The time scale of the facilitating versus impairing effects of emotional arousal on the processing of neutral information is an open question for the glutamate amplifies noradrenergic effects (GANE) model (sect. 7.2). The authors assert that “an emotionally salient word that impairs perception of a subsequent target word flashed in the same location 50 or 100 ms later can instead enhance perception of a target word flashed 1,000 ms later” (sect. 2.1). The specific time scales are likely to vary across experimental setups: for example, depending on the complexity of stimuli and the intensity of the arousal. The problem is that if the impact of arousal is not temporally bound, priority can be used to explain experimental effects in either direction, namely, both the impairing and the facilitating effects of arousal. Here we discuss how EEG data can provide crucial temporal dynamic information that can disambiguate GANE's predictions – evidence that Mather et al. did not consider.
According to GANE, arousing stimuli capture resources during their processing. Once their own processing is completed, the arousal they induce also facilitates the processing of subsequent stimuli. To test GANE we need to know in advance the duration of emotional stimulus processing. Previous work indicates that emotional pictures, a stimulus of choice in much of the human emotion–cognition literature, enhance a number of event-related potentials (ERPs). The most robust is the late positive potential (LPP). The LPP is thought to reflect attention allocation and maintenance of stimuli in working memory (Donchin & Coles Reference Donchin and Coles1988). The amplitude of the LPP 400–700 ms after stimulus presentation is higher when stimuli are emotional (Schupp et al. Reference Schupp, Flaisch, Stockburger and Junghofer2006), reflecting the additional resources allocated to such stimuli, in line with GANE. Emotion also enhances other components, including the positive slow wave, where amplitudes are higher up to 6 s post-stimulus. We can therefore conclude that the processing of neutral information presented within 6 s of emotional pictures may be attenuated.
Only with this temporal information can we put GANE to the test. We do so by comparing the ERPs associated with processing neutral stimuli presented on their own (blocked neutral condition) with those presented alongside emotional stimuli (mixed condition). Not only is the context more arousing in the mixed condition (Long et al. Reference Long, Danoff and Kahana2015), but also emotional stimuli increase arousal locally. When the interstimulus interval (ISI) is long, emotional and neutral stimuli are unlikely to compete for processing resources, and GANE predicts that the higher global arousal in the mixed condition should enhance neutral information processing. In contrast, Pastor et al. (Reference Pastor, Bradley, Löw, Versace, Moltó and Lang2008) used an ISI of 12 s and observed reduced LPPs for neutral stimuli in the mixed compared with the blocked condition. It is, however, possible that those emotional pictures were still being processed when the subsequent picture was displayed after 12 s. When the ISI is short, competition should be pronounced, so GANE predicts that the processing of neutral information should be impaired. In contrast, Schupp et al. (Reference Schupp, Schmälzle, Flaisch, Weike and Hamm2012) used an interval of 3 s and observed a null effect of context (blocked/mixed). It is, however, possible that the effect of emotion on resource allocation in that study was short-lived, for example, because of the orienting task. If emotional stimuli no longer attract attention when subsequent neutral stimuli are presented, the null effect is incompatible with GANE's predictions.
In our experiment (Barnacle et al. Reference Barnacle, Schaefer, Tsivilis and Talmi2015), 22 healthy adults viewed 16 lists of 14 pictures: 4 neutral, 4 emotional, and 8 mixed lists (50% emotional pictures). All pictures depicted people; emotional and neutral picture sets were equally semantically related, but the emotional pictures were more negative and arousing. Each picture was presented for 2 s with a jittered ISI of 4±0.5 s. Participants were asked to encode these pictures for a free-recall memory test, which followed each study list after a 60-second distractor task. EEG was recorded during encoding with a BioSemi Active Two (BioSemi, Amsterdam) using 64 electrodes conforming to the 10–20 system and preprocessed with SPM (www.fil.ion.ucl.ac.uk). Data were filtered between 0.1 and 25 Hz, downsampled to 125 Hz, and epoched between −200 and 4,000 ms. Eye-blink artifact was removed using an algorithm implemented in SPM. A threshold of 120 µV was used for trial rejection followed by robust averaging.
Following Schupp et al. (Reference Schupp, Schmälzle, Flaisch, Weike and Hamm2012), we extracted LPP and slow-wave component amplitude data, averaging across centroparietal electrodes (Cz, CPz, Pz, C1, C2, P1, P2, CP1, CP2) in three time windows: 400–700 ms, 1,000–2,000 ms, and 2,000–3,000 ms poststimulus. We compared emotional and neutral picture processing in the mixed condition at each window to ascertain the duration of the effects of arousal, using three one-tailed paired-sample t-tests (p<0.017 controlled for multiple comparisons). Emotion modulated ERPs in the 400–700 ms and 1,000–2,000 s windows, but not later (Figure 1). We then compared neutral picture processing in the mixed and blocked conditions at both these windows with two two-tailed t-tests (p<0.025). The LPP for neutral pictures was slightly, but not significantly attenuated in the mixed, compared with the blocked, condition.
Figure 1. Average ERP amplitude in time windows corresponding to the Late Positive Potential and to the early portion of the positive slow wave. Emotion modulation lasts up to 2,000 ms poststimulus.
Our data indicate that the duration of the effect of arousal is key for testing GANE in novel experimental setups. EEG data allowed us to determine how long emotional pictures attracted extra processing resources. Here the modulation lasted up to 2 s from stimulus onset, evident in the modulation of the early portion of the LPP and positive slow wave, but not later. This pattern suggests that arousing stimuli are no longer in competition for resources when neutral pictures are presented in the same sequence 3.5-4.5 s after the onset of the arousing pictures, and their prioritized processing should not detract from the resources allocated to neutral stimuli. On the contrary, because of the increased global arousal in the mixed condition, GANE predicts that neutral picture processing should be enhanced. In fact, our data provided evidence that the processing of neutral information is attenuated in that situation.
The three ERP data sets we reviewed appear to contradict GANE's predictions. The mature electrophysiology literature on the effect of emotion on perception, attention, and memory (e.g. Hajcak et al. Reference Hajcak, MacNamara and Olvet2010; Schupp et al. Reference Schupp, Flaisch, Stockburger and Junghofer2006) can provide crucial data for GANE until data on the times scales of norepinephrine–glutamate interactions are available.
The time scale of the facilitating versus impairing effects of emotional arousal on the processing of neutral information is an open question for the glutamate amplifies noradrenergic effects (GANE) model (sect. 7.2). The authors assert that “an emotionally salient word that impairs perception of a subsequent target word flashed in the same location 50 or 100 ms later can instead enhance perception of a target word flashed 1,000 ms later” (sect. 2.1). The specific time scales are likely to vary across experimental setups: for example, depending on the complexity of stimuli and the intensity of the arousal. The problem is that if the impact of arousal is not temporally bound, priority can be used to explain experimental effects in either direction, namely, both the impairing and the facilitating effects of arousal. Here we discuss how EEG data can provide crucial temporal dynamic information that can disambiguate GANE's predictions – evidence that Mather et al. did not consider.
According to GANE, arousing stimuli capture resources during their processing. Once their own processing is completed, the arousal they induce also facilitates the processing of subsequent stimuli. To test GANE we need to know in advance the duration of emotional stimulus processing. Previous work indicates that emotional pictures, a stimulus of choice in much of the human emotion–cognition literature, enhance a number of event-related potentials (ERPs). The most robust is the late positive potential (LPP). The LPP is thought to reflect attention allocation and maintenance of stimuli in working memory (Donchin & Coles Reference Donchin and Coles1988). The amplitude of the LPP 400–700 ms after stimulus presentation is higher when stimuli are emotional (Schupp et al. Reference Schupp, Flaisch, Stockburger and Junghofer2006), reflecting the additional resources allocated to such stimuli, in line with GANE. Emotion also enhances other components, including the positive slow wave, where amplitudes are higher up to 6 s post-stimulus. We can therefore conclude that the processing of neutral information presented within 6 s of emotional pictures may be attenuated.
Only with this temporal information can we put GANE to the test. We do so by comparing the ERPs associated with processing neutral stimuli presented on their own (blocked neutral condition) with those presented alongside emotional stimuli (mixed condition). Not only is the context more arousing in the mixed condition (Long et al. Reference Long, Danoff and Kahana2015), but also emotional stimuli increase arousal locally. When the interstimulus interval (ISI) is long, emotional and neutral stimuli are unlikely to compete for processing resources, and GANE predicts that the higher global arousal in the mixed condition should enhance neutral information processing. In contrast, Pastor et al. (Reference Pastor, Bradley, Löw, Versace, Moltó and Lang2008) used an ISI of 12 s and observed reduced LPPs for neutral stimuli in the mixed compared with the blocked condition. It is, however, possible that those emotional pictures were still being processed when the subsequent picture was displayed after 12 s. When the ISI is short, competition should be pronounced, so GANE predicts that the processing of neutral information should be impaired. In contrast, Schupp et al. (Reference Schupp, Schmälzle, Flaisch, Weike and Hamm2012) used an interval of 3 s and observed a null effect of context (blocked/mixed). It is, however, possible that the effect of emotion on resource allocation in that study was short-lived, for example, because of the orienting task. If emotional stimuli no longer attract attention when subsequent neutral stimuli are presented, the null effect is incompatible with GANE's predictions.
In our experiment (Barnacle et al. Reference Barnacle, Schaefer, Tsivilis and Talmi2015), 22 healthy adults viewed 16 lists of 14 pictures: 4 neutral, 4 emotional, and 8 mixed lists (50% emotional pictures). All pictures depicted people; emotional and neutral picture sets were equally semantically related, but the emotional pictures were more negative and arousing. Each picture was presented for 2 s with a jittered ISI of 4±0.5 s. Participants were asked to encode these pictures for a free-recall memory test, which followed each study list after a 60-second distractor task. EEG was recorded during encoding with a BioSemi Active Two (BioSemi, Amsterdam) using 64 electrodes conforming to the 10–20 system and preprocessed with SPM (www.fil.ion.ucl.ac.uk). Data were filtered between 0.1 and 25 Hz, downsampled to 125 Hz, and epoched between −200 and 4,000 ms. Eye-blink artifact was removed using an algorithm implemented in SPM. A threshold of 120 µV was used for trial rejection followed by robust averaging.
Following Schupp et al. (Reference Schupp, Schmälzle, Flaisch, Weike and Hamm2012), we extracted LPP and slow-wave component amplitude data, averaging across centroparietal electrodes (Cz, CPz, Pz, C1, C2, P1, P2, CP1, CP2) in three time windows: 400–700 ms, 1,000–2,000 ms, and 2,000–3,000 ms poststimulus. We compared emotional and neutral picture processing in the mixed condition at each window to ascertain the duration of the effects of arousal, using three one-tailed paired-sample t-tests (p<0.017 controlled for multiple comparisons). Emotion modulated ERPs in the 400–700 ms and 1,000–2,000 s windows, but not later (Figure 1). We then compared neutral picture processing in the mixed and blocked conditions at both these windows with two two-tailed t-tests (p<0.025). The LPP for neutral pictures was slightly, but not significantly attenuated in the mixed, compared with the blocked, condition.
Figure 1. Average ERP amplitude in time windows corresponding to the Late Positive Potential and to the early portion of the positive slow wave. Emotion modulation lasts up to 2,000 ms poststimulus.
Our data indicate that the duration of the effect of arousal is key for testing GANE in novel experimental setups. EEG data allowed us to determine how long emotional pictures attracted extra processing resources. Here the modulation lasted up to 2 s from stimulus onset, evident in the modulation of the early portion of the LPP and positive slow wave, but not later. This pattern suggests that arousing stimuli are no longer in competition for resources when neutral pictures are presented in the same sequence 3.5-4.5 s after the onset of the arousing pictures, and their prioritized processing should not detract from the resources allocated to neutral stimuli. On the contrary, because of the increased global arousal in the mixed condition, GANE predicts that neutral picture processing should be enhanced. In fact, our data provided evidence that the processing of neutral information is attenuated in that situation.
The three ERP data sets we reviewed appear to contradict GANE's predictions. The mature electrophysiology literature on the effect of emotion on perception, attention, and memory (e.g. Hajcak et al. Reference Hajcak, MacNamara and Olvet2010; Schupp et al. Reference Schupp, Flaisch, Stockburger and Junghofer2006) can provide crucial data for GANE until data on the times scales of norepinephrine–glutamate interactions are available.