The relation between generalized anxiety disorder symptoms and content-specific interpretation biases for auditory stimuli in children

Highlights

• Auditory Interpretation task: two words that differ by one phoneme are blended.

• Children with GAD symptoms show more negative interpretations.

• But only for words related to GAD and not to other fear-related words.

• This result was only found with the multiple-choice version.

• Children display cognitive biases specific for fear-relevant stimuli.

Abstract

Background and objectives

Cognitive theories of fear suggest that biases in interpretation are content-specific: Fearful children should only interpret materials negatively if they are specifically related to the content of their fear. So far, there are only a few studies available that report on this postulated content-specificity of interpretation processes in childhood fear. The goal of this study was to examine interpretation bias and its content-specificity in children with varying levels of Generalized Anxiety Disorder (GAD) symptoms.

Methods

In an Auditory Interpretation Task (AIT), two words that differ by one phoneme are acoustically blended so that one can hear only one of the words. In the current AIT, we included GAD-related blends, negatively-valenced fear-related blends and positive blends. Multiple-choice (n = 371) or open-ended (n = 295) responses were collected from 666 nonclinical children between 7 and 13 years of age.

Results

Children with higher levels of self-reported GAD showed significantly more negative interpretations of ambiguous GAD-related blends in the multiple-choice version than children with lower levels of GAD. There were no differences when interpreting the other ambiguous blends. This result was not found with the open-ended version.

Limitations

Effects were relatively small, some GAD-stimuli were sub-optimal, and the task was administered in a classroom setting. Even though we ensured that all children were able to hear all words clearly, this may have impacted the results.

Conclusions

The findings only partly support the idea that fearful children display cognitive biases specific for fear-relevant stimuli, and more research is needed to replicate the results and test the usability of the AIT.

Introduction

It is widely recognized that fearful children interpret the world as more threatening than their non-fearful peers. Cognitive theories indeed emphasize the importance of underlying automatic cognitive processes, such interpretation processes, in the onset and maintenance of fear (e.g., Beck, Emery, & Greenberg, 1985). The central assumption of these theories is that cognitive processes are driven by schemata. Schemata are cognitive structures of associations between knowledge elements that influence perception, attention, interpretation, and memory. In fearful individuals, schemata that are organized around the themes of threat and danger are chronically overactive, and many situations and stimuli are (automatically) associated with danger and fear, resulting in cognitive biases (e.g., Williams, Watts, MacLeod, & Mathews, 1997; for a schema-based theory of childhood fear, see; Daleiden & Vasey, 1997). These biases, such as biases in interpretation are thought to be content-specific: Fearful children should only interpret materials negatively if they are specifically related to the content of their fear and not to other-fear-related words or positive information. So far, there are only a few studies available that report on this postulated content-specificity of interpretation processes in childhood fear (see also Muris, 2010). Knowing more about the content-specificity of cognitive biases in children is important, because this could help us to improve treatment programs for fearful children. For example, would children profit more from generic treatments or is it important to focus on the specific fear cognitions of a child during treatment? Several authors have indeed expressed the need for more research on content-specificity in interpretation biases in fearful children (e.g., Muris, 2010; Stuijfzand, Dodd, Pearcey, Creswell, & Field, 2017). Therefore, the main goal of this study was to investigate the content-specificity of interpretation processes in a sample of children with varying levels of Generalized Anxiety Disorder (GAD) symptoms.

Research in fearful children has provided evidence of biased interpretation processes related to fear (for reviews, see Muris, 2010; Muris & Field, 2008; for a meta-analysis, see Stuijfzand et al., 2017). Interpretation bias refers to the phenomenon that fearful individuals have the tendency to interpret ambiguous situations as threatening. Interpretation biases are commonly assessed using direct measures, such as the Children's Negative Cognitive Error Questionnaire (CNEQ; Leitenberg, Leonard, & Carroll-Wilson, 1986). In direct measures, participants are asked directly about their feelings and opinions towards different situations. Advantages of these measures are that they are fast and easy to administer and that they are reliable in the sense that measurement errors are small. A limitation of direct measures is that participants only report what they are willing and able to report because responses are given in a controlled, deliberate manner (e.g., Nisbett & Wilson, 1977). As a result, answers may be biased by social desirability, limitations of introspection, or experimenter demands (e.g., Bijttebier, Vasey, & Braet, 2003). As a result, self-reports (direct measures) tap into more controlled processes related to fear, while indirect measure are needed to tap into more automatic processes, such as interpretation processes (e.g., Strack & Deutsch, 2004). Examples of indirect measures include reaction time tasks such as the Affective Priming Task (APT; De Houwer, 2003), the Approach Avoidance Task (AAT; Rinck & Becker, 2007), and the Emotional Expression Task (e.g., Richards, French, Nash, Hadwin, & Donnelly, 2007). It should be kept in mind, however, that reaction-based paradigms are usually less reliable than direct measures. This might especially be true in child samples, because children are generally distracted more easily and have relatively short attention spans (Huijding, Wiers, & Field, 2010).

To provide a compromise between direct and indirect measures, many studies of interpretation bias use variations of an “ambiguous scenario” paradigm. In this paradigm, children are asked to finish scenarios (short texts) about everyday situations. In this way, children are not directly asked how afraid they are, but are asked to finish stories about different situations related to fear. Commonly, responses to these scenarios involve a multiple-choice format in that children choose their ending from a number of provided possibilities. Alternatively, responses to scenarios can be open-ended in that children are asked to produce their own ending to the scenarios. Although the ambiguous scenarios task and its variations are not able to capture automatic aspects of cognitive processes as clearly as reaction time based paradigms, the general view is that this task relies less on insight than questionnaires or interviews as it does not directly ask how fearful a child is (see also, Muris, 2010).

So far, most studies using ambiguous scenarios found that fearful children show a tendency to interpret ambiguous situations in a negative way (for reviews, see Muris, 2010; Muris & Field, 2008). Some of these studies also explored whether interpretation biases are specific to the content of the fear, finding evidence for the specificity of interpretation biases related to fear (e.g., Bögels, Snieder, & Kindt, 2003; Klein et al., 2014). Also, Stuijfzand et al. (2017) found a moderating effect of the content of ambiguous scenarios in their meta-analysis; the relation between fear and interpretation bias was stronger when the ambiguous scenarios matched the fear subtype under investigation. However, the authors pointed out that this effect was mainly driven by studies that focused on social anxiety, and clearly more research is needed on other fears, such as generalized anxiety disorder.

Although the “ambiguous scenario” paradigm and its different variations are often seen as more indirect than questionnaires or interviews, this paradigm is still highly sensitive to experimenter demand and social desirability (e.g., Muris, 2010). Therefore, developing more sophisticated indirect measures may provide a clearer picture of biased interpretation processes that play a role in childhood fear. Examples of tasks that can be used to explore interpretation biases in children in a more indirect way without using reaction times include the homophone task (Hadwin, Frost, French, & Richards, 1997; Taghavi, Moradi, Neshat-Doost, Yule, & Dalgleish, 2000) and the Auditory Interpretation Task (AIT; Dearing & Gotlib, 2009; Lawson, MacLeod, & Hammond, 2002). In the homophone task, words with two different meanings are presented auditorily; the word has both a neutral meaning and a threatening meaning (e.g., pain versus pane). The child's task is to select a response (e.g., picture or sentence) that bests represents the meaning of the word. Several studies using this paradigm found that fearful children have a tendency to select homophones with a threatening meaning significantly more often than non-fearful control children (e.g., Hadwin et al., 1997; Taghavi et al., 2000). A disadvantage of the task is that in most languages other than English, it is close-to-impossible to find homophones with a neutral and a fearful meaning.

In the AIT, two words that differ by only one phoneme (e.g., “threat” and “thread”) are presented acoustically at the same time. Due to the fact that phoneme perception is categorical (e.g., Eimas & Corbit, 1973), listeners do not hear a blend of phonemes, but either one phoneme or the other. As a result, they will hear only one of the two words, usually without being aware of the fact that there might be a second possible interpretation of the perceived sound stream. When using this task to assess threat interpretation, one of the words is related to fear, while the other word has a neutral valence. The children's task is to simply listen to the words and report what they hear. The potential advantages of the AIT are manifold: The task is probably less sensitive to experimenter demand and social desirability than questionnaires and ambiguous scenarios as the child generally only hears one word without being aware that there are more words possible, it is more versatile than the homophone task, and it is possible to choose all kinds of different words. This task might therefore be a better alternative than the homophone task or the “ambiguous scenarios” paradigm.

To the best of our knowledge, only Dearing and Gotlib (2009) have used an AIT to study interpretation bias in children. In their study, daughters of mothers with a history of at least two depressive episodes or daughters of healthy mothers were asked to listen to 50 word pairs and choose the word they thought they heard out of two possible options. The test words were either related to depression, social situations, or positive situations. The two groups did not differ on current depression symptoms. Children of mothers with a history of depression identified significantly more depression-related words than the control group, but there was no difference in words related to social situations. The children of mothers with a history of depression also chose the positive words significantly less often than the control group.

The present study was based on the study by Dearing and Gotlib (2009), and was designed to investigate whether children with varying levels of Generalized Anxiety Disorder (GAD) symptoms display a negative interpretation bias, and whether this bias is specific for GAD-related materials. We created two different versions of the AIT; approximately half of the children in this study performed a multiple-choice version that was very similar to the study of Dearing and Gotlib (2009). The only two differences were the content of the words which were GAD-related, negatively-valenced fear-related or positively-related words, and we used a multiple-choice version that included four options instead of two options to reduce the effect of a possible response bias. Additionally, the other half of the participating children performed an open-ended version. In this version, children were asked to simply write down the word they heard. We created this version in order to make the task even more indirect to reduce possible bias due to the multiple-choice format.

Based on earlier studies on interpretation bias in fearful children, we hypothesized that children with higher levels of GAD symptoms would interpret GAD-related-neutral word blends as negative more often than children without symptoms of GAD. In line with the content-specificity hypothesis, we expected that children with higher levels of GAD would display an interpretation bias for GAD-related stimuli only, and not for the negatively-valanced fear-neutral word blends or the positively valenced-neutral word blends when compared to children with lower levels of GAD symptoms.