three patterns 1) Avoidance may be a purely automatic, reactive process for people whose predominant reaction to threat is avoiding it. 2) Avoidance may be a combination of an automatic process and regulatory assistance when AC is applied to prolong the automatic avoidance. In this case, AC can be maladaptive in the relation to anxiety. 3) Avoidance may be a pure regulatory process when AC is used to help direct attention away from threat for people who tend to get stuck on them. Actually, avoidance in this situation cannot be called AB in the strict sense, but rather an overcoming of the reactive biases.

As a third contribution, our model highlights the importance of distinguishing between the sub-components of EC in the study of anxiety for two reasons: First, the roles of IC and AC are not always the same (i.e., one is advantageous when another is disadvantageous), and second, IC and AC exert influences through different mechanisms. For example, what IC does for behaviorally inhibited children is affect the malleability of their initial reactive tendency, whereas AC indicates the extent of children’s ability to overcome that reactive tendency. Equally important for the model is our focus on self-regulation as a dynamic process, in that high EC is not necessarily ideal in all contexts. For instance, high IC can exacerbate the association between behavioral inhibition and maladaptive avoidance (Thai et al., 2016). Moreover, AC amplifies the correlation between threat avoidance and anxiety (Ho et al., 2017). Therefore, a flexible control system that is suitable for children’s temperamental characteristics and is motivated or directed by adaptive goals is essential for healthy development.

It is worth noting that there are many individual differences in the associations among fearful temperament, AB, EC, and anxiety. Thus, personalization of the cognitive tasks and trainings has great significance for leading to better assessment and intervention outcomes. In addition, there are children and adolescents who have a high level of anxiety showing no evidence of fearful temperament or atypical AB at earlier ages. Therefore, the model we propose here represents only one possible mechanism that is responsible for the development of anxiety in children and adolescents. Although we try to demonstrate a general pattern, it is important to keep in mind that there are substantial individual differences involved in this process. For example, there can be pathways to the development of anxiety composed of different risk factors (e.g., genetic and environmental influences), different operational mechanisms, and moderating influences. Although those different pathways are beyond what we examine in this review, it remains a critical task and a promising direction for future research to explore the intricate interactions among other risk factors and the mechanisms by which they lead to and maintain anxiety.

Limitations of Current Work

Measurement Challenges and Problems

Although relations among fearful temperament, AB, EC, and anxiety represent a research area that has been well studied, several limitations in the current work preclude the generation of definite conclusions and complicate the larger picture of processes associated with the development of anxiety. The biggest concern is related to the measurement of AB and EC. We address three of those challenges here: 1) the potential low psychometric properties of AB tasks, 2) the difficulties in testing automatic AB, and 3) the inconsistency in measuring the effects of the EC components of IC and AC.

Low psychometric properties of AB tasks.

It is tradition to measure AB with behavioral tasks that use reaction time as an index. The commonly used behavioral tasks include the dot-probe task, the spatial cueing task, the visual detection task, and the emotional Stroop task (e.g., Cisler & Koster, 2010). There are studies using the emotional Go/No Go task as well (e.g., Cohen Kadosh et al., 2014). Although widely used, the psychometric properties of the tasks are questionable. For example, it has been reported in some studies that the emotional Stroop and the dot-probe task display good split-half reliability (Bar-Haim et al., 2010; Dresler, Mériau, Heekeren, & van der Meer, 2009). There is evidence from most studies, however, indicating low split-half and test-retest reliability scores of reaction time tasks used to assess AB (Brown et al., 2014; Kappenman, Farrens, Luck, & Proudfit, 2014; Price et al., 2015).

The low and unstable psychometric properties of AB reaction time tasks may due to the multiple and complex cognitive and behavioral processes involved in task performance. One shared feature of AB reaction time tasks is that they all typically require participants to make behavioral responses according to the specific instructions. For example, in the dot probe and spatial cueing tasks, participants usually need to press a certain key to indicate the location or the direction of a target. In the emotional Stroop task, participants need to tell or select the color ink of the words. Threat is likely to slow or suppress the motor response thereby causing a “freezing” tendency. This makes it difficult to distinguish AB from the response slowing effect and difficult to determine how much AB is contributing to reaction time performance and how much cognitive processes are contributing (Mogg, Holmes, Garner, & Bradley, 2008). This is especially a problem with the emotional Stroop task and the emotional Go/No Go task because the slower response to threat words or faces may be caused by an attention capture at the early input stage or a slower behavioral response at the late output stage. Researchers should be cautious when using the spatial cueing task and comparing trials that include threat stimuli to those that only contain neutral stimuli because AB toward threat (i.e., slower reaction time to invalid threat trials compared with invalid neutral trials) may be partly confounded by a threat-related response slowing effect.

Moog and Bradley (2018) suggest that threat-distractor interference tasks (e.g., emotional Stroop task, emotional Go/No Go task) are different from orienting-based tasks (e.g., dot-probe task, spatial cueing task) in terms of the underlying mechanisms they are tapping. For example, in the emotional Stroop task, AB is indicated by the difference in reaction time or accuracy in response to either neutral or threatening words that always appear in the same location. In tasks like dot-probe, spatial cueing, and visual detection, stimuli appear at more than one location, incorporating visual orienting as part of reaction time. Thus, participants need to continuously orient to and disengage from threat or neutral cues that appear at different locations. Studies that examine the relation between dot probe and emotional Stroop tasks typically report nonsignificant correlations (Dalgleish et al., 2003; Mogg et al., 2000). There is also no correlation between dot probe and spatial cueing tasks, which are similar in that each are orienting tasks (Morales et al., 2017). It is problematic for the AB construct if task choice can affect the findings, especially if tasks are supposed to measure the same thing (i.e., AB in context of visual orienting).

The psychometric problem may also be related to the specific stimuli used in tasks. For example, socially anxious adolescents showed greater AB toward angry adult faces than angry adolescent faces, suggesting that adult faces may be more salient threat cues to adolescents (Wieckowski, Capriola-Hall, Elias, Ollendick, & White, 2018). Furthermore, although most studies use emotional pictures as threat stimuli, some studies use emotional and neutral words (Helzer et al., 2009; Lonigan & Vasey, 2009).

To improve the validity and reliability of AB measures without adversely affecting comparability across studies, it would be ideal to use other measures as a complement of the reaction time-based measurements, such as eye-tracking and psychophysiological measures, which tend to show greater reliability than reaction time (Kappenman et al., 2014; Price et al., 2015). An advantage of eye-tracking and psychophysiological measures is they can precisely detect eye movements and cognitive processes continuously rather than relying on a snapshot of the reaction-time based attention at a specific point (e.g., 500 ms during the trial). Psychophysiological measures can also reveal covert attention shifting and the underpinning neural activities underlying AB. Of course, the reaction time-based tasks have their own merits (e.g., easily analyzed, widely used, participants are highly engaged). A major scientific effort is needed to improve the psychometric properties of the reaction time-based tasks as well as to compare the results generated from the reaction time-based tasks with other measures, in order to make the interpretations of individual findings and the comparisons across studies more feasible and clear.

Difficulties in testing automatic AB.

The second measurement challenge is related to the testing of automatic AB. Specifically, there needs to be consideration of how long the stimulus exposure duration should be in order to better distinguish individuals whose dominant reaction is AB toward threat from those who tend to show a reactive attentional avoidance of threat. Researchers typically use a duration of 500 ms or shorter in the dot-probe task to indicate automatic AB. This is in contrast to the relatively longer exposure time required for EC components of AC and IC to come into play and potentially alter the initial reactive pattern of AB. Researchers should consider that individuals vary on how quickly they engage EC components; thus, AB measured at a specific time, such as 500 ms stimuli exposure in the dot probe task, may indicate an automatic AB process for some individuals but may also involve an EC regulatory process for others. To complicate matters, age may play a role here, as the efficacy and flexibly of using EC improve significantly across childhood (Luna et al., 2004). There is evidence that EC can affect the association between fearful temperament and the AB measured with a 500 ms stimuli exposure duration in the dot-probe task (Susa et al., 2014), which suggests that 500 ms duration cannot clearly dissociate early automatic from late regulatory attendonal processes. On the other hand, exposure time for threat stimuli cannot be too brief because time is required for avoiders to detect the threat and then avoid it.

Although theoretically it is feasible to distinguish automatic AB from the AB that can be influenced by EC, it is difficult to quantify a specific exposure duration which is exclusively automatic and can make an absolute distinction between individuals who show AB toward threat and those who show AB away from threat. Eye-tracking may offer some help to solve this problem. For instance, an individual’s automatic tendency can be inferred from the direction of their first fixation to either a threat or a neutral cue. In addition, the fixation duration on threat or neutral stimuli in a passive-viewing paradigm that diminishes the need of applying EC can also be an indicator of automatic AB. All in all, it remains a challenging task for future research to better detect individual differences in automatic AB, whether it is toward or away from threat.

Inconsistency in measuring EC and especially the AC component of EC.

The third difficulty relating to measurement issues is how to better test EC and its effect on altering the period of automatic AB. Different from the diversity and inconsistency in the measures of AB, studies examining EC have heavily relied on self-reports and maternal reports (Lonigan & Vasey, 2009; Pavlou et al., 2016; Susa et al., 2014), which can be a problematic issue. Laboratory assessments of EC and the parent-report questionnaire measures of EC are not always related because they may tap different aspects of EC. Specifically, laboratory behavioral tasks tend to reflect children’s EC processes of IC and AC in mildly stressful situations, whereas questionnaires focus on EC in typical daily environments (White et al., 2011). In addition, the results of maternal questionnaires are likely to be influenced by mothers’ own characteristics (Leerkes & Crockenberg, 2003). Exclusively depending on questionnaires is especially debatable when other variables in the study (e.g., anxiety) are measured via questionnaires as well because it may overestimate the correlations among the variables of interest due to common method biases. It is important for future research to complement questionnaires with other measurements, such as behavioral task (e.g., Wisconsin Card Sorting Task, Stroop task) and potentially eye-tracking and psychophysiological methods.

In assessing AC it is noteworthy that the degree of engagement required by participants varies widely. In some eye-tracking studies, participants are not required to identify the targets but merely passively view the stimuli, making it difficult to dissociate automatic tendency from capability. This becomes a problem when the main goal of the study is to examine the effect of AC because it remains unclear how much AC is actually applied in the absence of specific task-related goals. Does AC play a role during passive viewing? One can argue that AC may be used to regulate negative emotion, but without clear instructions to do so, we cannot know if participants are engaging in emotion regulation. The potential for wide variation in participant engagement due to the lack of standardization on the utilization of AC may confound the results. To facilitate the active role of AC, participants need a specific goal which does not necessarily require a behavioral response but may be as simple as identifying the shape of the targets in the mind (e.g., Pavlou et al., 2016).

Lastly, there is no answer to the question of how long the stimulus exposure should be in order to permit the effortful regulation of AB. Lonigan and colleagues (2004) claimed that the presentation time should be long enough to allow the strategic control of attention (i.e., 1250-1500ms). Nevertheless, results from empirical studies have indicated that AB measured with a 500 ms exposure duration can be affected by EC (Susa et al., 2014; Thai et al., 2016), even though such brief stimulus duration is too fast to allow a strategic control of the initial AB for the majority of individuals. However, it should also be noted that the exposure duration of the stimuli cannot be too long because then it is likely that most individuals, even those with low EC, can exert an effortful regulation on AB. In a word, either too brief or extremely long presentation duration of threat stimuli will reduce variability in AB. Here again, eye-tracking data may provide insight. For example, the effect of AC and IC may be indicated by the latency of orienting away from threat or away from neutral stimuli to target when there is a specific target-related goal. To sum up, the measurement of automatic AB and the strategic control of AB via EC require standardization of protocols to be most informative across research labs and age groups.

Lack of Longitudinal Studies

Another major limitation is the lack of longitudinal research. It is often the case that fearful temperament, AB, and anxiety are measured concurrently (e.g., LoBue & Pérez-Edgar, 2014; Morales et al., 2015), which prevents our understanding of the developmental patterns of AB and the inference of directionality of the observed effects.

Among the few longitudinal studies, the concurrent AB at age 7 but not the AB at age 5 moderates the association between behavioral inhibition and anxiety at 7 years. This finding may reflect the effect of AB in the manifestation of current anxiety symptoms instead of the onset of anxiety at a later time point. On the other hand, the association between AB and anxiety may become stronger across development and is likely not to be established until middle childhood (White et al., 2017). Importantly, there was little evidence for stability of AB across development. It is possible that a certain combination of internal and external factors, such as parenting and the development of executive functions, may contribute to the changes of AB within individual (White et al., 2017). Therefore, it remains an important direction for future research to include more frequent and multiple-level assessments of AB to better capture the developmental trajectories of AB and the nature of the association between AB and anxiety at different stages of development.

To justify the causal relationship between AB and anxiety, Kindt and Van Den Hout (2001) took a developmental perspective in examining the associations between the two. As proposed in our review, early AB is part of the normal developmental phenomena that occurs in each individual. With age, children with low levels of anxiety are able to modulate the biases by developing better cognitive control abilities. Anxious children or children who are prone to anxiety would not learn how to inhibit AB. It is the continuation of AB rather than AB itself that is involved in the onset and maintenance of anxiety at a later stage. Similarly, in a review, Van Bockstaele and colleagues (2014) suggested that the relation between AB and anxiety can be bi-directional and mutually informing.

In our model, we stress the one-way direction from fearful temperament to AB and to anxiety because our main goal is to examine potential early risk factors and mechanisms that underlie the onset and maintenance of anxiety in later development. It remains a promising avenue for future research to examine the relation between AB and anxiety with longitudinal designs and sophisticated statistical analyses (e.g., cross-lagged panel analysis). For instance, high levels of anxiety caused by sustained AB may further perpetuate the AB pattern. For studies that aim to examine and prevent the development of anxiety, using longitudinal designs will allow cause-and-effect conclusions in terms of the predictive effects of fearful temperament, AB, and EC on anxiety. This will provide stronger support for these complex associations as well as the potential for specific advice for early screening and interventions.

Deficiency of Cross-cultural Studies

The lack of cross-cultural research on AB is a major limitation of the current research literature. Cultural differences are not fully discussed and represented, with the majority of the participants in AB studies from Western cultures (e.g., White et al., 2017). Attention to cultural differences is particularly important in research on children’s socioemotional development, given the great influence culture has on parents’ and teachers’ expectations and attitudes toward children’s behaviors. For example, in Eastern culture that is characterized by collectivistic and socially restrained norms, being fearful or shy is not regarded as deviant as it is in Western culture (Sanson, Hemphill, & Smart, 2004). Moreover, children with high IC are more likely to be valued and commended in Eastern culture, given the links between IC and compliance. As such, there may be greater opportunities for fearful children to receive social support from parents, teachers, and peers (Eggum-Wilkens et al., 2016). Therefore, the effects or the degree of effects of fearful temperament and EC on anxiety may not be the same for children from Western and Eastern cultures. The mechanisms by which they affect anxiety are also likely to be different. More cross-cultural studies are needed to examine if the relational patterns of fearful temperament, AB, EC, and anxiety found in one culture hold in another one.

Questions for Future Research

What determines the direction and stability of the two patterns of automatic AB (i.e., toward or away from threat) and how they are involved in the development of anxiety?

Although it has been well established that AB is associated with child and adolescent fearful temperament and anxiety, previous research is inconsistent regarding the impact of direction of AB. There are studies reporting that fearful or anxious children are characterized by AB toward threat (Pérez-Edgar et al., 2010; Waters et al., 2010), with others reporting that fearful or anxious children are characterized by AB away from threat (LoBue & Pérez-Edgar, 2014; Shechner et al., 2017). The two patterns of AB may represent separate mechanisms of anxiety experienced by different individuals. Thus, the first question that needs to be the focus of future research is which individual characteristics are responsible for the different patterns of AB.

We proposed in this review that different levels of early fearful temperament may elicit different patterns of atypical AB, which is involved in the develop different intensities of anxiety. In addition to this, other individual characteristics may moderate the association between fearful temperament and AB. For example, adults with greater defensiveness, a high need for social approval and an unwillingness to report negative emotions, tend to show avoidance of social threat (Mogg et al., 2000). In addition, individuals who show high fear and trauma symptoms tend to show threat avoidance (Mogg, Bradley, Miles, & Dixon, 2004; Mogg & Bradley, 2018). This is consistent with the findings that threat avoidance is associated with fear-related disorders (i.e., specific phobia, social anxiety, post-traumatic stress disorder). Evidence from psychophysiological studies show that parietal alpha electroencephalogram (EEG) asymmetry is correlated with the direction of AB. Specifically, right parietal EEG asymmetry predicts AB threat avoidance and left parietal asymmetry predicts AB toward threat (Pérez-Edgar, Kujawa, Nelson, Cole, & Zapp, 2013). Similarly, among individuals who have right frontal EEG asymmetry, those with low right parietal activity show AB toward threat, and those with high right parietal activity show AB away from threat (Grimshaw, Foster, & Corballis, 2014). Thus, it is important for future research to test what individual factors affect or moderate the direction of AB.

Another major task for future research is to test the stability of the direction of AB concurrently and longitudinally. It has been suggested that reliable patterns of AB tested concurrently across tasks, regardless of whether it is AB toward or away threat, are related to high levels of anxiety (Morales et al., 2017). This raises the probability that stable concurrent AB itself may be an important indicator relating to fearful temperament and anxiety. To the best of our knowledge, however, no study has examined how individual differences in longitudinal stability of AB pattern affect the risk of anxiety. Children and adolescents who show stable AB either toward or away from threat longitudinally may develop different levels of anxiety compared to those who show changeable patterns. There is a great need for future research to test if AB patterns toward and away from threat shift as a function of development and whether shifting pattern changes might track the level of anxiety.

In addition, it has been reported in two studies that AB toward or away from threat represent two mechanisms that underlie different types of anxiety problem (Salum et al., 2013; Waters et al., 2014). Specifically, children who show AB toward threat are at higher risk for having distress-related anxiety disorders (e.g., general anxiety disorders), whereas children who demonstrate AB away from threat tend to have fear-related anxiety disorders (e.g., specific phobia, social anxiety). Both studies measured AB with the dot probe task. More research is needed to replicate the findings with other measurement methods and different samples of children.

What are the mechanisms by which different components of EC affect the relation between fearful temperament and anxiety through AB?

There are continuing efforts to examine the effect of EC in the relations among fearful temperament, AB, and anxiety. Nevertheless, previous research has not been adequate to generate conclusions. Questions and findings remaining to be answered and replicated include: Do the roles of IC and AC differ in terms of exacerbating or alleviating the adverse effect of automatic AB? Can AC be maladaptive in the development of anxiety? And if so, under what circumstances that can happen? Is IC adaptive or nonadoptive in the development of anxiety? What affects the role of IC? How does motivation affect the role of AC in the process?

According to previous research, the effect of IC is different for children with and without fearful temperament. Specifically, rather than indicating an adaptive self-regulation ability for children at no or lower risk (Liu et al., 2018), IC appears to contribute to an inflexible control system that freezes and prolongs the initial reactive response for fearful children (Henderson & Wilson, 2017; White et al., 2011). More work and replications are needed in revealing the role of IC in the development of anxiety. How IC affects the association between fearful temperament and anxiety through AB and whether IC buffers against or increases risks in the context of AB needs further empirical examination.

Although AC is generally regarded as a beneficial cognitive ability, recent findings that high AC combined with attentional avoidance of threat predicts higher levels of anxiety should be noted. This work needs replication, especially with supports from longitudinal studies (Grafton et al., 2016; Ho et al., 2017; Taylor, Eisenberg, & Spinrad, 2015). The nature of the relation between AB and anxiety seems vary by AC. Specifically, at low levels of AC, greater difficulties in disengaging attention from social threat are correlated with increasing anxiety. In contrast, at high levels of AC, faster attentional disengagement from threat are associated with higher levels of anxiety (Taylor et al., 2015). Therefore, there might be two attentional mechanisms maintaining the development of social anxiety. First, poor AC interferes with efficient attentional disengagement from social threat; or second, high AC facilitates maladaptive avoidance of threat and thus precludes adequate emotional processing of the threat. In this situation, sustained avoidance of threat can be regarded as a combination of an automatic tendency and a regulatory assistance, rather than a purely automatic process.

In addition, it has been suggested that the effect of AC on AB may reflect motivation or goal priorities under particular circumstances (Mogg & Bradley, 2016). For example, depending on the initial direction of the automatic AB (i.e., either toward or away from threat) when the primary goal to is to reduce distress or discomfort, AC may help to disengage attention from threat or maintain attention on neutral stimuli. When the major goal is to identify task-related targets, AC may help to disengage attention from threat or from neutral stimuli and orient to the target. Future studies should test how motivational goals affect the role of AC in this process.

How do environmental factors affect the association between AB and anxiety?

Associations between atypical AB and anxiety may be moderated by environmental factors. For example, children from cohesive families and disengaged families show a similar early-stage AB toward threat. Cohesive families have the highest level of emotional closeness and independence, which represents an optimal family relationship. Disengaged families have the lowest level of emotional closeness and independence, which indicates emotional distant and high conflict relationships among family members. It may be that similar AB toward threat exhibited by children from such disparate types of families serves different functions. Particularly, in low-risk families (e.g., cohesive families), sustained vigilance to threat is considered to be adaptive, involving the appraisal of the stimulus as low threatening and representing prosocial functions (e.g., sensitivity to others’ distress). In high-risk families (e.g., disengaged families), however, consistent vigilance to high risks indicates prolonged threat rumination and may become the main source of developing psychopathology (Lindblom et al., 2017). Family violence predicts greater anxiety and trauma symptoms for children (age 3-5) who showed AB toward threat (Briggs-Gowan et al., 2015). In addition, in a community sample of adolescents (age 9-15), negative parenting (e.g., authoritarian style and negative affect) predicts greater social anxiety through AB toward threat (Gulley, Oppenheimer, & Hankin, 2014). Future research may want to examine how environmental factors come into play and relate to AB and anxiety.

What are the implications for interventions?

Children and adolescents who are characterized by different patterns of automatic AB show different responses to interventions (i.e., ABM, cognitive behavioral therapy or CBT; Price, Tone, & Anderson, 2011). Furthermore, recent findings suggest that ABM may not be as effective in treating anxiety as expected given that it is designed to specifically target AB toward threat (Ollendick et al., 2018). Perhaps efforts are needed to distinguish between individuals whose dominant automatic AB is toward or away from threat and examine how individuals in those two groups are different in responding to different types of treatments. If there is differential response, how can we adjust the treatments to satisfy those who do not respond well to the traditional treatments? For example, it has been reported CBT is more effective in reducing anxiety for children who show AB toward threat compared with those who tend to avoid threat, perhaps because the latter have limited chances to fully process the threat (Waters et al, 2012; but see Legerstee et al., 2009). Thus, for children with automatic AB away from threat as the main vulnerable factor of anxiety, rather than training them to look more to neutral stimuli using the traditional ABM, a treatment that facilitates orienting toward threat may be more efficacious for increasing exposure to threat, accelerating habituation, and reducing the aversive salience of threat. In a word, it is important to customize interventions to meet the needs of individuals characterized by different patterns of AB.

Another promising way to intervene may be directly targeting on the abnormal threat detection that precedes both vigilance and avoidance. For example, what CBT does is to increase the exposure to threat with the learning of adaptive cognitive reconstructing and coping skills, thus providing children opportunities to realistically evaluate threat and enhance their thresholds for detecting threat. Todd and colleagues (2012) suggested that rather than training a child to orient away from the threat that has already been detected, more efforts are needed in changing children’s pre-tuning to threat that precedes the stimulus presentation. For example, efforts should be focused on training children to locate the smiling faces among an array of faces with negative expressions, thus increasing sensitivity to positive stimuli and away from negative stimuli.

Finally, there may be great value in teaching children to focus on cognitive reconstruction or reappraisal as a strategy to regulate negative emotion associated with threat. For children who show consistent AB toward threat, reappraisal helps them to replace the negative thoughts with more neutral or positive ones, thus reducing the negative influence caused by sustained vigilance to threat. For children who tend to avoid threat and wish to regulate distress, AC may not assist in maintaining the maladaptive avoidance. Instead, AC may direct attention to threat if children habitually use reappraisal to regulate negative emotion, which requires fully evaluation of negative stimuli, thus facilitating the habituation of threat.

Concluding Remarks

We have highlighted three key points in the study of threat-related AB. First, AB varies between threat vigilance and avoidance, which represent distinct implications linking fearful temperament and anxiety. Second, the roles of IC and AC in moderating AB and anxiety depend on individuals’ temperament-based characteristics, dominant automatic biases, and goal priorities. Thus one cannot simply state that higher IC or AC is unconditionally ideal in socioemotional development. Third, developmental effect is involved in the process as AB may change over time depending on the level of fearful temperament and EC. Considering these key points, we proposed a two-mechanism developmental model suggesting that biologically determined fearful temperament during early life leads to atypical AB (toward or away from threat). Over the course of development, the continuation of fearful temperament interacts with AB, determining its severity and longitudinal duration. As the prefrontal cortex continuously develops, EC comes into play and exhibits an increasingly stronger effect during late childhood and adolescence by moderating the association between initial AB and sustained AB. The perseverance of the sustained biases across development triggers the formation of anxiety.

Our review sheds light on the current literature by demonstrating a more nuanced developmental pattern of associations among fearful temperament, AB, EC, and anxiety. We have offered suggestions to promote a more detailed and in-depth understanding of AB and EC. Specifically, we focused on the need for: 1) improving and standardizing research-related tasks, 2) conducting longitudinal and cross-cultural studies, 3) examining the role of motivational goals and environmental factors in the process, and 4) developing interventions with higher pertinence and flexibility. There are complex associations among the different processes associated with the development of anxiety. Much more work is required before effective interventions can be developed.