The Fast and the Furious: Does Affect Modulate Temporal
Switching of Attention?
Floortje Bouwkamp
Student number: 1140654 Supervisor: Dr. Hans Phaf Date: 22-01-2016 Word count: 5705 Abstract word count: 135
2 Abstract
To investigate whether affect modulates temporal switching or spatial breadth of attention performance was measured on a visual search task. Positive and negative affect were induced for forty-eight participants with autobiographical recall and priming with emotional faces. Participants were more negative after the negative induction relative to after the positive induction. The slopes of the search curves were in the direction the temporal switching hypothesis predicted, but both a pop-out effect and an order effect occluded results. When comparing only
well-practised participants without a pop-out strategy, however, the difference between the slopes was significant with a considerable effect size. More eccentric
targets were harder to detect, but mood did not influence this effect. These results support the notion that positive affect facilitates serial search via
3 The Fast and the Furious: Does Affect Modulate Temporal Switching of
Attention?
In 1993 Rauscher, Shaw and Ky found that the ability of abstract spatial reasoning of students increased after listening to Mozart for only ten minutes. This groundbreaking ‘Mozart effect’ unleashed a trend of early childhood
exposure to classical music that is still seen today. Parents filled with the desire to push their child’s development purchase ‘Baby Einstein’ products to enhance the cognitive abilities of their little one. But research showed that the Mozart effect depended on whether you preferred to listen to it (Nantais & Shellenberg, 1999) and that it was an artifact of arousal and mood (Thompson & Schellenberg, 2001). Any type of music can increase positive affect (because you like it) and increase performance (Lesiuk, 2005). This research may have debunked the Mozart effect, however, it did show the power of positive affect.
There is ample evidence for the influences positive affect has on our cognitive abilities (Ashby & Isen, 1999, Isen, 1999). It increases creative thinking (Isen, Daubman & Nowicki, 1987), enhances problem solving and decision making (Isen, 2001) and facilitates memory (Isen, Shalker, Clark & Karp, 1978). In spite of all this evidence, there are few theories on how positive affect exerts its influence on cognition, one the Broaden-and-Build theory by Fredrickson. This theory argues that positive emotions broaden your action repertories and increase cognitive flexibility (Fredrickson, 2004).
Attention is a vital aspect of cognition, acting like a filter often described as a ‘spotlight’: what falls into its scope is processed further leaving the rest to be ignored. This spotlight can also be directed towards stimuli that need processing. Affect modulates attention as well (Phelps & Carrasco, 2006), but it remains
4 unclear what the underlying mechanism is. The aforementioned
Broaden-and-Build theory by Fredrickson incorporates affective modulation of both temporal and spatial attention. The narrowing of the scope of attention by negative affect (Easterbrook, 1959) can be seen as complementary to the Broaden-and-Build framework. This narrowing effect is illustrated by the well-established ‘weapon focus’ phenomenon (Loftus, Loftus & Messo, 1987, Steblay, 1992). In contrast to negative affect, positive affect is thought to widen the attentional scope. This is supported by the finding that positive affect promotes global processing (Gasper & Clore, 2002) and increases interference of surrounding stimuli (Rowe, Hirsch & Anderson, 2007). In the latter experiment participants completed an Erikson Flanker Task after a positive, negative or neutral mood induction procedure. A central target was surrounded by flankers that where either compatible or incompatible and were placed at a near, medium or far distance. When
compatible and incompatible flankers were compared, the flankers after positive induction had a higher impact on performance than after a neutral and sad
induction. Moreover, the most eccentric flankers had the greatest impact after the positive induction relative to the sad and neutral induction. Participants
completed a Remote Associates Test (RAT) as well. A word triad was offered and participants had to answer which fourth word connected these words. It appeared that positive affect increased performance on this task. Taking the word ‘remote’ literally they argued that spatial and conceptual representations are very much alike. The widening of the attentional scope would therefore enable the ‘reach’ for remote associates.
However, an increase in interference of flankers with positive affect is not always replicated (Huntsinger, 2012, Bruyneel et al., 2013). Phaf (2015) even found that when the flankers were masked the interference reversed as a
5 function of mood. This suggests a temporal instead of a spatial account, because the masking limits the duration of the flanker. This is supported by findings within the attentional blink paradigm which is a temporally based phenomenon: when a stimulus at a certain point in time (T1) is attended, a stimulus presented
successively, but with a certain lag (T2), is often missed. Smith, Most, Newsome and Zald (2006) found that stimuli at T1 that were aversely conditioned, causing negative affect, induced an attentional blink. Additionally, positive affect has been found to reduce the attentional blink (Olivers & Nieuwenhuis, 2006), presumably because switching to the second target was facilitated.
An explanation for a temporal effect is an affective modulation of switching of attention (Phaf, 2015). This hypothesis was derived from evolutionary
simulations in which, unexpectedly, neuronal competitive networks began to oscillate. These oscillations appeared to differ in frequency depending on whether an agent encountered food or a predator. It was postulated that high frequency oscillations were related to fast switching and positive affect (finding food) and low frequency oscillations were related to slow switching and negative affect (encountering a predator). This makes sense from an evolutionary point of view. If you encounter a threat in the form of a predator it is best to stay engaged and not become distracted. But when you are foraging for food you should not become fixated on one target, but be able to switch to a suddenly appearing predator (“It is better to miss dinner than to be dinner”). This is supported by research of Dreisbach and Goschke (2004) who found that positive affect promotes more flexible, but also more distractible behavior.
Not only does temporal switching make sense in an evolutionary way, it also clearly supports the notion of positive affect increasing cognitive flexibility. Moreover, it can explain many findings on affective modulation of attention that
6 were thought to be spatial. To begin with the weapon focus effect, which can be explained by a slower switching of attention due to negative affect. It is not so much the narrowing of attentional focus, but the impairment of disengaging attention from stimuli causing negative arousal. This is reflected in the finding that people suffering from depression appear to have trouble disengaging from a negative word while reading (Phaf & Khan, 2007). Additionally, Compton (2000) found that the ability to disengage in an orientation task predicted negative affect. This idea of disengagement can be transferred to the aforementioned attentional blink: one needs to disengage from T1 to attend to T2.
Recent findings on affective modulation of global versus local processing can also be more readily explained by cognitive flexibility and temporal switching of attention. Positive affect has been found to increase the ability to overcome global precedence and switch to local processing if the task demands it
(Baumann & Kuhl, 2005). Actually, the effect depends on your personal bias as positive affect will facilitate switching to your less preferred way of processing information (Heerebout, Todorovic, Smedinga & Phaf, 2013; Tan, Watson & Jones, 2009) whether this be global or local.
Temporal switching explains the increase in performance on a RAT (Rowe, Hirsch & Anderson, 2007) with the idea that positive affect increases switching of attention, allowing a higher activation build up of and better access to these words. This seems to be a more straightforward explanation than regarding conceptual representations as spatial and thus influenced by modulation of the attentional scope.
Positive affect decreased performance on an Erikson flanker task (Rowe, Hirsch & Anderson, 2007). This can be explained by time spent at flankers
7 outside the attentional scope. If positive affect increases switching of attention, the amount of time spent at the flankers will increase relative to the time spent at the target (there are more flankers than targets) which leads to more interference and thus to a decrease in performance. Furthermore, if this task makes use of different stimulus onset asynchronies (SOA’s) the temporal switching hypothesis predicts that faster disengagement due to a short SOA will result in a shorter time spent at flankers and thus to less interference of flankers with positive affect. This reversal of the effect of positive affect as a function of SOA is exactly what Phaf (2015) found. Based on these explanations, the suggestion could be made that temporal switching is the basic mechanism underlying the increase in cognitive flexibility due to positive affect, perhaps renders the spatial broadening aspect of affective modulation redundant.
The experimental methods of investigating attentional modulation of positive affect are, however, biased to the spatial broadening hypothesis, as they mainly investigate the spatial breadth of attention. Moreover, a dissociation of the two hypotheses would allow a stronger inference, which was not the case with the flanker task. How the two hypotheses compare when they are investigated in a task emphasizing the switching of attention, like a visual search task, is of our current interest. When searching an array for a target that shares properties with surrounding distractors, called a conjunction search, items are searched one by one (i.e., sequentially) until the target is found. For trials containing the target, this yields a linear relationship when plotting reaction time against set size (Treismann & Gelade, 1980). This task investigates a temporally based serial search process and has a spatial layout, making it very well suited to test both the spatial
broadening and temporal switching hypothesis. The array can be seen as a field of distractors that interfere with finding a target (spatial hypothesis) or as a field
8 of competing items wherein switching between items takes place until the target is found (temporal hypothesis). Both views explain an increase in reaction time with a bigger set size but they differ in the expectation of how affective
modulation influences this visual search process.
In the current set-up positive and negative affect were induced by asking participants to write about a personal experience that made them feel either happy or angry. This autobiographical recall method seems successful (Jallais & Gilet, 2010) and was also used by Phaf (2015). Additionally, Phaf (2015) used music as a secondary mood induction procedure (MIP), which Rowe et al. (2007) used as their primary MIP. However, in both experiments the negative condition did not differentiate from neutral condition sufficiently. Perhaps music did not bring about a strong enough change in mood when it comes to sadness which was the targeted emotion in the negative condition. We replaced sadness with anger and instead of listening to music we refreshed the induced moods during the visual search task by priming with happy and angry faces. Priming with emotional faces was used successfully as a primary MIP in the experiment
investigating affective modulation of attentional switching in a global-local task by Heerebout, Todorovic, Smedinga and Phaf (2013). Even subliminally, priming can effect behavioral bias induced by affect for up to 24 hours (Sweeny, Suzuki & Paller, 2009). Moreover, using two MIP's is thought to yield a stronger result as long as one method is more in the background (Gilet, 2008). Mood was checked before the autobiographical recall and after the task in each condition with the self assessment manikin (SAM, Bradley & Lang, 1994).
The two competing hypotheses lead to opposing expectations (Figure 1) on affective modulation on two aspects: the shape and the steepness (slope) of the relationship between set size and reaction time. Regarding the shape the
9 temporal switching hypothesis predicts a linear relationship: more items means more switches which makes you slower in finding the target. The spatial
hypothesis predicts a linear relationship as well, but one that flattens when the spatial extent of the array exceeds the scope of attention, and this flattening will be stronger with negative than positive affect because the scope of attention is expected to be smaller. Regarding the slopes, the temporal switching hypothesis predicts positive affect will increase switching and negative affect will decrease switching. The difference in search curve will increase with set size because with more switches an increase in switching will be more beneficiary, resulting in a steeper slope for negative than positive affect. The spatial broadening hypothesis, however, argues that positive affect will broaden attentional scope which will lead to more interference from distractors, whereas negative affect will narrow
attentional scope and lead to less interference from distractors. The difference will, again, increase with set size because there are more interfering items within or outside of the attentional scope. This would result in a steeper slope for positive affect than for negative affect.
Figure 1. Expected plots of reaction time for different set sizes for positive (green line) and negative affect (red line) under the temporal switching and spatial broadening hypothesis. For reference the dotted line represents the expected line with neutral affect.
RT setsize negative affect positive affect (neutral) 4 10 18 30
Temporal Switching Hypothesis
RT setsize negative affect positive affect (neutral) 4 10 18 30
10 To further strengthen the efforts to contrast the temporal switching and
spatial broadening hypotheses expectations on eccentricity are compared as well. Though often ignored, eccentricity is a very robust effect. Targets in the periphery are harder to detect than targets near fixation (Wolfe, 1998). This might be due to the fact that the receptive fields in the periphery are much larger than in the fovea, but Wolfe (1998) argues that eccentricity also reflects an attentional field which would make the eccentricity effect prone to the influence of affect.
Meinecke and Donk (2002) found that within a visual search task this eccentricity effect differs per set size. When eccentricity ranged from 0 to 12 visual degrees, the relationship between hit rate and eccentricity was linear at the medium set sizes (below 7 and at 49 elements the relationship was shaped differently). If we analyze only arrays with the largest set size (30 items), with items placed on 4 concentric circles (Figure 2) we can contrast predictions on mean reaction time per circle. The circles function as a degree of eccentricity as distance from the centre increases.
Figure 2. Example of an array of set size 30 with items on 4 concentric circles. The Black circle is the target and is presented at the second target ring.
11 Temporal switching predicts faster switching with positive affect and
slower switching with negative affect which will respectively facilitate or slow down performance in finding targets. This effect will be strongest with targets that are more eccentric because bigger switches are needed for more eccentric
targets. This would result in a steeper slope for negative affect than for positive affect. Spatial broadening, however, leads to the opposite expectation. If positive affect widens the attentional scope a more eccentric target will be found more easily, but will also lead to more interference from distractors. A smaller attentional scope due to negative affect will make it harder to find a more eccentric target but easier to identify once found because there is less
interference from surrounding distractors. Positive affect would then decrease performance on all targets, but dampen the eccentricity effect, while negative affect will increase performance and strengthen the eccentricity effect. These predictions are illustrated in Figure 3.
Figure 3. Expected plots of reaction time for different degrees of eccentricity (target ring) for positive affect (green line) and negative affect (red line) under the temporal switching and spatial broadening hypothesis. For reference the dotted line represents the expected line with neutral affect. Note: the lines of the spatial broadening hypothesis (right) might also cross at some point.
RT target eccentricity negative affect positive affect (neutral) 1 2 3 4
Temporal Switching Hypothesis Spatial Broadening Hypothesis
RT target eccentricity negative affect positive affect (neutral) 1 2 3 4
12 Method
Participants
Forty-nine healthy participants (age 18-36 years, M=23.79, SD=4.93) with normal of corrected-to-normal vision participated of which 31 volunteered and 17
received course credit in return. All spoke Dutch fluently and signed informed consent before participating.
Design
The visual search task had a 2x4 within-subjects factorial design. Mood (happy or angry) was the first independent variable and the order of induction was
counterbalanced across participants. The second independent variable was set size (4, 10, 18 or 30 items) and trials with different set sizes were presented in a pseudorandom order. All levels were combined creating 8 conditions. Reaction time on target trials with a correct response was the dependent variable. Outliers trials were removed per condition and per participant with the interquartile range method. Mean reaction time per participant and per condition were also checked with the interquartile range method. Error rates on both target and non-target trials were registered and participants with a rate higher than 10% were considered unmotivated and their data was excluded.
Materials and apparatus
Visual Search Task. Participants performed the Visual Search Task (VST) on a computer connected to a 23-inch Asus VG236H monitor with a 1920x1080 resolution and a 120 Hz refresh rate. Responses for seeing a target were
recorded with a button-box on the right hand side (placed on the left hand side for left handed people). The items consisted of black or white circles 60 px in
13 diameter (1,325 visual degrees) and black or white squares that were matched in surface area to the circles. The target was determined pseudorandomly per participant. Only distractors of the same color or with same shape were used. By doing so we bias our experiment towards the spatial broadening hypothesis because these distractors would lead to more interference than the ‘neither‘ distractors (different color and shape) would. Research done by Kim and Cave (1995) shows that spatial attention is feature driven. Distractors with the same color attract the most attention followed by distractors with the same shape. This set-up will strengthen the support for the temporal switching hypothesis if results are not as the spatial broadening hypothesis predicts.
Four set sizes were used: 4, 10, 18 and 30 items. All items were evenly distributed on concentric circles and each increase in set size meant a circle was added. Starting with 4 items on a circle of 80 px ø (1,77 visual degrees) in
diameter and each added circle had a diameter increased with 120 px (2,65 visual degrees). Participants went through 260 trials per condition, including 20 practise trials. Because the effect of mood was expected to be stronger with the bigger set sizes, number of trials were increases as well. Starting with 12 trials (12 blank trials and 12 target trials) for set size 4, 24 trials for set size 10, 36 trials for set size 18 and 48 trials for set size 30. The circle on which the target appeared was selected pseudorandomly, except for set size 30. Because we wanted a minimum of 12 trials per circle to analyze eccentricity the 48 trials of set size 30 were counterbalanced over the 4 circles, but the order in which they were offered was pseudorandom.
Autobiographical recall. To induce positive or negative affect participants were asked, after reading an exemplary story (see Appendix), to recall a personal experience that had made them either happy or angry and write them down.
14 Karolinska set. The Karolinska Directed Emotional Faces database (KDEF) designed by Lundqvist, Flykt and Öhman (1998) contains images of 70
individuals expressing 7 different emotional expressions. This database is regarded as a valid set of affective facial pictures (Goeleven, De Raedt, Leyman, & Verschuere, 2008). These trained amateur actors were between 20 and 30 years of age and did not have a beard or moustache, wear earrings or eyeglasses nor used visible make-up. Only the faces in a frontal position from the happy and angry set were used.
SAM. To measure mood the Self Assessment Manikin (SAM) was used. This non-verbal assessment technique makes use of graphic depictions of three dimensions of mood on a 5-point likert scale: valence, intensity and dominance (Bradley & Lang, 1994). Only the valence (positive/negative) and intensity (aroused/calm) dimensions were used. Participant had to indicate which picture fitted their current mood the most.
Procedure
Participants were explained that they were participating in an experiment that investigated the influence of happy and angry moods on attention. After reading the information brochure, getting verbal instructions and signing informed consent, participants were seated behind the desk with a chinrest installed at a distance of 60 cm to the screen. Participants started with indicating their baseline mood on the SAM, after which they were given at least 20 trials to practise. After affirming the task was clear to them, participants were instructed to read the exemplary story on the screen and asked to think of a personal experience that made them feel happy or angry, depending on the current condition, and write them down. Participants were assured these stories would never be read and they could take them home afterwards. They were also encouraged to feel the targeted
15 emotion and get into a happy or angry mood. When the story was finished the
experimental task commenced. Each trial (see figure 4) started with a fixation cross presented for 1 second, followed by a brief presentation of a happy or angry face (12 ms). After the prime the visual search array was presented for 2500 ms (blank trials) or until a response (target trials), followed by a screen with the instruction to press the button when participants wished to continue. After completing 240 trials participants completed another SAM indication of mood. They the had a short break (5-10 minutes) after which the second block and the remaining mood induction began following the same procedure as the first block. At the end of the experiment participants were asked for general impressions, whether they had seen the faces and their emotions, if they had used specific strategies, if they thought the mood inductions were successful and how they felt currently. If participants felt too negative the experimenter made efforts to
alleviate their mood.
Figure 4. Timeline of a trial with a negative affect prime (angry face) and a visual search array of 30 items. Fixation cross 1 sec Prime 12 ms Array 2500 ms / until response Response druk op de knop om verder te gaan
16 Data analysis
The effect of the mood induction was checked with a dependent t-test on the SAM scores before and after the task per induction type (positive and negative affect). For the main analysis a general slope was calculated from the different reaction times per set size using the least-squares method. These slopes were analyzed with a repeated measures ANOVA with mood induction as independent and slope as dependent variable. The same procedure was followed for the eccentricity effect, but now the reaction times per circle or degree of eccentricity were transformed into slopes.
Results
All participants met the requirement of a < 10% error rate. When using the
interquartile-range method on average reaction time per condition one participant had a very slow reaction time both conditions. Combined with the results from the exit-interview this participant was regarded as unmotivated and was excluded from further analysis. This resulted in a sample size of N=47 (male=14, female=33), with no difference in proportion male/female between the two groups with a different order of induction (χ2(1)=0.30, p=0.412). The
interquartile-range method was used iteratively to exclude outlier trials per participant and per condition. On average, this led to exclusion of 7% of the trials.
The scores on the Self-Assessment Manikin (SAM) (Table 1) revealed that participants became more negative after the negative induction (t(46)= -8.11, p<0.001, r=0.35) but stayed approximately the same after the positive induction (t(46)= -0.52, p= 0.607, r=0.08). Participants also became more aroused after the negative induction (t(46)= 2.55, p= 0.014, r= 0.35). Participants were more negative after the negative induction (m=2.77, SD=0.81) then they were after the
17 positive induction (m=2.11, SD=0.79) so the induction did create a difference in
mood (t(46)=-5.55, p<0.001, r=0.63). All valence scores (both baseline and post induction) were biased to the positive side and all valence scores were biased to the calmer side (see also Figure 1).
Table 1. Mean score on the valence and arousal scale of the Self-Assessment Manikin (SD), per induction and per order of induction. A higher score on the valence dimension, scored 1-5, represents a more negative mood and a higher score on the arousal
dimension, scored 1-5, represents a calmer state.
SAM Valence SAM arousal before after before after Positive induction N/P 2.09 (0.73) 2.13 (0.82) 4.00 (0.80) 3.83 (0.94) P/N 2.00 (0.66) 2.08 (0.78) 3.75 (0.85) 3.75 (0.95) Total 2.04 (0.69) 2.11 (0.79) 3.89 (0.81) 3.81 (0.92) Negative induction N/P 2.00 (0.67) 2.83 (0.72) 3.78 (0.90) 3.61 (0.99) P/N 1.88 (0.61) 2.71 (0.91) 4.04 (0.86) 3.54 (0.88) Total 1.94 (0.65) 2.77 (0.81) 3.91 (0.88) 3.57 (0.93)
18
Figure 1. Mean scores on both the valence (A) and arousal dimensions (B) of the Self-Assessment Manikin (SAM) with standard errors (error bars), per type of induction.
On average participants were faster (F(1,18)=77.32, p<.001, r=0.90) in the second block (m=457.09, SD=53.43) compared to the first block (m=506.86, SD=63.62). This order effect was present 97.6% of the participants (n=40). As expected within a serial search process, reaction time (Table 2) monotonically increased as the set size became larger (Huyn-Feldt<0.7, Box’s test p>0.05, multivariate test with Pilai’s Trace F(3,43)=28.54, p<0.001) and this relationship was linear (polynomial contrast F(1)=79.42, p<0.001). Due to the
aforementioned order effect, the effect of the inductions on reaction time
depended on whether it was offered in the first or second block (F(1,45)=111.28, p<0.001). While the search curve (Figure 2) of the negative induction had a steeper slope (M=2.33, SD=1.80) than the search curve of the positive induction (M=2.02, SD=1.84), this difference was not statistically significant (F(1,45)=2.21, p=0.144, r=0.21), although, this effect depended on the order of the inductions (F(1,45)=10.06, p=0.003). A flattening of both slopes is visible as set size increases. For the positive induction this flattening starts at a smaller set size compared to the negative induction. See Table 3 for all slope values.
19
Table 2. Mean reaction time per set size (SD) per induction and per order of induction (N=negative, P=positive). set size 4 10 18 30 Positive N/P 425.36 (48.61) 437.75 (43.03) 446.57 (43.10) 463.31 (59.92) induction P/N 468.78 (60.94) 510.74 (77.98) 520.29 (81.88) 544.03 (90.63) Total 447.53 (58.89) 475.02 (72.10) 484.21 (75.02) 504.52 (86.53) Negative N/P 454.92 (50.77) 474.62 (53.14) 501.22 (55.77) 517.91 (69.22) induction P/N 428.23 (55.67) 449.30 (56.11) 471.71 (72.68) 487.31 (73.25) Total 441.29 (54.46) 461.69 (55.57) 486.15 (65.96) 502.28 (72.21)
Figure 2. Mean reaction time and standard error (error bar) per set size and per induction for all participants
20
Table 3. Mean slope (SD) of the reaction time plots, calculated with the least squares method per induction and per order of induction, for all participants.
Slopes (RT)
Positive induction Negative induction Negative/Positive 1.41 (1.24) 2.43 (1.93) Positive/Negative 2.61 (2.13) 2.24 (1.71) Total 2.02 (1.84) 2.33 (1.80)
Unexpectedly, some participants had a negative slope value, which is a sign of a (non-serial) feature search or ‘pop-out’ effect (Treismann & Gelade, 1980). Because the hypotheses only apply to serial search, the slopes were re-analysed after removing the participants displaying this pop-out strategy, as indicated by a negative slope in one of the conditions (n=7, 8 blocks with slope<0, 5 after negative and 3 after positive induction, 4 in each block). After exclusion, both the negative (M=2.72, SD=1.59) and the positive slope (M=2.33, SD 1.76) increased, moreover, the difference between these slopes increased as well (F(1,38)=3.75, p=0.06, r=0.30). When analysing the reaction times upon which these slopes are based, the interaction between mood and set size became significant (F(3,114)=2.83, p=.039), suggesting that a positive mood accelerates the serial search process.
Due to the order effect, the slopes were compared separately within the first block (Figure 3A) and within the second block (Figure 3B). Within the first block the difference between the positive slope (m=2.91, SD=2.03) and the negative slope (m=2.87, SD=1.73) disappeared (t(38)=0.07, p=0.945, r=0.01).
21 In contrast, within the second block the difference between the positive slope
(m=1.69, SD=1.17) and the negative slope (m=2.60, SD=1.49) increased considerably (t(38)=-2.13, p=0.04, r=0.33). In other words, when participants were well-practised, the difference between the slopes due to the induction seemed to be enhanced.
Table 4. Mean slopes (SD) and the difference between the slopes (SE) after exclusion of participants with a negative slope, per block and for both blocks.
Slopes
Positive slope Negative slope Difference Both blocks 2.33 (1.76) 2.72 (1.60) 0.39 (0.25) First block 2.91 (2.03) 2.87 (1.73) 0.04 (0.60) Second block 1.69 (1.17) 2.60 (1.49) 0.91 (0.43)
Figure 3. Mean reaction time plotted against set size per induction (error bars represent Standard Errors) for the first block (A) and second block (B) after exclusion of participants with a negative slope.
22 All trials of set size 30 were analysed separately to see if there was an
eccentricity effect and whether the inductions influenced this effect. Figure 4 shows mean reaction time plotted against eccentricity (target ring). See Table 5 for corresponding values. Reaction time monotonically increased with eccentricity (Huyn-Feldt<0.7, Box’s test p>0.05, multivariate test with Pilai’s Trace
F(3,45)=32.50, p<0.001) and as expected this relationship was linear
(polynomial contrast, F(1,45)=73.05, p<0.001). There was no statistical evidence for an effect of mood on eccentricity (F(1,45)=0.561, p=0.644), however, the effect of mood was dependent on whether the induction was offered in the first or second block (F(1,45)=89.27, p<0.001).
Table 5. Mean reaction time per target ring (SD), per induction and per order of induction (N=negative, P=positive) for all participants. A higher target ring number means a more eccentric target. target ring 1 2 3 4 Positive N/P 438.34 (58.77) 447.25 (54.97) 476.06 (65.80) 495.73 (76.07) induction P/N 508.29 (80.19) 526.88 (83.90) 554.41 (102.67) 595.59 (121.80) Total 474.06 (78.20) 487.91 (81.14) 516.07 (94.38) 546.72 (112.83) Negative N/P 488.75 (69.97) 494.95 (63.61) 537.39 (84.86) 559.94 (83.68) induction P/N 460.83 (63.63) 465.58 (62.82) 493.06 (80.12) 530.94 (110.06) Total 474.50 (67.57) 479.96 (64.25) 514.73 (84.60) 545.13 (98.08)
23
Figure 4. Mean reaction time with standard error (error bars) plotted against degree of eccentricity per induction for all participants.
The slopes of both induction types (see Table 6) did not differ
(F(1,45)<0.01 p=.994, r<0.01), and the effect of order of induction was no longer there (F(1,45)=0.24, p=0.629, r=0.07). Removing participants with a negative search slope (pop-out effect) did not change either the positive or the negative eccentricity slope, nor on the difference between those slopes (F(1,38)=0.06, p=0.812, r=0.04). This made sense considering a pop-out target is still under the influence of eccentricity (Meinecke & Donk, 2002).
24
Table 6. Mean slopes (SD) of the eccentricity plots calculated with the least squares method per induction and per order of induction of all participants
Slopes (eccentricity)
Positive induction Negative induction Negative/Positive 0.35 (0.35) 0.33 (0.32) Positive/Negative 0.30 (0.23) 0.32 (0.30) Total 0.32 (0.29) 0.33 (0.30)
The eccentricity slopes were analysed separately within the first and second blocks, but this time we left the negative search slopes included. Within the first block (Figure 5A) there was no difference between the slopes
(t(45)=0.51, p=0.612, r=0.08). More practise did dampen the slopes of eccentricity, but there was no difference between the slopes (t(45)=-0.55, p=0.588, r=0.08) within the second block (Figure 5B). See Table 7 for the corresponding numbers.
Figure 5. Mean reaction time plotted against degree of eccentricity per induction (error bars represent Standard Errors) for the first block (A) and second block (B) for all participants.
25
Table 7. Mean slopes (SD) of eccentricity and the difference between the slopes (SE) for both blocks, only the first block, and only the second block for all participants.
Slopes (eccentricity)
Positive slope Negative slope Difference Both blocks 0.32 (0.29) 0.33 (0.30) 0.00 (0.04) First block 0.38 (0.34) 0.34 (0.24) 0.04 (0.08) Second block 0.26 (0.22) 0.31 (0.36) 0.05 (0.09)
Discussion
Our findings provide more support for the temporal switching hypothesis than they do for the spatial broadening hypothesis. The flattening of the search curve after a positive mood induction relative to the search curve after the negative mood induction was as the temporal hypothesis predicted. However, there were several factors that may have limited the accuracy of the measurements, and this could have caused the non-significant difference between the slopes. Firstly, many participants experienced a pop-out effect which may have occluded the results. Secondly, there was a large order effect that lead to the conclusion that the data of the first block was less reliable. Thirdly, the mood induction procedure may not have been effective enough.
When excluding participants that displayed a non-serial search curve (negative slope) due to a pop-out effect, the difference between the positive and negative slope increased from r=0.21 to r=0.30, the latter being a medium effect
26 size. There were equal numbers of these pop-outs in the positive condition
compared to the negative condition, and this can also be seen as evidence against the spatial broadening hypothesis. A wider scope of attention due to positive affect should create more opportunity for a pop-out effect than a
narrower attentional scope. But the exclusion of the participants with a negative slope did not ensure the data was covering purely serial search. It is likely that other participants experienced the pop-out effect as well (this was also reported by more participants than those excluded) dampening their search curve even though a positive slope remained. It is therefore of importance that in future research precautions are taken to prevent grouping and specifically pop-out effects. This can be accomplished by making the visual search more difficult. In the current set-up, only two types of distractors were used, and increasing this number will make the task more complex and diminish the chance of a pop-out effect. Another option is to use shades of grey could for the stimuli instead of black & white. Another factor that was considerably different in our setup compared to that of others investigating an attentional field was is the size of stimuli and the array itself. Both Rowe, Hirsch and Anderson (2007) and Gasper and Clore (2002) used smaller arrays and stimuli. The bigger stimuli might have contributed to the the pop-out effect and a suggestion is to enlarge the difference between the size of the array and the stimuli by making the items smaller, for instance 8-10 mm like in the global-local task of Gasper and Clore (2002).
Sufficient practise appeared to be a precondition for the affective modulation of the search curve to occur, as there was a large order effect (r=0.90) that was seen across participants. The slower overall searches in the first block compared to the second block may have obscured the effect of mood. When comparing the slopes within the first block the difference was negligible
27 (r=0.01). In contrast, when comparing the slopes within the second block a clear
and significant difference between the slopes emerged with a medium effect size (r=0.32). Although these results are under the influence of individual differences, the plot from the second block closely resembled the expected plots for the temporal hypothesis. The order effect could have been prevented by starting with a larger (neutral) practise block so all participants would have been sufficiently familiarized with the task.
The scores on the Self-Assessment-Manikin (SAM) revealed a successful negative induction but an unsuccessful positive induction. Our positive induction was evidently not strong enough to yield an effect on subjectively
reported mood. This could explain our small overall effect, as this difference in mood reflected a neutral versus negative state, instead of a positive versus a negative state. To our surprise quite often even a more negative mood was reported after a positive induction. Perhaps this as a sign of fatigue, biasing the SAM scores to the negative side, which would have made the positive induction seem less strong and the negative induction stronger than it actually was. However, a more negative mood was still reported after the negative induction than after the positive induction.
There is a possibility that the positive induction was successful without being evident in the SAM scores. The second measurement was not taken directly after the induction, but after the entire first block. Experimentally induced
emotional states are short-lived in general and the manipulation check of a Mood Induction Procedure (MIP) is therefore more often (81% of research) carried out immediately after the manipulation (Gerrards-Hesse, Spies & Hesse, 1994). The positive induction may have worked, but it’s effect might not have lasted long enough to be measured while the effect of the negative induction did. The fact
28 that baseline mood was biased towards the positive side, a common finding
(Phaf, 2015), could account for this longer lasting effect of negative induction compared to the positive induction. It takes more effort, especially in an
experimental setting, to get in a more extreme state (very happy) than it takes to get in a mild state (slightly angry). This would leave the question whether priming with faces was an adequate procedure to maintain the effect on mood since it would only have worked in the negative condition. As music appears to be better suited for maintaining a positive mood (Rowe, Hirsch & Anderson, 2007, Phaf, 2015) perhaps these MIPS should be combined. It is, however, strongly advised to take a post induction measurement of mood in the future as well to gain a better understanding of the effectiveness of the Mood Induction Procedure.
The expectation that the affective modulation of the slopes would be the largest with the bigger set sizes because there are more switched to be made, was not evident in the results. There was even a slight flattening of both slopes between the higher set sizes, supporting the spatial broadening hypothesis on this account. However, the flattening of the positive slope started sooner than the negative slope, which contradicts the spatial broadening hypothesis that would predict less interference when the array exceeds the attentional field (bigger set size), and thus more flattening, and at smaller set sizes with a smaller attentional field due to negative affect. An alternative explanation is the use of a grouping strategy within the visual search task. Many participants (n=12) reported to cluster all items of the opposing colour and focus only on the items of the right colour. This strategy becomes more effective with larger set sizes, resulting in a dampening of the slope between those set sizes. But the differences in steepness of the slopes between set sizes should be interpreted with caution as the amount of trials increased as set size increased.
29 Eccentricity appeared not to be under the influence of mood and removing negative slopes did not change these results, nor did a separate analysis of the blocks. Though these results should be interpreted with caution because even less trials were used (12 trials per target ring) for these analyses, the most straightforward explanation for this is that the eccentricity effect is caused by a difference in size of receptive fields in the periphery, and not as Wolfe (1998) states, “a reflection of an attentional field”. Therefore, mood would have no influence on the eccentricity effect, as confirmed by the current findings.
While not conclusive, our results are more in line with the temporal switching hypothesis compared to the spatial broadening hypothesis. Moreover, the difference in steepness between the slopes of the search curves, both with and without negative slopes, did have considerable effect sizes. If measurement procedures would have been more precise, these effects are expected to become significant. These results are therefore a first step in the investigation of what mechanism underlies the affective modulation of attention within a visual search task. More general, the results add weight to the notion that positive affect modulates temporal switching, and not the breadth of attention.
Parents can safely return their Baby Einstein gadgets and turn of Mozart. To spawn creative offspring with a flexible mind set and improved memory, who are able to swiftly switch their attention making them good decision makers and problem solvers, parents ‘simply’ have to make their children happy.
30 References
Ashby, F. G., Isen, A. M., & Turken, U. (1999). A neuropsychological theory of positive affect and its influence on cognition. Psychological Review, 106, 529-550.
Baumann, N., & Kuhl, J. (2005). Positive affect and flexibility: Overcoming the precedence of global over local processing of visual information.
Motivation and Emotion, 29, 123-134.
Bradley, M. M., & Lang, P. J. (1994). Measuring emotion: the self-assessment manikin and the semantic differential. Journal of Behavior Therapy and
Experimental Psychiatry, 25, 49-59.
Bruyneel, L., van Steenbergen, H., Hommel, B., Band, G. P., De Raedt, R., &
Koster, E. H. (2013). Happy but still focused: Failures to find evidence for a mood-induced widening of visual attention. Psychological Research, 77, 320-332.
Compton, R. J. (2000). Ability to disengage attention predicts negative affect.
Cognition & Emotion, 14, 401-415.
Dreisbach, G., & Goschke, T. (2004). How positive affect modulates cognitive control: reduced perseveration at the cost of increased distractibility.
Journal of Experimental Psychology: Learning, Memory, and Cognition, 30,
343.
Easterbrook, J. A. (1959). The effect of emotion on cue utilization and the organization of behavior. Psychological Review, 66, 183.
Fredrickson, B. L. (2004). The broaden-and-build theory of positive emotions.
Philosophical transactions of the Royal Society B: Biological Sciences, 359, 1367-1377.
31 versus local processing of visual information. Psychological Science, 13,
34-40.
Gerrards-Hesse, A., Spies, K., Heese, W. (1994). Experimental inductions of emotional states and there effectiveness: A review. Britisch Journal of
Psychology, 85, 55-78.
Gilet, A. L. (2008). Mood induction procedures: a critical review. L'Encephale,
34, 233-239.
Goeleven, E., De Raedt, R., Leyman, L., & Verschuere, B. (2008). The Karolinska directed emotional faces: a validation study. Cognition and Emotion, 22, 1094-1118.
Heerebout, B. T., & Phaf, R. H. (2010). Good vibrations switch attention: An affective function for network oscillations in evolutionary simulations.
Cognitive, Affective, & Behavioral Neuroscience, 10, 217-229.
Heerebout, B. T., Todorović, A., Smedinga, H. E., & Phaf, R. H. (2013). Affective modulation of attentional switching. The American Journal of Psychology,
126, 197-211.
Huntsinger, J. R. (2012). Does positive affect broaden and negative affect narrow attentional scope? A new answer to an old question. Journal of
Experimental Psychology: General, 141, 595.
Isen, A. M. (1999). Positive affect. In T. Dagleish & M. Power (Eds.). The Handbook
of Cognition and Emotion. (pp 521-539). New York: Guilford press.
Isen, A. M. (2001). An influence of positive affect on decision making in complex situations: Theoretical issues with practical implications. Journal of
Consumer Psychology, 11, 75-85.
Isen, A. M., Daubman, K. A., & Nowicki, G. P. (1987). Positive affect facilitates creative problem solving. Journal of Personality and Social Psychology,
32
52, 1122.
Isen, A. M., Shalker, T. E., Clark, M., & Karp, L. (1978). Affect, accessibility of material in memory, and behavior: A cognitive loop?. Journal of Personality
and Social Psychology, 36, 1.
Jallais, C., & Gilet, A. L. (2010). Inducing changes in arousal and valence: Comparison of two mood induction procedures. Behavior Research
Methods, 42, 318-325.
Kim, M. S., & Cave, K. R. (1995). Spatial attention in visual search for features and feature conjunctions. Psychological Science, 6, 376-380.
Lesiuk, T. (2005). The effect of music listening on work performance. Psychology
of Music, 33, 173-191.
Loftus, E. F., Loftus G. R., & Messo, J. (1987). Some facts about "weapon focus."
Law and Human Behavior, 11, 55-62.
Lundqvist, D., Flykt, A., & Öhman, A. (1998). The Karolinska Directed Emotional Faces - KDEF, CD ROM from Department of Clinical Neuroscience, Psychology section, Karolinska Institutet, ISBN 91-630-7164-9. Meinecke, C., & Donk, M. (2002). Detection performance in pop-out tasks:
Non-monotonic changes with display size and eccentricity. Perception, 31, 591-602.
Nantais, K. M., & Schellenberg, E. G. (1999). The Mozart effect: An artifact of preference. Psychological Science, 10, 370-373.
Olivers, C. N., & Nieuwenhuis, S. (2006). The beneficial effects of additional task load, positive affect, and instruction on the attentional blink. Journal of
Experimental Psychology: Human Perception and Performance, 32, 364.
Phaf, R. H. (2015). Attention and positive affect: Temporal switching or spatial broadening? Attention, Perception, & Psychophysics, 77, 713-719.
33 Phaf, R. H., & Kan, K. J. (2007). The automaticity of emotional Stroop: A meta-
analysis. Journal of Behavior Therapy and Experimental Psychiatry, 38, 184-199.
Phelps, E. A., Ling, S., & Carrasco, M. (2006). Emotion facilitates perception and potentiates the perceptual benefits of attention. Psychological science, 17, 292-299.
Rowe, G., Hirsh, J. B., & Anderson, A. K. (2007). Positive affect increases the breadth of attentional selection. Proceedings of the National Academy of
Sciences, 104, 383-388.
Smith, S. D., Most, S. B., Newsome, L. A., & Zald, D. H. (2006). An emotion- induced attentional blink elicited by aversively conditioned stimuli.
Emotion, 6, 523-527.
Steblay, N. M. (1992). A meta-analytic review of the weapon focus effect. Law and
Human Behavior, 16, 413-424.
Sweeny, T. D., Grabowecky, M., Suzuki, S., & Paller, K. A. (2009). Long-lasting effects of subliminal affective priming from facial expressions.
Consciousness and cognition, 18, 929-938.
Tan, H. K., Jones, G. V., & Watson, D. G. (2009). Encouraging the perceptual underdog: Positive affective priming of nonpreferred local–global processes. Emotion, 9, 238-247.
Thompson, W. F., Schellenberg, E. G., & Husain, G. (2001). Arousal, mood, and the Mozart effect. Psychological Science, 12, 248-251.
Treisman, A. M., & Gelade, G. (1980). A feature-integration theory of attention.
Cognitive Psychology, 12, 97-136.
Wolfe, J. M. (1998). Visual search. In H. Paschler (Ed.), Attention (pp. 13-73). London, UK: University College London Press
Appendix
Exemplary story positive affect (Happy):
"Ik heb leren paardrijden op de manege vanaf mijn 8e en al snel begon ik met wedstrijden rijden. Op een gegeven moment was het veulen van de pony van mijn zus groot genoeg geworden om ingereden te worden: eindelijk mijn eigen pony! Ons eerste wedstrijdseizoen was een heel heftige tijd. Mijn pony was nog jong en erg pittig: ze bokte sprong alle kanten op als het haar niet beviel. Dat was vaak erg frustrerend, maar ik bleef volhouden en langzaam gingen we vooruit. Aan het einde van het seizoen waren de kringkampioenschappen. Het was een warme dag en we moesten 2 x een proef rijden. Door het warme weer was mijn pony rustiger dan normaal, de eerste proef ging al goed en de 2e zelfs beter. Ik mocht daardoor meedoen aan de finale: wat was ik blij verrast! Nu al trots op mezelf en mijn pony reed ik de finale proef. Ik vond het heel fijn gaan, maar ik had verder geen verwachtingen. Tijdens de prijsuitreiking reed ik door de grote ring
afwachtend op de plaatsingen die afgeroepen worden. Ze gaan dan van de laatste plaats naar de eerste. Maar mijn naam werd maar niet omgeroepen en ik raakte helemaal in de war toen ze bij nummer drie waren: ze waren me
vergeten… Maar dat was niet zo, want ik was kampioen geworden! Iedereen was zo blij voor me maar ik was het blijst van iedereen. Ik heb dagen lang niet kunnen stoppen met glimlachen. Ik mocht door naar de Nederlandse kampioenschappen en daar werd ik derde. Een ongelofelijke prestatie als je naar het start van het seizoen kijkt. Dit is een van de fijnste herinneringen die ik heb."
Exemplary story negative affect (Angry):
In het tweede jaar van psychologie had ik een vak, het gesprekspracticum. Hiervoor mocht je eigenlijk geen lessen missen om het vak te halen. Je moest wekelijks een volle dag (10:00 tot 18:00) en een halve dag aanwezig zijn en meedoen met de oefeningen. Het vak duurde twee maanden. We kwamen als gehele groep vaak te laat en daar was onze begeleider in het begin erg coulant over. Later werd dit minder, het te laat komen werd ook minder. Maar af en toe kwam er nog iemand te laat binnen, hiermee werd de grens van onze begeleider opgezocht. Toen ik op een gegeven moment - echt door overmacht – te laat kwam, werd ik direct het vak uitgezet. Terwijl iemand anders eerder een hele dag had gemist en mocht wel blijven! Voor mijn gevoel hadden wij de grens als
collectief bereikt. Onze begeleider wilde nu een statement maken naar de groep om het overwicht te bewaren en daarvan werd ik de dupe. En dit gebeurde een week voor het einde van het vak, de eindopdracht had ik al gedaan! Nu moest ik het vak opnieuw doen in het tweede semester en hierdoor doe ik een half jaar langer over mijn bachelor. Daarnaast vond ik het verschrikkelijk om het vak helemaal opnieuw te doen. Inhoudelijk was het voor de tweede keer erg saai en het was voor mij al irrelevant omdat ik met mijn brein & cognitie specialisatie nooit iets zal gaan doen met klinische gespreksvaardigheden. Het kostte alleen maar heel veel van mijn tijd en het voelde compleet zinloos. Ik ben er nog steeds boos over."
