A frontal positive slow wave in the ERP
in the context of emotional slides
Oliver Diedrich 1 , Ewald Naumann 2 , Stefanie Maier 2 , Gabriele Becker 2 , Dieter Bartussek 2
1
Institute of Medical Psychology and Behavioral Neurobiology, University of Tübingen, Tübingen, Germany
2
Dept. of Psychology, University of Trier, Trier, Germany
Supported by Grant Ba 926/4-1 from the Deutsche Forschungsgemeinschaft (DFG)
We thank Renate Freudenreich, Helmut Peifer, and Klaus Schmitt for the technical work in the laboratory.
Journal of Psychophysiology, in press
Abstract
] [Introduction
Methods
] [Results
Discussion
] [References
Event-related potentials (ERPs) were recorded from subjects watching emotionally negative, neutral, and positive color slides. Subjects had to attend either to the emotional content (emotion-focused processing) or to a structural feature of the slides (str uctural processing). It was expected that (1) the late positive slow wave (SW) demonstrates a frontal amplitude maximum in the emotion-focused processing group; (2) SW effects are different at the frontal and parietal electrode sites; (3) amplitudes in the P3 time range are larger to negative and positive slides, compared to neutral ones. The results indicate that (1) both processing groups exhibit frontally maximal SW amplitudes from 1000 ms after stimulus onset on, and that emotion-focused processing lead s to more positive SW amplitudes at all locations from 600 ms after slide onset on; (2) wave form, the effect of stimulus valence and hemispheric lateralization in the SW time range (from 500 ms on) are different at the frontal and parietal electrode sites ; (3) the amplitudes between 600 and 800 ms after slide onset, surrounding a positive, parietally largest peak with a latency of 720 ms, demonstrate the expected effect of more positive amplitudes to negative and positive stimuli. These findings are discus sed in the context of a model that interprets slow potentials in the ERP to represent the regulation of cortical excitability, with positive slow potentials signifying a state of lowered excitability. In the context of emotional slides, the positive slow w ave with frontally largest amplitudes may index motor inhibition as a counter-regulation of automatically initiated emotional behavior.
Abstract
] [Introduction
Methods
] [Results
Discussion
] [References
In the last two decades there has been a growing interest in the psychophysiology of emotions. There are some basic features of emotions that are widely accepted: Emotions can be described on a valence dimension ranging from unpleasant to pleasant, which i s connected to an action disposition organizing behavior along an avoidance-approach dimension (Lang et al., 1990 ). Emotions provide a link between the person and the environment, they yield information about the relation between a stimulus and the indivi dual, and they tend to lead to motoric, autonomic, and humoral responses (LeDoux, 1989, 1992 ). In contrast with moods, emotions are evoked by stimuli, and they are characterized by responses on three levels: subjective experience (usually measured via ver bal reports), overt behavior, and physiological changes (Lang, 1993 ). The elicitation of these responses by an emotional stimulus seems to be an automatic and preattentive process which does not require a conscious processing of the stimulus (Öhman, 1988
One potent class of emotional stimuli that can be used in a laboratory setting are pictures with an emotional content. Bradley et al. (1993) have demonstrated that the verbal reports of emotional experience elicited by emotional slides can be described us ing two underlying dimensions: valence, ranging from unpleasant to pleasant, and arousal, ranging from calm to excited. These dimensions are linearly independent, but exhibit a quadratic relationship: slides receiving extreme valence ratings (either positi ve or negative) are rated higher in arousal. Lang et al. (1993) interpret valence and arousal as motivational parameters, integrated in subcortical centers, that organize emotions and underlie a strategic behavioral disposition to approach or avoid stimula tion (positive or negative valence) with differing intensity (corresponding to arousal).
In spite of the general acceptance of the important role of centralnervous processes in the generation of emotions, only few studies have examined the effects of emotional stimuli using ERP methodology. It is well known that the ERP is influenced by a wide range of cognitive processes (for overviews, see Birbaumer et al., 1990 ; Hillyard and Picton, 1987 ; Rösler and Heil, 1991 ; Rohrbaugh et al., 1990 ), but little attention has been paid to the impact of 'affective' processes onto the ERP.
Begleiter et al. (1983) used a paradigm in which subjects had to respond to two stimuli by pressing two different buttons. The response either had no consequence, resulted in the winning of one dollar, or was - in the case of a false or slow response - pu nished by the loss of one dollar. Data were analyzed with a principal component analysis. A factor with highest loadings between 270 and 430 ms after stimulus onset and parietally maximal factor scores - interpreted as P3 - exhibited higher amplitudes to s timuli signalling a potential winning or loss, compared to stimuli with no consequences. This effect was most prominent at the parietal electrode site. The authors interpret their finding in terms of stimulus significance as an important variable influenci ng P3 amplitude; stimulus significance may be established by task relevance (as in the standard oddball paradigm) or - as demonstrated in this study - by subjective motivational features of the stimulus.
Johnston et al. (1986) extended the finding of Begleiter et al. (1983): If the stimulus' subjective value is a relevant variable influencing P3 amplitude, what is the nature of the original value system from which these learned values are derived? They pr opose the existence of an emotional evaluation structure which is activated by emotionally significant stimuli - regardless of this emotional significance either being learned (as in the Begleiter et al., 1983, study), or stimulus inherent. To test this hy pothesis, they presented color slides belonging to five different categories: dermatological cases and same sex nudes as unpleasant stimuli, opposite sex nudes and babies as pleasant stimuli, and neutral slides of ordinary people. When viewing the slides, subjects either had to learn associations between these slides and senseless patterns of three letters by predicting the emotional value of the next slide after having been shown a letter pattern, or had to count the number of different letter patterns and slides.
A PCA of the ERPs evoked by the slides yielded three factors with highest factor loadings between 250 and 400 ms (labeled P3), 400 and 700 ms (labeled P4), and 700 and 1000 ms (labeled slow wave). The two early factors exhibit similar valence effects, with smallest amplitudes following the presentation of neutral people slides, and demonstrate a parietal amplitude maximum. The valence effect of the P4 factor is most prominent at the parietal sites; this factor is influenced by the type of the task, too. For the late slow wave factor, the valence effect is restricted to the parietal site. The topography of this factor differs from the two early factors: parietally, smallest amplitudes are found, while the largest amplitudes are present at the central electrod e site. The authors interpret these results in terms of a serial information processing model in which P3 and P4 both reflect stimulus properties, but P4 is additionally influenced by the processing task. The slow wave however reflects a very different pro cess of final evaluation, leading to a different scalp distribution and different effects of the experimental variables.
Johnston and Wang (1991) found similar results using the same slides with women in different phases of the menstrual cycle who were instructed to watch the slides but did not have to perform a specified processing task. Again, neutral slides of ordinary p eople elicit smallest amplitudes in a PCA factor with highest loadings between 300 and 600 ms after slide onset and parietally maximal factor scores. As expected, this valence effect is affected by the endocrine status of the women defined by their menstru al phase. A second factor with highest factor loadings between 600 and 1000 ms is not influenced by the emotional value of the slides; this factor exhibits frontally maximal amplitudes.
Vanderploeg and al. (1987) let their subjects judge the subjective emotional value of faces and words. The ERP data were analyzed via PCA. The resulting slow wave factor with loadings from 500 ms on exhibits factor scores of the same magnitude at the fron tal and parietal leads for both classes of stimuli, with emotional slides eliciting larger amplitudes than the neutral ones. This valence effect was not found for an earlier, parietally maximal factor with highest loadings between 250 and 450 ms after stim ulus onset.
Taken together, these studies demonstrate that the emotional valence of stimuli influences the late positive complex in the P3 time range between 300 and 700 ms after stimulus onset, with larger (more positive) amplitudes following the presentation of emot ional stimuli. This time range usually demonstrates parietally maximal amplitudes; sometimes, the valence effect is more prominent at the parietal sites. Because the crucial difference in amplitudes is always found between emotional and neutral stimuli wit hout a differentiation between positive and negative stimuli, this effect seems to reflect the arousal dimension of the stimuli. The enhanced amplitudes in this time range are interpreted as an effect of the higher significance of emotional stimuli.
The late positive slow wave in the time range from about 600 to 700 ms on does not seem to be influenced by the emotional valence of the stimuli. Usually, the positive slow wave in complex stimulus processing tasks is characterized by the same parietal amp litude maximum found for the preceding P3 component (e.g., Rösler et al., 1990 ; Ruchkin et al., 1990 ; for overviews see Birbaumer et al., 1990; Ruchkin et al., 1988 ). This seems to hold true for pictorial stimuli, too, as demonstrated by Sommer et al. ( 1991) for neutral human faces, or by Kok and Rooyakkers (1986) for line drawings. In the studies using emotional stimuli however, there is a tendency towards a fronto-central amplitude maximum of the late positive slow wave or at least similar amplitudes at the frontal and parietal electrode sites.
An explanation for the unusual topography in the context of emotional stimuli is proposed by Naumann et al. (1992) who varied two aspects of emotionality: First, the emotional content of the material was varied by using emotionally negative, neutral, and positive adjectives. Second, subjects were instructed either to rate the subjective emotional valence (attention directed towards the emotional content; 'affective' processing group) or to count the number of letters of the words (attention distracted from the emotional content; 'structural' processing group). The authors analyzed P3 (with a latency of 350 ms) and slow wave (defined as the averaged amplitude in the time window 700 to 1200 ms after stimulus onset).
An effect of the emotional content of stimuli was found only for the P3 amplitude, with higher amplitudes following negative and positive stimuli, compared to neutral ones. This effect was present in both processing groups, regardless of the subject's atte ntion being directed towards or being distracted from the emotional content of the stimuli. For the positive slow wave an interaction between processing task and topography was found: The 'structural' processing group exhibits the usual parietal amplitude maximum of the slow wave, while directing the subject's attention towards the emotional content elicits a slow wave with amplitudes of the same magnitude at all electrode sites. Differences between the processing groups are present only at the frontal and central, but not at the parietal sites. This effect is attributed to an additional frontal slow positive wave in the 'affective' processing group that is interpreted as a consequence of the action of specific neuronal structures which are activated when su bjects have to attend to the emotional content of the stimuli, but are not activated when only a further cognitive processing of stimuli is required.
If this explanation holds true, the same phenomenon should be elicited by different kinds of emotional stimuli, if their emotional content has to be attended; and it should be possible to demonstrate that parietal and frontal slow positive waves reflect di fferent aspects of information processing, such leading to different effects of experimental variables at the frontal and parietal electrode sites. The present study addresses these questions by using color slides as powerful emotional stimuli, and impleme nting two different processing tasks by varying the attention paid to the emotional content of these stimuli.
We expect that (1) focusing the subjects' attention to the emotional content of the slides will lead to a late slow positive wave with a fronto-central amplitude maximum from about 700 ms after stimulus onset on, resulting in an effect of processing mainly at the frontal electrode sites; (2) effects of other variables onto the slow wave will be different at the frontal and parietal electrode sites; (3) emotional slides will elicit larger amplitudes in the P3 time range between 300 and 700 ms after stimulus onset, irrespective of the kind of processing task.
Abstract
] [Introduction
Methods
] [Results
Discussion
] [References
Subjects were 46 students which were paid for their participation. They were randomly assigned to two groups with different processing tasks (group 1: n = 23, 17 female, mean age 21.4 ± 2.3; group 2: n = 23, 16 female, mean age 22.6 ± 2.8). Group 1 was ins tructed to attend to the emotional content of the slides (emotion-focused processing), while group 2 had to attend to structural features of the slides (structural processing).
Stimuli were 60 slides taken from the International Affective Picture System (Lang, Öhman, and Vaitl, 1988 ) with 20 slides each being emotionally negative, neutral, or positive. Extremely unpleasant slides (mutilated bodies) were omitted because these sli des seem to form a special category of contents with extreme values not only in valence, but also in other slide qualities like familiarity or complexity. Examples for negative slides are dead animals or weapons; the neutral category contains common object s or ordinary faces; examples for positive slides are animals or babies. Additionally, one, two, or three thin lines that did not disturb the content of the slides were inserted. Slides were presented for 500 ms by two tachistoscopes in a random series wit h at most three successive slides having the same emotional valence or the same number of lines inserted. Tachistoscopes were placed outside the EEG cabin; the slides were back projected on a 80 x 80 cm screen, resulting in a visual angle of 9.5°.
After arriving in the laboratory, subjects were seated in a comfortable chair in the EEG cabin where the electrodes were attached. All subjects first performed a visual oddball task (a slide of a cup (30 %) and a slide of a plate (70 %), presented for 500 ms; task: counting the number of presentations of the cup). This task was intended to habituate subjects to the experimental setting and to control for pre-experiment group differences.
After a short rest, the 60 slides were presented in a random order. According to the processing group, subjects either had to judge the subjective emotional valence of the slide (emotion-focused processing) or had to count the number of lines inserted (str uctural processing). Subjects responded by pressing one of three keys, assigned with the meanings 'unpleasant', 'neutral', and 'pleasant' in condition 1 or with the number of lines (1, 2, 3) in condition 2. Before starting the EEG recording, subjects were given the opportunity to have six training trials. The processing task was followed by a replication of the oddball task.
After the experiment all subjects were presented the same slides in the same order as in the experiment, but without the additional lines. Now all subjects had to judge the subjective affective value of the slides on a scale ranging from -100 (very unpleas ant) to +100 (very pleasant). These ratings were used to determine the subjective emotional value of each slide for each subject.
A single experimental trial had the following sequence: A small cross was presented for one second in the center of the screen. Subjects were instructed to fixate the cross and to avoid excessive eye movements or blinks from this on. After a random interva l ranging between 1000 and 3000 ms, the slide (with the inserted lines) was presented for 500 ms. Two seconds after the start of slide presentation, the cross was shown again, signalling the subject to give the answer using the keypad. After receiving the answer, the next trial started with the presentation of the fixation cross.
Recordings took place in a sound attenuated, electrically shielded cabin. EEG was recorded with an ECI electrocap system from 300 ms before until 1500 ms after slide onset from the scalp locations F3, C3, P3, Fz, Cz, Pz, F4, C4, and P4. Additionally, verti cal and horizontal eye movements were recorded with Ag/AgCl electrodes placed above and below the left eye, and beside the left and the right eye, respectively. Linked mastoids served as reference for the EEG electrodes; impedances were kept below 3 kOhms.
EEG and EOG data were recorded using a Nihon Kohden amplifier with time constant of 10 s, digitized with a sampling rate of 100 Hz and stored on a hard disk. Prior to digitalisation, a 35 Hz low pass filter was applied. A/D conversion, storing of data, and stimulus presentation were controlled by a microcomputer.
EEG recordings were averaged according to the subjective emotional value of the slides: For each subject, the slides were divided in three tercils (emotionally negative, neutral, and positive) based on the valence ratings obtained at the end of the experim ent. Prior to averaging, a correction for ocular artifacts was performed for vertical and horizontal EOG, extending the method proposed by Gratton et al. (1983): After removing blink artifacts with the regression approach of Gratton et al. (1983) in a fir st step, two additional regression equations were builded. The first regression removes influences of vertical eye movements outside the blinks onto the EEG; the second regression subtracts the influence of horizontal eye movements. In the oddball tasks, E RPs were averaged for the rare and the frequent stimulus.
In the slide ERPs, mean amplitudes were computed for time intervals of 100 ms. Starting with the time window 310 to 400 ms, there resulted 11 averaged amplitudes for the time windows 310 to 400 ms, 410 to 500 ms, ..., 1310 to 1400 ms. All amplitudes were c omputed as differences to the averaged amplitude 200 ms before slide onset. Prior to computation, a 12 Hz low pass filter was applied. These averaged amplitudes offer several advantages, compared to a traditional peak analysis: They are more appropriate wi th regard to our hypotheses expecting different effects in different time ranges of the ERP; they are more suitable in quantifying the slow wave in which we are most interested; and they offer the opportunity to observe the time course of experimental effe cts and to differentiate time ranges in the ERP on the basis of a differing topography or different impacts of experimental variables.
To separate topography effects into hemispheric differences and differences between the frontal, central, and parietal electrodes, the nine electrode sites were orthogonalized into two factors hemisphere (left, midline, right) and 'frontality' (frontal, ce ntral, parietal). For each of the averaged amplitudes, there resulted a four way analysis of variance with the factors processing group (emotion-focused vs. structural processing), subjective emotional value of the stimulus (negative, neutral, positive), h emisphere , and frontality , with repeated measurement on the last three factors.
In order to assure that EEG effects are not caused by eye movements not completely removed by the eye artifact correction algorithms, vertical and horizontal EOG were averaged according to the subjective emotional value of the slides. Averaged amplitudes w ere computed as described for the EEG and analyzed in a two way ANOVA with the factors processing group and subjective emotional value
For all analyses including repeated measurement factors Huynh-Feldt corrections were applied. The a error was adjusted to 5 %. For significant effects, the effect size w 2 is reported.w 2 indicates the amount of variance accounted for by the independent vari able(s) and can be interpreted in the same way as the determination coefficient (Cohen, 1988; Hager and Möller, 1986 ). This measure allows statements concerning the relative size of experimental effects; Cohen (1992) proposes conventions for small, medi um, and large effect sizes (w 2 =.02, .13, and .25).
Abstract
] [Introduction
Methods
] [Results
Discussion
] [References
Correlations were computed within each subject between the ratings given after the experiment and the normative ratings provided by Lang et al. (1988). All correlations are significantly positive, ranging from 0.49 to 0.93 with a median of 0.79 in the emot ion-focused processing group, and from 0.39 to 0.88 with a median of 0.73 in the structural processing group.
In order to compare the ratings in the two processing groups, an analysis of variance was performed for the ratings of the emotional value of the slides, with the factors emotion category (negative, neutral, positive slides) and processing group. The mean ratings for the three emotion categories are different (negative: -51.8, neutral: 11.4, positive: 49.6; F (2, 88) = 384.54, p < .01); but neither the main effect of processing group nor the interaction between processing group and emotional category reache s significance (both p > .10).
Neither vertical nor horizontal EOG exhibit a main effect of processing group and subjective emotional value or an interaction between these variables (all p > 0.10). The mean number of trials with a blink following the presentation of emotionally negative , neutral and positive slides is 8.7, 9.7, and 8.5 (range 4 to 15) for the emotion-focused processing group, and 6.7, 6.6, and 7.0 (range 2 to 11) for the structural processing group. There is a tendency towards a higher number blinks in the emotion-focuse d processing group (F (1, 44) = 3.59, p < 0.10, w 2 = 0.06), but this does not lead to a group difference in the averaged vertical EOG; so it is unlikely that group differences in the EEG are caused by the higher number of blinks in the emotion-focused proc essing group.
The grand means of the two processing groups in the oddball tasks before and after the experiment are given in figure 1 (13 kByte) . A statistical analysis with the factors processing group, stimulus (rare, frequent), frontality, and hemisphere for the oddball P3 did not reveal any significant effects under participation of processing group in any of the two oddball runs.
The grand means of the two processing groups are presented in figure 2 (18 kByte) . In the emotion-focused processing group, there develops a more positive slow wave leading to significant main effects of processing group for all time windows between 600 and 1400 ms a fter slide onset (F values and effect sizes are listed in table 1 ; all p < .01). This main effect is not superimposed by any interaction with other variables: The amplitude differences between the processing groups are of the same magnitude at all electrod e locations and are not influenced by the emotional value of the stimuli. As can be seen from the effect sizes given in table 1, the largest effects are found in the time range from 700 to 1000 ms.
Significant main effects of the subjective emotional value of the slides are present in the time range from 410 to 700 ms (F (2, 88) = 6.10, e = 1.0, p < .01; 8.60, e = 1.0, p < .01; 3.95, e = 1.0, p < .05 for the three averaged amplitudes;w 2 varies betwe en .04 and .10): In all three time windows, positive slides elicit the most positive amplitudes, such leading to a linear valence effect in this time range. There is a significant interaction between emotional value and processing group for the averaged am plitudes 410-500 ms (F(2, 88) = 7.26, e = 1.0, p < .01, w 2 = .08) and 510-600 ms (F(2, 88) = 4.78, e = 1.0, p < .01, w 2 = .05): While there are no differences between the amplitudes for the three emotion categories when the subject's attention is distracte d from the emotional content of the stimuli by the structural processing task, the subjects who have to attend to the emotional content exhibit a quite consistent pattern of a linear valence effect with most positive amplitudes following the presentation o f emotionally positive stimuli. Figure 3 presents the amplitudes for the three emotion categories of both groups in the time range affected.
The only other effect including the factor subjective emotional value is an interaction between frontality and emotional valence that reaches significance for the time windows 610-700 ms, 710-800 ms, 910-1000 ms, and 1010-1100 ms (F (4, 176) between 3.05 a nd 4.49; e between .60 and .67; all p < .05; w 2 between .02 and .03; see figure 4 ): Between 600 and 800 ms after stimulus onset, there is always the well-known arousal effect of larger amplitudes following the presentation of emotional stimuli (either nega tive or positive), compared to neutral ones at the parietal and central leads. The amplitudes at the frontal leads however demonstrate the linear valence effect that was found at all electrode sites in the earlier time windows.
Figure 5 demonstrates that there is a reversal from a parietal to a frontal amplitude maximum in the time range between 900 and 1000 ms after stimulus onset: While the amplitudes at the parietal and central electrodes reach their maximum about 700 ms, ther e is a strong positive going wave at the frontal electrodes starting about 450 ms after stimulus onset, with most positive amplitudes from 1000 ms on. Accordingly, there are significant main effects of the factor frontality for all averaged amplitudes exce pt the time window between 910 and 1000 ms (all F (2, 88) > 9.0, all p < .01, Huynh-Feldt e between .58 and .71, w 2 between .10 and .65; 910-1000 ms: F(2, 88) = 0.94, p > .30). The interaction between frontality and processing group does not reach signific ance for any of the time windows (all p > .10).
In order to differentiate between hemisphere effects caused by a trivial midline amplitude maximum and hemispheric asymmetries, an analysis without the midline electrodes was computed, such leading to a hemisphere factor with only two levels (left, right). Significant main effects caused by different activity over the left and right hemisphere were found only in the late time region from 1100 ms after stimulus onset on. The F values (df = 1, 44) for the three time windows affected are 4.69 (1110-1200 ms; p < .05, w 2 = .04), 6.06 (1210-1300 ms; p < .05, w 2 = .05), and 8.44 (1310-1400 ms; p < .01, w 2 = .08), respectively. The effect consists in an increasing asymmetry with more positive amplitudes over the right hemisphere, as reflected in increasing effect si zes of the main effect hemisphere over time.
This main effect is superimposed by an interaction between hemisphere and frontality which reaches significance for all time windows from 700 ms after stimulus onset on. Figure 6 demonstrates that the usual midline amplitude maximum is present in all time windows at the parietal electrode sites. Frontally however, there are no hemisphere effects between 700 and 1000 ms, but an increase of amplitudes from left to right from 1000 ms on. The F values in the time range affected vary between 7.53 and 18.53 (df = 2, 88; e between .85 and .99; all p < .01; w 2 grows over time from .05 to .11).
Abstract
] [Introduction
Methods
] [Results
Discussion
] [References
The effects of the emotional value of stimuli are only partially in line with the results of Begleiter et al. (1983), Johnston and coworkers (Johnston et al., 1986; Johnston and Wang, 1991), and Naumann et al. (1992): In the P3 time range between 400 and 6 00 ms, there dominates a distinct linear valence effect with most positive amplitudes following the presentation of positive slides and smallest amplitudes following the presentation of negative slides. This result clearly contradict the results of Johnsto n and coworkers who found a quadratic relationship between emotional value and ERP amplitudes for P3 factors in the time range of 300 to 600 ms following the presentation of emotional slides; and it is also contradictory to the observation of Naumann et al . (1992) who found the same quadratic P3 effect of the emotional content of word stimuli (with a P3 latency of 350 ms) with comparable processing tasks. This linear valence effect between 400 and 600 ms after stimulus onset seems to require powerful emotio nal stimuli (as the color slides used in this study) and a processing task that focuses on the emotional content of these stimuli.
The time range from 600 to 800 ms after stimulus onset however exhibits the result pattern reported for the P3: The largest amplitude is found at Pz; centrally and parietally, the amplitudes are larger following the presentation of emotional slides, compar ed to neutral ones. This time range seems to reflect the enhanced stimulus significance of emotional stimuli, as proposed by Johnston et al. (1986) for P3 amplitude. At 720 ms, there is a distinct peak with largest amplitudes at the parietal electrode site which might be interpreted as a P3-like component: Kok and Rooyakkers (1986) report a positive component with a latency of 720 ms following the presentation of line drawings in a semantic categorization task; they discuss this component to be comparable t o the classical P3 component with regard to the effects of experimental variables as well as topography. In a memory study by Noldy et al. (1990) , a P3-like component with a latency of 600 ms was elicited by simple drawings. It seems that pictorial stimul i elicit a P3 component with an unusually high latency. In this study however, this component might be confounded with the P2 elicited by switching off the slides 500 ms after slide onset; so an unequivocal interpretation is not possible.
The task of judging the subjective emotional value induces a linear valence effect in the time range 400 to 600 ms after slide onset; no differences between the electrocortical responses to negative, neutral, and positive slides are present in the structur al processing group. This time range seems to reflect a different processing of the emotional qualities of the stimuli, depending on the amount of attention paid to the emotional content. After this time range, amplitudes are generally more positive in the emotion-focused processing group, regardless of the emotional value of the stimuli. This difference between the processing groups, remaining until the end of the measurement period, might reflect the enhanced difficulty of the valence judgement.
About 450 ms after stimulus onset, a positive going wave starts at the frontal electrode sites, exhibiting the largest increase in frontal positivity between 600 and 700 ms and leading to a frontal amplitude maximum from 1000 ms on. However, the expected i nteraction between processing and frontality was not found: in both processing groups, the slow wave exhibits a frontal amplitude maximum; and the amplitude difference between the processing groups is of the same size at all electrode sites. This result cl early contradicts the results of Naumann et al. (1992) who found a difference in slow wave amplitudes between the emotion-focused and the structural processing group only at the frontal and central, but not at the parietal electrode sites.
An explanation for this contradiction might be that the emotional content of the stimuli is more dominant for slides than for words: Even when the processing task requires to focus the attention to other features than the slides' emotional content, it is i mpossible to ignore this emotional content; as a consequence, the same processes are initiated in both processing group to some extent, such eliciting a slow wave with a frontal amplitude maximum in both groups. It is possible that the structural processin g task of counting inserted lines was not suitable to completely distract the subjects' attention from the emotional content of the slides.
There are three indices for a dissociation of frontal and parietal processes. First, the wave forms are very different at the frontal and parietal sites: At the frontal leads, a positive going wave starts about 450 ms after stimulus onset, reaches its maxi mum one second after stimulus onset, and holds on for the complete measurement period; at the central and parietal electrodes however, the amplitudes decrease after reaching a maximum with the P720 peak, thus leading to a change in the scalp distribution o f the late slow positive wave from a parietal to a frontal maximum about 900 ms. Second, the 'P3' effect of more positive amplitudes for emotional stimuli, compared to neutral ones, is restricted to the parietal and central leads. Third, only the parietal leads demonstrate the usual midline amplitude maximum, while there is an asymmetry with larger amplitudes over the right hemisphere at the frontal leads after the frontal slow wave having reached its maximal amplitude at 1000 ms after stimulus onset.
Taken together, these results yield the impression that frontal and parietal slow wave may reflect different aspects of information processing: Additionally to a parietal slow wave indexing further processing of complex stimuli, the slides elicit a frontal positive wave, starting about 450 ms after stimulus onset and holding on for a long time. Of course, this differentiation is purely descriptive. An unequivocal demonstration of a dissociation of frontal and parietal processes requires either the recording of more electrodes in order to perform a dipole analysis, or evidence of different functionality, e.g. by finding experimental manipulations that selectively influence the parietal and the frontal slow wave.
A promising approach for such an experimental dissociation might be derived from the results of Falkenstein et al. (1995) , Schupp et al. (1994) , and Roberts et al. (1994), which point to a possible functional explanation of the phenomenon of a frontal s low wave in the context of emotional stimuli. These studies consistently found with Go/No-Go paradigms, that the inhibition of a prepared response after a No-Go stimulus leads to more positive P3 amplitudes at the frontal and central, but to equal or small er amplitudes at the parietal electrode sites, compared to the Go P3. In the Schupp et al. (1994) study, this fronto-central positivity was associated with decreased startle responses following auditory and visual startle probes. Fronto-central positivity is discussed to constitute a prerequisite for motor inhibition.
The idea of motor inhibition following the presentation of emotional stimuli makes sense in the light of emotion theories which stress behavioral responses as one important part of the emotional response: If an emotional stimulus initiates a motor response as proposed by LeDoux (1989) or Öhmann (1988), it is functional to inhibit this emotional behavior if it is not situationally adequate in order to guarantee a coherent behavior.
The data from this experiment do not allow to decide about the hypothesis that the frontal positive wave is an index of a counter-regulation, consisting in the inhibition of an automatically elicited emotional response. There are some points that need to b e clarified:
First, the time course of the frontal positive wave, starting about 450 ms after stimulus onset, is confounded with stimulus presentation time of 500 ms. From the hypothesis outlined above, it can be derived that this phenomen should be independent of stim ulus presentation time: Whenever a stimulus is able to elicit an automatic emotional response, the process of counter-regulation should take place with a similar time course.
Second, the data for the dissocation between the frontal and the parietal slow wave gathered in this experiment are rather weak. If the slow wave reflects different processes at the frontal and the parietal sites, it should be possible to find experimental variables that differentially affect these two processes.
A third question refers to the proposed functional significance of the frontal positive wave. Rockstroh et al. (1992) and Schupp et al. (1994) have demonstrated that cortical positivity is associated with slower reaction times and decreased startle respons es which both are interpreted as indices of motor inhibition. It has to be shown that this also holds true for the frontal positive wave in the context of emotional stimuli.
A last objection applies to the fundamental assumption that the frontal positive wave is exclusively emotion-related. As so far, there are no indications of this phenomen reflecting other aspects of information processing, but it has to be ruled out clearl y that the frontal slow wave in the context of pictorial stimuli is an index of other than emotion-related processes.
Abstract
] [Introduction
Methods
] [Results
Discussion
] [References
Begleiter, H., Porjesz, B., Chou, C.L., & Aunon, J.I. (1983). P3 and stimulus incentive value. Psychophysiology, 20,
Birbaumer, N., Elbert, T., Canavan, A., & Rockstroh, B. (1990). Slow brain potentials of the cerebral cortex and behavior. Physiological Review, 70,
Bradley, M.M., Greenwald, M.K. & Hamm, A.O. (1993). Affective picture processing. In N. Birbaumer & A. Öhman (Eds.). The structure of emotion (pp. 48-68). Seattle: Hogrefe & Huber.
Cohen, J. (1988). Statistical power analysis for the behavioral science. (2nd ed.) Hillsdale, NJ: Erlbaum.
Cohen, J. (1992). A power primer. Psychological Bulletin, 112,
Falkenstein, M., Koshlykova, N.A., Kiroj, V.N., Hoormann, J., & Hohnsbein, J. (1995). Late ERP components in visual and auditory Go/Nogo tasks. Electroencephalography And Clinical Neurophysiology, 96,
Gratton, G., Coles, M.G.H., & Donchin, E. (1983). A new method for off-line removal of ocular artifacts. Electroencephalography And Clinical Neurophysiology, 55,
Hager, W. & Möller, H. (1986). Tables and procedures for the determination of of power and sample sizes in univariate and multivariate analyses of variance and regression. Biometrical Journal, 28,
Hillyard, S.A. & Picton, T.W. (1987). Electrophysiology of cognition. In F. Plum (Ed.). Handbook of phyiology, vol. 5: Higher functions of the brain (pp. 519-584). Bethesda: American Physiological Society
Johnston, V.S., Miller, D.R., & Burleson, M.H. (1986). Multiple P3s to emotional stimuli and their theoretical significance. Psychophysiology, 23,
Johnston, V.S. & Wang, X.T. (1991). The relationship between menstrual phase and the P3 component of ERPs. Psychophysiology, 28,
Kok, A. & Rooyakkers, J.A.J. (1986). ERPs to laterally presented pictures and words in a semantic catgorization task. Psychophysiology, 23,
Lang, P.J. (1993). The three-system approach to emotion. In N. Birbaumer & A. Öhman (Eds.). The structure of emotion (pp. 18-30). Seattle: Hogrefe & Huber
Lang, P.J., Bradley, M.M., & Cuthbert, B.N. (1990). Emotion, attention, and the startle reflex. Psychological Review, 97,
Lang, P.J., Greenwald, M.K., Bradley, M.M., & Hamm, A.O. (1993). Looking at pictures: affective, visceral, and behavioral reactions. Psychophysiology, 30,
Lang, P.J., Öhman, A. & Vaitl, D. (1988). The International Affective Picture System. Gainesville: University of Florida, Center for Research in Psychophysiology.
LeDoux, J.E. (1989). Cognitive-emotional interactions in the brain. Cognition And Emotion, 3,
LeDoux, J.E. (1992). Brain mechanisms of emotion and emotional learning. Current Opinion In Neurobiology, 2,
Naumann, E., Bartussek, D., Diedrich, O., & Laufer, M.E. (1992). Assessing cognitive and affective information processing functions of the brain by means of the late positive complex of the event-related potential. Journal of Psychophysiology, 6,
Noldy, N.E., Stelmack, R.M., & Campbell, K.B. (1990). Event-related potentials and recognition memory for pictures and words: the effects of intentional and incidental learning. Psychophysiology, 27,
Öhman, A. (1988). Preattentive processes in the generation of emotions. In V. Hamilton, G.H. Bower & N.H. Frijda (Eds.). Cognitive Perspectives on Emotion and Motivation (pp. 127-143). Dordrecht, Netherlands: Kluwer Academic Publishers
Öhman, A. & Birbaumer, N. (1993). Psychophysiological and cognitive-clinical perspectives on emotion: introduction and overview. In N. Birbaumer & A. Öhman (Eds.). The structure of emotion (pp. 3-17). Seattle: Hogrefe & Huber
Roberts, L.E., Rau, H., Lutzenberger, W., & Birbaumer, N. (1994). Mapping P300 waves onto inhibition: go/no-go discrimination. Electroencephalography And Clinical Neurophysiology, 92,
Rockstroh, B., Müller, M., Cohen, R. & Elbert, T. (1992). Probing the functional brain state during P300-evocation. Journal of Psychophysiology, 6,
Rohrbaugh, J.W., Parasuraman, E. & Johnson, R.J. (1990). Event-related brain potentials: Basic issues and applications. New York: Oxford University Press.
Rösler, F. & Heil, M. (1991). Towards a functional categorization of slow waves: taking into account past and future events. Psychophysiology, 28,
Rösler, F., Schumacher, G., & Sojka, B. (1990). What the brain tells when it thinks: event-related potentials during mental rotationand mental arithmetic. The German Journal of Psychology, 14,
Ruchkin, D.S., Johnson, R.J., Canoune, H., & Ritter, W. (1990). Short-term memory storage and retention: an event-related brain potential study. Electroencephalography And Clinical Neurophysiology, 90,
Ruchkin, D.S., Johnson, R.J., Mahaffey, D., & Sutton, S. (1988). Towards a functional classification of slow waves. Psychophysiology, 25,
Schupp, H.T., Lutzenberger, W., Rau, H., & Birbaumer, N. (1994). Positive shifts of event-related potentials: a state of cortical disfacilitation as reflected by the startle reflex probe. Electroencephalography And Clinical Neurophysiology, 90,
Sommer, W., Schweinberger, S.R., & Matt, J. (1991). Human brain potential correlates of face encoding into memory. Electroencephalography And Clinical Neurophysiology, 79,
Vanderploeg, R.D., Brown, W.S., & Marsh, J.T. (1987). Judgement of emotion in words and faces: ERP correlates. International Journal of Psychophysiology, 5,