The effect of positive affect on attention: A contrast between
spatial broadening and temporal switching
Name: Vera Tesink Student-ID: 10563881 Date: 27-05-2016 Supervisor: Hans Phaf Word count: 6114
Abstract
Positive affect causes the processing of more (task-) irrelevant and non-target stimuli, in comparison to negative affect. This was generally thought to be a result of a spatial broadening of the attentional scope. However, more studies now seem to find support for the assumption that not spatial broadening of the attentional scope, but temporal flexibility of attention causes this effect. The present study contrasted these two hypotheses using a visual search task, with the aim of determining which hypothesis is most accurate. Positive and negative mood induction were applied and distances between the stimuli and number of stimuli in the visual search task were varied. Participants in the positive mood condition were noticeably faster in finding a target at small distance arrays compared to participants in a negative mood. At large
distances, no evident effect was found. These results indicate that no concrete support can be assigned to either of the hypotheses. The small effects that were found,
The effect of positive affect on attention: A contrast between spatial
broadening and temporal switching
Witnesses of an armed robbery often fail to accurately describe the robber, whereas the description of the weapon generally seems to be undisputed. A frequently
occurring phenomenon, called the ‘weapon-focus effect’, which many believe can be explained by the Easterbrook hypothesis (Easterbrook, 1959). Easterbrook suggested that a high level of arousal (i.e., negative affect, in which affect is the subjective expression of an emotional state) causes the attentional focus to narrow. Therefore, the attention is narrowed towards the weapon, leaving the robber relatively
unattended. It was thus suggested by Easterbrook that negative affect causes a narrower attentional scope, a hypothesis on which a considerable amount of research is focused.
Whereas the focus long remained towards attention modulation and negative emotions, Fredrickson (2001) emphasizes the role of positive emotions in general and in attention modulation. She introduced her Broaden-and-Build Theory, suggesting that certain positive emotions lead to the broadening of people’s thought-action repertoires and the enabling of flexible and creative thinking. The consideration of the influence of positive affect led to a variety of studies to determine its role in attention modulation.
Based on earlier research (Easterbrook, 1959; Fredrickson, 2001) it has been proposed that positive affect leads to the broadening of attention, while negative affect leads to the narrowing of attention. This is a spatial hypothesis, since it merely focuses on the spatial aspect of the effect of affect on attention modulation. Various studies found evidence to support this hypothesis. A prominent study in this field is a
study by Gasper and Clore (2002), where the influence of affect on attentional focus was tested. Their results showed that positive affect induces a global (i.e., broad) focus and negative affect induces a local (i.e., narrow) focus. This result is based on examination with emotional stimuli (i.e., drawing of emotional pictures). It might, however, also be informative to study affective attention modulation regarding non-emotional stimuli, considering this could arouse different effects (i.e., excite emotions irrespective of the stimuli). The latter was conducted by Rowe, Hirsh and Anderson (2007), who used an Eriksen Flanker task (Eriksen and Eriksen, 1974), and found that more interference was caused by flankers in positive affect conditions than in
negative affect conditions. Rowe et al. thus concluded that the increased flanker incompatibility was a result of the broadening of the attentional scope due to positive affect, since the broadening of attention enables the processing of the flankers as well as the target. The influence of affect on attention can not only be observed in target detection, but also in internal systems such as memory, as positive affect causes improved memory for irrelevant (and non-emotional) information (Biss & Hasher, 2011). This effect can again be, according to Biss and Hasher, attributed to
broadening of the attentional scope.
There are some drawbacks concerning the results mentioned above. For
example, Bruyneel et al. (2013) could not replicate the findings by Rowe et al. (2007), even as they used an identical replication of the Eriksen Flanker task as performed by Rowe and his colleagues. Hence, this study severely weakens the results found by Rowe et al. (2007). Moreover, the Eriksen Flanker task that was used is too spatially oriented and therefore not sufficient to fully explore affective attention modulation, as it merely focuses on the spatial aspect of attention modulation, and thus does not allow for possible temporal aspects of attention modulation. Therefore, a task that
5 allows for a possible temporal aspect of attention modulation to be investigated
should be used, such as the attentional blink task. Studies using this attentional blink task established that attentional blink reduced as a result of positive affect (Olivers & Nieuwenhuis, 2006; Srivastava & Srinivasan, 2010). Since the stimuli in an
attentional blink task are presented sequentially at a consistent spatial location, a spatial explanation of broadening of the attentional scope is not sufficient to explain these findings.
The three studies mentioned above (Bruyneel et al., 2013; Olivers & Nieuwenhuis, 2005; Srivastava & Srinivasan, 2010) weaken the findings of the studies that provide evidence for the spatial hypothesis (Gasper & Clore, 2004; Rowe et al, 2007; Biss & Hasher, 2011). More importantly, these studies (Bruyneel et al., 2013; Olivers & Nieuwenhuis, 2005; Srivastava & Srinivasan, 2010) insinuate a different explanation for the differences in attention modulation due to affect. A temporal hypothesis is proposed, suggesting that positive affect does not broaden the attentional scope, but causes attentional flexibility. This hypothesis suggests an increase in switching between stimuli over time due to positive affect. This temporal flexibility in attention was developed early in evolution, according to Heerebout and Phaf (2010), who investigated the attentional flexibility using evolutionary
simulations in which agents were supposed to collect food and avoid predation. The mechanisms involved in attention modulation were measured. Higher oscillation frequencies were found in positive affect within these mechanisms compared to negative affect, resulting in more speed and efficacy in attentional switching. These findings suggest a neurobiological and evolutionary based connection between affect and attentional switching.
neurobiological relation between affect and attentional switching, the temporal hypothesis is also supported by a variety of behavioural studies. Several studies have shown that positive affect facilitates switching to non-dominant biases (e.g., from global to local bias), whereas negative affect induces interference in switching (Heerebout et al., 2013; Baumann & Kuhl, 2005). This demonstrates attentional flexibility between local and global biases due to positive affect, which is inconsistent with the implication made by Gasper and Clore (2002) stating that positive affect consequently causes a global bias. These results (Heerebout et al., 2013; Baumann & Kuhl, 2005) therefore undermine the findings by Gasper and Clore (2002).
Comparably, positive affect plays an important role in switching from former task-relevant stimuli to present task-task-relevant stimuli (Dreisbach & Goschke, 2004). This finding corresponds to the findings of the study by Biss and Hasher (2011). However, Dreisbach and Goschke interpreted the results in a temporal manner (i.e., temporal flexibility), whereas Biss and Hasher interpreted the results in a spatial manner (i.e., spatial broadening). This inconsistency marks the indistinctness concerning the effect of affect on attention towards non-emotional stimuli and the accuracy of either of the two hypotheses.
As previously stated, there is a lot of uncertainty as to which of the affective attention modulation hypotheses is most valid. To acquire this certainty, a contrast should be made between the two hypotheses. Phaf (2015) contrasted the broadening and the spatial hypothesis using an Eriksen Flanker task with different flanker-target intervals (created by the masking of flankers), which allow for a temporal component. He found that flanker-target intervals strongly influenced the interference of the flankers in positive and negative affect, even reversing the effect, implicating a strong influence of a temporal component. Phaf thus inferred this to be the result of
7 attentional flexibility. Moreover, the masked flankers led to more interference in negative moods than in positive moods, which is inconsistent with the findings of Rowe et al. (2007). Although Phaf used the same paradigm as Rowe et al. (2007) (i.e., Eriksen Flanker task), the results of both studies do not correspond, which can be attributed to the allowance of a temporal aspect in Phaf’s modified version of the Eriksen Flanker task.
The studies discussed above all find effects of affect on attention towards non-emotional stimuli that can either be explained by attentional broadening and temporal flexibility, or effects that can only be explained by temporal flexibility, insinuating that this temporal flexibility plays an important role in attentional modulation towards non-emotional stimuli either way. The aim of our study is to investigate the role of this temporal flexibility in attention modulation as a result of positive affect, to establish what positive affect exactly causes in regard to attention (in contrast with negative affect). We will investigate this by contrasting the temporal hypothesis and the spatial hypothesis. Phaf (2015) already contrasted the two hypotheses, yet his findings are not sufficient to draw a definitive conclusion with regard to the
legitimacy of either of the two hypotheses, as the adding of the flanker-target intervals to the Eriksen Flanker task is an artificial way of adding a temporal aspect. It is
therefore preferable to use a task in which both the spatial and temporal aspects are standard characteristics, such as the visual search task (Treisman & Gelade, 1980). The visual search task requires serial searching in varying diameters, which allows for both temporal and spatial aspects of attention modulation.
Assuming the spatial hypothesis is true, it is expected that participants in a positive mood are slower at finding a target than participants in a negative mood, because broadening of the attentional scope results in more interference of distractors
(Figure 1). The slope for participants in a negative mood will be smaller than for participants in a positive mood, which signifies that there is less interference per distractor for participants in a negative mood. Assuming the temporal hypothesis is true, it is expected that participants in a positive mood are faster at finding a target than participants in a negative mood, because increased attentional flexibility results in faster switching between stimuli (Figure 2). The slope for participants in a positive mood will be smaller than for participants in a negative mood, which signifies that less time is needed to switch between stimuli. A third assumption can be made, being the accuracy of both hypotheses. If this is the case, it is expected that participants in a negative mood are faster at finding a target at small distances, based on the spatial hypothesis, using the narrow attentional scope caused by negative affect. However, once distances become larger, the attentional scope will not be large enough to perceive all stimuli, resulting in switching between the stimuli. Therefore, this third assumption expects participants in a positive mood to be faster at large distances, based on the temporal hypothesis. This third assumption accounts for variation due to shifting distances (Figure 3).
The visual search task was performed after positive and negative emotion induction, which was counterbalanced over participants. The emotion induction was implemented by autobiographical recall (positive and negative) while listening to positive or negative music. Subjective reports were given for verification of the mood induction.
9 Figure 1. Expectations regarding the outcome of the experiment, assuming the spatial hypothesis is true. The solid line represents the slope of the positive mood condition and the dotted line represents the slope of the negative mood condition. The lines express the velocity (reaction time on y-axis) at which participants can find a target relative to different set sizes (4, 10, 18 or 30 items on x-axis).
Figure 2. Expectations regarding the outcome of the experiment, assuming the temporal hypothesis is true. The solid line represents the slope of the positive mood condition and the dotted line represents the slope of the negative mood condition. The
4 10 18 30 R eac ti on t im e Set size Spatial hypothesis
small & large distances
Positive Negative 4 10 18 30 R eac ti on t im e Set size Temporal hypothesis
small & large distances
Positive Negative
lines express the velocity (reaction time on y-axis) at which participants can find a target relative to different set sizes (4, 10, 18 or 30 items on x-axis).
Figure 3. Expectations regarding the outcome of the experiment, assuming that both the spatial and temporal hypotheses are true. The solid line represents the slope of the positive mood condition and the dotted line represents the slope of the negative mood condition. The lines express the velocity (reaction time on y-axis) at which
participants can find a target relative to different set sizes (4, 10, 18 or 30 items on x-axis). 4 10 18 30 R eac ti on ti m e Set size Spatial hypothesis small distances Positive Negative 4 10 18 30 R eac ti on ti m e Set size Temporal hypothesis large distances Positive Negative
Method
Participants
Sixty participants, consisting of forty-five students from the University of Amsterdam (mean age = 20.9), participated (46 female). Forty-five participants were rewarded with course credit, whereas fifteen participants volunteered. The participants had either normal vision or corrected-to-normal vision (i.e., glasses or contacts). Participants with an error rate higher than 10% on the visual search task were
removed from the dataset, since a high error rate indicates an insufficient motivation, in which case the data are inadequate to examine the contemplated effect.
Design
The visual search task had a 2 x 2 x 4 within-participants factorial design. The first independent variable was mood induction (positive and negative). The order in which the participants were given the induction was counterbalanced over participants (every odd participant number started with negative mood induction). The second independent variable concerned the distance between items (array size), consisting of small and large distances. The final independent variable concerned the set size of the items in the search field, which consisted of four grades (4, 10, 18 and 30 items). The dependent variable was the reaction time to find a target in trials in which the target was present and a correct response was given.
Material
Visual Search Task. The stimuli were presented at a Dell P2412H screen (24” and 1920 x 1080 resolution), and the distance of the participants to the screen was
approximately 60 centimetres. The programme used for the visual search task was Presentation 18.1. Participants responded to the stimuli with two button boxes (left and right of the monitor), which were used to register the reaction times. Two separate programmes were made to correct for right-handed and left-handed participants (in which the dominant hand was correctly registered).
The background of the screen was grey (RGB = 0x0x128). The stimuli were circles and squares (0.5 cm in diameter), which were either light grey (RGB =
195x195x195) or dark grey (RGB = 60x60x60). The difference in luminance between the stimuli and the background was equal for both light and dark stimuli. Half of the trials consisted of large arrays with a radius of 80 pixels (4.227 degrees of visual angle) and the other half of small arrays with a radius of 40 pixels (2.114 degrees of visual angle), containing a varying amount of stimuli. In each trial, all items were divided over four concentric circles, and with each additional circle the set size expanded (4, 10, 18 or 30 items). With each expansion of the set size the radii of the circles also expand, which can be calculated by multiplying the pixels of the array (small or large) with 1 (set size 4), 2.5 (set size 10), 4 (set size 18) or 5.5 (set size 30). The size of the stimuli remained constant regardless of expansion of the radii. The order of the trials was randomized by the used programme (Presentation 18.1), in a manner that every set size and array size was equally represented. The assignation of a target was randomly chosen by the used programme, and was consistent during the whole experiment. It was also ensured that the target and the distractors were similar on one dimension (form or colour). During each trial, the fixation point was visible for 1000 ms and the arrays were visible for 2650 ms. The start of a successive trial was initiated by the pressing of either of the buttons-boxes or the expiration of the time window of 2650 ms.
13 Figure 4. The course of one trial. Examples of a small array (top) and a large array (bottom) with a large set size (30 items).
Emotion induction. For the first part of the emotion induction (i.e., choosing and subsequently listening to sounds) a Samsung Galaxy S4 mini and IMG MD-5000DR headphones were used (Appendix 1 contains sound alternatives). For the second part of the emotion induction (i.e., the autobiographical recall) pen and paper were used, since it was strictly confidential. Further, participants were given an example of a positive/negative memory as a reference, to facilitate the process of formulating a memory of their own (Appendix 2).
Self-Assessment Manikin. Mood was tested using subjective rapports in the form of the Self-Assessment Manikin, or SAM (Bradley & Lang, 1994). The SAM is a self-assessment scale in the form of graphic pictures, in which the participants can indicate their mood (positive or negative) and level of arousal (high or low intensity).
Procedure
information brochure. They were informed that the influence of mood on attention was investigated, and that this involved emotion induction and a visual search task. Participants carried out a total of 680 trials: two blocks consisting of 320 trials, and a practice block consisting of 40 trials. The whole experiment lasted generally one hour.
Initially participants were asked to sign an informed consent, stating that all aspects of the experiment were adequately elaborated. Subsequently, the task on the computer started. Participants indicated their mood on a SAM, after which they started with 40 practice trials. After the 40 practice trials the participants were given the occasion to practice more, or to continue. After the practice trials, the participants were instructed to choose either a positive or negative sound for the emotion
induction (depending on their participant number), which would be continuously played during the first block. After the selection of a sound, the participants were instructed to write down a memory that would evoke the targeted emotion. A small duration of time was granted to formulate the memory, after which the participants were asked to put on their headphones and write down the memory while listening to the selected sound. The content of their autobiographical recall remained strictly confidential. Participants were instructed to directly continue the experiment on the computer after the writing of their autobiographical recall was finished. The first block started with an indication of their mood using SAM, and thereafter the first 320 trials began. After the 320 trials, participants indicated their mood again using SAM. The participants then had a ten-minute break. After the break, participants again rated their mood using SAM. For the second emotion induction participants were subjected to the same procedure as described above, except with induction of the opposite emotion. Subsequently participants rated their mood using SAM and began the
15 second block of 320 trials. After the second block, participants rated their mood in the final SAM. Lastly, participants were asked about their experience, expectancies and strategies in an exit interview, and whether they obtained the corresponding mood through the emotion induction.
Data analysis
To measure the inducted moods, the SAM test was used. To check if the manipulation was successful, the baseline SAM-score (before emotion induction) was subtracted twice from the sum of the SAM-scores post induction and post block for both positive and negative mood separately for each participant, after which the outcome of this subtraction for the positive mood was subtracted from the outcome of the subtraction for the negative mood. If the outcome of this last subtraction was a negative number or null, the manipulation was labelled unsuccessful. The outliers were removed based on the interquartile range method before the analysis was conducted.
The mean reaction times of all participants on all the set sizes were calculated, after which the slopes from the reaction times per set size were determined. This was calculated using the least squares method, which minimalizes the difference of the values of the slope given by a model and the observed data (since a model never comprises all observed data accurately). The slopes of all participants were averaged in the positive and negative mood condition and subsequently visualized in graphs (separately for small and large distances), which would allow us to visually interpret the velocity at which participants were able to find a target in positive and negative moods. Additionally the effect sizes of the reaction times were reviewed to determine the extent of the effect.
ANOVA’s was performed (with reaction time or slope as dependent variables and mood as independent variable). The ANOVA’s were performed with the objective of verifying whether a significant difference was present between the values of the reaction times and the slopes for positive and negative conditions.
Given that an effect of affect on reaction time is found in the results, weight is assigned to either one of the two hypotheses (or both). However, given that no effect of affect on reaction time is found, it is not valid to state that no effect is present, since the absence of a significant result does not necessarily ascertain that no effect is actually present. Hence, the emphasis of our analysis was on the interpretation of the results, and less on the statistical significance. Applied to our research this means that certain results can align more or less with one of the two hypotheses, irrespectively of the significance of the results.
17
Results
Fourteen participants were excluded from the data, resulting in analysis of forty-six participants (34 woman, mean age = 20.9). Six of these participants scored above the 10% error rate and appeared not properly motivated. The mood induction of the other eight excluded participants appeared unsuccessful (based on calculations with SAM-scores described in Data Analysis).
Table 1.
Mean scores of the SAM for positive and negative moods before the induction, after the induction and after the block.
Minimum Mood Maximum Mood Mean Mood Mean Arousal Std. Deviation Positive Pre induction 1 4 2.07 3.70 0.712 Positive Post induction 1 3 1.39 3.54 0.537 Positive Post block 1 3 1.93 3.70 0.680 Negative Pre induction 1 4 2.13 3.65 0.859 Negative Post induction 1 5 3.39 3.22 0.774 Negative Post block 2 5 3.70 2.85 0.813
Table 1 shows the mean scores of the SAM, where a low score (1) represents a positive mood and a high score (5) represents a negative mood. The baseline mood in the positive mood condition and the negative mood condition did not significantly differ. However, a small bias towards a positive mood was observed, seeing that the baseline scores lie beneath the median of 3. The mean scores on the SAM post induction and post block considerably differed between the positive mood condition
and the negative mood condition. Large effect sizes were found for mood post induction (Cohen’s d = -2.372, 95% CI [-2.250, -1.750]) and for mood post block (Cohen’s d = -1.859, 95% CI [-2.042, -1.480]), insinuating that the mood induction was successful. Moreover, this assumption was supported by a statistical test. A paired samples t-test was performed, which showed that the scores on the SAM after induction between participants in the positive mood condition and the negative mood condition were significantly different (t(45) = 16.086, p < 0.001), as well as the scores on the SAM after the block (t(45) = 12.609, p < 0.001). The level of arousal
accompanying every mood was relatively stable throughout the experiment (Table 1).
Table 2.
Overall mean reaction times with standard deviations and mean reaction times per set size (all in milliseconds) for positive and negative mood conditions at small and large distances.
Mean Std.
Deviation Mean Set size 4 Mean Set size 10 Mean Set size 18 Mean Set size 30
Large distance Positive 859.36 107.49 692.76 830.88 958.52 955.26 Large distance Negative 858.56 122.52 693.89 826.46 965.05 948.85 Small distance Positive 660.25 82.81 566.59 640.18 703.93 730.93 Small distance Negative 680.07 91.21 576.96 657.65 727.94 757.69
Reaction Times. The overall mean reaction times and the reaction times per set size were calculated and visualized in graphs (Table 2 and Figure 5). Participants in the positive mood condition were approximately equally fast in finding a target in large distance arrays (N = 859.36 ms ± 107.49) as participants in the negative mood condition (M = 858.56 ms ± 122.52). These mean reaction times did not appear to be
19 to be faster in finding a target in small distance arrays (M = 660.25 ms ± 82.81) than participants in the negative mood condition (M = 680.06 ms ± 91.21). A multivariate ANOVA was executed, which revealed a significant main effect of mood on reaction time at small distance arrays (F(45) = 15.516, p < 0.001). Notwithstanding this observed difference in reaction time, a small effect size was found (Cohen’s d = -0.312, 95% CI [-38.70, -0.93]), which signifies that the effect of mood on reaction time is limited.
Figure 5. The mean reaction times per set size for the positive mood condition and the negative mood condition at small and large distances.
The mean reaction times per set size for small distances ascend with every expansion of set size in a linear relation (Figure 5), which is indicative of serial search (i.e., with every added stimulus the search time increases). The mean reaction times per set size for large distances, however, show a linear relation up to set size 18, after which changes in inclination occur from set size 18 to set size 30 (Figure 5). The overall mean reaction times for small and large distances, irrespective of mood, also indicate that serial searching was implemented, since the mean reaction times are larger for large distance arrays than for small distance arrays (i.e., in large distance
arrays a larger distance has to be covered, which overall causes an increase in reaction time).
A repeated measures ANOVA was executed to evaluate whether the reaction times of participants in the positive mood condition and participants in the negative mood condition over small and large distance arrays were significantly different. No significant main effect of mood on reaction time was found (F(45) = 1.054, p = 0.310, η2
= 0.023). Moreover, no significant interaction effect was found between distance of the arrays and mood (F(45) = 3.457, p = 0.070, η2 = 0.071)
Table 3.
Mean slopes (in milliseconds/item) with standard deviations and mean intercepts for participants in the positive mood condition and the negative mood condition for both large and small distances.
Mean Slope Std. Deviation Mean Intercept Large distance Positive 9.79 5.82 707.59 Large distance Negative 9.62 6.67 709.46 Small distance Positive 6.10 3.03 565.68 Small distance Negative 6.74 3.51 575.60
Slopes. The mean slopes of the positive mood condition and the negative mood condition for both small and large distances were calculated (Table 3), to draw inference with regard to the velocity at which participants in a positive mood and a negative mood were able to find a target.
21 Figure 6. The slopes of the reaction times per set size of the positive and negative mood conditions for large and small distance arrays.
Figure 6 shows that participants in the positive mood condition and
participants in the negative mood condition were practically equally fast in finding a target in large distance arrays. No discernible difference was observed in the slopes of the positive mood condition (M = 9.79 ms/item ± 5.82) and the negative mood
condition (M = 9.62 ms/item ± 6.67), indicating that neither a positive nor a negative mood causes facilitation or interference (i.e., less or more interference of distractors) regarding the search for a target in large distance arrays. The slopes representing the small distance arrays are comparatively more divergent. The slope of the positive mood condition (M = 6.10 ms/item ± 3.03) is less steep than the slope of the negative mood condition (M = 6.74 ms/item ± 3.51), which shows that participants in the positive mood condition were faster at finding a target in small distance arrays than participants in the negative mood condition. This observation aligns with the temporal hypothesis, as this hypothesis states that participants in positive moods are faster at finding a target (at both small and large distances) than participants in negative moods due to the faster switching between stimuli. The slopes in the small distance arrays
therefore probably describe the velocity of switching between stimuli over time. However, the effect size is considerably small (Cohen’s d = -0.150, 95% CI [-1.902, 0.625]), which disables us to formulate firm conclusions. Notwithstanding the small effect size, a multivariate ANOVA revealed a significant main effect of mood on slopes of small distance arrays (F(45) = 6.795, p = 0.003, η2 = 0.240).
Intercept. A large difference was observed between the intercepts of the positive mood condition (b = 707.59 ms) and the negative mood condition (b = 709.46 ms) for the large distance arrays and the intercepts of the positive mood condition (b = 565.68 ms) and the negative mood condition (b = 575.60 ms) for the small distance arrays (Table 3). This indicates that participants initiated the search for the target in large distance arrays later than in small distance arrays, since the
intercept is irrespective of mood.
Explorative analyses
Since the experiment consisted of two blocks performing the same task, an order effect might occur. Two univariate ANOVA’s, one for each mood condition, were executed to check for an order effect. No order effect was present in the positive mood condition (F(45) = 0.596, p = 0.444, η2 = 0.013) or in the negative mood
condition (F(45) = 2.228, p = 0.143, η2 = 0.048), signifying that the order of the mood inductions did not influence the performance on the task.
23 Table 4.
The mean reaction times and mean slopes with corresponding standard deviations of the different targets.
Target ID Mean RT Std. Deviation RT
Mean Slope Std. Deviation Slope Dark grey circle 825.9630 25.98542 10.6475 2.78307 Light grey circle 693.4504 24.92347 7.3581 1.71005 Dark grey square 776.6991 21.58411 7.5371 3.46775 Light grey square 725.6415 19.54462 6.4971 2.58813
Furthermore, there appeared to be a difference in reaction time (in both positive and negative mood conditions) due to diversity in the targets. The reaction times of participants responding to a light grey target were substantially smaller than the reaction times of participants responding to a dark grey target (see Table 4 and Figure 7). Identically, the slopes representing the light grey targets were less steep than the slopes representing the dark grey targets (Table 4). There appeared to be a significant difference in reaction time in response to the different targets (F(45) = 5.271, p = 0.004, η2
= 0.274), as well as a significant difference in slopes for the different targets (F(45) = 4.226, p = 0.011, η2 = 0.232).
Figure 7. The mean reaction times for each different target, including confidence intervals of 95%.
Discussion
People in positive moods are moderately faster in finding a target at small distances than people in negative moods, which is expressed in a linear relation between reaction time and set size. People in positive moods and negative moods do not vary in regard to the velocity at which they can find a target at large distances. Seeing that no difference in reaction time is observed in large distance arrays, and only a small difference is observed in the small distance arrays (i.e., positive mood condition slightly faster than negative mood condition), no concrete support can be assigned to either one of the hypotheses. It can, however, be argued that the found reaction times and slopes in small distance arrays align more with the temporal hypothesis than with the spatial hypothesis. The steeper slope of the positive mood condition, compared to the negative mood condition, is in accordance with the expectancies of the temporal hypothesis, although the effect is considerably small. Thus, in the case of small distances, our results are in accordance with the results of Phaf (2015).
Besides the accuracy of either the spatial or the temporal hypothesis, a third possibility was described in the introduction, being the accuracy of both hypotheses. Assuming both hypotheses are true, it was expected that a negative mood would facilitate the search for a target in small distance arrays (seeking with small
‘spotlight’) and that a positive mood would facilitate the search for a target in large distance arrays (‘spotlight’ too small, continue to switching). A difference in slopes due to small and large distances was indeed observed, though the slopes of the small distance arrays show an opposite outcome (positive mood faster than negative mood) as to what was predicted for the small distance arrays (negative mood faster than positive mood) assuming both hypotheses are true. Thus, the different outcomes due
to distance do not provide support for the assumption that both hypotheses are true. Rather, as stated before, the slopes for small distances provide more support for the temporal hypothesis.
The slopes in general were, relatively speaking, not very steep, and are therefore more characteristic for feature search than for the intended conjunction search (Treisman & Gelade, 1980). This might be the consequence of our choice of stimuli, since the luminance difference between the light grey and dark grey stimuli might have been too pronounced. As a result, participants possibly first searched for contrast differences (i.e., light grey or dark grey) resulting in cluster formation, and subsequently searched for differences in form (i.e., circle or square). This is a form of feature search, since single features are considered in succession, instead of the aimed conjunction search, where multiple features are considered at the same time. Hence, different stimuli should be used in future research. A possibility would be the use of shades of grey with less luminance difference, impeding the differentiation between stimuli and discouraging the feature-based cluster formation.
Another unintended consequence of cluster formation is the decrease of the set size. If participants first form a cluster based on colour (i.e., light grey or dark grey), the (for them) irrelevant colour will subsequently be ignored, causing the halving of the set size in which the participant is deemed to search. This decrease in set size causes a decrease in reaction time, since fewer stimuli have to be evaluated. This decrease in reaction time was observed in the large distance arrays, as the mean reaction times for the largest set size (30) were smaller than the mean reaction times for a smaller set size (18), resulting in discontinuance of the linear relation (which signifies serial search). This being the case, the reaction times on large set sizes at large distances are not representative for these large set sizes, since in actuality the
27 search concerns a smaller set size.
The forming of clusters might also explain why the intercepts for large distance arrays are much higher than the intercepts for small distance arrays. In the large distance arrays, it is beneficial to form clusters (i.e., reducing of the set size), while in small distance arrays it is no necessity (i.e., target is easier found in smaller surface). The forming of clusters takes time (i.e., time to ‘prepare’). Hence, the higher intercepts in the large distance arrays might be the consequence of cluster formation. As stated earlier, this problem can be prevented using shades of grey with less luminance difference (discouraging the formation of contrast-based clusters). An evident difference was found in the mean reaction times and slopes between different targets. The light grey targets appeared to be more easily
recognized, since the mean reaction times in response to these targets were smaller and the mean slopes less steep. Therefore, the reaction times and slopes regarding the different targets cannot be validly compared. Further research should consider using the same target for every participant, preventing these differences in reaction times. The main deficiency in the experiment was the choice of stimuli, to which a variety of notable outcomes can be ascribed. The selection of different stimuli in future research would benefit more representative outcomes of the examined effects. Future research should provide more evident clarification as to which of the two hypotheses is most accurate. As to our findings, no firm conclusion concerning the accuracy of either of the two hypotheses can be drawn, though the small effects that were found align more with the temporal hypothesis than with the spatial hypothesis. With reference to the weapon-focus effect, a positive mood will in any case improve the description of the robber, regardless of the validity of the spatial or the temporal hypothesis. Whether people in positive moods switch faster between stimuli or have a
29
References
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.
Biss, R. K., & Hasher, L. (2011). Delighted and distracted: positive affect increases
priming for irrelevant information. Emotion, 11, 1474.
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.
Dhinakaran, J., De Vos, M., Thorne, J. D., Braun, N., Janson, J., & Kranczioch, C.
(2013). Tough doughnuts: affect and the modulation of attention. Frontiers in
human neuroscience, 7.
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.
Eriksen, B. A., & Eriksen, C. W. (1974). Effects of noise letters upon the
identification of a target letter in a nonsearch task. Perception &
psychophysics, 16, 143-149.
Fredrickson, B. L. (2001). The role of positive emotions in positive psychology: The
broaden-and-build theory of positive emotions. American psychologist, 56,
218.
versus local processing of visual information. Psychological science,13, 34-
40.
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.
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.
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.
Srivastava, P., & Srinivasan, N. (2010). Time course of visual attention with
emotional faces. Attention, Perception, & Psychophysics, 72, 369-377.
Treisman, A. M., & Gelade, G. (1980). A feature-integration theory of
31
Appendix 1
Table 4.
List of positive and negative sounds of which participants chose (all songs were instrumental).
Positive music Negative music
Bonobo - Kong Baby Crying Sound
Todd Terje - Inspector Norse Mosquito Flying Sound Imagine Dragons - On Top of the World
Appendix 2
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 twee 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 gechilled 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 acht had.’
‘Twee weken geleden had ik mijn laatste tentamen van mijn bachelor om negen 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 vijf minuten te laat. Twintig euro armer rende ik nog snel naar het lokaal, maar van de docent mocht ik niet meer naar binnen. Later bleek dat het tentamen tien minuten later 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!’
