Affective
Monitoring
Eliciting Positive Affect from Non-Affective Stimuli Cindy C. Gomes Master thesis Clinical Neuropsychology February, 24 2014 Supervisor: Dr. R.H. Phaf Brain & Cognition
Faculty of Social and Behavioral Science University of Amsterdam
Amsterdam The Netherlands
1 Table of contents Table of contents ... 1 Abstract ... 2 Introduction ... 3 Method ... 10 Results ... 16 Discussion ... 21 Acknowledgements ... 25 References ... 26 Appendices ... 30
2
Abstract
Conflict is well known to raise negative effect, but is it also instrumental in eliciting positive affect? A basic premise of the Affective Monitoring (AM) hypothesis is that conflict is needed to elicit both positive and negative affect from neutral stimuli. Negative affect follows from persistent conflict, whereas positive affect arises if it is swiftly resolved. Experimental support for the second part of the AM hypothesis is, however, scarce. To create the initial conflict in the present experiment, participants had to decide whether a face in a morphed male-female picture was of a male, or a female. A second picture was then shown to either solve, or sustain the conflict. Gender evaluation of a single neutral face served as baseline. To measure the affective priming by the conflict participants were asked to judge, as quickly as possible, whether words had a positive or negative valence. The results supported the expectation that resolved conflicts resulted in faster positive decisions, and slower negative decisions than baseline. Likewise sustained conflicts resulted in slower positive decisions, and faster negative decisions than baseline, but not reliably so. The current results provide support for the AM hypothesis that conflict can also result in positive affect.
3 Introduction
“A man who was having heart trouble went to the doctor to see what his options were. Naturally, the doctor recommended a heart transplant. The man reluctantly agreed, and asked if there were any hearts immediately available, considering that money was no object. "I do have three hearts," said the doctor. "The first is from an 18-year old kid, non-smoker, athletic, swimmer, with a great diet. He hit his head on the swimming pool and died. It's $100,000. The second is from a marathon runner, 25 years old, great condition, very strong. He got hit by a bus. It's $150,000. The third is from a heavy drinker, cigar smoker, steak lover. It's $500,000. "Hey, why is that heart so expensive? He lived a terrible life!”. "Yes, but it's from a lawyer. It's never been used." (http://www.jokes.com/funny-work-jokes/zx57bc/the-heart-of-the-matter).
A well-placed joke may illustrate very well how positive affect is elicited. First tension is slowly built up, which after it is released rapidly, is followed by laughter, as described in the tension-release hypothesis by Sroufe and Waters (1976). This hypothesis represents one of the very few attempts to analyze affect in terms of constituent mechanisms, instead of only relying on introspective notions of positive and negative feelings. Still, studying the dynamics of affect elicitation is essential, because it is considered one of the most basic aspects of emotions in nearly all emotion theories (e.g., Frijda, 1986). More recently, new ideas about affect have developed that placed it in an evolutionary context “Brains have evolved to generate pleasant, or unpleasant feelings pertaining to those environmental aspects that were
4
a consistent benefit, or threat to gene survival in ancestral environments” (p.173, Johnston, 2003; see also Heerebout & Phaf, 2010). Even more recently, these two hypotheses about affect elicitation have been joined in the affective monitoring framework by Phaf and Rotteveel (2012). In this view, neural competition, or ‘conflict’ signals fitness-reducing situations, whereas the rapid resolution of conflict signals fitness-increasing situations. There is already considerable evidence that sustained conflict may elicit negative affect (for a review, see Dreisbach & Fischer, 2012a). The empirical evidence for the counterintuitive hypothesis that conflict can also elicit positive affect seems, however, much more scarce. This study aims at showing that the fast resolution of conflict raises positive affect.
Phaf and Rotteveel (2012) conceptualized the Affective Monitoring (AM) hypothesis in a connectionist neural framework, which will not be discussed in great detail here. These networks continuously process stimuli in a distributed fashion and match them with stored representations and with other stimuli. Conflict can arise locally when there is an insufficient match with other representations in that region. Conflict is implemented in neural networks by competition between mutually inhibiting nodes or quasi-neurons, which can stand for representations of stimuli. Affective monitoring has also been implemented mechanically (i.e., in terms of connections between nodes) in such networks by determining the level of competition in such a region. The same neural competition has been postulated to form the basis of attentional selection (Duncan, 1996; Phaf, van der Heijden, & Hudson, 1990). AM
5 thus integrates attentional and affective processes. The sensory input to the system
always leads to competition at many different levels of the neural network. Attentional selection takes place through the resolution of competition and the determination of ‘winners’ at these levels. According to the AM-hypothesis (Phaf & Rotteveel, 2012), the speed or fluency of competition resolution is simultaneously monitored, and leads to positive affect with predominant fluent resolution at these different levels and to negative affect with disfluent resolution.
In AM affective states are only evoked by match-mismatch processing after competition (i.e., a conflict) has first arisen (Phaf & Rotteveel, 2012; cf. Murre, Phaf, & Wolters 1992). When the competition is sustained, this raises negative affect (Dreisbach & Fischer, 2012ab; Phaf & Rotteveel, 2012; van der Heijden, 2013). This contrasts to the basic assumptions of the fluency account that positive affect from non-emotional stimuli is merely elicited by fluent information processing also in the absence of initial competition (Reber, Winkielman, & Schwarz, 1998; Willems & Van der Linden, 2006). The classical mere exposure effect (for a review, see Zajonc, 2001) may be the prime example of affect elicitation by mere fluency manipulations. In this paradigm, the unreinforced presentation of neutral stimuli leads to an increased preference for these stimuli relative to novel stimuli. Even here though, it can be argued that conflict is raised by the simultaneous presentation of familiar and novel stimuli, which can be solved readily by the fluent processing of the familiar stimulus. In many other examples of positive affect elicitation by fluency manipulations (for an
6
overview see Winkielman, Schwarz, Fazendeiro, & Reber, 2003), it is unclear whether conflict preceded the fluent processing. What is thus needed is the
demonstration of the elicitation of positive affect following an explicit manipulation of conflict.
One of the few attempts (to our knowledge, the only other attempt was made by Aarts, De Houwer, & Pourtois, 2012) to compare fluent conditions with and without preceding conflict was made by van der Heijden (2013). The conflict was created by hybrid (i.e., morphed) pictures, which were comprised of male and female faces. In a gender-decision task, participants had to decide whether the second stimulus was male, or female. The second stimulus could consist of one of the two constituent pictures of the hybrid and thus solve the conflict, or be a continuation of the hybrid, so that the conflict would be sustained. The control condition consisted of the continuous presentation of a non-morphed picture of a male of female face. To determine the extent of affective priming, the reaction times to evaluations of subsequent happy, neutral, and angry faces were measured. Positive priming is characterized by a shift to faster evaluations of positive stimuli and a slower
evaluations of negative stimuli (cf. Dreisbach & Fischer, 2012b), whereas negative priming corresponds to the reverse shifts.
The results corroborated the findings of Dreisbach and Fischer (2012b) that sustained conflict served as a negative signal, but relative to the control condition negative priming was also obtained after the resolution of conflict (van der Heijden,
7 2013). Van der Heijden (2013) had obtained nearly as much negative affect in the
resolved-conflict condition, as in the sustained-conflict condition. It can be
questioned, however, whether the conflict had actually been solved in the resolved conflict condition. The participants did report marked indecisiveness in the resolved conflict condition. Van der Heijden (2013) discussed three factors which could have caused this. The first factor was that participants experienced difficulty in
determining the gender of particular male and female faces due to androgynous features of some models. The inability to determine the gender of these faces may have prolonged the conflict instead of solving it. Another factor which could have played a role was that the choice for faces as affective priming targets may not have been optimal. Participants were instructed to indicate whether the happy, neutral, or angry face presented had a positive, or a negative facial expression. Conflict may have been raised unwittingly when the participants had to determine if a neutral face had a positive, or a negative expression, arguably making also the decisions on positive and negative faces less certain. The last factor was that participants were uncertain when they had to make their choices, (a) in part because of the fast pace of the subsequent responses, and (b) in part because the cue to make a selection was not evident to all participants. In many trials the conflict may thus have remained,
explaining why negative priming was found instead of positive priming in the resolved conflict condition (van der Heijden, 2013).
8
The recommendations for improvements in follow-up research made by van der Heijden (2013) were implemented in the current study. This was made concrete with (a) the elimination of the ambiguity of the sexes of the models used as primes and first targets, (b) the replacement of emotion faces by emotional words for the affective priming, (c) and the adaptation of the trial sequencing. To begin with, any doubts about the gender of the faces were eliminated by giving a familiarization phase with all models prior to the practice trials. The emotion faces which served as second targets in the study by van der Heijden (2013) were substituted by emotion words in the present study. To maximize affective priming the strongest emotion words were selected from Phaf, Van der Leij, Stienen, and Bierman (2006). In addition, the positive and negative categories were matched on word-length and high word frequency (Keuleers & Brysbaert, 2010). The experimental trials had been adapted in length, and in sequencing. Another addition was the confirmation of the right answer in the resolved conflict and baseline condition, or a question mark in the sustained conflict condition after the responses during 100 ms. Also the inter-trial intervals were randomized to reduce trial predictability.
To answer the research question whether positive affect can be elicited by fluent processing of neutral stimuli after initial conflict most aspects of the test-procedure and stimuli of van der Heijden (2013) were also incorporated in the present study. To create the initial conflict in this experiment, participants again had to decide whether a face in a morphed picture was of male, or female. A second face picture
9 was then shown to either solve, or sustain the conflict. Gender evaluation of a single
unmorphed neutral face served as the baseline condition in the first part of a trial. To measure affective priming, participants were asked to judge as quickly as possible with two response buttons whether highly frequent emotion words had a positive or negative valence. To eliminate any remaining conflict in the resolved conflict condition, only the ten fastest (correct) responses on the first target were included in the analysis of the second response. The gender ambiguity of the models is probably solved by the familiarization prior to the trials, but there remain many other sources of conflict in such speeded visual presentations, such as distractions, eye blinks, and eye movements. This, moreover, corresponds to a somewhat similar procedure used by Aarts et al. (2012) for the inclusion of fast conflict resolution trials.
The expectations for the results of the present experiment are that gender decisions will take longer in the sustained conflict (SC) condition than in the resolved conflict (RC) condition, which in turn will take longer than in the baseline (BL) condition. Sustained conflict will result in slower positive decisions and faster
negative decisions than with baseline. On the other hand, resolved conflict will result in faster positive decisions and slower negative decisions than with baseline.
10
Method Participants
The 52 participants in the experiment were mainly first-year psychology students from the University of Amsterdam (UvA), who participated for course credit, or a €10 financial compensation. Participants were right handed, had normal, or corrected-to-normal vision, and Dutch as their first language. Participants with error rates higher than 10% on the easy task of evaluating the emotion words were excluded from the analyses.
Design
The experiment had a 3 (Blank screen – Neutral; Neutral – Morphed; Morphed– Morphed) x 2 (Negative, Positive words) within-subjects factorial design. Per condition, only the trials with the ten fastest correct responses in the gender-decision task were analyzed with respect of the evaluation responses. The reaction-times (RT) for the correct responses in these ten trials for the evaluation task served as the
dependent variable. Outlier RTs in the evaluation task shorter than 250 ms and longer than 1000 ms were removed from these ten trials and replaced by trials with the next fastest responses in the gender-decision task. In the evaluation task, RT differences between positive and negative responses (RTpos –RTneg) were calculated and
subjected to a single factor ANOVA to test the main expectation of the experiment. A similar ANOVA was performed on the RTs of the gender-decision task.
11 For the first target in the gender-decision task the response buttons for male
and female gender were counterbalanced between subjects. For one half of the participants the left-button represented male, and the right-button female, for the other half the left-button represented female, and the right-button male. The buttons for positive and negative valence pertaining to the word target were kept constant across participants, the left button for negative, and the right button for positive. For right-handed participants this is the most compatible placement of the response buttons (Casasanto, 2009). Counterbalancing also the affective response buttons could increase the error variance and decrease the accuracy of the measurements.
Material and apparatus
This research took place a normally lit, laboratory room equipped with a desk, a Windows pc with, two pc-screens, a chin-rest and a two-button response box. The pc screen used for the experiment was a 23 inch (51 cm width, and 28 cm height) 3D LCD 1920x1080 Full HD monitor. The distance in between the buttons on the response box was 1 cm. To keep the distance between all participants and the pc-screen constant the chin-rest was placed at 60 cm distance from the pc-screen. Presentation® software (Neurobehavioral Systems Inc., USA) was used to run the experiment. Images and texts on the screen were presented on a gray background.
For the gender-decision target the same 15 male and 15 female models were used from the Radboud faces database (Langner et al., 2010) as in van der Heijden’s
12
(2013) study. The picture resolutions were in terms of pixels: 681 width x 1024 height x 24 depth. The images were presented in the center of the screen, and had an actual height of 36 cm, and width of 14 cm. The 30 male and female models in the pictures all had neutral facial expressions. To create the conflict hybrids or ‘morphs’ were constructed by superimposing male, and female pictures using the Sqirlz Morph program (Xiberpix, 2002 – 2008). Three different types of morphs were construed, (a) a morphed face with a 50/50 ratio of a superimposed male and female face, and (b) two versions of 60/40 morphs were created. In half of these morphs the ratios were 60 percent male, and 40 percent female, and in the latter half 60 percent female, and 40 percent male. In each trial one of the two pictures (male or female) forming the male-female hybrid was used as target. Each face was coupled only once to another face in the hybrid set. In total sixty 50/50 ratio morphs were construed, fifteen 60/40 male-to-female ratio morphs, and fifteen 60/40 female-to-male ratio morphs. A few examples of the models used, the 50/50, and 60/40 morphs can be viewed in Appendix 1. Each of the 30 models had been used three, up to six times as primes, and as targets. The further specifications of which models were used, the frequency of use, and the morph percentages can be found in Appendices 2 and 3.
The 110 Dutch words used for the experiment were obtained from the EmoClar database by Phaf, Van der Leij, Stienen and Bierman (2006). The
corresponding word-frequencies (WF) of all the 55 positive, and 55 negative words was obtained from the SUBTLEX-NL database (Keuleers & Brysbaert, 2010).
13 Emotion words with the highest contextual (WF), and matching word-lengths (WL)
were chosen. The average WL of both the 55 positive, and 55 negative words was 7.93 letters, and the WF percentage is 18.042. Twenty of these words were selected to be used in 20 practice-trials. The remaining 90 words were presented at random in the experimental trials. The emotion words were displayed in Calibri font style, and font size 40 in the center on the screen. The list of emotion words used were added to the second appendage. The corresponding affective valence, WF, and WL of each word used can be found in Appendix 4.
Trial description: Familiarization, practice trials, and experimental trials For the purpose of gender disambiguation, participants are first given a
familiarization block. Each of the 30 models was presented four times in this pre-experimental gender-decision task. Every face was presented for 100 ms on the pc screen. As soon as the face was removed from the screen, a blank screen appeared. As soon as this screen appeared the participants had a maximum of 500 ms to decide if the model was male, or female. A short beep sounded to indicate when the participant gave an incorrect answer. After this phase 20 practice-trials are given, using four female, and four male faces for T1, and the 20 pre-selected positive and negative words. Six BL, seven RC, and seven SC practice trials were given.
Each practice and experimental trial consisted of two phases. The first phase of a trial served to create the initial conflict. This was presented in a gender-decision
14
task, consisting of a Prime (P) – Target 1 (T1) combination. The Prime in the baseline (i.e. Non-Conflict; BL) condition was a blank screen, and a Conflict-Stimulus in the two other experimental conditions. The 50/50 morphs served as primes to elicit the initial conflict. In the condition that aimed to resolve the conflict (Resolved-Conflict, RC) the morph P is followed by the male, or the female neutral face (i.e., T1) used in the prime. In the condition that aimed to prolong conflict (Sustained-Conflict, SC) the morph P is followed by a 60/40 morph models used in the P (i.e., T1). In addition, feedback was given after the participants entered their choice for T1. In the RC and BL conditions the target-response is confirmed by the presentation of the word “MAN” or “VROUW”. In the SC condition a question-mark was displayed leaving the gender of the target undisclosed. In total 90 trials were given in two blocks, with 30 trials per condition.
The second phase of a trial consisted of a measurement of affective priming. A word-valence decision task was performed in which Dutch words with highly positive or negative emotional valences were presented.
15 The sequencing of each trial starts with a fixation cross displayed on the
screen (see Figure 1) for 500 ms. This is followed by the P in the RC, and SC during 1000 ms, except in the BL condition. The fixation point in the BL condition remained on screen for 2000 ms. In all three conditions the text ‘Man of Vrouw?’ or ‘Vrouw of Man?’ appears depending on the counterbalanced condition the participant is in. Then T1 appears during 1000 ms. Subsequently, a blank screen appears during 1000 ms, during which the participants have to select their answer and is immediately followed by feedback during 500 ms. Then the second phase of the trial starts with a screen with the text ‘Negatief of Positief?’ for 500 ms, followed by an emotion-word for 1000 ms. The trial ended with a blank screen to select the answer to T2 for the last 500 msec. The inter-trial interval was jittered between 1000 -2000 ms to prevent participants from responding in a fixed rate.
Procedure
Participants were informed that the experiment investigated how their decision speed was influenced by a prior decision. They received instructions to make fast but accurate decisions, if necessary based on their intuition. After signing informed consent, participants were seated comfortably behind the desk and requested to place their heads on the chin-rest in front of the computer screen.
16
First, the familiarization session of 120 trials was given of all 30 models. This was followed by 20 practice trials in the experimental format. For the practice trials three female, and three male faces were used combined with the 20 words selected for this purpose. Practice consisted of six NC trials, seven RC trials, and seven SC trials. After a brief pause two sets of 90 trials were given using all 30 faces, and the 90 words selected for the experimental trials. Per set 30 BL, 30 RC, and 30 SC were given. Each word was displayed only once per block. Between blocks, the
participants were given the opportunity to take a short break. When the experiment was done, an exit-interview was taken. The experimenter enquired how the
participant experienced the experiment, whether they applied any particular
strategies, or whether they made any observations during the tasks. They were also asked to tell something about the strategies they used while performing the tasks.
Results
From the 52 participants, 11 were excluded from the analyses. One of the female participants had to be excluded from the analyses due to severe pain associated with muscle-inflammation, which hindered her in giving speeded button responses. One male participant was excluded because of a very strong suspicion he was under the influence of cannabis. Three female, and one male, participants had a high error percentage on the second target – higher than 10%. Other participants, Four female, and one male participants were excluded because of the large number of missing data
17 (< 10%) due to slow responding. Of the remaining 41participants 26 were female, 15
were male, seven were left-handed, and 34 were right-handed. Their average age was 23.2 years (SD = 4.3).
Conflict in the Gender evaluation task
For the manipulation check the average of ten fastest correct gender decisions was calculated (see Table 1, and Figure 2). When the evaluation of the second target was incorrect, longer than 1000 ms or shorter than 250 ms, this trial was excluded and the next fastest response to Target1 was included in the average. This resulted in the RT averages being calculated over the same trials for Target1 as for Target2.
Table 1: Target 1 average reaction-times (ms) with corresponding standard deviations as a function of conflict condition.
Average reaction time SD
Resolved Conflict 522 59
Baseline 563 65
18
The pattern of results supported the conclusion that conflict had been raised by the morphs and that it was successfully solved by the unmorphed male or female face. In a 3 (RC, BL, SC) x 2 (positive vs. negative Target2) ANOVA the main effect of the conflict condition: F(2, 80) = 14.34, p < 0.0001, ηp 2 = 0.527) was indeed statistically
reliable. As the responses to Target1 could not be influenced by Target2, which only appeared after Target1, all effects with the valence of Target2 were of course non-significant (F < 1). Reaction times were longer in the SC condition than in the RC condition (t(40) = -5.02, p < 0.0001, Cohen’s d = 0.72), which in turn was shorter than in the control condition (t(40) = 5.22, p < 0.0001. d = 0.67). The difference between reaction-times in the SC condition with those in the baseline condition proved unreliable (t(40) = -1,36, p = 0.180, d = 0.21).
19 Conflict and Affective Priming
Conflict and its resolution influenced the speed of evaluating positive and negative words (see Table 2 and Figure 3). The 3 (conflict) x 2 (Target2 valence) ANOVA on the Target 2 RTs revealed significant main effects for target valence (F(1, 80) =5.71, p < 0.05, ηp 2 = 0.114) and conflict (F(2, 80) = 3.31, p < 0.05, ηp 2 = 0.100) and an
interaction effect of conflict and valence (F(2, 80) = 4.47, p < 0.05, ηp 2 = 0.101). The
faster reactions to positive words (M = 646 ± 75 ms) than to negative words (M = 661 ± 77 ms) represents a standard finding with affective stimuli (e.g., Taylor, 1991). A clearer picture of the priming patterns can be obtained from the differences between positive RTs and negative RTs (i.e., the priming index: RTpos – RTneg), where this
standard effect cancels.
Table 2: Target 2 average reaction-times for positive, and negative words with standard deviations (ms) as a function of conflict condition.
Positive Negative
Average RT SD Average RT SD
Resolved Conflict 637 74 674 73
Baseline 640 60 647 73
20
The priming index showed strong positive priming in the RC condition (I = -37 ± 65 ms) and some negative priming in the SC condition (I = -4 ± 65 ms) relative to the BL condition (I = -6 ± 53 ms)(See Figure 4). The main effect in the single-factor
ANOVA was reliable (F(2, 80) =4.47, p < 0.05, ηp 2 = 0.101). Planned t-tests revealed
that the RC-SC difference was significant (t(40) = -2.56, p < 0.05, d = 0.51), as was the RC-BL difference (t(40) = -2.85, p < 0.01, d = 0.53), but the negative priming in the SC condition relative to baseline did not reach conventional levels of significance (t(40) = -0.14, p = 0.89, d = 0.03).
21 To investigate whether some participants experienced more conflict than others and
whether this would be reflected in the affective priming, the correlations between RTs to Target1 and the priming index were calculated for each of the conflict conditions. For the BL condition r = -0.17, for SC condition r = 0.14, and for the RC condition r = -0.003. All correlations were far removed from significance. Within conditions, thus, all participants seemed to be experiencing approximately the same level of conflict which led to the same level of affective priming.
Discussion
Conflict was successfully raised and solved in the experiment. More importantly, the conflict manipulation in the processing of affectively neutral stimuli had affective consequences with sustained conflict leading to slightly more negative affect and resolved conflict to appreciably more positive affect than in the no-conflict condition.
22
Similar to Aarts et al. (2012), trials with disfluent processing due to other factors than the conflict manipulation (e.g., distractions, eye blinks, eye movements) were
removed from the averages, and only the ten fastest trials with gender decision were involved in the analyses. This appeared to result in relatively homogeneous levels of conflict per condition and may thus have increased the chances of finding this relatively small effect.
Finding positive affect after resolving conflicts, with only little negative affect after sustained conflicts in this experiment seems almost the opposite of the findings in the study van der Heijden (2013). In the latter study strong negative priming was obtained in both sustained and resolved conflicts conditions relative to baseline. Here, however, also the baseline condition differed from the present study. Because prime and Target1 were the same unmorphed face, a high level of fluency, presumably without preceding conflict, may have been reached with baseline in the study of van der Heijden. A simple fluency account may thus also explain these results. A fluency account cannot serve to explain the positive priming by resolved conflict relative to baseline in the present study, because it would predict negative priming instead. The present results thus seem to contradict the fluency account and alternatively support the affective monitoring hypothesis that conflict is needed also to raise positive affect.
A further difference between the van der Heijden study and this study concerned the selection of the ten fastest gender-decision trials. It is, however, unlikely that in view of the large differences a similar selection in the former study
23 would have reversed the priming effect in the resolved conflict condition. The
alternative explanation that much conflict remained in the RC condition of this study, due to androgynous faces and the fast pace of stimulus presentation, seems more plausible. On the other hand, the rather limited extent of negative priming after sustained conflict may well be explained by the selection. The shortest RTs in the SC condition probably corresponded with the least conflict elicited by the morphs and/or the easiest resolution of the conflicts by the 60/40 morphs. It should be noted,
however, that this explanation still relies on the affective monitoring account.
The most supportive pattern of results for the affective monitoring hypothesis would be the simultaneous finding of strong positive priming in a resolved conflict condition and strong negative priming in a sustained conflict condition relative to a baseline condition that entails no obvious conflict. How can this be achieved? To start with the baseline condition, it can perhaps be simplified to a simple reaction task. The participants would then only have to press a button when a probe (e.g., a dot or a cross) appears after a variable interval. Because the probe appears in every trial, this presents as little conflict as possible and would thus constitute the most suitable control condition. An obvious improvement would be to increase the number of trials per condition and exclude not only the fastest, but also the slowest, gender decision trials.
Another recommendation for research would be to incorporate one the procedures done in the study by Aarts et al (2012). Aarts et al. (2012) hypothesized
24
that individuals with high anxiety traits are more sensitive to negative information, which in turn might lead to speeded responses to negative words, and errors, or conflict. Even though the results did not reach significance, which might be attributed to the small size of the sample it is still a valid argument. Giving the participants questionnaires to measure state anxiety, and comparing if the RTs of highly anxious, to non-anxious participants do differ significantly. Another suggestion for future study would be to repeat this experiment and comparing different conflict stimuli. Using and comparing the neutral, the resolved, and sustained conflict RTs elicited by the Stroop-task (Dreisbach & Fischer, 2012), the Go-NoGo task (Aarts et al. 2012), and the gender-evaluation task (van der Heijden, 2013) and the affective priming using emotion words.
Empirical support for affective monitoring would lead us one step further away from introspective notions of affect. It might even be questioned whether affect always corresponds to conscious feelings (e.g., see Winkielman & Berridge, 2004). According to affective monitoring (Phaf & Rotteveel, 2012; see also Johnston, 2003) at least, positive affect represents a neural code for fitness-enhancing conditions, and negative affect a neural code for fitness-decreasing conditions, but are these codes only rarely transformed into conscious feelings. In this study also it is unlikely that the participants were aware of their speed differences in reacting to positive or negative words, or of the role conflict played in affect. These preliminary results support the premise that conflict is instrumental in eliciting both negative and positive
25 affect, but cannot of course show that initial conflict is always required for positive
affect. This conflict-resolution, or tension-release (Sroufe & Waters, 1976)
mechanism conforms largely to the processes underlying a joke, but apparently does not necessarily has to lead to conscious positive feelings. Even if the participants in this study were not rolling on the floor laughing out loud after successful gender decisions, their positive affect could be measured implicitly.
Acknowledgements
First and foremost I want to express my immense gratitude to my mentor dr. R.H. Phaf. Without whose knowledge, guidance, encouragement and understanding it would not have been possible for me to complete this thesis. I would also like to show my appreciation to Jasper Wijnen, Ien van den Berg and Manon Slockers for their contribution to the completion of this process. My profound gratitude also goes to Hubert Eleonora for his invaluable support. And last I want to thank Sonja
Houkooper and the Exam Commission of the University of Amsterdam for still providing me the opportunity to obtain my Master’s degree in Psychology. My deepest thanks to you all.
26
References
Aarts, K., De Houwer, J., & Pourtois, G. (2012). Evidence for the automatic evaluation of self-generated actions. Cognition, 124, 117-127.
Casasanto, D. (2009). Embodiment of abstract concepts: good and bad in right-and left-handers. Journal of Experimental Psychology: General, 138, 351-367. Dreisbach, G., & Fischer, R. (2012a). The role of affect and reward in the conflict-triggered adjustment of cognitive control. Frontiers in Human Neuroscience 6: 342. doi: 10.3389/fnhum.2012.00342
Dreisbach, G., & Fischer, R. (2012b). Conflicts as aversive signals. Brain and Cognition, 78, 94-98. doi:10.1016/j.bandc.2011.12.003
Duncan, J (1996). Cooperating brain systems in selective perception and action. Inui, T. and McClelland, J.L. (Eds), (1996). Attention and performance 16:
Information integration in perception and communication. Attention and performance., pp. 549-578. Cambridge, MA, US: The MIT Press, xvii, 680 pp.
Frijda, N.H. (1986). The Emotions: Studies in emotion and social interaction. Cambridge University Press.
Heerebout, B.T., & Phaf, R.H. (2010). Good vibrations switch attention: an affective function for network oscillations in evolutionary simulations. Cognitive, Affective, and Behaviorial Neuroscience, 10, 217-229.
27 Johnston, V.S. (2003). The origin and function of pleasure. Cognition & Emotion, 17,
167-179.
Keuleers, E., & Brysbaert, M. (2010). SUBTLEX-NL: A new measure for Dutch word frequency based on film subtitles. Behavior Research Methods, 42, 643-650.
Langner, O., Dotsch, R., Bijlstra, G., Wigboldus, D.H.J., Hawk, S.T., & van Knippenberg, A. (2010). Presentation and validation of the Radboud Faces Database. Cognition & Emotion, 24, 1377-1388.
Murre, J.M.J., Phaf, R.H., & Wolters, G. (1992). CALM: categorizing and learning module. Neural Networks, 5, 55-82.
Phaf, R.H., van der Heijden, A.H.C. & Hudson, P.T.W. (1990). SLAM: A connectionist model for attention in visual selection tasks. Cognitive Psychology, 1990 July 22 (3): 273–341.
Phaf, R.H., van der Leij, A.R., Stienen, B.M.C., & Bierman, D. (2006). Positieve, neutrale en negatieve woorden bij minimale aanbieding: Een ordening door perceptuele clarificatie [Positive, neutral, and negative words at minimal presentation levels: Ordering by perceptual clarification], Amsterdam, The Netherlands: Technical Report, Universiteit van Amsterdam
Phaf, R.H., & Rotteveel, M.(2012). Affective monitoring: A generic mechanism for affect elicitation. Frontiers in Psychology, 3:47. doi:
28
Reber, R., Winkielman, P., & Schwarz, N. (1998). Effects of perceptual fluency on affective judgments. Psychological Science, 9, 45-48.
Sqirlz Morph software program (Xiberpix 2002-2008).
Sroufe, L.A., & Waters, E. (1976). The ontogenesis of smiling and laughter: a perspective on the organization of development in infancy. Psychological Review, 83, 173-189.
Taylor, S.E. (1991). Asymmetrical effects of positive and negative events: The mobilization-minimization hypothesis. Psychological Bulletin, 110, 67-85. Van der Heijden, J. (2013). Conflict as positive signals. Master thesis, Brain &
Cognition, Psychology Department, Faculty of Social and Behavioral Sciences. University of Amsterdam, The Netherlands.
Willems, S., & Van der Linden, M. (2006). Mere exposure effect: a consequence of direct and indirect fluency preference links. Consciousness & Cognition, 15, 323-341.
Winkielman, P., & Berridge, K.C. (2004). Unconscious Emotion. Current Directions in Psychological Science, 13, 120-123.
Winkielman, P., Schwarz, N., Fazendeiro, T.A., & Reber, R. (2003). The hedonic marking of processing fluency: Implications for evaluative judgment. In J. Musch & K.C. Klauer (Eds.), The psychology of evaluation (pp. 189-217). Mahwah, NJ: Erlbaum.
29 Zajonc, R.B. (2001). Mere exposure: A gateway to the subliminal. Current Directions
30
Appendices
Appendix 1: Examples of models used for gender evaluation
Male and female models with neutral facial expressions (Langner, O., Dotsch, R., Bijlstra, G.et al. 2010)
50/50 Morphs of male and female models using Squirlz Morph
31 Appendix 2: Frequency of usage of models
Practice trials
Female Prime T1 Totaal
12 4 3 7
13 4 2 6
14 3 2 5
15 3 3 6
14 10
Male Prime T1 Totaal
12 3 2 5
13 4 3 7
14 3 2 5
15 4 3 7
14 10
Aantal keren gebruikt:
Aantal keren gebruikt:
Experimental trials
Female Prime T1 Totaal Male: Prime T1 Totaal
1 3 4 7 1 4 4 8 2 3 5 8 2 4 5 9 3 3 5 8 3 3 5 8 4 4 4 8 4 3 5 8 5 4 4 8 5 4 5 9 6 4 4 8 6 4 4 8 7 4 4 8 7 4 4 8 8 4 4 8 8 4 4 8 9 4 4 8 9 3 5 8 10 4 4 8 10 5 4 9 11 3 6 9 11 6 3 9 12 5 3 8 12 4 3 7 13 5 3 8 13 4 3 7 14 5 3 8 14 4 3 7 15 5 3 8 15 4 3 7 60 60 120 60 60 120
32
Appendix 3: Stimulus set specified for each experimental trial
Practice trials
Conditie Trialnummers Prime T1 T2
SC 1 t/m 7 morph 50/50 morph 60/40 OEFENW
RC 8 t/m 14 morph 50/50 100/0 OEFENW
BL 15 t/m 20 100/0 OEFENW
Trial Prime (P) Target 1 (TA) Target 2 (TB)
50% 50% 60% 40% 1 t/m 20 1 F12 M13 F12 M13 OEFENW 2 F13 M15 F13 M15 OEFENW 3 F15 M12 M12 F15 OEFENW 4 F12 M12 F12 M12 OEFENW 5 F13 M15 M15 F13 OEFENW 6 F12 M14 M14 F12 OEFENW 7 F14 M15 F14 M15 OEFENW
Trial Prime (P) Target 1 (TA) Target 2 (TB)
50% 50% 100% 1 t/m 20 8 F15 M13 M13 OEFENW 9 F13 M14 M14 OEFENW 10 F14 M13 F14 OEFENW 11 F12 M13 M13 OEFENW 12 F15 M12 F15 OEFENW 13 F14 M15 M15 OEFENW 14 F13 M14 F13 OEFENW
Trial Prime (P) Target 1 (TA) Target 2 (TB)
100% 1 t/m 20 15 F15 OEFENW 16 M15 OEFENW 17 F12 OEFENW 18 M13 OEFENW 19 F15 OEFENW 20 M12 OEFENW
33 Experimental trials
Conditie Trial Prime T1 T2
1 1 t/m 30 morph 50/50 morph 60/40 POSNEG WORD
2 31 t/m 60 morph 50/50 100/0 POSNEG WORD
3 61 t/m 90 0 100/0 POSNEG WORD
Conditie SC Trial Prime (P) Target 1 (TA) Target 2 (TB)
50% 50% 60% 40% Words 21 t/m 110 1 1 F02 M15 F02 M15 POSNEG WORD 1 2 F04 M14 F04 M14 POSNEG WORD 1 3 F06 M13 F06 M13 POSNEG WORD 1 4 F08 M12 F08 M12 POSNEG WORD 1 5 F10 M11 F10 M11 POSNEG WORD 1 6 F11 M01 M01 F11 POSNEG WORD 1 7 F12 M02 M02 F12 POSNEG WORD 1 8 F13 M03 M03 F13 POSNEG WORD 1 9 F14 M04 M04 F14 POSNEG WORD 1 10 F15 M05 M05 F15 POSNEG WORD 1 11 F02 M11 F02 M11 POSNEG WORD 1 12 F04 M12 F04 M12 POSNEG WORD 1 13 F06 M13 F06 M13 POSNEG WORD 1 14 F08 M14 F08 M14 POSNEG WORD 1 15 F10 M15 F10 M15 POSNEG WORD 1 16 F11 M06 M06 F11 POSNEG WORD 1 17 F12 M08 M08 F12 POSNEG WORD 1 18 F15 M09 M09 F15 POSNEG WORD 1 19 F14 M10 M10 F14 POSNEG WORD 1 20 F13 M05 M05 F13 POSNEG WORD 1 21 F01 M15 F01 M15 POSNEG WORD 1 22 F03 M14 F03 M14 POSNEG WORD 1 23 F05 M13 F05 M13 POSNEG WORD 1 24 F07 M12 F07 M12 POSNEG WORD 1 25 F09 M11 F09 M11 POSNEG WORD 1 26 F11 M06 M06 F11 POSNEG WORD 1 27 F12 M07 M07 F12 POSNEG WORD 1 28 F13 M08 M08 F13 POSNEG WORD 1 29 F14 M09 M09 F14 POSNEG WORD 1 30 F15 M02 M02 F15 POSNEG WORD
34
Conditie RC Trial Prime (P) Target 1 (TA) Target 2 (TB)
50% 50% Female Male Words 21 t/m 110
2 31 F01 M01 F01 POSNEG WORD 2 32 F03 M10 F03 POSNEG WORD 2 33 F05 M11 F05 POSNEG WORD 2 34 F07 M01 F07 POSNEG WORD 2 35 F08 M11 F11 POSNEG WORD 2 36 F12 M06 M06 POSNEG WORD 2 37 F09 M07 M07 POSNEG WORD 2 38 F15 M08 M08 POSNEG WORD 2 39 F14 M09 M09 POSNEG WORD 2 40 F13 M10 M10 POSNEG WORD 2 41 F02 M06 F02 POSNEG WORD 2 42 F04 M07 F04 POSNEG WORD 2 43 F06 M08 F06 POSNEG WORD 2 44 F08 M10 F08 POSNEG WORD 2 45 F10 M11 F10 POSNEG WORD 2 46 F09 M01 M01 POSNEG WORD 2 47 F04 M02 M02 POSNEG WORD 2 48 F05 M03 M03 POSNEG WORD 2 49 F06 M04 M04 POSNEG WORD 2 50 F07 M05 M05 POSNEG WORD 2 51 F01 M11 F01 POSNEG WORD 2 52 F03 M12 F03 POSNEG WORD 2 53 F05 M13 F05 POSNEG WORD 2 54 F07 M14 F07 POSNEG WORD 2 55 F09 M15 F09 POSNEG WORD 2 56 F10 M02 M02 POSNEG WORD 2 57 F12 M03 M03 POSNEG WORD 2 58 F13 M04 M04 POSNEG WORD 2 59 F14 M05 M05 POSNEG WORD 2 60 F15 M07 M07 POSNEG WORD
35
Conditie BL Trial Prime (P) Target 1 (TA) Target 2 (TB)
Female Male Words 21 t/m 110
3 61 F11 POSNEG WORD 3 62 F10 POSNEG WORD 3 63 F09 POSNEG WORD 3 64 F08 POSNEG WORD 3 65 F07 POSNEG WORD 3 66 M01 POSNEG WORD 3 67 M10 POSNEG WORD 3 68 M09 POSNEG WORD 3 69 M08 POSNEG WORD 3 70 M07 POSNEG WORD 3 71 F01 POSNEG WORD 3 72 F02 POSNEG WORD 3 73 F03 POSNEG WORD 3 74 F04 POSNEG WORD 3 75 F05 POSNEG WORD 3 76 M01 POSNEG WORD 3 77 M02 POSNEG WORD 3 78 M03 POSNEG WORD 3 79 M04 POSNEG WORD 3 80 M05 POSNEG WORD 3 81 F06 POSNEG WORD 3 82 F11 POSNEG WORD 3 83 F09 POSNEG WORD 3 84 F03 POSNEG WORD 3 85 F02 POSNEG WORD 3 86 M03 POSNEG WORD 3 87 M06 POSNEG WORD 3 88 M04 POSNEG WORD 3 89 M11 POSNEG WORD 3 90 M09 POSNEG WORD
36
Appendix 4: Emotion words
Forty-five Positive Dutch words used in the Experimental trials
word VALCAT RT mean SUBTLEXCD Word length
1 sexy positief 3852.89 12.0942 4 2 wijs positief 3890.61 15.7869 4 3 zacht positief 3765.82 10.0867 5 4 prima positief 4348.81 56.0223 5 5 geluk positief 4495.66 46.5675 5 6 mazzel positief 3537.73 6.6171 6 7 vrijen positief 3539.18 8.0421 6 8 muziek positief 3696.63 26.5304 6 9 aardig positief 3796.94 48.8476 6 10 lekker positief 3868.16 57.8067 6 11 gezond positief 3871.1 13.2962 6 12 zuiver positief 4110.3 4.5353 6 13 vriend positief 4185 76.0967 6 14 cadeau positief 4214.73 10.8674 6 15 passie positief 4334.38 6.0843 6 16 gelukt positief 4338.07 23.9157 6 17 vredig positief 4243.64 3.3333 6 18 plezier positief 3696.82 40.5081 7 19 vrolijk positief 3739.36 8.9839 7 20 vermaak positief 4123 5.3284 7 21 applaus positief 4187.47 6.1958 7 22 knuffel positief 4191.8 5.0682 7 23 grappig positief 4199.1 37.4226 7 24 respect positief 4222.5 25.1425 7 25 prettig positief 4330.94 11.4746 7 26 gevoelig positief 3362.05 9.0954 8 27 vakantie positief 3767.38 17.6952 8 28 geslaagd positief 3880.63 6.3941 8 29 bevriend positief 3897.93 7.1499 8
37 30 speciaal positief 3912.33 19.8141 8 31 geweldig positief 4082.88 61.4126 8 32 glimlach positief 4171.72 6.3693 8 33 paradijs positief 4264.9 5.5638 8 34 gezellig positief 4326.44 11.1648 8 35 kerstmis positief 4378.2 6.6791 8 36 aangenaam positief 4215.5 21.5366 9 37 verrassen positief 4229.22 5.886 9 38 spelletje positief 4535 15.3779 9 39 vertrouwd positief 4236.23 3.8166 9 40 romantiek positief 4544.33 3.2466 9 41 belangrijk positief 3985.26 50.3098 10 42 romantisch positief 4095.55 9.3185 10 43 uitstekend positief 4178 15.9232 10 44 verrassing positief 4324.88 22.7014 10 45 ontspannen positief 4824.52 8.9591 10
Ten Positive Dutch words used in the Practice trials
46 enthousiast positief 3857.84 4.9442 11 47 gemakkelijk positief 3913.84 14.969 11 48 schitterend positief 3936.47 12.1066 11 49 fantastisch positief 3963.05 32.8501 11 50 betrouwbaar positief 4093.78 4.1388 11 51 feliciteren positief 3968.64 3.2962 11 52 beeldschoon positief 4393.27 3.1475 11 53 aantrekkelijk positief 3500.77 7.4349 13 54 vriendinnetje positief 4241.73 4.0025 13 55 gefeliciteerd positief 4431.21 26.5056 13 Averages 4078.076182 17.97207091 8.0363636
Forty-five Negative Dutch words used in the Experimental trials
word VALCAT RT mean SUBTLEXCD Word length
1 ziek negatief 4208.1 36.2454 4
38 3 straf negatief 4205.05 13.6183 4 4 graf negatief 4098.58 11.8959 4 5 vals negatief 4314.35 10.3346 4 6 kwaad negatief 4321.82 44.8575 5 7 wapen negatief 4228.5 30.4957 5 8 ruzie negatief 4481.11 21.9455 5 9 wreed negatief 3859.35 6.6791 5 10 slecht negatief 4191.35 65.3532 6 11 gewond negatief 4253.95 22.2305 6 12 vijand negatief 4312.52 16.8278 6 13 paniek negatief 4105.94 15.7497 6 14 geweld negatief 3961.05 14.7708 6 15 ziekte negatief 4165.7 11.8959 6 16 lelijk negatief 3939.21 11.3259 6 17 ongeluk negatief 4037.7 30.0248 7 18 verlies negatief 3886.94 20.7187 7 19 misdaad negatief 4073.58 14.5849 7 20 leugens negatief 4569.88 10.6072 7 21 schamen negatief 4174.33 7.5093 7 22 waanzin negatief 3597.58 7.4969 7 23 verraad negatief 4441.83 7.1004 7 24 klootzak negatief 4004 37.4473 8 25 gestolen negatief 4115.26 27.6456 8 26 gebroken negatief 4579.94 25.0682 8 27 gestoord negatief 4013.88 19.8265 8 28 gevangen negatief 4457.1 18.5378 8 29 begraven negatief 4346.38 17.3854 8 30 afscheid negatief 4519.73 13.8662 8 31 wanhopig negatief 4022.11 8.2032 8 32 gewapend negatief 4074.05 6.6914 8 33 vreselijk negatief 3684.16 34.2379 9 34 zelfmoord negatief 4051.75 17.7819 9 35 ontslagen negatief 4444.5 17.0136 9 36 vervelend negatief 4335.22 13.9281 9 37 leugenaar negatief 4384.72 13.7546 9 38 schrikken negatief 3905.57 12.7261 9
39 39 gespannen negatief 4176.16 7.9554 9 40 doodsbang negatief 3903.72 7.72 9 41 overgeven negatief 4454.94 7.3606 9 42 noodgeval negatief 4347.47 6.7782 9 43 vermoorden negatief 4630.44 45.4895 10 44 gevaarlijk negatief 3769.84 33.0607 10 45 gevangenis negatief 3960.94 25.3903 10
Ten Negative Dutch words used in the Practice trials
46 begrafenis negatief 3929 11.9579 10 47 afgesloten negatief 4195.82 10 10 48 ongelukkig negatief 3971.58 7.4845 10 49 duisternis negatief 4309.21 7.026 10 50 belachelijk negatief 4104.11 22.5155 11 51 slachtoffer negatief 4176.75 15.886 11 52 nachtmerrie negatief 4445.38 12.3296 11 53 aangevallen negatief 4498.27 11.3135 11 54 bewusteloos negatief 4423.72 6.8278 11 55 afschuwelijk negatief 4115.55 8.3519 12 Averages 4178.004364 18.11333091 7.81818182
Average RT EMOCLAR Positive 4078.0762
Negative 4178.0044
4128.0403 ms
SUBTLEX NL word frequency Positive 17.972071
Average Negative 18.113331
18.042701 % frequency
Average word length Positive 8.0363636
Negative 7.8181818
