Late processing stage (300-600ms)

The present ERP data did not find any statistically significant differences in waveforms for emotional versus neutral faces during the third time-window.

To ensure no late activity was missed, the time-window got extended to the first 900ms after stimulus onset, but again, no significant brain activity related to the specific processing of emotions was recorded over posterior regions. According to the majority of past ERP studies investigating emotion recognition in faces, emotion effects are consistently reported to occur within the first 200-300ms of processing (e.g. Eimer & Holmes, 2007; Sato et al., 2001).

Commonly, within the first 400ms, a late posterior negativity or frontal positivity for emotional compared to neutral faces is found, which possibly reflects projections from the amygdala back to the visual cortex (Sato et al., 2001).

One possible explanation for the lack of emotion effects in faces could be that the used stimuli were not salient enough to elicit a significant ERP signal. Especially the ERP signal around 300 and 450ms seems to be higher for high-arousing compared to low-arousing faces (Rozenkrants & Polich, 2008). Similarly, between 400 and 800ms, the late positive potential (LPP) was found to be enhanced for images with a high compared to low degree of arousal (Leite, Carvalho, Galdo-Alvarez, Alves, Sampaio, & Goncalves, 2012). Hence, although not statistically tested, it is possible that differences in arousal levels between neutral and emotional stimuli in the present study are too subtle and did not get picked up in the ERP signal.

Secondly, the current study did not find different waveforms for angry compared to happy faces during the final time-window. The lack of emotion-specific effects could, again, be attributed to the possibility of similar degrees of arousal in happy and angry faces. However, lack of emotion- specific effects in the late time-window is not uncommon: for example Eimer and Homes (2007) reported global emotion effects that lasted up to 1 second but were unaffected by the specific emotion category. According to Eimer and Holmes (2007), this lack of emotion-specific effects could be evidence for one general emotion core network of face processing. For example, the amygdala has been associated with not only with fear processing (Morris et al., 1998) but also equally with

happiness and anger processing (Fitzgerald, Angstadt, Jelsone, Nathan, & Phan, 2006). However, it is to note that whilst Eimer and Holmes (2007) compared brain waves for six individual basic emotions, the current study only compared one positive with one negative emotion category and due to the current study design, these emotions were judged implicitly rather than explicitly.

Further, it is possible that the current design of the study additionally limited emotion effects between angry and happy stimuli by balancing the number of angry and happy stimuli that expressed emotions with open mouths and teeth visible.

192 Previously, it has been proposed that happiness superiority effects in faces may be the confounding result of global visual features such as displaying white teeth (Schyns et al., 2009). Hence, it is possible that for the current study, visual features were comparable across both emotions which in turn results in comparable ERP waveforms.

5.4.3.2. Voices

During the later stage of emotion processing, a late positive component over fronto-central sites was found for the discrimination of emotional versus neutral as well as for angry versus happy vocalisations. The late stage of processing in frontal brain regions may reflect conscious evaluation of emotions (Eimer & Holmes). It has for example been stated that the amygdala is functionally

connected with the medial prefrontal cortex (Kim, Gee, Loucks, Davis, & Whalen, 2010) which may couple bottom-up emotion experience and top-down emotion appraisal (Whalen et al., 2013).

Secondly, for voices within the third time-window, there were separate waveforms for angry and happy voices that differed in their mean latencies but not in mean amplitudes. This late positivity seemed to apply especially to frontal brain areas. According to Hypothesis 2, this hierarchical processing of emotions supports Schirmer and Kotz’s (2006) stage model which suggests broad emotion discrimination before narrow emotion-specific processing in the third stage. The present findings are supported by Iredale et al. (2013) who suggested that at the third processing stage

between 400 and 650ms, sentences spoken with angry or happy prosody elicited different brain waves at frontal sites. As for the second time-window, angry vocalisations had earlier mean latencies

compared to happy vocalisations, but at the third stage, this comparison was statistically more significant. Again, brain activity for emotion-specific processed seemed to move from a posterior negativity in the second time-window to a frontal positivity in the third time-window.

Interestingly, damage to frontal lobe regions has also been associated with impaired emotion recognition in faces as well as in affect vocalisations (Hornak, Rolls, & Wade, 1996). Although not all participants were impaired across both modalities, Hornak et al.’s (1996) data supports the current findings that emotion recognition abilities may indeed be linked to pre-frontal activity. Alternatively, it is possible that the fronto-central activation seen in the current study reflects processes in deeper brain structures such as the amygdala which is commonly associated with emotion processing. For example, the amygdala is often activated when listening to non-verbal exclamations of laughter and crying (Sander & Scheich, 2005).

Overall, faces did not show any emotion-effects for either emotional versus neutral nor for angry versus happy discrimination during late stages of emotion processing. For voices, emotional versus neutral as well as happy versus angry comparisons yielded significant differences in fronto- central regions. This suggests that angry vocalisations had earlier mean latencies than happy vocalisations at 300ms after stimulus onset.

193 It is possible that in the present within-subject design, participants perceived vocalisations are more arousing that static faces which results in the null-effect seen for the face condition.