Article
Reference
Implicit happiness and sadness are associated with ease and difficulty: evidence from sequential priming
LASAUSKAITE, Ruta, et al.
Abstract
Three experiments tested the hypothesis of implicit associations between happiness and the performance ease concept, and between sadness and the performance difficulty concept. All three studies applied a sequential priming paradigm: participants categorized emotion words (Experiment 1) or facial expressions (Experiment 2) as positive or negative or as referring to ease or difficulty (Experiment 3). These targets were preceded by briefly flashed ease- or difficulty-related words or neutral non-words (Experiments 1 & 2) or by happy, sad, or neutral facial expressions (Experiment 3) as primes. As predicted, all three experiments revealed increases in reaction times in the sequential priming task from congruent trials (happiness/ease and sadness/difficulty) over neutral trials to incongruent trials (sadness/ease and happiness/difficulty). The findings provide evidence for implicit associative links of happiness with ease and sadness with difficulty, as posited by the implicit-affect-primes-effort model (Gendolla, 2012, 2015).
LASAUSKAITE, Ruta, et al. Implicit happiness and sadness are associated with ease and difficulty: evidence from sequential priming. Psychological Research, 2017, vol. 81, no. 1, p.
321-331
DOI : 10.1007/s00426-015-0732-3
Available at:
http://archive-ouverte.unige.ch/unige:124225
Disclaimer: layout of this document may differ from the published version.
1 / 1
Implicit Happiness and Sadness are Associated with Ease and Difficulty:
Evidence from Sequential Priming
Ruta Lasauskaite, Guido H. E. Gendolla, Mylène Bolmont, and Laure Freydefont University of Geneva
In press: Psychological Research (future DOI: 10.1007/s00426-015-0732-3) 6492 words (main text + notes)
40 references
Accepted final version:
Lasauskaite, R., Gendolla, G.H.E., Bolmont, M., & Freydefont, L. (2017). Implicit happiness and sadness are associated with ease and difficulty: Evidence from sequential priming. Psychological Research, 81, 321-331. doi: 10.1007/s00426-015-0732-3.
Mailing Address: Guido H.E. Gendolla Geneva Motivation Lab FPSE, Dept. of Psychology University of Geneva 40, Bd. du Pont-d’Arve
CH-1211 Geneva 4, Switzerland Telephone: +41 22 379 92 31 Fax: +41 22 379 92 16
E-mail: [email protected]
Abstract
Three experiments tested the hypothesis of implicit associations between happiness and the performance ease concept, and between sadness and the performance difficulty concept. All three studies applied a sequential priming paradigm: participants categorized emotion words (Experiment 1) or facial expressions (Experiment 2) as positive or negative or as referring to ease or difficulty (Experiment 3). These targets were preceded by briefly flashed ease- or difficulty- related words or neutral non-words (Experiments 1 & 2) or by happy, sad, or neutral facial expressions (Experiment 3) as primes. As predicted, all three experiments revealed increases in reaction times in the sequential priming task from congruent trials (happiness/ease and
sadness/difficulty) over neutral trials to incongruent trials (sadness/ease and happiness/difficulty).
The findings provide evidence for implicit associative links of happiness with ease and sadness with difficulty, as posited by the implicit-affect-primes-effort model (Gendolla, 2012, 2015).
Keywords: implicit affect, implicit associations, sequential priming, implicit-affect-primes-effort model
Implicit Happiness and Sadness are Associated with Ease and Difficulty:
Evidence from Sequential Priming
A recent series of studies has revealed replicated evidence that briefly flashed affective stimuli that are processed during task performance systematically influence effort mobilization, operationalized as cardiac response. The best replicated finding in this research is that
participants who processed sadness primes during moderately difficult cognitive tasks mobilized more resources than participants who processed happiness primes (e.g., Gendolla & Silvestrini, 2011; Lasauskaite, Gendolla, & Silvestrini, 2013; Silvestrini & Gendolla, 2011a; see also Blanchfield, Hardy, & Marcora, 2014). According to the implicit-affect-primes-effort (IAPE) model (Gendolla, 2012, 2015)—the theoretical framework guiding this research—this systematic impact of affect primes on effort mobilization is explained by their impact on experienced task demand. Affect primes are supposed to activate mental representations of affective states (i.e.
implicit affect). Accessible and applicable knowledge in terms of features of the activated affect concept can then influence behavior.
Given that resource mobilization is guided by the general principle of resource
conservation (Brehm & Self, 1989), especially accessible information about performance ease or difficulty is applicable to the self-regulation of effort. The IAPE model suggests that people have acquired knowledge about performance ease and performance difficulty as typical features of different affective states. Rendering this available knowledge accessible results in experiences of lower or higher subjective demand, which influences effort. According to motivational intensity theory (Brehm & Self, 1989), people mobilize resources in proportion to subjective demand as long as success is possible and the necessary effort is justified (see Gendolla, Wright, & Richter, 2012 for a review). Some affect primes like happiness activate the ease concept, while others like sadness activate the difficulty concept as features of their mental representation. That way,
different affect primes have different systematic effects on effort mobilization due to their impact on subjective task demand. That is, the core idea of the IAPE model is that of the existence of implicit associative links between individuals’ representations of different affective states and performance ease or difficulty.
In support of the IAPE model logic, some of our studies have found that participants who processed sadness primes during task performance rated subjective task demand indeed as higher than participants who processed happiness primes (Gendolla & Silvestrini, 2011; Lasauskaite et al., 2013; Silvestrini & Gendolla, 2011b; see also Lasauskaite Schüpbach, Gendolla, &
Silvestrini, 2014). But although all of our published studies revealed the predicted affect prime effects on effort mobilization, some of those experiments did not find significant prime effects on post-performance ratings of subjective demand (Chatelain & Gendolla, 2015; Freydefont &
Gendolla, 2012; Freydefont, Gendolla, & Silvestrini, 2012; Silvestrini & Gendolla, 2011a). One could argue that null effects, like nonsignificant prime effects on demand ratings, are hardly interpretable and may be ignored. But also for the obtained significant affect prime effects on subjective demand evaluations it is to note that those ratings were retrospective judgments, which are vulnerable for a number of biases (Robinson & Clore, 2002). Most relevant, the IAPE model posits that affect primes influence subjective task demand directly during performance. This is hardly assessable with self-report measures for which it would be necessary to interrupt task performance. Thus, more direct measures of associative links between implicit sadness and difficulty and between implicit happiness and ease are called for (see De Houwer, Teige- Mocigemba, Spruyt, & Moors, 2009; Nosek, Hawkins, & Frazier, 2011 for overviews). The present research addressed this issue.
The Present Research
Three experiments tested the hypothesis of implicit associative links between individuals’
representations of two types of affect and the concepts of ease and difficulty, which are central for the IAPE model explanation for affect prime effects on effort mobilization. We did so by applying a sequential priming paradigm (see Wentura & Degner, 2010). Here, the presentation of one stimulus—the prime—facilitates (or impairs) the accessibility and thus the categorization of a subsequent stimulus—the target—to the degree to which the stimuli-related concepts are associated (or dissociated) in memory. Since we were interested in associative links rather than implicit evaluations, which have been frequently investigated with this paradigm (see Klauer &
Musch, 2003), we tried to minimize the evaluative process in sequential priming by our choice of prime and target stimuli. In Experiments 1 and 2, we did so by using non-emotional concepts—
performance ease and difficulty words—as primes and the affective stimuli—sadness and
happiness words and expressions—as targets (in evaluative sequential priming the stimulus order is usually inversed). In Experiment 3, we used emotional stimuli as primes, but asked participants to categorize demand-related target words as referring to ease or difficulty—i.e. non-emotional response categories.
We tested the idea that performance ease is part of the emotion knowledge about happiness and that performance difficulty is part of the emotion knowledge about sadness (cf.
Niedenthal, 2008). If the IAPE model logic about implicit associative links of sadness with difficulty and of happiness with ease is correct, target categorization should be faster for congruent prime-target configurations (ease/happiness and difficulty/sadness) than for incongruent trials (ease/sadness and difficulty/happiness).
Experiment 1
Participants categorized happiness- and sadness-related words that were preceded by difficulty- or ease-related prime words or neutral non-words. Based on the IAPE model
(Gendolla, 2012, 2015), we expected a linear increase in response times from congruent trials (ease/happiness and difficulty/sadness), reflecting high accessibility of the targets, over neutral control trials (neutral/sadness and neutral/happiness) to incongruent trials (ease/sadness and difficulty/happiness), reflecting inhibited accessibility of the targets.
Method Participants and Design
Twenty-four first year psychology students participated for course credit. All participants were fluent in French. Four participants were excluded from the data analysis due to technical problems during the data collection (2 participants), lacking responses (1 participant), or an error rate above 50% (1 participant), leaving a final sample of N = 20 (18 women, average age 20 years). The experiment was run in a 3 (prime: ease, neutral, difficulty) 2 (target: happy, sad) 3 (time: block 1, block 2, block 3) within-participants design.
Stimuli and Randomization
The study was run in French; therefore the administered prime and target words were also in French. Based on a pretest, we selected 4 ease-related and 4 difficulty-related words as ease and difficulty primes. For the pretest, other university students than those who participated in the main experiment rated two lists of the same 12 words on 8-point scales regarding how far each word related to difficulty (list 1) and ease (list 2). The 4 words that were rated as being most strongly associated with ease (translation: light, relaxed, holidays, obvious) and the 4 words rated as being most strongly associated with difficulty (translation: hard, exam, effort, test) were chosen as prime words for the experiment. Neutral primes were 4 non-words created by jiggling the letters of 2 words from the ease- and 2 words from the difficulty primes (taken from the original lists in French) (emaxem, edivent, cofgé, edpeuve). Target words were 4 happiness-
(translation: gaiety, happy, joyful, pleasant) and 4 sadness-related words (translation: painful, overwhelmed, downcast, sad)1. Each of the 3 experimental blocks consisted of 96 trials—i.e., randomized sets of all possible 12 × 8 prime-target combinations.
Procedure
The experiment was run in individual sessions. The procedure was computerized and instructions were displayed on the computer monitor. After having signed a consent form, participants had to correctly categorize the target words as quickly as possible as negative or positive by pressing respective response keys. Participants responded by pressing respective response keys (“A” or “L”) on a keyboard with the fingers of their choice using both hands. To control for potential learning effects, we administered 3 blocks that were separated by 30 sec.
breaks. Trials started with a fixation cross (500 ms), followed by a prime word (53 ms), which was backward-masked with a series of „X“ (133 ms), resulting in an SOA of 186 ms. Then a target word appeared for maximally 2000 ms during which participants had to respond. All stimuli appeared in the center of the computer screen (in all three experiments reported here).
Inter-trial-intervals were 3000 ms minus participants’ response time. This way, all blocks had the same duration for all participants.
Before the task, participants performed 20 practice trials with neutral primes and different target words than in the main task. In those training trials participants received feedback about categorization correctness, their response time, and the percentage of correctly classified words.
Feedback was absent during the main experimental task to avoid possible affective reactions (e.g., Kreibig, Gendolla, & Scherer, 2012) that could interfere with the primes and target words.
Results Response Times
The response time range for correct target categorizations was 200 - 2000 ms, making outlier cutoffs unnecessary. Given that the response time distribution was skewed, we analyzed square-root-transformed data. To facilitate the interpretation of the findings, we report the results, however, in milliseconds. We considered only correct responses in the reaction time analyses in this and the other two experiments.
Congruency. As outlined by Wentura and Degner (2010), sequential priming effects are best expressed as differences between congruent and incongruent prime-target configurations.
Therefore we calculated response time scores for congruent (ease/happiness; difficulty/sadness), incongruent (ease/sadness; difficulty/happiness), and neutral control trials (non-word/sadness;
non-word/happiness). This resulted in a congruency factor as independent variable (congruent trials vs. neutral trials vs. incongruent trials). We then tested our hypothesis about a linear increase in response times from the congruent over the neutral to the incongruent trials with contrast analysis, which is the most powerful statistical test of predicted patterns of means (Rosenthal & Rosnow, 1985; Wilkinson and the Task Force for Statistical Inference, 1999).
Specifically, we tested if the linear congruency contrast was significant in the context of a 3 (congruency) 3 (block) repeated measures ANOVA. This analysis revealed a significant and strong linear congruency contrast effect, F(1, 19) = 24.98, p < .001, p2 = .568. The orthogonal quadratic contrast failed significance (p = .057) and the block main effect and the congruency block contrast interactions were not significant (ps > .29).
As depicted in Figure 1, response times increased, as expected, from the congruent trials (M = 626.42, SD = 107.09) via the neutral-prime control trials (M = 628.71, SD = 108.74) to the incongruent trials (M = 641.36, SD = 112.04). This reflects the predicted sequential priming effect: congruent trials composed of difficulty-related primes/sadness-related targets and ease-
related primes/happiness-related targets were processed faster than incongruent trials, consisting of difficulty-related primes/happiness-related targets and ease-related primes/sadness-related targets. Moreover, additional comparisons with paired t-tests revealed that the incongruent condition differed significantly from both the congruent and neutral conditions, ts(19) > 3.89, ps
< .001, Cohen’s ds > .81.2 The difference between the congruent and neutral condition was, however, not significant (p = .245).
Decomposed prime target analysis. Additionally to the prime-target congruency effect analysis, we further examined the reaction times according to the experimental design to evaluate if the above-discussed sequential priming effect was carried by specific prime-target
configurations. A 3 (prime) 2 (target) 3 (block) repeated measures ANOVA of the reaction times found a significant target main effect, F(1, 19) = 41.01, p < .001, p2 = .683: participants responded generally faster to positive targets (M = 610.46, SD = 103.43) than to negative targets (M = 653.86, SD = 116.71). More relevant, the prime target factor interaction was significant, F(2, 38) = 8.14, p = .003, p2 = .300. As anticipated, sadness-related words were faster classified in the difficulty-prime condition (M = 645.14, SD = 108.87) than in the ease-prime condition (M
= 664.53, SD = 127.72), t(19) = 2.45, p = .012, Cohen’s d = .67. The neutral-prime condition fell in between (M = 651.92, SD = 116.96) but differed only significantly from the ease-prime
condition, t(19) = 1.87, p = .038, Cohen’s d = .42, but not from the difficulty-prime condition (p
= .152). By contrast, happiness-related targets were classified more slowly in the difficulty-prime condition (M = 618.18, SD = 102.12) than in the neutral-prime condition (M = 605.50, SD = 103.45), t(19) = 2.45, p = .012, Cohen’s d = .62. Reaction times in ease-prime condition (M = 607.69, SD = 107.93) differed significantly from the difficulty-prime condition, t(19) = 1.71, p = .026, Cohen’s d = .41, but not from the neutral-prime condition (p = .370). The prime target
block interaction was not significant (p = .471).
Response Accuracy
Participants responded correctly within the response time window in M = 96.98% (SE = 0.49) of the trials. Neither a 3 (congruency) 3 (block) repeated measures ANOVA that
corresponded to the above discussed analysis of response times (ps > .572), nor a 3 (prime) 2 (target) 3 (block) repeated measures ANOVA, corresponding to the response times analysis in the above decomposed analysis, found any significant effects on response accuracy (ps > .277).
Only the block main effect approached significance, F(2, 38) = 3.21, p = .065, p2 = .145, reflecting a trend to a learning effect due to slightly lower accuracy in the first block (M = 96.25, SE = 0.65) than in the second (M = 97.55, SE = 0.50; p = .05) and the third blocks (M = 97.14, SE = 0.57; p = .122), according to LSD post-hoc tests. No other ANOVA effects approached significance (ps > .279).
Discussion
The present results lend first support to the IAPE model idea about implicit associative links between difficulty and sadness and between ease and happiness (Gendolla, 2012, 2015):
Congruent trials (ease/happiness and difficulty/sadness) led to faster response times than
incongruent trials (ease/sadness and difficulty/happiness) and the neutral prime control trials fell in between these two conditions—albeit direct comparisons revealed only a significant difference with the incongruent condition. However, the results supported the overall linear pattern and, most relevant, response times in the congruent condition were significantly faster than those in the incongruent condition, though the incongruent trials produced a stronger effect.
Additionally to the above-discussed expected congruency effects, the more detailed analysis of the different prime-target configurations on the target categorization response times
revealed a target main effect: happiness-related target words were in general faster categorized than sadness-related targets. This higher general accessibility of happiness-related targets was not predicted a priori. Nevertheless, faster processing of positive information is a rather common observation. The facilitated categorization of positive stimuli, reflecting high accessibility of positive information, can be explained as being caused by higher density of positive information in memory (Unkelbach, Fiedler, Bayer, Stegmüller, & Danner, 2008). However, the target main effect merely added a constant to the expected prime target interaction. Altogether, the results of Experiment 1 lend initial support our prediction of implicit associative links of happiness with ease and sadness with difficulty.
Experiment 2
In order to generalize our findings and to test if other stimuli produced more symmetrical effects of congruent and incongruent prime-target configurations, we conceptually replicated our first study by administering (1) sad and happy facial expressions instead of emotion-words as targets and (2) a new list of difficulty- and ease-related words as primes. We also (3) increased the number of target items and run only two blocks, because the response time analysis of Experiment 1 had not revealed that the sequential priming effect involved a strong learning component and occurred only in later blocks. Finally, we (4) run a bigger sample to increase statistical power in order to still better protect against possible false positive effects (see McShane, & Bockenholt, 2014; Simmons, Nelson, & Simonsohn, 2011). We expected again a linear increase in response times from congruent trials (ease/happiness and difficulty/sadness) via neutral trials (neutral/sadness and neutral/happiness) to incongruent trials (ease/sadness and difficulty/happiness).
Method
Participants and Design
Forty first-year psychology students (36 women) participated for course credit. Seven students were not native French speakers and consequently excluded from the analysis, leaving N
= 33 participants as final sample (average age 20 years). The design was 3 (prime: ease, neutral, difficulty) 2 (target: happy, sad) 2 (time: block 1, block 2) within-participants.
Stimuli and Randomization
We ran a new pretest for selecting new prime words. Participants (other than the
experimental sample) rated on 7 point-scales, ranging from easy (0) to difficult (6), in how far 20 French ease- and difficulty-related words referred to ease or difficulty. We selected the 4 words that were rated as being most strongly associated with ease (translation: easy, evident, relaxed, simple) and the four words that were most strongly associated with difficulty (translation:
difficulty, complex, complicated, testing) as primes, and created 4 neutral primes by jiggling the letters of the prime words (aina, sornel, sempale, epreviens).2
Targets were 6 pictures of sad and 6 pictures of happy facial expressions of 3 women (no.
16, 22, 31) and 3 men (no. 10, 46, 71) taken from the Radboud Faces Database (RaFD). The chosen happiness and sadness expressions had the highest emotion recognition scores in the database validation study (Langner et al., 2010). One experimental block consisted of 144 trials, i.e. all possible combinations of 12 primes 12 targets, which were pseudo-randomized in that the same prime and the same target were not presented successively.
Procedure
The study was again run in individual sessions. After having signed consent, participants had to correctly categorize as quickly as possible if the target faces expressed a rather positive or rather negative emotional state by pressing respective response keys (“4” or “6”) on a number-
pad keyboard with the fingers of their choice of their dominant hand (the same applies to Experiment 3). Each of the two blocks, which were separated by a 30 sec. break, took about 10 minutes. Trials started with a fixation cross (500 ms), followed by a prime word (53 ms), which was backward-masked with a series of “X” letters (133 ms). Then a target picture appeared for maximally 1500 ms during which participants had to respond (the response time window was shorter than in Experiment 1, because facial expressions can be faster categorized than words).
After the response, the message “response entered” appeared for 2000 ms minus participants’
response time. All stimuli appeared in the middle of the computer screen. The inter-trial interval randomly varied between 500 and 1500 ms. The task was preceded by 16 training trials with correctness feedback, consisting of neutral primes and other sadness and happiness expressions as targets than those presented in the task. No feedback was given during the main task.
Results Response Times
Due to a skewed distribution, we analyzed again square-root-transformed response times of correct target categorizations but report, for the sake of easier interpretation, results in
milliseconds. Response time scores for congruent, incongruent, and neutral control trials were created as in the first study.
Congruency. A 3 (congruency) 2 (block) ANOVA revealed the expected significant linear congruency contrast effect, F(1, 32) = 4.91, p = .034, p2 = .13. As expected and depicted in Figure 2, response times increased from the congruent (M = 599.17, SD = 88.35) via the neutral-prime control trials (M = 602.94, SD = 86.62) to the incongruent trials (M = 606.69, SD = 88.51). The orthogonal quadratic contrast was not significant (p = .983). Neither the bloc main effect (p = .200) nor the block congruency contrast interactions (p > .168) approached
significance. This reflects again a significant sequential priming effect: congruent trials,
composed of difficulty-related primes and sadness-related targets and of ease-related primes and happiness-related targets, were faster processed than incongruent trials, consisting of difficulty- related primes and happiness-related targets and ease-related primes and sadness-related targets.
Additional comparisons with paired t-tests revealed that the incongruent condition differed significantly from congruent condition t(32) = 2.13, p = .021, Cohen’s d = .34. The differences between the neutral condition and the congruent (p = .158) and incongruent
conditions (p = .141) were, however, not significant although the response times depicted an even linear pattern across the conditions.
Decomposed prime target analysis. To test if the above-discussed sequential priming effect was carried by specific prime-target configurations, we further analyzed the response times with a 3 (prime) 2 (target) 2 (block) repeated measures ANOVA. This analysis revealed again a target main effect, F(1, 32) = 11.63, p = .002, p2 = .267. As in Experiment 1, participants generally categorized happiness targets faster (M = 594.74, SD = 88.85) than sadness targets (M = 611.13, SD = 87.61). Neither the prime target interaction (p = .144) nor the prime target block interaction (p = .571) were significant.
Response Accuracy
Participants responded correctly and without exceeding the response time window in M = 97.21% (SD = 4.59) of the trials. A 3 (congruency) 2 (block) ANOVA, corresponding to the above analysis of congruency effects on response times, revealed no significant effects (ps >
.111). A 3 (prime) 2 (target) 2 (block) ANOVA of the accuracy scores corresponding to the above decomposed prime target analysis found a significant prime main effect, F(2, 64) = 3.69, p = .031, p2 = .103, due to slightly higher response accuracy after the difficulty primes (M =
98.14, SD = 2.21) than after both the ease (M = 97.41, SD = 3.42; p = .065) and the neutral primes (M = 97.06, SD = 3.69, p = .03), according to LSD post hoc tests. No other ANOVA effect approached significance (ps > .20).
Discussion
The present findings conceptually replicate the core results of Experiment 1 with different sets of ease- and difficulty-related prime words, emotional expressions rather than sadness- and happiness-related words as targets, and a bigger sample. The sequential priming effect of faster correct target categorizations in congruent than in incongruent trials reflects again the implicit associations of ease with happiness and of difficulty with sadness. The differences between the neutral-prime condition and both the congruent and incongruent conditions were not significant, but, obviously, the neutral-prime condition fell in between the other two, producing an even linear pattern of response times. Thus, we interpret the findings as conceptual replication of the first study providing additional evidence for implicit associative links of sadness with difficulty and happiness with ease, as posited in the IAPE model (Gendolla, 2012, 2015).
Experiment 3
So far we have found that ease- and difficulty-related prime words can systematically influence the categorization speed of sadness- and happiness-related targets. This provides evidence for associative links of ease with happiness and difficulty with sadness—as posited by the IAPE model. However, participants’ task was categorizing targets as positive or negative.
Considering that ease-related words may have a positive connotation, while difficulty-related words may have a negative connotation (see Dreisbach & Fisher, 2012), a critical reader could argue that the findings of our first two experiments reflect affective priming (Fazio, 2001) or more generally response priming instead of sematic links (see Wentura & Rothermund, 2014).
Accordingly, positive and negative primes facilitated congruent positive and negative responses.
We addressed this issue in this third study by changing the experimental procedure. (1) We used sad and happy facial expressions as primes and ease- and difficulty-related words as targets. (2) Most relevant, we asked participants to categorize the targets as referring to ease vs.
difficulty instead of positivity vs. negativity. This divergence between primes (emotion
categories) and responses (demand categories) should make simple response priming unlikely.
(3) We chose emotional facial expressions from another database than in Experiment 2 in order to further generalize the expected findings. We predicted again a linear increase in response times from congruent trials (sad/difficulty and happy/ease) via neutral trials (neutral/difficulty and neutral/ease) to incongruent trials (sad/ease and happy/difficulty).
Method Participants and Design
Forty first-year psychology students (33 women; average age 21 years) participated for course credit. All students were native French speakers. The design was 3 (prime: sadness, neutral, happiness) 2 (target: difficulty, ease) 2 (time: block 1, block 2) within persons.
Stimuli and Randomization
Averaged sad (FSAS, MSAS), neutral (FNES, MNES), and happy (FHAS, MHAS) male and female faces from the Averaged Karolinska Directed Emotional Faces (AKDEF) database (Lundqvist & Litton, 1998) were used as primes. These pictures were chosen for this experiment because they correspond to primes in other experiments of our lab testing the IAPE model hypothesis on mental effort. Targets were the 4 difficulty-related and 4 ease-related words that were employed as primes in Experiment 2. Forty-eight combinations—i.e., all possible
combinations of 6 primes × 8 targets—were pseudo-randomized in that the same prime and the
same target were not presented successively. Each experimental block consisted of 144 trials (3 sets of 48 pseudo-randomized combinations).
Procedure
The experiment was again run in individual sessions. After having signed the consent form, participants had to correctly categorize as quickly as possible if a target word was related to difficulty or ease by pressing respective response keys. Each of the 2 blocks, which were
separated by a 30 sec. break, took about 10 minutes. Trials started with a fixation cross (500 ms), followed by a prime picture (27 ms). The prime presentation time was decreased due to
facilitated priming with pictures than words (e.g., Carr, McCauley, Sperber, & Parmelee, 1982) and our pretests. The primes were backward masked with a picture of the size showing scattered black and white dots (133 ms). A target word appeared for maximally 2000 ms during which participants had to respond with the fingers of their choice of their dominant hand on a number pad. After the response, the message “response entered” appeared for 2000 ms minus
participants’ response time. Again, all stimuli appeared in the middle of the computer screen. The inter-trial interval randomly varied between 500 and 1500 ms. The task was preceded by 16 training trials with correctness feedback, consisting of neutral primes and other difficulty and ease words as those presented in the main task. No feedback was given during the main task.
Results Response Times
Due to skewed distributions, we analyzed again square-root-transformed response times of correct target categorizations but report, for the sake of easier interpretation, results in milliseconds. We created again response time scores for congruent, incongruent, and neutral control trials.
Congruency. A 3 (congruency) 2 (block) repeated measures ANOVA revealed a block main effect, F(1, 39) = 21.91, p < .001, p2 = .360. Participants were slower in the first block (M
= 706.70, SD =) than in the second block (M = 672.33, SD =), reflecting a learning effect. The linear congruency contrast was not significant (p = .228). However, there was a significant linear congruency contrast block interaction, F(1, 39) = 8.43, p = .006, p2 = .18. We decomposed this interaction effect by running contrast analyses for each block. Neither the linear congruency contrast (p = .199), nor the orthogonal quadratic contrast (p = .341) was significant in block 1 (means response times were as follows: congruent M = 711.52, SD =102.71, neutral M = 703.14, SD = 95.59, incongruent M = 705.45, SD = 102.95). However, the linear congruency contrast was significant in block 2, F(1, 39) = 6.78, p = .013, p2 = .15, while the orthogonal quadratic contrast was not (p = .775). As depicted in Figure 3, response times increased from the congruent trials (M = 664.42, SD = 103.87) via the neutral trials (M = 671.611, SD = 105.75) to the incongruent trials (M = 680.96, SD = 113.50).
Additional comparisons with paired t-tests found that the incongruent condition differed significantly from the congruent condition, t(39) = 2.61, p = .007, Cohen’s d = .39, and
tendencially from the neutral condition, t(39) = 1.60, p = .059, Cohen’s d = .26. The difference between the congruent and neutral conditions was not significant (p = .123) though the response times depicted an even linear pattern across the conditions.
Decomposed prime target analysis. We tested again if the above-reported effects were carried by specific prime-target configurations. A 3 (prime) 2 (target) 2 (block) repeated measures ANOVA of the reaction times found, of course, the above-reported significant block effect (p < .001). The target main effect just failed significance (p = .053; p2 = .093). Ease- related words (M = 683.44, SD = 97.37) only tended to be categorized faster than difficulty-
related words (M = 695.60, SD = 104.61). Most relevant, the ANOVA also revealed a significant three-way interaction, F(2, 78) = 4.01, p = .023, p2 = .093. Further analysis revealed that the prime target interaction was not significant for block 1 (p = .512), but for block 2, F(2, 78) = 4.39, p = .018, p2 = .10.
In block 2, ease-related words were faster categorized in the happy-prime condition (M = 658.24, SD = 108.54) than in the sadness-prime condition (M = 678.75, SD = 123.87), t(39) = 2.17, p = .018, Cohen’s d = .38, with the neutral-prime condition falling in between (M = 667.23, SD = 108.78), but not differing significantly from the happy- and sadness-prime conditions (ps >
.11). By contrast, classification of difficulty-related words was significantly faster in the sadness-prime condition (M = 670.60, SD = 105.35) than in the happiness-prime condition (M = 683.18, SD = 111.91), t(39) = 1.70, p = .049, Cohen’s d = 26. The neutral-prime condition (M = 675.99, SD = 110.94) fell in between but did not differ significantly from the happiness-prime and sadness-prime conditions (ps > .101).
Response Accuracy
Participants responded correctly and within the response time window in M = 96.50% (SD
= 0.04) of the trials. Neither a 3 (congruency) 2 (block) repeated measures ANOVA, corresponding to the above congruency effect analysis of response times (ps > .454), nor a 3 (prime) 2 (target) 2 (block) ANOVA of the accuracy scores corresponding to the above decomposed prime target analysis revealed any significant effects (ps > .378).
General Discussion
Using a sequential priming paradigm, the three present experiments support the idea of existing implicit associative links between individuals’ representations of sadness and the concept of performance difficulty as well as between representations of happiness and the
performance ease concept. Corresponding effects were found using different sets of ease- and difficulty-related prime words and different types of target stimuli—happiness- and sadness- related words in Experiment 1 and happy or sad facial expressions in Experiment 2. Finally, Experiment 3 used emotional expressions as primes and asked participants to categorize target words referring to ease vs. difficulty. Considering these different manipulations, all three studies revealed corresponding results: Response times for participants’ target categorizations increased from the congruent trials (sadness-difficulty/happiness-ease) via the neutral prime control trials to the incongruent trials (sadness-ease/happiness-difficulty). Although focused comparisons
revealed that the neutral-prime condition did not differ significantly from the congruent
condition, and in Experiment 2 also not from the incongruent condition, the pattern of response times was clearly linear in all three experiments, as also supported by the significant linear contrast effects in absence of significant effects for the orthogonal quadratic contrasts. Thus, we conclude that the present studies revealed convergent evidence for the hypothesized associative links between representations of affect and demand. However, in Experiment 3, the congruency effect appeared only in the second block, indicating a learning effect, which is not uncommon for obtaining reliable priming effects (Wentura & Degner, 2010).
We are aware that some authors have criticized sequential priming procedures as
suboptimal to test for semantic associations between categories and their mental representation in terms of concepts. The reason is that sequential priming procedures could in fact trigger response priming instead of activating semantic associations. In response priming, response speed and response correctness benefit from compatibility between a prime and a response (see Wentura &
Rothermund, 2014)—for example because prime and response share a response-relevant critical feature like the same positive or negative valence as in the case of affective priming (Fazio, 2001). That is, in response priming a prime facilitates a reaction like categorizing a stimulus as
positive because the prime activated this response.
However, we think that an explanation of the reported effects in terms of response priming is not straightforward. First, sequential priming may activate responses rather than semantic associations in some cases, but this does not mean that sequential priming is always in fact response priming. Second, the responses in Experiment 3 were demand-related, while the primes were emotion-related. That is, in contrast to Experiments 1 and 2, participants made no evaluative categorizations in terms of positivity or negativity of the targets. This divergence between prime categories and response categories speaks for semantic links between sadness and difficulty and between happiness and ease instead of facilitated responses because primes and responses belonged to the same categories. Third, it is of note that all three studies presented all stimuli in the center of the computer screen. That is, primes and responses did not share a feature that the target stimuli did not share, as it for example would have been the case if the primes would have appeared at locations that corresponded to the response keys. Consequently, the primes could not directly prime a motoric response due to compatibility between prime location and response location—like for example in the Simon effect (Simon, 1969). Hence, if primes would have directly primed responses, this could only be because of the recoding of the responses in terms of the feature used to categorize the target stimuli—such as valence
(Experiments 1 & 2) or ease/difficulty (Experiment 3). This speaks for a semantic link between sadness and difficulty and between happiness and ease. However, we acknowledge that the present effects are limited to one experimental paradigm—sequential priming. We leave it to future studies to test if compatible effects can be obtained with different priming procedure, as for example a lexical decision task to answer the question of the present effects are generalizable to other paradigms (cf. Wentura & Rothermund, 2014).
Looking at the sequential priming literature, it appears that categorizing the valence of
stimuli is the easiest categorization response (e.g., Murphy & Zajonc, 1993; Zajonc, 1980).
Maybe because of this, categorizing the valence of target stimuli appears to be the most frequently applied task in sequential priming (see Klauer & Musch, 2003). In fact, researchers have just recently started to ask participants for other, emotion-specific rather than valence- specific categorizations (e.g., Neumann & Lozo, 2012; Rohr, Degner, & Wentura, 2012).
However, associative links between two concepts exist—by definition—in both ways. If ease and difficulty are features of individuals’ mental representations of happiness and sadness, as the IAPE model suggests, ease and difficulty should activate the happiness and sadness concepts and vice versa. Also, if it is conceivable that these associative links differ in strength—maybe a sadness prime can easier activate the difficulty concept than the other way around—those links should be always present. Consequently, assessing the link between non-affective primes and affective target stimuli—as we did in Experiments 1 and 2—is an appropriate measure of the associative strength between both concepts. Experiment 3, with the task to categorize target stimuli regarding their reference to ease or difficulty adds to the just emerging evidence of sequential priming effects in paradigms that ask for semantic rather than valence-based categorization.
The present studies were based on the IAPE model (Gendolla, 2012, 2015) that explains how implicit affect influences mental effort. Accordingly, affect primes activate mental
representations of affective states (i.e. implicit affect), which in turn render associated knowledge about performance ease vs. difficulty accessible. Because task performance needs resources and resource mobilization is guided by a resource conservation principle (Brehm & Self, 1989), this activated knowledge is especially applicable in achievement situations. People avoid mobilizing more resources than necessary and therefore look for hints about task demand (see Gendolla et al., 2012 for a recent review). According to the IAPE model, implicit affect informs about
subjective demand, because it activates representations of performance ease or difficulty. The present studies provide support for this assumption.
Additionally to the predicted congruency effects, more detailed analyses also found target main effects in Experiment 1 and 2: happiness-related target stimuli were in general categorized faster than sadness-related targets. These significant effects were strong, as evident in large effect sizes. In Experiment 3, a corresponding effect for ease- vs. difficulty-related targets did not attain significance and was rather weak. Although we had not predicted higher general accessibility of the happiness concept in Experiments 1 and 2, those main effects are not too surprising. In accordance with the afore mentioned density hypothesis (Unkelbach et al., 2008), positing that positive information is represented in higher density than negative information, also other studies have found that positive stimuli are faster recognized than negative stimuli. Research on the recognition of facial expressions of emotions has, for example, revealed that happiness expressions are faster processed than sadness, anger, disgust, or even neutral expressions (see Kirita & Endo, 1995; Leppänen & Hietanen, 2003). Concerning faces, it has been argued that this
“happy face recognition advantage” occurs primarily at cognitive processing states (Leppänen, Tenhunen, & Hietanen, 2003) and that it may reflect a higher-level asymmetry in the recognition and categorization of emotionally positive and negative signals in general. This is compatible with the fact that the faster accessibility of the happiness concept occurred not only with faces as target stimuli (Experiment 2), but also with word stimuli (Experiment 1). That is, the happiness advantage effect applies well to faces, but it is apparently not limited to this type of social stimuli. Nevertheless, the target main effect in the present studies did not interfere with our predictions about implicit associations of sadness with difficulty and happiness with ease.
However, the divergence in the strength of target main effects between Experiments 1 and 2, in which participants categorized happiness- and sadness related targets, and Experiment 3, in
which they categorized demand-related targets, may also be regarded as support for the idea that the ease- and difficulty-related target words were not processed as positive and negative stimuli.
Conclusions
This present research provides the first direct evidence for implicit associations between the concept of difficulty and the representation of sadness and between ease and happiness, as predicted by the IAPE model (Gendolla, 2012, 2015). Demonstrating the existence of those links is essential to explain how sadness- and happiness primes could influence experienced task demand and effort, which was found in our previous studies (e.g., Gendolla & Silvestrini, 2011;
Lasauskaite et al., 2013; Silvestrini & Gendolla, 2011b; see also Lasauskaite Schüpbach et al., 2014). Thus, the present findings lend support to the idea that individuals’ emotion knowledge involves associations between sadness and difficulty and between happiness and ease.
References
Blanchfield, A., Hardy, J., & Marcora, S. (2014). Non-conscious visual cues related to affect and action alter perception of effort and endurance performance. Frontiers in Human
Neuroscience, 8, 967. doi:10.3389/fnhum.2014.0096
Brehm, J. W., & Self, E. A. (1989). The intensity of motivation. Annual Review of Psychology, 40, 109–131. doi:10.1146/annurev.ps.40.020189.000545
Carr, T. H., McCauley, C., Sperber, R. D., & Parmelee, C. M. (1982). Words, pictures, and priming: On semantic activation, conscious identification, and the automaticity of information processing. Journal of Experimental Psychology: Human Perception and Performance, 8, 757–777. doi:10.1037/0096-1523.8.6.757
Chatelain, M., & Gendolla, G. H. E. (2015). Implicit fear and effort-related cardiac response.
Biological Psychology, 111, 73-82. doi:10.1016/j.biopsycho.2015.08.009
De Houwer, J., Teige-Mocigemba, S., Spruyt, A., & Moors, A. (2009). Implicit measures: A normative analysis and review. Psychological Bulletin, 135, 347–368.
doi:10.1037/a0014211
Dreisbach, G., & Fischer, R. (2012). Conflicts as aversive signals. Brain and Cognition, 78, 94–
98. doi:10.1016/j.bandc.2011.12.003
Fazio, R. H. (2001). On the automatic activation of associated evaluations: An overview.
Cognition & Emotion, 15, 115–141. doi:10.1080/0269993004200024
Förster, J., & Liberman, I. (2007). Knowledge activation. In A. W. Kruglanski & E. T. Higgins (Eds.), Social psychology: Handbook of basic principles (2nd ed.). (pp. 201-31). New York, NY: Guilford Press.
Freydefont, L., & Gendolla, G. H. E. (2012). Incentive moderates the impact of implicit anger versus sadness cues on effort-related cardiac response. Biological Psychology, 91, 120 127. doi:10.1016/j.biopsycho.2012.04.002
Freydefont, L., Gendolla, G. H. E., & Silvestrini, N. (2012). Beyond valence: The differential effect of masked anger and sadness stimuli on effort-related cardiac response.
Psychophysiology, 49, 665–671. doi:10.1111/j.1469-8986.2011.01340.x
Gendolla, G. H. E. (2012). Implicit affect primes effort: A theory and research on cardiovascular response. International Journal of Psychophysiology, 86, 123-135.
doi:10.1016/j.ijpsycho.2012.05.003
Gendolla, G. H. E. (2015). Implicit affect primes effort: Basic processes, moderators, and boundary conditions. Social and Personality Psychology Compass, 1–14.
doi:10.1111/spc3.12208
Gendolla, G. H. E., & Silvestrini, N. (2011). Smiles make it easier and so do frowns: Masked affective stimuli influence mental effort. Emotion, 11, 320–328. doi:10.1037/a0022593 Gendolla, G. H. E., Wright, R. A., & Richter, M. (2012). Effort intensity: Some insights from the
cardiovascular system. In R. M. Ryan (Ed.), The Oxford handbook of human motivation (pp. 420–440). New York, NY: Oxford University Press.
Kirita, T., & Endo, M. (1995). Between face advantage in recognizing facial expressions. Acta Psychologica, 89, 149–163. doi:10.1016/0001-6918(94)00021-8
Klauer, K. C., & Musch, J. (2003). Affective priming: Findings and theories. In J. Musch & K. C.
Klauer (Eds.), The psychology of evaluation: Affective processes in cognition and emotion (pp. 7–49). Mahwah, N.J.: Lawrence Erlbaum Associates.
Kreibig, S. D., Gendolla, G. H. E., & Scherer, K. R. (2012). Goal relevance and goal
conduciveness appraisals lead to differential autonomic reactivity in emotional responding
to performance feedback. Biological Psychology, 91, 365-375.
doi:10.1016/j.biopsycho.2012.08.007
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. doi:10.1080/02699930903485076
Lasauskaite, R., Gendolla, G. H. E., & Silvestrini, N. (2013). Do sadness-primes make me work harder because they make me sad? Cognition & Emotion, 27, 158-165.
doi:10.1080/02699931.2012.689756
Lasauskaite Schüpbach, R., Gendolla, G.H.E., & Silvestrini, N. (2014). Contrasting the effects of suboptimally versus optimally presented affect primes on effort-related cardiac response.
Motivation and Emotion, 38, 748-758. doi:10.1007/s11031-014-9438-x
Leppänen, J. M., & Hietanen, J. K. (2003). Positive facial expressions are recognized faster than negative facial expressions, but why? Psychological Research, 69, 22–29.
doi:10.1007/s00426-003-0157-2
Leppänen, J. M., Tenhunen, M., & Hietanen, J. K. (2003). Faster choice-reaction times to positive than to negative facial expressions: The role of cognitive and motor processes.
Journal of Psychophysiology, 17, 113–123. doi:10.1027//0269-8803.17.3.113 Lundqvist, D., & Litton, J. E. (1998). The Averaged Karolinska Directed Emotional Faces -
AKDEF. CD ROM from Department of Clinical Neuroscience, Psychology Section, Karolinska Institutet.
McShane, B. B., & Bockenholt, U. (2014). You cannot step into the same river twice: When power analyses are optimistic. Perspectives on Psychological Science, 9, 612-625.
doi:10.1177/174569161454851
Murphy, S. T., & Zajonc, R. B. (1993). Affect, cognition, and awareness: Affective priming with
optimal and suboptimal stimulus exposures. Journal of Personality and Social Psychology, 64, 723–739. doi:10.1037/0022-3514.64.5.723
Niedenthal, P. M. (2008). Emotion concepts. In M. Lewis & J. M. Haviland-Jones (Eds.), Handbook of emotions (3rd ed., pp. 587–600). New York, NY: Guilford Press.
Neumann, R., & Lozo, L. (2012). Priming the activation of fear and disgust: Evidence for semantic priming. Emotion, 12, 223-228. doi:10.1037/a0026500
Nosek, B. A., Hawkins, C. B., & Frazier, R. S. (2011). Implicit social cognition: from measures to mechanisms. Trends in Cognitive Sciences, 15, 152–159.
doi:10.1016/j.tics.2011.01.005
Robinson, M. D., & Clore, G. L. (2002). Belief and feeling: Evidence for an accessibility model of emotional self-report. Psychological Bulletin, 128, 934-960. doi:10.1037/0033- 2909.128.6.934
Rosenthal, R., Rosnow, R., L. (1985). Contrast analysis: Focused comparisons in the analysis of variance. New York: Cambridge University Press.
Rohr, M., Degner, J., & Wentura, D. (2012). Masked emotional priming beyond global valence activations. Cognition & Emotion, 26, 224-244. doi:10.1080/02699931.2011.57685 Silvestrini, N., & Gendolla, G. H. E. (2011a). Do not prime too much: Prime frequency effects of
masked affective stimuli on effort-related cardiovascular response. Biological Psychology, 87, 195–199. doi:10.1016/j.biopsycho.2011.01.006
Silvestrini, N., & Gendolla, G. H. E. (2011b). Masked affective stimuli moderate task difficulty effects on effort-related cardiovascular response. Psychophysiology, 48, 1157–1164.
doi:10.1111/j.1469-8986.2011.01181.x
Simon, J. R. (1969): Reaction toward the source of stimulation. Journal of Experimental Psychology, 81, 174-176. doi:10.1037/h0027448
Simmons, J. P., Nelson, L. D., & Simonsohn, U. (2011). False-positive psychology: Undisclosed flexibility in data collection and analysis allows presenting anything as significant.
Psychological Science, 22, 1359-1366. doi:10.1177/0956797611417632
Unkelbach, C., Fiedler, K., Bayer, M., Stegmüller, M., & Danner, D. (2008). Why positive information is processed faster: The density hypothesis. Journal of Personality and Social Psychology, 95, 36–49. doi:10.1037/0022-3514.95.1.36
Wentura, D., & Degner, J. (2010). A practical guide to sequential priming and related tasks. In B.
Gawronski & B. K. Payne (Eds.), Handbook of implicit social cognition (pp. 95–116).
New York, NY: Guilford.
Wentura, D., & Rothermund, K. (2014). Priming is not priming is not priming. Social Cognition, 32, 47-67. doi:10.1521/soco.2014.32.supp.47
Wilkinson, L., & The Task Force on Statistical Inference (1999). Statistical methods in
psychology journals: Guidelines and explanations. American Psychologist, 54, 594-604.
doi:10.1037/0003-066X.54.8.594
Zajonc, R. B. (1980). Feeling and thinking: Preferences need no inferences. American Psychologist, 35, 151–175. doi:10.1037/0003-066X.35.2.151
Notes
1 French words that were originally used as ease-related prime words: léger, relax, congé, evident; difficulty-related prime words: dur, examen, effort, épreuve; positive target words: gaité, heureux, joyeux, agréable; negative target words: pénible, accablé, abattu, triste.
2 Given that we had directed hypotheses, the p-values of the paired t-tests are one-tailed.
3 French words that were originally used as ease-related prime words: aisé, evident, relax, simple; difficulty-related prime words: ardu, complexe, compliqué, éprouvant.
Author Note
Guido H. E. Gendolla, Ruta Lasauskaite, Mylène Bolmont, and Laure Freydefont, University of Geneva, Switzerland.
This research was supported by research grants from the Swiss National Science Foundation (SNF 100014-131760, 100014-140251) awarded to Guido H. E. Gendolla.
Correspondence concerning this article should be addressed to Guido H. E. Gendolla, Geneva Motivation Lab, University of Geneva, FPSE, Department of Psychology, 40, Bd. du Pont d’Arve, CH-1211 Geneva 4, Switzerland. E-mail: [email protected]
Figure Captions
Figure 1. Reaction times in milliseconds in the congruent (ease/happiness and difficulty/sadness), incongruent (ease/sadness and difficulty/happiness), and neutral-prime conditions in Experiment 1. Error bars indicate standard errors of the means.
Figure 2. Reaction times in milliseconds in the congruent (ease/happiness and difficulty/sadness), incongruent (ease/sadness and difficulty/happiness), and neutral-prime conditions in Experiment 2. Error bars indicate standard errors of the means.
Figure 3. Reaction times in milliseconds in the congruent (ease/happiness and difficulty/sadness), incongruent (ease/sadness and difficulty/happiness), and neutral-prime conditions in Experiment 3. Error bars indicate standard errors of the means.
Figure 1
Figure 2
Figure 3
