Experiment 3: Implicit Emotion Detection
6 General Discussion
6.2.2 Theme 2: Methods
The second theme revolves around the aim of exploring experimental
techniques beyond the dot-probe which allow a closer investigation into attention bias in social anxiety. The lack of clear-cut findings from the experiments conducted within this thesis might easily mislead one to consider the tasks unsuitable for further study into attention bias. However, it is important to keep in mind that none of these experiments have been frequently used to explore attention bias in social anxiety. As such, there is much to learn from these experiments that can inform further
investigation using these tasks. With this in mind, the following section will focus on what the methods offered as used here as well as methodological limitations and suggestions for future work to ensure that these methods are used to their full potential.
By far the largest limitation of the attentional blink experiments was that the tasks were too easy. Especially in the child study, but also in experiments 1 and 2 in adults, there was a strong ceiling effect of performance. This was despite extensive piloting and adjustments to the paradigm. It is likely that this ceiling effect stems from human’s expertise for, and sensitivity to, faces (Kanwisher, 2000), as well as the tasks’ targets and distractors perhaps not being similar enough (Müsch et al., 2012). To achieve larger and more normally distributed variance in the data to analyse individual differences better, future attentional blink studies with faces should aim for more task difficulty, for example by backward-masking the targets. Experiment 3 of chapter 2, which chronologically took place after the child study, implemented these suggestions. Indeed, the data for experiment 3 met assumptions for analysis. A further consideration is that disgusted faces may be less threatening than angry faces, and are
therefore less successful for investigating threat bias. In conclusion, the attentional blink, with the right adjustments, has the benefit to potentially unlock the time course of attentional bias toward emotional faces in relation to facilitated attention as well as disengagement from threat in both children and adults but careful consideration must be given to task parameters to ensure an appropriate level of difficulty.
The saccadic curvature paradigm has the potential to look at early stages of facilitated attention to threat, later processing stages of delayed disengagement, as well as top-down driven spatial avoidance. To unlock this potential, a few suggestions are proposed. Firstly, the degree with which distractor items capture attention should be increased. As noted in chapter 4, a previous study was conducted with smaller images which were more peripherally located. This study found no effects. It can therefore be argued, that bottom-up saliency could be improved further by increasing the image size and placing the images more centrally, closer to the vertical saccade trajectory. A second suggestion for future studies is procedural. Whilst this design was unique in that it looked at fast-onset versus slow-onset saccades for each participant individually, the cue, target, and distractors appeared simultaneously on every trial. By varying the stimulus onset asynchrony as in previous studies (e.g. Nummenmaa, Hyönä, & Calvo, 2009; McSorley & Van Reekum, 2013), saccades driven by task-characteristics could be analysed alongside individual onset-times.
The ssVEP study was designed to investigate sustained attention toward threat under different task-demands (see theme 3). A benefit of the paradigm was that the evoked response emerged after brief stimulation, which makes this design powerful and efficient. Given how novel the method is, the results are insightful, but further improvements to the task could be made. One of the suggested adjustments would be implementing a different face condition. Currently, the stimulus changed after 30
flickers. Including a condition during which the face is changed at every flicker (see for example Gerlicher et al., 2013), would provide an individual baseline for each participant against which adaptation to repeated emotional stimuli could be measured. Furthermore, the current design used conditional randomisation of stimulus
presentation order. The conditions were that relative to the presented stimulus, the subsequent stimulus had to be either the same identity, or express the same emotion. After stimulus presentation, it takes a few seconds for the ssVEP response to establish (Rossion, 2014; Rossion & Boremanse, 2011). For this reason, the first trial was omitted from the analysis. Due to the randomisation of order, there was no control over which stimulus was presented in the first trial, essentially leading to missing data for every participant. Future designs could include a neutral face at the beginning, which only serves for the ssVEP response to be evoked. With these minor adjustments to the approach the ssVEP offers a powerful and exciting tool for examining whether sustained attention toward threat is maintained in social anxiety,
Reflecting on this second theme of methods, with a few adjustments as highlighted the methods used in this thesis could offer unique insights into how emotional face processing exactly is impaired or preferentially allocated in social anxiety and to contribute to the fields of individual differences and social cognition more broadly.