Investigations of anxiety-related attentional biases for threatening information in adults have consistently demonstrated that anxious adults preferentially direct attentional resources towards threat-related information. However, in contrast to the robust anxiety-related effect amongst adults, studies conducted to date with children have produced mixed results, with some studies indicating that a threat-related attentional bias may be common to children in general, regardless of anxiety status. In light of the inconsistencies in the childhood domain, the present research was designed to investigate the normative developmental parameters of children's attentional processing of differentially valenced stimulus materials, and to examine whether children with anxiety disorders, like their adult counterparts, display patterns of attentional processing that differ compared to the normative data. Three experimental studies were conducted to achieve these aims. In each of the studies, four well-established cognitive and psychophysiological procedures were employed, including the dot-probe detection task, the array tasks (i.e., A9 task: 3 x 3 matrices; A4 task: 2x2 matrices), the emotional Stroop task, and the startle eyeblink modification paradigm.
The first experiment was conducted with a sample of 48 non-selected adult participants (27 females; 21 males) aged between 17 and 44 years who participated on an informed consent basis. The purpose of conducting this experiment was to assess the adequacy of the experimental tasks and to determine whether the predicted patterns of processing could be reliably obtained in these tasks. In recent years, cognitive accounts of anxiety disorders have become increasingly more complex reflecting the integration of evolutionary perspectives of fear and anxiety within information processing models. In recognition that a threat-related bias during the early stages of attentional processing may serve as an evolutionarily-driven adaptive device, these more recent formulations propose that an attentional bias for high threat stimuli may be common to individuals in general, with the magnitude of this bias being more pronounced and generalised for anxious individuals. Based on these recent cognitive formulations, it was predicted in the first experiment that non-selected adults would display an attentional bias for high threat pictures compared to neutral and pleasant pictures. The results of the A9 task (i.e., 3 x 3 matrices) supported the hypothesis by demonstrating that non-selected adults were faster to detect fear-relevant target pictures within backgrounds of eight fear-irrelevant stimuli than to locate fear-irrelevant targets within backgrounds of eight fear-relevant distractors. However, when the number of distracting stimuli was reduced from eight to three in the A4 task (i.e., 2 x 2 matrices), the difference in reaction-times for detecting fear-relevant targets compared to fear-irrelevant targets amongst opposite backgrounds was not significant. The differences in results from the A9 task and the A4 task are discussed in terms of the differential impact of stimulus complexity on the processing of fear-relevant and fear-irrelevant target pictures. The results from the dot-probe detection task provided support for the hypothesis of preferential processing of threat-related stimuli by demonstrating that non-selected adults were faster to locate dot-probes presented after threat pictures than after neutral or pleasant pictures. The results from the emotional Stroop task revealed that non-selected adults were faster to name the colour that both threat and pleasant pictures were presented in compared to neutral pictures. The unexpected finding of faster rather than delayed colour-naming of threat and pleasant pictures is discussed in terms of the differential ease of processing pictorial versus word stimuli, which are the stimuli most commonly used in emotional Stroop tasks. The finding of faster colour-naming of both threat and pleasant pictures is discussed in terms of the attentional characteristics (e.g., interest value or arousal) rather than emotional characteristics (e.g., valence) of the stimuli driving responding under single stimulus conditions. The results of the startle eyeblink task revealed an attentional account of long lead interval startle modification in that startle-magnitude was facilitated during threat and pleasant pictures and inhibited during neutral pictures relative to the baseline. This pattern of results suggests that the attentional characteristics (i.e., interest value or arousal) rather than the emotional characteristics (i.e., valence) of the stimuli may have been more influential in modulating startle reflexes during long lead intervals.
The second study was conducted with a sample of 105 children (63 girls; 42 boys) aged between 9 and 12 years who were recruited from a local primary school on an informed consent basis. As the purpose of the second experiment was to assess the normative developmental parameters of children's information processing, the children in the second experiment represented a large sample of non-selected Australian children. Children were only excluded from the experiment if they were reported by parents to have an intellectual impairment, a learning disability or if they did not speak English as their first language. An assessment of children's anxiety status as reported by children and by their parents revealed levels of anxiety that were within the normal range for non-clinical children. The results from the cognitive and psychophysiological tasks revealed a pattern of results very similar to that obtained for the non-selected adults. In the A9 task, children were faster to locate fear-relevant target pictures amongst fear-irrelevant distractors than fear-irrelevant targets amongst backgrounds of fear-relevant distractors. Similar to the results in the first experiment for the A4 task, there was no significant difference in children's speed to detect fear-relevant versus fear-irrelevant targets amongst the opposite backgrounds. In the dot-probe detection task, children were faster to locate dot-probes presented after threat pictures than after neutral or pleasant pictures at the right probe-position, but were more efficient in responding to dot-probes after both threat and pleasant pictures at the left-probe position. These results were discussed in terms of children's ability to regulate the processing of differentially valenced picture stimuli. Similar to the results in the first experiment, children were faster to colour-name both threat and pleasant pictures compared to neutral pictures in the emotional Stroop task. In the startle eyeblink task, the opposite pattern of long lead interval startle modification effects was obtained compared to the results observed with adults in the first experiment. Children's startle-magnitude facilitation was larger during neutral pictures compared to both threat and pleasant pictures at the long lead intervals. These contrasting results were discussed in terms of developmental differences between adults and children in their ability to regulate the processing of differentially valenced stimulus materials. A significant differential effect of picture valence was also observed at the 60 ms lead interval, with startle eyeblink significantly facilitated during threat pictures and inhibited during neutral and pleasant pictures. This finding was discussed as potentially reflecting a bias at the very early stages of threat stimulus processing.
The third study was conducted with a sample of 23 clinically-anxious children (9 girls; 14 boys) aged between 9 and 12 years and data from 23 children selected from the second experiment matched for age and gender. In addition to meeting the inclusion criteria in Experiment 2, children in the Anxiety Group were reported by parents to have no existing psychological diagnoses and to have not received psychological or psychopharmacological treatment for anxiety. Children's anxiety status was determined using the parent version of a semi-structured anxiety disorders interview schedule. Child- and parent-report measures supported the clinically-significant nature of the anxiety status of children in the Anxiety Group. Consistent with an increasing body of evidence suggesting that a threat-related attentional bias may be common to children in general, the results from the cognitive and psychophysiological tasks indicated that both the Anxiety Group and the Control Group preferentially processed pictorial stimuli depicting threat scenes relative to pictures representing neutral or pleasant scenes. In the A9 task, both groups of children were more efficient in detecting fear-relevant targets amongst fear-irrelevant backgrounds than in locating fear-irrelevant targets amongst fear-relevant backgrounds. The results from the A4 task continued to be inconsistent with the hypotheses, with no significant differences emerging in both groups of children. However, post-hoc analyses revealed that children in the Anxiety Group displayed shorter reaction-times overall compared to the Control Group. The dot-probe detection task demonstrated that both groups of children were faster to locate dot-probes presented after threat pictures than after neutral or pleasant pictures. Post-hoc analyses revealed that anxious children remained more vigilant in responding to threat and pleasant pictures throughout the task whereas the reaction-times of the Control Group slowed as the task progressed except for right-probed threat pictures. Furthermore, anxious children tended to show faster reaction-times overall compared to the Control Group. The results from the emotional Stroop task indicated no significant effect of picture valence. These results were discussed in terms of the basic valence effect being unstable in the emotional Stroop task. However, the Anxiety Group was faster overall in colour-naming pictures compared to the Control Group. Post-hoc analyses also revealed that across the three valence categories, the colour-naming latencies of the Control Group were significantly longer during the last half than the first half of the trials whereas the Anxiety Group showed no significant differences in colour-naming latencies across the two halves of the task. In accordance with the results from the second experiment, the startle eyeblink task revealed that both groups of children displayed larger startle-magnitude facilitation during threat pictures than during neutral or pleasant pictures at the 60 ms lead interval. Startle-latency shortening was also significantly longer at the 60 ms lead interval for control children than for anxious children. Since it remains largely unclear what psychological processes are reflected in eyeblink modification at lead intervals of 100 ms or less, it is premature to consider the psychological significance of the differential effect of children's anxiety status upon startle eyeblink latency. Taken together, the results of the third experiment indicate that an attentional bias for threatening pictorial stimuli may be common to children in general, regardless of anxiety status. These findings are discussed in terms of the suggestion raised in previous research that children may lack the cognitive ability to regulate the processing of emotional information. Extending on the well-accepted contention that 'normal' childhood fears play a functional role in children's lives, it was postulated, as it has been in the adult literature, that an attentional bias for objectively threatening stimuli may serve as an adaptive device for all children to promote the rapid detection of, and subsequent escape from, objectively threatening and dangerous situations. The findings that anxious children showed faster reaction-times overall and continued to remain vigilant in responding throughout the tasks are discussed in terms of possible early cognitive characteristics that may distinguish anxious and non-anxious children. A discussion is also offered about the differential developmental progression of the threat-related attentional bias into adulthood for anxious and non-anxious children.