Affective modulation of attention: Spatial broadening of
temporal switching?
Bachelor’s thesis psychology – brain and cognition 2015/2016
University of Amsterdam Supervised by R.H. Phaf & L. Mulder
Anne-Wil Kramer 10325999
May, 2016
Table of Contents Abstract 3 Introduction 4 Method 9 Results 12 Discussion 20
Abstract
Evolutionary reasoning suggests that affect influences our attentional system. This can also be applied to affectively neutral stimuli. The influence of positive affect on visual processing has been explained in terms of spatial broadening and temporal flexibility using a Flanker task paradigm. In this study, a visual search task was used to determine which of these two
hypotheses would be more consistent with the affective modulation of attention during visual processing of affectively neutral stimuli. Target detection was faster during positive moods when processing stimuli in small arrays, contrary to what would be expected according to a broadening account. It is argued here that these results lean more towards a flexibility account, although the obtained effects were small.
Introduction
Imagine a crowded pub when suddenly a fire breaks out. Would everyone direct their attention to the emergency exits and flee safely or would they all focus on the main exit thereby disallowing most people to get out? Unfortunately, the latter is what happened during a fire in a pub in 2001 in the Netherlands, resulting in 13 deaths and many injured (Alders, Belonje, van den Berg, ten Duis, & Hoelen, 2001). What was it that made these individuals not notice the emergency exits? What role did their emotions play in their attentional processes? From an evolutionary perspective, it has been proposed that one important function of emotion is to serve as a marker which determines towards which emotional stimuli we direct our attention (Compton, 2003; Damasio, Everitt, & Bishop, 1996). This hypothesis has also found empirical support (Öhman, Flykt, & Esteves, 2001). How does
emotion, on the other hand, influence our attentional processes to non-emotional stimuli, such as an exit? According to the broaden-and-build theory, positive affect broadens spatial attention and expands an individual’s action repertory in reaction to non-emotional stimuli (Frederickson, 2004). However, in the same article, Frederickson (2004) argues that positive affect also increases flexibility between action tendencies. The aim of this study is to
distinguish between spatial- and flexibility hypotheses, and determine which of the
hypotheses mentioned above is more consistent with data from the current study in order to investigate how we react to non-emotional stimuli under the influence of different emotions (positive versus negative).
Substantial empirical support has been found for the spatial hypothesis of affective modulation of attention. In this view, attention is often being compared to the beam of a spotlight (LaBerge, 1983). Physiological evidence suggests that attention does have a
measurable spatial scope by showing that attention directed to a specific target location in the visual field produced cortical enhancement in the visual cortex. The locations of these
enhancements corresponded with cortical representations of the attended target when presented in isolation (Brefczynski, DeYoe, 1999). Furthermore, Gasper and Clore (2002), found that positive affect fostered a broadened/global perspective whilst negative affect narrowed the spatial focus to a local perspective. Additionally, Rowe, Hirsch, and Anderson (2007) found that positive affect increased interference by incongruent flankers in a Flanker task, because attention would be broadened, which would cause more interference from distracters. In a Flanker task, several arrows are present, surrounding the target arrow in the middle. The surrounding arrows are called ‘flankers’ and can either be congruent or
incongruent with respect to the direction of the target arrow (Eriksen, 1996). Notable, however, similar results to the Flanker task were also obtained for a semantic
remote-associates task. Here, participants were asked to override typical semantic associations to find semantically distant or remote associations in terms of one word related to all three stimulus words. Participants performed better at this task when in a positive mood and this was explained as positive mood facilitating access to remote semantic associates. It seems questionable whether these semantic concepts can be regarded as spatially oriented and thus as support for a spatial hypothesis of visual attention. Moreover, for Rowe’s et al. (2007) explanation of their Flanker task results, alternatives have been proposed in terms of flexibility.
The flexibility hypothesis holds that positive affect increases cognitive flexibility, which causes more temporal switching between stimuli (Baumann, & Kuhl, 2005; Heerebout, Todorović, Smedinga, & Phaf, 2013). Each stimulus that receives a saccade, builds up
activation which competes with activation of the target, and with more switching, this would result in a longer RT under the influence of positive affect in a Flanker task (Schwarz & Mecklinger, 1995). This can serve as an alternative explanation for the results of Rowe et al. (2007). The flexibility hypothesis is supported by Heerebout et al. (2013) who found that
masked happy faces facilitated attentional switching between non-emotional stimuli, but masked angry faces did not. They concluded that affective modulation of attentional
switching may have developed early in evolution and may therefore have transferred to other more complex information processing. This conclusion is based on the finding that this modulation can be found in the simplest of tasks requiring a mere shift of attention, and, in addition, on the finding that affective modulation of flexibility is more basic than the
modulation of spatial attention (Heerebout et al., 2013). Furthermore, Phaf (2015) found in a masked Flanker task that interference of the flankers in- or decreased as a function of flanker-target interval. This implies a temporal component in these mood effects (Phaf, 2015). The reversed affective modulation of flanker interference with simultaneous presentation of flankers and target even squarely contradicted the spatial hypothesis. Additionally, Baumann and Kuhl (2005, see also Tan, Jones, & Watson, 2009) also found that positive affect
increased cognitive flexibility, and negative affect even decreased cognitive flexibility. They found that the typical precedence for global over local processing, observed after neutral and negative prime words, was reversed after positive prime words. They found this in a global-local shape detection task. When looking at the individual differences from their study, it seemed that one group of participants with dominant local processing, became faster at detecting global stimuli in positive moods. An opposite pattern was shown by the other group of participants with dominant global processing. This observation forms strong evidence for the assumption that positive affect indeed involves more cognitive flexibility rather than only global processing.
Despite all of aforementioned literature, no concluding answer has been found on the question whether positive mood promotes temporal switching of attention between stimuli or more spatial broadening of attention. The results from Rowe et al. (2007) did not exclude a temporal explanation. They presented the flankers simultaneously with the target. Slower RTs
in positive moods could also have resulted from more switching between flankers and target, which might have caused more interference by building up flanker activation, instead of more interference from a wider attentional scope.The results from Phaf (2015), on the other hand, showed a reversed effect, but gave a rather artificial explanation and therefore no concluding evidence. An answer to this question might also help explain why individuals might not notice an emergency exit when under the influence of negative emotions.
The current study aims to investigate if positive affect causes spatial broadening or more temporal switching of attention, by using a visual search task. In most of
aforementioned studies, a Flanker task was used (see Eriksen, 1995). In a Flanker task, the target will always be in the centre, which makes it an unsuitable test for serial search. Moreover, temporal explanations of flanker interference (cf, Phaf, 2015), appear rather artificial due to this spatial bias. In a visual search task, array sizes, target locations, and distances between stimuli can be manipulated in such a way that subjects are forced to
conduct a serial search, and this task is thus very useful for detecting temporal effects. On the other hand, this task also allows for detecting spatial effects, because the size of the arrays in which the stimuli are presented can be varied in such a way that they can either fall within or outside the attentional scope. According to Treisman and Gelade (1980), searching for a target that differs on two dimensions from distracters requires attention and serial search. This results in a longer RT with more distracters. In this way, a comparison can be made between RTs on larger and smaller arrays with an increasing number of distracters in positive versus negative moods. To promote the spatial hypothesis, distracters that differ on two dimensions are excluded, because this would cause pop-out effects. Inducing positive versus negative moods allows for discovery of affective modulatory effects on attention in a simple way instead of focusing on a specific emotion, which would be hard to define. Also, when only comparing positive mood versus baseline mood, no large effect would be expected, because
baseline moods might be biased to the positive side.
If positive affect broadened attentional scope, the slope of RT per distracter would be steeper for positive than negative affect (figure 1). In absolute terms, interference would increase with an increasing number of distracters. If positive affect increased temporal switching between stimuli, the slope of RT per distracter would be steeper for negative affect in comparison to positive affect (figure 2). If the slope of RT per distracter would be steeper for positive affect for small arrays (figure 3) and the slope of RT per distracter would be steeper for negative affect for large arrays (figure 4), then both hypotheses might be true, but on different levels. For when attention would have a spatial scope, using small arrays would shorten RTs in a negative mood, because attention would be narrowed which allows for faster target detection and less interference from distracters according to the spatial hypothesis (Figure 3). On the other hand, when using large arrays in which serial search is necessary and stimuli might fall outside the attentional scope, while assuming that positive affect stimulates temporal switching, RTs would be shorter under the influence of positive affect according to the temporal hypothesis (Figure 4).
Figure 1. Spatial hypothesis. Figure 2.Temporal hypothesis.
0 2 4 6 8 Rea ct io nt im e ( m s) Number of stimuli Positive Negative 0 2 4 6 8 Rea ct io nt im e ( m s) Number of stimuli Negative Positive
Figure 3. Spatial hypothesis when using Figure 4. Temporal hypothesis when using large small arrays. large arrays.
X-axes reflect increasing number of stimuli (distracters) and Y-axes reflect reaction time to detect a target. Slopes reflect the time needed to detect the target for each additional distracter.
Method
Participants
Sixty students, mostly from the University of Amsterdam, participated either voluntarily or for course credit. Participants who made more than 10% errors on the visual search task were removed from the analyses.
Design
This visual search task had a 2 x 2 x 4 within-participants design. Mood induction served as the first independent variable. The order of the two mood inductions was randomized across participants. The second independent variable concerned array sizes, which could either be large or small. The target was randomly allocated out of four different possibilities. The target did not vary within participants, but was randomized across them. Finally, number of stimuli had four levels (i.e. 4, 10, 18, 30). Trial order was randomized by the computer, with number of trials per condition equalized. The dependent variable concerned reaction time.
0 2 4 6 8 Rea ct io nt im e ( m s) Number of stimuli Positive Negative 0 2 4 6 8 Rea ct io nt im e ( m s) Number of stimuli Negative Positive
Materials and apparatus
The stimuli were presented against a gray background on a Dell P2412H monitor (24’’ and 1920 x 1080 resolution). Distance between participant and screen was approximately 60 cm. Music was played on IMG MD-5000DR headphones from a Samsung galaxy S4 mini. Reaction times were registered from two different buttons. Participants were told to press a button with their dominant hand when a target was present and to press the other button with their non-dominant hand when a target was not present.
Stimuli were 0.5 cm squares and circles with 0.37 visual degrees in dark gray (RGB 60) and white (RGB 195) against a light gray background (RGB 127.5). Target and distracters always differed on two dimensions. When, for example, the target would be a white circle, then distracters would be dark gray circles and white squares. Which figure would be the target was randomized across participants, but remained the same within participants. Number of stimuli (target and distracters) could either be four, 10, 18 or 30. This differed randomly across trials. Also, large and small arrays were used. The inner circle in small arrays had a radius of 40 pixels with 2.11 visual degrees and for large arrays the radius was 80 pixels with 4.23 visual degrees. The outer circle in small arrays had a radius of 220 pixels with 11.59 visual degrees and for large arrays the radius of the outer circle was 440 pixels with 22.95 visual degrees.
Moods were induced by asking the participant to write a story about a personal event when they were either happy or angry. During the writing and during the experiment, either happy music or annoying sounds were played to maintain the mood induction. Participants were asked to indicate their mood by a Self-Assessment Manikin (SAM), an affective rating system devised by Lang (1980) (see Appendix).
Procedure
information brochure and to sign an informed consent. Next, they were asked to indicate their mood. They were first presented with 40 practice trials, after which they indicated if they wanted to practice again or wanted to continue to the experiment. Next, they made a choice between either three happy songs or two annoying sounds (see Appendix), depending on mood condition. Subsequently, participants were given a story as an example for the story about a personal event they were then asked to write (see Appendix). Participants were given some time to think about the story they wanted to write, and when they indicated they knew what to write about, either the corresponding music or sound they chose earlier started to play and was repeated until the end of the first block. Whether participants started with the positive or negative condition was randomized. After the mood induction, participants were again asked to indicate their mood and after that, the first block of the experiment started which consisted of 320 trials. Before each trial, participants fixated on a cross in the middle of the screen for 1000 ms. Each trial ended after 2650 ms or when the participant pressed a button. At the end of the first block, participants were asked to indicate their mood again. Then, participants were given a short break of approximately 5 minutes. After that, they were asked to indicate their mood, to ensure they were in a neutral mood again. Next, they were asked to choose between either five happy songs or four annoying sounds, depending on mood
condition. Participants were given an example story and were asked to write a story about a personal event again. When indicating they were ready to write, the corresponding music or sound started to play until the end of the second block. After writing the story, participant were asked to indicate their mood, after which the second block started. This block also consisted of 320 trials, and ended with asking participants for a mood indication.
small array (2650 ms or until response) or:
1000 ms
large array (2650 ms or until response)
Figure 5. Timeline of a trial (with set size 30) for small and large arrays.
Beforehand, participants were instructed to respond as quickly as possible, because RTs were recorded. After the experiment, the experimenter conducted a semi-structured exit-interview with the participant. In this exit-interview the experimenter tried to make sure that the participant was not in a negative mood anymore. Also, questions about the possible use of strategies and impressions were asked, as well as participants thoughts on the possible effects of the mood induction on their visual search task (e.g. number of mistakes or speed).
Results
Mood induction
From the initial 60 participants, 14 were excluded because they made more than 10% errors on the visual search task, indicated they did not follow the instructions, or the mood reports indicated that the mood induction had not been successful. Successful mood induction was calculated by summing the SAM scores before and after each block and subtracting from this twice the baseline SAM score. Subsequently, the negative value was subtracted from the positive value and if this value was zero or higher, it meant the mood induction had not been successful and the participant was then excluded. Results of the SAM are presented in Figure
6 and Figure 7. The baseline mood, as measured by the SAM, in the positive and negative condition did not differ (M = 2.07 ± 0.71 and M = 2.13 ± 0.86 respectively) and were biased to the positive side. Mood inductions were successful with a difference between SAM scores after positive versus negative mood inductions (M = 1.39 ± 0.54 and M = 3.39 ± 0.77 respectively). A T-test revealed a significant difference (t(45) = 16.09, p < .01, d = 2.37). Mood inductions were also successfully maintained as indicated by the post block SAM scores after the positive block (M = 1.93 ± 0.68) and after the negative block (M = 3.70 ± 0.81) with a significant difference between the moods (t(45) = 12.61, p < .01, d = 1.56). In absolute terms, participants felt more negative after the negative mood induction and felt more positive after the positive mood induction. They maintained this mood throughout the corresponding block.
Arousal was also measured by the SAM and differed between the mood conditions with lower scores representing more arousal. In absolute terms, participants felt more aroused after the negative induction than after the positive induction and felt even more aroused after the negative block compared to immediately after induction. Baseline arousal in the positive and negative condition did not differ (M = 3.70 ± 0.92 and M = 3.64 ± 0.88 respectively). Arousal did differ after the mood inductions between the positive and
negative condition (M = 3.54 ± 0.96 and M = 3.22 ± 0.87 respectively). A T-test revealed a significant difference between degree of arousal immediately after the different mood
inductions (t(45) = 2.34, p < .05, d = 0.36). Arousal also differed after the blocks ended with participants being more aroused in the negative than in the positive condition (M = 2.85 ± 0.94 and M = 3.70 ± 1.07 respectively). A T-test revealed a significant difference (t(45) = 4.98, p < .01, d = 0.76). No order effect of mood induction was found.
Figure 6. Mean mood SAM scores before induction, after induction and after blocks. ‘Pos’ represents positive mood induction whilst ‘neg’ represents negative mood induction. Lower scores represent more positive moods, and higher scores represent more negative moods.
Figure 7. Mean arousal SAM scores before induction, after induction and after blocks. ‘Pos’ represents positive mood induction whilst ‘neg’ represents negative mood induction. Lower scores represent more arousal, and higher scores represent less arousal.
Visual search task Reaction time
After removal of incorrect responses, trials without target and outliers, 34 percent of trials remained available for analysis. Reaction times are presented in Table 1. Average reaction times were higher in the positive as well as in the negative condition for large arrays (M = 859.36 ms ± 107.50 and M = 858.56 ms ± 122.52 respectively) than in the positive and negative condition for small arrays (M = 660.25 ms ± 82.81 and M = 680.06 ms ± 91.21 respectively). Reaction times generally increased with increase in set size. However, for large arrays, a decrease in RTs was present between set size 18 and 30 in both mood conditions. In small arrays a small effect from mood has been found for the difference in RTs between the positive and negative condition (d = -0.31, 95% CI [-38.71, -.93]). To see how much importance could be attached to this result, a repeated measures ANOVA was conducted (F(1, 45) = 4.47, p < .05). A T-test confirmed the small significant difference (t(45) = 2.11, p < 0.05) with faster target detection in the positive condition. This indicates more temporal switching instead of spatial broadening (see Figure 3 and 4). For large arrays, no difference was found.
Table 1.
Average and mean reaction times and standard deviations (in ms) per condition for each mood induction, set size and array size.
Affect Set size Small arrays Large arrays
Positive 4 566.69 (71.72) 692.76 (81.85)
10 640.18 (87.81) 830.88 (110.71)
18 703.93 (100.88) 958.52 (142.86)
Mean 660.25(82.81) 859.36 (107.50) Negative 4 576.96 (82.83) 693.89 (104.26) 10 657.65 (107.70) 826.46 (126.45) 18 727.94 (108.57) 965.05 (164.52) 30 757.69 (121.43) 948.85 (192.86) Mean 680.06 (91.21) 858.56 (122.52)
Slopes and intercepts
The time needed to detect the target for each additional stimulus (distracter) present was calculated by the method of least squares and resulted in slopes. The average slopes for large arrays for positive and negative mood did not differ (M = 9.79 ms/item ± 0.86 and M = 9.62 ms/item ± 0.98 respectively) as can be seen in Table 2. The average slopes for small arrays for positive and negative mood were smaller and showed a difference (M = 6.10 ms/item ± 0.45 and M = 6.74 ms/item ± 0.52 respectively) as can be seen in figure 7, 8 and Table 2. In absolute terms, small arrays produced faster responses and within the small array condition, positive moods produced faster target detection. In small arrays, a small effect was found for the difference in slopes between both mood conditions (d = -0.15, 95% CI [-1.90, 0.63]) with less time needed to detect the target for each additional stimulus in positive moods. To see how much importance could be attached to this result, a repeated measures ANOVA was conducted (F(1, 45) = 1.03, p = 0.32). In small arrays, a small effect was also found for the difference in intercepts (preparation time) between both mood
conditions (d = -0.14, 95% CI [-30.41, 10.56]) with shorter preparation time in the positive condition. Again, to see how much importance could be attached to this result, a repeated measures ANOVA was conducted (F(1, 45) = 0.95, p = 0.33). In large arrays, no difference was found.
Table 2.
Mean slopes, intercepts and standard deviations (in ms) for each mood induction and array size.
Condition Large Small
Positive Slope 9.79 (0.86) 6.10 (0.45)
Intercept 707.59 (84.94) 565.68 (73.76)
Negative Slope 9.62 (0.98) 6.74 (0.52)
Intercept 709.46 (120.12) 575.60 (88.75)
Figure 7. Slopes for small arrays. Slopes indicate time needed to detect the target for each additional distracter. The X-axis represent increasing number of stimuli (distracters) and the Y-axis represent time needed to detect the target (in ms).
Also, figure 8 and 9 show that time needed to detect the target for each additional distracter (set size) increases in mostly a linear way. This indicates that participants conducted serial search. However, as can be seen in figure 7, reaction times decreased in the negative condition at set size 30 for large arrays. In absolute terms, participants were faster at detecting the target in set size 30 than in set size 18 in the negative condition for large arrays, which is unexpected. The fact that this kink is found for set size 30 might indicate
500 550 600 650 700 750 800 850 900 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 rea ct io n tim e ( in m s) set size positive negative
that participants narrowed their spatial scope and this result might therefore lean towards a spatial hypothesis. However, according to the spatial hypothesis, the slope should have been steeper for positive affect and this was not found (see Figure 3). For small arrays, even the opposite was found. This result can therefore serve as evidence against the spatial hypothesis. Another possibility might be that the attentional scope does not cover the space of large arrays at all and an explanation might be found on terms of strategy participants used. As can be seen in Table 2, intercepts increased for large arrays relative to small arrays. In the current study, intercept equals time needed to detect the target with zero distracters – or – preparation time. Because larger intercepts were found in large arrays, this might indicate participants used a different strategy with more preparation time needed.
Figure 8. Slopes and intercepts for small distances. Slopes indicate time needed to detect the target for each additional distracter (in this case set sizes).
Figure 9. Slopes and intercepts for small distances. Slopes indicate time needed to detect the target for each additional distracter (in this case set sizes).
Explorative analyses
Finally, some explorative analyses were conducted on the possible effect of the target ID (colour and form) on reaction time. In both conditions, white targets were detected faster than dark targets (see Figure 10 and 11). An ANOVA revealed a significant effect of target ID on RTs in the positive and negative conditions for large arrays (F(2, 42) = 6.71, p < 0.01). This might indicate that participants experienced more pop-out or used a strategy in which they only focused on colour when their target was white. Effects of target ID on time to detect the target for each additional stimulus (slope) and preparation times (intercepts) were also explored. As can be seen in Table 3, slopes were substantially higher when the target
concerned a dark circle, especially for large arrays. This might indicate that participants used another strategy when looking for a dark circle compared to searching for other targets. This might have influenced the overall results. Also, intercepts were (generally) higher for dark targets than for white targets. This indicates participants needed more preparation time and therefore might have used different strategies when searching for dark targets compared to white targets
Figure 10. Relation between target ID’s and average RTs for large arrays in the positive condition. X-axis represents target ID, Y-axis represents reaction time.
Figure 11. Relation between target ID’s and average RTs for large arrays in the negative condition. X-axis represents target ID, Y-axis represents reaction time.
Target ID Condition Large slope Large intercept Small slope Small intercept White circle Positive 8.66 (5.22) 650.10 (91.06) 4.94 (3.02) 516.24 (57.48) White square 6.40 (4.02) 694.49 (73.48) 6.19 (2.02) 543.04 (60.06) Dark circle 15.06 (4.13) 705.68 (57.08) 7.01 (3.43) 603.35 (75.63) Dark square 8.78 (5.96) 746.19 (88.69) 6.04 (3.24) 579.45 (73.70) White circle Negative 9.75 (4.30) 626.23 (86.33) 5.97 (2.30) 527.80 (81.10) White square 6.02 (7.46) 724.85 (174.73) 7.04 (2.22) 548.77 (90.86) Dark circle 13.60 (6.88) 731.50 (126.78) 7.66 (3.54) 608.02 (93.78) Dark square 8.88 (6.30) 731.12 (81.73) 6.40 (4.54) 594.14 (80.39) Table 3. Means and standard deviations for slopes and intercepts per target ID.
Discussion
In this study affective modulation of attention has been studied. A small effect of mood has been found in the visual search task: for small arrays, participants in a positive mood were faster at detecting the target than participants in a negative mood. Also, time needed to detect the target was shorter for each additional distracter (slope) in small arrays for positive affect. In small arrays, in which all stimuli were expected to fall inside the assumed spatial scope of attention, faster target detection in positive moods implicates more switching between stimuli. When attention would have a spatial scope, using small arrays would allow for faster target detection in a negative mood, because attention would be narrowed which provides for less interference from distracters, according to the spatial hypothesis. Consequently, time needed to detect the target for each additional stimulus should have been shorter in small arrays in a negative mood, according to the spatial hypothesis (see Figure 3). However, the opposite pattern was found, which implies that the obtained results lean towards a temporal hypothesis, meaning that a positive mood facilitated switching between items and therefore faster target detection.
The fact that the observed slopes in both mood conditions differed between large and small arrays might indicate that both hypotheses are true. However, if both hypotheses are true, participants should have been faster in large arrays in positive moods (see Figure 4) according to the temporal hypothesis. Though, no effect of mood on target detection has been found for large arrays. Target detection was slower in large than in small arrays. This
indicates participants conducted serial searching, which was the purpose of using large arrays, because this would allow for detecting temporal effects under the assumption that large arrays might not be completely covered by the attentional scope and switching would be required. Also preparation time (intercept) was longer in large arrays, indicating participants used different strategies when searching large arrays compared to small arrays.
When looking at the observed slopes compared to earlier literature, the idea of different strategies became more clear. In the study from Treisman and Gelade (1980), time needed to detect a target (slope) when conducting conjunction search in a visual search task was about 70 ms for each additional distracter. In the current experiment, this often did not exceed 9 ms. This might indicate participants did not always conduct conjunction search, but instead first searched for colour and afterwards for form or vice versa. In this way, it might have been possible participants clustered the stimuli according to one feature as an
explanation for the small observed slopes, especially in large arrays. By clustering, stimuli with other features than those focused on, are ignored. In this way, set size can actually be reduced, making it easier to detect a target. This notion can be supported by the observed faster target detection when targets were white. Thus, it seems plausible that participants clustered stimuli by colour and not by form. Interestingly, stimuli were presented against a gray background which equally differed in contrast with respect to the dark and white targets. Therefore, faster target detection of white targets seemed unexpected. However, this is
luminance of a target produced faster detection of that target. The idea of clustering might also serve as explanation for the longer preparation times in large arrays: it seems plausible participants would use this strategy in large arrays rather than in small arrays. Also, clustering might have been the source of faster target detection in set size 30 compared to set size 18 in large arrays in negative moods (see Figure 9).
This negative linear relationship between set size 18 and 30 in large arrays might also indicate individual differences between participants in their way of processing visual stimuli. According to Kozhevnikov, Kosslyn, and Shephard (2005), individuals can be distinguished in their way of processing visual stimuli into two groups: verbalizers and visualizers.
Verbalizers are a homogenous group, whilst visualizers can be distinguished into object visualizers who process stimuli more globally and look for object features such as colour, and spatial visualizers who process stimuli more locally and look for analytical- and spatial relationships. When individuals differ in their way of processing visual stimuli, this can influence visual search tasks. Spatial visualizers are then expected to be faster in conjunction search whilst object visualizers are expected to be faster in non-conjunction search and might therefore experience more pop-out effects. This might explain the negative slope from set size 18 to 30 in large arrays. It could have been possible that participants who were object
visualizers started to conduct feature search (clustering) to colour when they saw a large amount of stimuli (set size 30), which might be supported by the longer preparation times observed in large arrays. It is therefore recommended to conduct, for example, an Embedded Picture Task in future research to test whether participants are more globally or locally oriented, in order to be sure what visual processing preference participants have. In this way, participants can be segregated into different groups and then results might show differential mood effects depending on processing preference.
were detected faster than dark targets and it might have been possible participants clustered by colour. An important recommendation for future research might therefore be to use targets which cannot easily be clustered into groups by colour. Thus it would be recommended to use stimuli in the same colour without high luminance, such as black. Also, to prevent clustering by form, stimuli should not be easily distinguishable, such as inverted T’s or L’s.
Another recommendation for future research is to let participants choose all of the music or sounds they will be hearing during the experiment before the experiment has started. Now, participants chose before each block. When participants were in the negative condition in the second block, they already knew they had to hear the ‘annoying’ sound they were to choose for at least 10 minutes. Then, they might not have been completely honest about which sound they thought most irritating, which might have caused the negative mood induction to be less successful than it could have been. Also, use of the SAM is subjective to answering in a social desirable manner, which makes it hard to ensure the mood inductions were actually successful. For future research, it is therefore recommended to use a more objective measure of mood and arousal, such as skin conductance or EEG.
Because effect sizes were quite small, no conclusive statements can be made regarding the affective modulation of attention in terms of spatial broadening or temporal switching. However, time needed to detect the target for each additional distracter and average reaction times were shorter in small arrays for positive affect, which implies a more attentional process in terms of temporal switching rather than spatial broadening. The fact that not all results were significant does not mean they are not meaningful. Significant testing often reveals more about measurement accuracy than about actual differences (Johnson, 1999).
Altogether, it seems that positive affect might increase (attentional) flexibility, which is also reported by Rowe et al. (e.g., semantic remote-associates task, 2007). Therefore, negative affect might decrease this flexibility, or narrow one’s spatial attentional scope. What
caused the individuals during the 2001 fire to not notice the emergency exits remains mostly unanswered, but negative affect most likely modulated their attentional processes to become less flexible or more fixated on this one main door.
Literature
Alders, J. G. M., Belonje, M. M., van den Berg, A., ten Duis, H. J., & Doelen, A. (2001). Cafébrand Nieuwjaarsnacht: Eindrapport. Retrieved May, 17, from
http://nationaalbrandweerdocumentatiecentrum.nl/wp-content/uploads/2015/01/Eindrapport.pdf
Baumann, N. & Kuhl, J. (2005). Positive affect and flexibility: overcoming the precedence of global over local processing of visual information. Motivation and Emotion, 29, 123-133.
Brefczynski, J. A., & DeYoe, E. A. (1999). A physiological correlate of the ’spotlight’ of visual attention. Nature Neuroscience, 2, 370-374.
Compton, R. J. (2003). The interface between emotion and attention: A review of evidence from psychology and neuroscience. Behavioral and Cognitive Neuroscience Reviews, 2, 115–129.
Damasio, A. R., Everitt, B.J., & Bishop, D. (1996). The somatic marker hypothesis and the possible functions of the prefrontal cortex. Philosophical Transactions of the Royal Society B ,351, 1413-1420.
Eriksen, C. W. (1995). The flankers task and response competition: A useful tool for investigating a variety of cognitive problems. Visual Cognition, 2, 101-118.
Fredrickson, B. L. (2004). The broaden-and-build theory of positive emotions. Philosophical Transactions of the Royal Society B: Biological Sciences, 359, 1367–1377.
Gasper, K., & Clore, G. L. (2002). Attending to the big picture: Mood and global versus local processing of visual information. Psychological Science, 13, 34-40.
Haines, R. F., Gross, M. M., Nylen, D., & Dawson, L. M. Peripheral visual response time to colored stimuli imaged on the horizontal meridian. Washington: National Astronautics and Space Administration (US); 1974, Jun. 17 p. Report No:NASA-TM-X-3086.
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.
Johnson, D. H. (1999). The insignificance of statistical significance testing. The Journal of Wildlife Management, 63, 763-772.
Kozhevnikov, M., Kosslyn, S., & Shephard, J. (2005). Spatial versus object visualizers: A new characterization of visual cognitive style. Memory & Cognition, 33, 710-726. LaBerge, D. (1983). Spatial extent of attention to letters and words. Journal of Experimental
Psychology, 9, 371-379.
Lang, P. J. "Self-assessment manikin." Gainesville, FL: The Center for Research in Psychophysiology, University of Florida (1980).
Ohman, A., Flykt, A., & Esteves, F. (2001). Emotion drives attention: Detecting the snake in the grass. Journal of Experimental Psychology, 130, 466-478.
Phaf, R.H. (2015). Attention and positive affect: Temporal switching or spatial broadening? Attention, Perception, & Psychophysics, 77, 713-719.
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 Schwarz, W., & Mecklinger, A. (1995). Relationship between flanker identifiability and
compatibility effect. Perception & psychophysics, 57, 1045–1052.
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.
Triesman, A. M., & Gelade, G. (1980). A feature-integration theory of attention. Cognitive Psychology, 12, 97-136.
Appendix 1: Songs and annoying sounds participants could choose from
Positive condition Negative condition
Imagine dragons – On top of the world (instrumental)
Annoying mosquito sounds
Todd Terje – Inspector Norse Crying baby sounds
Bonobo - Kong
Appendix 2: Example stories participants could use for writing their own personal stories (Dutch)
Positive example story
Na mijn laatste tentamen, zaten mijn vrienden al op het terras met een biertje. De zon scheen en de zomervakantie was eindelijk daar! Op een gegeven moment hoorden wij geschreeuw vanuit een bootje op de gracht; het bleken 2 goede vrienden van ons te zijn die een boot hadden gehuurd met een barbecue! Een beter begin van de vakantie kon ik me niet wensen, we hebben de hele middag rondgevaren en heel lekker gegeten. Ik was zelfs een beetje bruin geworden en genoot echt van de zon in mijn gezicht. We hebben nog tot in de late uurtjes gechilld en het maakte niets uit, want ik kon voor het eerst die maand lekker uitslapen! Toen een week later de cijfers bekend werden, bleek dat ik een 8 had.
Negative example story
2 weken geleden had ik mijn laatste tentamen van mijn bachelor om 9 uur ‘s ochtends. Ik had al een aantal dagen erg hard geleerd en was goed voorbereid. Op de dag zelf bleek dat mijn wekker niet afgegaan was en ik om half 9 pas wakker werd. Ik rende snel naar de bus en
zwaaide nog naar de buschauffeur, maar hij reed voor mijn neus weg. Ik kon die man wel wat aandoen. Uit wanhoop belde ik een taxi, maar door het verkeer kwam ik 5 minuten te laat. 20 euro armer rende ik nog snel naar het lokaal, maar van de docent mocht ik echt niet meer naar binnen. Later bleek dat het tentamen pas 10 minuten te laat was begonnen, dus ik had nog wel naar binnen gemogen. Hierdoor moest ik dit tentamen herkansen in augustus, waardoor ik mijn vakantie naar Italië moest afzeggen. Ik ben nog nooit zo boos geweest!
Appendix 3: Self-Assessment Manikin (SAM)
The affective rating system used to assess mood. The upper images can be used to rate mood whilst the lower images can be used to rate arousal.
