Abstract
In rapid serial visual presentation of pictures, an early ERP component shows enlarged negativity over occipital regions for emotional, compared to neutral, pictures.
The present study examined the processing of emotional target pictures as a function of the valence of a preceding prime picture. Dense sensor ERPs were measured while subjects viewed a continuous stream of pleasant, neutral, and unpleasant pictures, presented for 335 ms each. Two main findings were observed: First, replicating previous results, emotional target pictures were associated with a larger posterior negativity, compared to neutral pictures. Second, the emotional content of the prime picture affected target processing, with pleasant or unpleasant primes resulting in reduced negativity of the following picture, irrespective of its emotional valence.
These findings are discussed within the framework of competition for neural representation in visual processing.
Introduction
Preparation for action has been suggested as key function of emotion (Frijda, 1986; Lang et al., 1997; Öhman et al., 2000a). However, because of the flexible tailoring of responses to environmental contingencies in humans, the organization of successful approach/avoidance behaviors is fostered by the accurate perceptual analysis of the environment (Lang et al., 1997; Öhman et al., 2000a). Accordingly, stimuli signifying evolutionary relevant topics such as food, mating partners, or signals of threat are effective cues to capture attention. Behavioral support for this hypothesis is provided by research utilizing the emotional Stroop, dot probe, and visual search tasks (Mathews & MacLeod, 1994; Mogg & Bradley, 1999; Öhman et al., 2000a). In addition, neuroimaging studies of the processing of complex emotional scenes consistently observed increased activations by emotional pictures in posterior brain areas involved in visual information processing (e.g., Junghöfer et al., 2005;
Lang et al., 1998b; Sabatinelli et al., 2004; Sabatinelli et al., 2005).
Event-related brain potential studies have also assisted in detailing the temporal dynamics of emotion processing in the visual brain. When pictures are presented at very rapid rates (e.g., 3 Hz), a difference in processing emotional (pleasant and unpleasant), compared to neutral, pictures is shown by a larger negative potential over temporo-occipital sites developing around 150 ms after stimulus onset and lasting until about 300 ms. Furthermore, highly arousing pleasant and unpleasant pictures elicit more pronounced occipital negativity compared to pictures of the same valence that are rated lower in arousal (Junghöfer et al., 2001; Schupp et al., 2004b).
Accordingly, the early posterior negativity was suggested to reflect the facilitated encoding of visual scenes depicting information of emotional significance.
Functional/evolutionary considerations suggest the benefit of the selective processing of emotional (pleasant and unpleasant) cues. In a world where various stimuli compete for attentional resources, the fast and reliable detection of positive and negative reinforcers facilitates adaptive behavior, finally promoting survival and reproductive success (Cacioppo et al., 1999; Lang et al., 1997; Öhman et al., 2000a).
Determining the boundary conditions of selective emotion processing, a number of recent fMRI- and ERP studies presented stimuli simultaneously in visual spatial attention tasks. Overall, it was found that (task-irrelevant) selective emotion
processing was attenuated with increasing primary task demands (cf., Holmes et al., 2003; Pessoa et al., 2002b; Vuilleumier et al., 2001). Here, we focused on another aspect of attentional orienting, the impact of successively presented pictures. In the real world, stimuli often appear not in isolation but follow each other, raising the question to what extent the selective encoding of emotional pictures varies as a function of the emotion category of a preceding “prime” picture. For instance, is the processing of an erotic scene facilitated when preceded by an image of a baby rather than a scene of threat? To examine this issue, we utilized a modified version of the rapid serial presentation paradigm (cf., Junghöfer et al., 2001; Potter, 1976) presenting pleasant, neutral, and unpleasant pictures as prime and target picture categories.
Considered from the perspective of motivated attention (Lang et al., 1997), processing resources are automatically captured and sustained by emotional cues (cf., Cuthbert et al., 2000; Keil et al., 2002). Accordingly, one might assume that an emotional prime picture drawing heavily from limited processing resources is hampering the processing of subsequently presented stimuli. To the extent that emotional target pictures are already known to efficiently draw processing resources compared to neutral images, the motivated attention hypothesis assumes that target picture processing will vary as a function of processing resources devoted to both prime and target pictures. Statistically, this hypothesis predicts main effects of prime and target emotion category. On the one hand, emotional target pictures are associated with larger posterior negativity compared to neutral images. And on the other hand, emotional prime pictures, themselves associated with increased posterior negativity, should reduce the posterior negativity of subsequent target pictures. Consistent with this hypothesis, neuroscientific and behavioral studies suggest that stimuli presented in sequence may evince features of competition for attention and processing resources (cf., Keysers & Perrett, 2002; Potter et al., 2002) extending competition concepts derived from concurrent stimulus presentations (e.g., Pessoa et al., 2002a).
While congruence in hedonic valence between a prime and a target picture has no special status in a motivated attention framework, such effects might be predicted from the perspective of affective priming. A typical finding in behavioral priming tasks is that target words are evaluated faster when preceded by a short-duration (e.g., 300 ms) prime of the same hedonic valence (for reviews, see Bargh, 1997; Klauer & Musch, 2003). In order to examine affective priming effects, we
utilized in the present study complex emotional scenes known to reliably engage measurable emotional responses (for an overview see Bradley & Lang, 2000). An affective priming hypothesis predicts an interaction of prime and target picture category due to the facilitated processing of target pictures preceded by prime pictures of the same valence category. The finding of affective priming at the level of early visual processing would provide strong empirical support for variants of spreading activation accounts (cf., Ferguson & Bargh, 2003).
Method
Participants
24 subjects (13 female) participated in the study. 22 received course credit for an introductory psychology class at the University of Florida, 2 were post-graduate laboratory employees. Participants were between the ages of 18 and 22 years (M = 19.0).
Stimulus Materials and Procedure
One hundred pleasant (erotic couples, babies, sports and adventure scenes), neutral (neutral people, household objects), and unpleasant pictures (mutilations, violence, attack) from the International Affective Picture System (IAPS; Lang et al., 1999) were presented. The three valence categories differed significantly from each other in their normative valence ratings (M = 6.5, 5.1 and 2.4 for pleasant, neutral, and unpleasant contents on a 1-9 scale). Mean arousal levels for emotional categories were significantly higher than for neutral contents (M = 5.8, 3.0 and 6.4 for pleasant, neutral, and unpleasant contents, respectively). All pictures were colored, had simple figure/ground distinctions and a small white fixation cross in their center. Physical parameters such as brightness, contrast, complexity did not differ across categories (cf., Junghöfer et al., 2001).
Pleasant, neutral, and unpleasant images were shown in a perceptually random sequence. Pictures were presented as a continuous stream of images on a 21 inch CRT-monitor located approximately 100 cm in front of the participant, without perceivable inter-stimulus intervals (75 Hz refresh rate). Each individual picture was displayed for 335 ms (see Fig. 4.1 A). The picture set (n=300) was repeated in different orders twenty times, resulting in a total of 6000 picture presentations. Picture presentation lasted for 37 minutes with a short break in between to allow for posture adjustments.
Subjects were instructed to simply view the pictures without any further instruction. In addition, subjects were asked to fixate their eyes on the cross at the center of the pictures.
Figure 4.1
(A) Illustration of the rapid serial visual presentation paradigm. Pleasant, neutral, and unpleasant pictures are shown for 335 ms without perceivable inter-stimulus gap.
(B) Illustration of the prime-target-picture combinations examined in the present study.
Apparatus and Data Analysis
Electrophysiological data were collected from the scalp using a 256-channel system (Electrical Geodesics, Inc., Eugene, OR). Scalp impedance for each sensor was kept below 50 kΩ. Due to the high input impedance of the E0 amplifier, scalp impedances below 50 kΩ are recommended to ensure optimal signal to noise ratio for this amplifier system. The EEG was recorded continuously with a sampling rate of 250 Hz, the vertex sensor as reference electrode, and on-line bandpass filtered from .01 – 100 Hz. Continuous EEG data were low-pass filtered at 40 Hz before stimulus synchronized epochs were extracted. A statistical approach as described in Junghöfer and colleagues (2000) was applied for artifact correction including the transformation of the ERP data to an average reference. Before statistical analysis, a baseline correction was applied to the data incorporating a 40 ms interval preceding prime picture onset (see Fig. 4.5, left side).
Separate average waveforms were calculated for the 9 experimental cells (3 picture categories for the prime and current image, respectively) for each sensor and participant (see Fig. 4.1 B). Effects depending on the “prime” and “target” picture category were observed in bilateral clusters over temporo-occipital as well as frontal regions. The EPN amplitude was scored as mean activity in the time interval from 248
to 288 ms collapsed over temporo-occipital sensors with EGI sensor numbers 146, 147, 135, 136, 123, 124, 137, 115, 116, 125, 126, 117, and 118 on the left, and sensors 166, 157, 167, 158, 168, 149, 159, 150, 169, 139, 160, 151 and 140 on the right hemisphere. Over frontal areas, the lateralized sensor clusters were comprised of sensors # 36, 37, 32, 33, 26, 38, 34, 27, 39, 35, 28, 22, 29 and 23 on the left, and sensors # 18, 11, 19, 12, 20, 3, 4, 13, 225, 226, 5, 14, 216 and 6 on the right side (see Fig. 3.2).
The EPN amplitude was submitted to repeated-measures analysis of variance (ANOVA) including the factor prime Category (pleasant vs. neutral vs. unpleasant), target Category (pleasant vs. neutral vs. unpleasant), and Laterality (left vs. right).
For effects involving repeated measures, the Greenhouse-Geisser epsilon was utilized to correct for violations of sphericity.
Figure 4.2.
Illustration of the frontal and posterior sensor clusters entering statistical analysis.
Results
Looking at the effect of picture category exclusively for the target picture, emotional as compared to neutral pictures were associated with a larger negativity over occipital sensors (see Fig. 4.4). This emotional modulation over posterior areas was accompanied by a corresponding positivity over centro-frontal sites. The time course of this modulation is illustrated in Figure 4.3 (top), showing a representative right-occipital sensor (# 150). When the category of the prime picture is kept constant (see Fig. 4.3, lower panel), the affective modulation due to the category of the target picture can be observed consistently starting at around 150 ms with a peak around 260 ms post picture onset. Comparing pleasant and unpleasant pictures, pleasant ones elicited a relatively stronger negativity over posterior sites which was also accompanied by a reversed polarity over frontal areas. For both pleasant and unpleasant pictures, this occipital negativity tended to be more pronounced over the right hemisphere.
When keeping the category of the target picture constant, a consistent positive shift over occipital sensors was observed after emotional as compared to neutral prime pictures (see Figs. 4.6 & 4.7). Mirroring the effect of picture category of the target picture, this positivity was reversed over frontal areas and was stronger after pleasant than after unpleasant prime pictures. Only after pleasant prime pictures, this effect seemed to be more prominent over right hemisphere occipital sensors.
Entering all factors into statistical analysis, highly significant interactions for category of the prime, as well as the target picture with sensor cluster location (F(2, 46) = 26.96; p < .001 and F(2, 46) = 52.85; p < .001, respectively) justified subsequent analyses using reduced models for both the frontal and the occipital sensor clusters.
Figure 4.3.
Upper panel: ERP waveforms for a representative right-occipital sensor (#150) serve to illustrate the main effect of target valence. The upper panel illustrates enlarged posterior negativities for emotional compared to neutral target pictures averaged across pleasant, neutral, and unpleasant prime pictures.
Lower panel: Detailed illustration of the target effect by presenting waveforms separately following pleasant, neutral, and unpleasant prime pictures.
Analyzing the occipital sensor clusters separately, no interaction between the categories of the prime and the target picture was apparent (F(4, 92) = 0.87; ns.).
However, the main effects for picture category were highly significant, both for the prime and the target picture (F(2, 46) = 21.99; p < .001 and F(2, 46) = 44.22; p <
.001). Post-Hoc comparisons confirmed that both pleasant and unpleasant categories of the target picture elicited a higher negativity over occipital areas than neutral pictures (F(1, 23) = 59.49; p < .001 and F(1, 23) = 16.76; p < .001). Moreover, this
negativity was stronger for pleasant than for unpleasant pictures (F(1, 23) = 45.78; p <
.001). Accordingly, comparing the categories of the prime picture, pleasant and unpleasant as compared to neutral pictures were followed by a highly significant positivity during the processing of a target picture (F(1, 23) = 33.24; p < .001 and F(1, 23) = 14.97; p < .001). Again, this effect was stronger for pleasant than for unpleasant prime pictures (F(1, 23) = 11.08; p < .01). Finally, there was also a significant interaction between the category of the target picture and sensor cluster laterality.
Following up this interaction, there was a trend of a right-lateralized effect of picture category, specifically for pleasant pictures (F(1, 23) = 2.36; ns.).
Figure 4.4.
Scalp potential maps of the difference waves (Pleasant – Neutral) and (Unpleasant – Neutral) for the interval used in statistical analysis reveal the topographical distributions of the target valence effects. To derive these brain maps, voltages were interpolated to the scalp surface using spherical splines and back-projected to a back view of the model head. Please note the different scales for the two maps.
Figure 4.5.
Upper panel: ERP waveforms for a representative right-occipital sensor (#150) serve to illustrate the main effect of prime valence. The upper panel illustrates reduced posterior negativities following emotional compared to neutral prime pictures averaged across pleasant, neutral, and unpleasant target pictures.
Lower panel: Detailed illustration of the prime effect by presenting waveforms separately for pleasant, neutral, and unpleasant target pictures.
Analyzing the frontal sensor clusters separately, no interaction between the categories of the prime and the target picture was found (F(4, 92) = 0.3; ns.).
Corresponding to the occipital sensor cluster analysis, there were highly significant main effects as a function of the category of both the prime and the target picture (F(2, 46) = 28.35; p < .001 and F(2, 46) = 58.8; p < .001). Post-Hoc analyses revealed a greater positivity for the pleasant and unpleasant as compared to the neutral category of the target picture (F(1, 23) = 78.59; p < .001 and F(1, 23)=17.3; p<.001). This
positivity was more pronounced for pleasant than for unpleasant pictures (F(1, 23) = 86.54; p < .001). Comparing the categories of the prime picture to each other, both pleasant and unpleasant pictures were related to a greater negativity over frontal sensors as compared to neutral pictures (F(1, 23) = 52.68; p < .001 and F(1, 23) = 18.51; p < .001). Again, this effect was more pronounced for pleasant than for unpleasant pictures (F(1, 23) = 10.29; p < .01). In contrast to the occipital sensor clusters, there was a significant 3-way-interaction between picture category of the prime and the target picture and Laterality (F(4, 92) = 3.5; p < .05). However, subsequent analyses revealed no interaction between the categories of the prime and the target picture, neither for the left- nor for the right-lateralized frontal sensor clusters, respectively (F(4, 92) = 1.148; ns., and F(4, 92) = 0.51; ns.). In general, the overall structure of the higher-order analysis was confirmed in both lateralized sensor clusters and no meaningful decomposition of this higher-order interaction could be achieved.
Figure 4.6.
Scalp potential maps of the difference waves (Pleasant – Neutral) and (Unpleasant – Neutral) reveal the topographical distributions of the modulation of target picture processing as a function of prime valence.
Finally, there was a prime picture category by laterality interaction (F(2, 46)
= 5.78; p < .01) which was followed up as being due to a more positive left-lateralized potential, specifically for neutral pictures (F(1, 23) = 8.6; p < .01).