The Influence of Inhibition of Return on the Evaluation of Emotion Words
J.J. Staats
Masterthese Sociale Psychologie Studentnumber: 6340938 Supervisor: M. Rotteveel Date: 10-6-2015
Index
Abstract 3
The Effect of Inhibition of Return on the Evaluative Rating of Words 4
Attention on Affect 5
Inhibition of Return Effect 7
Method 10 Participants 10 Design 10 Materials 11 Procedure 14 Results 15 Participants 15 Reaction Time 16 Evaluative Ratings 17
Correlation between Inhibition of Return and Evaluative Ratings 20
Catch 21 Emotional Contagion 22 PANAS 23 Three x 2 Design 25 Reaction Time 3x2 25 Evaluative Ratings 3x2 27 Discussion 28 References 33
Appendix I: Analysis with Different Outlier Detection 35
Appendix II: Word Stimuli 41
Abstract
This research tried to determine if the inhibition of a task-relevant visual attention-shift could influence the evaluative rating of word stimuli. Participants conducted a reaction time task with Inhibition of Return (IOR) trials, where the attention shifts to a previously cued spatial location, and Non-IOR trials, where attention shifts to an opposite spatial location than the previously cued location. After each reaction time trial participants rated word stimuli, consisting of negative, neutral, or
positive words. It was expected that words would be evaluated more negatively after an IOR effect has occurred. The results indicate that an IOR effect does exist, but it does not affect the evaluative ratings of words. However, a directional effect was found on the evaluative ratings of neutral words, people tend to rate neutral words more positive when attention shift from left-to-right than from right-to-left. Explanations for the results are discussed.
The Influence of Inhibition of Return on the Evaluation of Emotion Words
Attention enables people to focus their cognitive resources on relevant information. A lot of
visual information is available to us, however, our processing system is limited in capacity
(Broadbent, 1958, 1971; Posner, 1978; In Britton & Tesser, 1982), which means we have to
be selective in our attention. In day-to-day life it is required to focus attention on targets that
are relevant while ignoring other objects (Kastner & Ungerleider, 2000). This attention
selection may have an influence on other aspects of human life, such as how we feel about
objects.
How we feel was traditionally seen as post-cognitive, it was a reaction to one’s
personal interpretation of a situation (i.e. appraisal, Lazarus, 1980). However, Robert Zajonc
(1980) proposed that affect and cognition could be seen as two separate systems, that could
influence each other. Many researchers showed that affect can have an influence on cognition.
For example: it was shown that an individual’s mood can influence cognition; individuals that
are in a sad mood are more likely to remember sad memories, and attend more to sad
materials (Bower, 1981). In addition, individuals in a bad mood recognize more negative acts
when watching a video of a previous interaction, and recognize more positive acts when in a
good mood (Forgas, Bower, & Krantz, 1984). Positive emotions can also lead to a broader
scope of attention, which was shown in a study by Fredrickson and Branigan (2005).
Participants were shown either a video clip that provoked positive, negative, or neutral
emotions, after which they performed an 8-item global-local visual processing test. In the
global-local processing test the participants were shown a standard figure on top and two
comparison figures below it. The participants then circled either one of the two comparison
figures to indicate which of the two comparison figures was more similar to the standard
figure. One of the two comparison figures represented a more global scope of attention, while
the other one represented a more local scope of attention. Findings showed that when primed
with positive emotions, participants had a more global scope of attention. The current feeling
of individuals can thus influence the way people perceive stimuli, and draw for instance
attention to stimuli, or change their scope of attention, congruent with how they’re currently
feeling. The question that remains unanswered to some extent is whether or not attention can
influence affect and maybe feelings.
Attention & Affect
There is some evidence indication that attention can influence affect. Raymond, Fenske, and
Tavassoli’s (2003) study showed that participants evaluated previously ignored patterns
differently than previously attended patterns on an emotional evaluation scale. Participants
were asked to respond as fast as possible on in advance specified stimuli by clicking the left
arrow button or the right arrow button, depending on which side of the screen the stimulus
was presented. These were neutral stimuli. Each stimulus was presented simultaneously with
another neutral stimulus, which had to be ignored. At the end of each trial half the participants
were asked to rate the stimulus as ‘cheery’ on a three-point scale, the other half were asked to
rate the stimulus as ‘dreary’ on a three-point scale, from 1 (not cheery / dreary) tot 3 (cheery /
not dreary). It was found that previously ignored patterns were evaluated more negatively, so
drearier, and less cheery, than previously attended stimuli. In a later experiment by Fenske et
al. (2005) it was found that when shown a No-Go sign, which meant that they were required
to restrain from pressing any key, participants rated face stimuli as less trustworthy than when
shown a Go sign. This shows that when people experience inhibition they tend to evaluate a
stimulus more negatively than when not inhibited. Veling, Holland, and Van Knippenberg
(2008) showed that response inhibition not only affects trustworthiness, but also overall
affective evaluation. In the experiment by Veling et al. (2008) a Go/No-Go experimental
paradigm was used with negative, neutral, and positive stimuli. After the participants finished
the Go/No-Go trials, the participants were asked to rate the stimuli for future use in research
on a 9-point scale on attractiveness. Their findings showed that only the positive stimuli that
were presented with a No-Go cue were evaluated more negatively. This effect was not found
with neutral or negative stimuli. Veling et al. (2008) proposed that devaluating positive
stimuli after experiencing inhibition is useful in social interactions where approaching
positive stimuli is not appropriate, thus devaluating a positive stimuli will decrease the
approach tendency. It is however interesting that Veling et al. (2008) did not find an effect on
the negative stimuli, as avoiding negative stimuli can also be inappropriate in social
interactions.
In response to the conclusions from Raymond et al. (2003) and Fenske et al. (2005),
which propose that inhibition is the cause of the devaluation of stimuli, Dittrich and Klauer
(2012) propose that the devaluation of the stimuli were caused by the labels and instructions
given during the experiment. They propose that the devaluation is caused by the so called
evaluative coding (Eder and Rothermund, 2008). Evaluative coding is explained as:
‘‘evaluative implications of action instructions and action goals assign affective codes to
motor responses on a representational level that interact with stimulus evaluations on a
response selection stage’’. Simply put: by labeling a stimulus as a to-be ignored stimulus, a
negative affective code might be created that will lead to a negative attitude towards the to-be
ignored stimulus, which in turn leads to a more negative evaluation of that stimulus. A study
by Dittricht and Klauer (2012) showed a different effect for stimuli labeled as task-relevant,
meaning that they were relevant somewhere during the experiment, and task-irrelevant,
meaning that they should be ignored throughout the experiment. There was only a devaluating
effect for task-irrelevant stimuli, while there was no devaluation effect for task-relevant
stimuli in the trial block where participants were required to ignore them. This is not in line
with the previously mentioned devaluation-by-inhibition effect, which states that stimuli will
be devaluated when participants are required to ignore them, and thus inhibit their response.
Although the results from Dittrich and Klauer (2012) do imply that the devaluation of
the stimuli is not caused by an inhibition effect, it should be noted that the experiments were
somewhat different than experiments from for example Raymond et al (2003) and Fenske et
al. (2005). By implying a stimulus to be relevant or not, it might be possible that Dittrich and
Klauer (2012) created a pre-trial evaluation of that stimulus, which was not present at the
experiments from Raymond et al. (2003) and Fenske et al. (2005). A relevant stimulus might
be perceived as more positive, which might compensate for the devaluation effect of
inhibition, which would explain why Dittrich and Klauer (2012) did not find an effect of
inhibition on the evaluation of stimuli for task-relevant stimuli. Even though criticism on
Dittrich and Klauer’s (2012) experiment has to be taken seriously, evaluative coding does
point out an important problem within the research on selective attention and affect, as it can
undermine the devaluation-by-inhibition conclusions.
Inhibition of Return Effect
The problem of the so-called Evaluative Coding can be solved when an experiment uses a
setting in which participants are not asked to inhibit some kind of response. An example of an
inhibition effect that does not require participants to restrain from responding is a specific
attention effect called the Inhibition of Return (IOR) effect.
When a second stimulus is presented on the same spatial position shortly after another
stimulus, it is perceived faster than when positioned on another spatial position. However,
when the second stimulus is presented after a short delay, it is perceived slower when it is
positioned at a previous cued spatial position than when it is presented at another spatial
position, which is called the Inhibition of Return (IOR) effect. The IOR effect has initially
been studied by Posner and Cohen (1984). In Posner and Cohen’s (1984) experiment it was
shown that when stimuli were presented at the same spatial location in an experiment with
three horizontally aligned outline boxes with a short delay (approximately 150ms), people
responded faster than when presented on the opposite side. However, when the delay was
longer (approximately 300ms), people responded faster to the stimulus that was presented on
the opposite side of the previously cued box than to the stimulus presented at the previously
cued box. An explanation for this effect is that it can be useful when we are looking for
information (Tipper, Grison, & Kessler, 2003). Our first response is to look at the previously
cued location, but when the information we are looking for is not present, the attention shifts
away, explaining why the IOR effect is found only after a short delay.
Research shows that the IOR effect is stronger under certain conditions such as from
left-to-right, a tendency that only occurs for those who read from left to right, and is reversed
for those who read from right to left (Spalek & Hammad, 2005). Another directional factor
that influences the effect of IOR is top to bottom (Spalek & Hammad, 2004), which results in
a stronger IOR effect than bottom to top. This is said to be the result of learned regularity that
is associated with objects moving from top to bottom due to gravity. Also the time of onset,
when facilitation turns into inhibition, can vary, for example with task difficulty. The harder
the task, the later inhibition starts (Folk, Remington, & Johnston, 1992). This might be so
because the harder the task, the more intense attention the tasks requires. In which case
attention stays focused at the cue, leading to a delay in shifting to another stimuli (LaBerge,
1973, In: Klein, 2000).
So far it is clear when the IOR effect occurs (Posner & Cohen, 1984), and what
conditions may influence the amplitude of the effect. The direction seems to influence the
strength of the effect (Spalek & Hammad, 2004, 2005), and the onset of the IOR effect is
vulnerable to task-difficulty (Folk, Remington, & Johnston, 1992). It is however unclear how
this attention effect can influence affect.
There is support for the assumption that inhibition can influence the affective
evaluation of stimuli (Fenske et al., 2003, 2005; Raymond et al., 2003; Veling et al. 2008).
There is, however, some discussion on the underlying mechanism of the influence of
inhibition on affective devaluation. There is evidence that inhibition can lead to a devaluation
of stimuli, the question is whether or not this is caused by the inhibition effect, or because of
the negative evaluative coding (Dittrich & Klauer, 2012). In previous research, a Go/No-Go
cue is often used explicitly to activate attentional inhibition. However, inhibition can possibly
also be activated by the Inhibition of Return effect. In an IOR experiment stimuli cannot be
labeled affectively because the inhibition is subconsciously. Which, when the affective
devaluation of stimuli will be found, will make the evaluative coding explanation of Dittrich
and Klauer (2012) less likely. A few studies conducted by Mark van Weert (2009) showed the
influence of the Inhibition of Return effect on affective evaluation of words. Three types of
word stimuli were used, positive, neutral, and negative. These stimuli were shown in the
middle of the screen during a trial. Participants were asked to respond as fast as possible when
the target, a square, was presented. The target could either be presented on the left or on the
right side of the screen. In some trials participants were not shown a target stimulus, these are
so-called ‘catch-trials’. At the end of all the trials participants were asked to evaluate the
stimuli presented to them during the previous trials on a 7-point scale. Half of the participants
rated the stimuli from negative to positive, the other half from positive to negative. His
findings showed that only negative words were devaluated, while no effect was found for the
neutral or positive words. This is in contrast with previous findings by Veling et al. (2008),
which stated that only positive stimuli were devaluated.
In the study conducted by Van Weert (2009) an affective devaluation of negative
stimuli was found when the participants were asked to evaluate word stimuli after they
completed all the trials. His finding was a somewhat accidental finding. The devaluation
effect found in Van Weert’s (2009) studies might be stronger when participants evaluate the
word stimuli within each trial, so directly after the participant experiences inhibition of return.
This can possibly lead to a devaluation not only of the negative stimuli, but also of the
positive and neutral stimuli. That is why in the present study participants will evaluate stimuli
within each trial separately. Expected is that when people experience inhibition of return, thus
when stimuli appear on the same spatial position as the previous cue, they tend to evaluate
stimuli more negatively than when the cue and the target are presented on opposite spatial
positions.
Method Participants
For this experiment 40 participants were acquired. These students were acquired via the University of Amsterdam using DPMS. The participants were informed, signed informed consent, and received course credits for their participation after the completion of the experiment.
Design
The research design is a replication of the experiment used by Van Weert (2009), except for the moment of affective evaluation of the words. In this experiment the participants were asked to evaluate the word stimuli within each trial instead of at the end of all the trials.
This experiment is a 3 (word valence: positive, neutral, or negative) x 2 (position cue: left, right) x 2 (position target in contrast to the cue: same vs. different) factorial design. One dependent variable was the participant’s reaction time of pressing the response button after being presented with the target; this will be measured in milliseconds. The other dependent variable was the evaluative
rating of the words, a score that can range from 0 to 100, with 0 being very negative to 100 being very positive.
The analysis was done using IBM SPSS Statistics 19. The results of the participants were analyzed to check for outliers. Participants that had a reaction time of more than 2.5 SD above or below average were considered as outliers, this was done for each condition; this method was also used by Spalek (2005). Also a second outlier detection method was conducted, to see if the results would differentiate from Spalek’s (2005) method. The second outlier detection considers single observations of reaction times lower than 100ms and higher than 900ms as cut-off points. The results for the outlier detection method with cut-off point 100 and 900ms are summarized in Appendix I, footnotes in the Results will highlight the findings using the cut-off points 100 and 900ms method. To check if the IOR effect was present at the experiment, a repeated measures ANOVA was conducted on the latencies. To find the influence of the attentional shifts on the evaluative rating of the words, an identical repeated measures ANOVA was conducted. A bivariate correlation was used to analyze possible correlations between the latencies and the evaluative ratings. Independent-sample t-tests were conducted to spot the effect of participant’s emotional contagion and positive or negative affect (PANAS) on the evaluative ratings of the words. Furthermore repeated measure ANOVA’s were conducted to analyze the reaction times and evaluative ratings further in a more simple 3x2 design.
Materials
For this experiment, participants were placed in a dimly lit room where the participants would sit alone. The participants were placed in front of a computer screen where the stimuli were presented. The participants sat at a distance of 60cm from a 17-inch monitor (1024 x 768 pixels). The software Inquisit 4 was used to design and conduct the experiment.
The participants were shown three visible outlines (10° apart, which is 10,58 cm) of squares, so called placeholders (each sized at a 4° horizontal and vertical visual angle, which is 4.20 cm), during the reaction time task. These placeholders were horizontally aligned in the center of the screen against a white background. The attention task consisted of six practice trials and 96 experimental
trials. At the start of each trial only the three placeholder squares were visible. This was followed by a 500-ms display of placeholders with the word stimulus shown in the middle placeholder. The cue (a black circle) was then shown in either the left or right placeholder for the duration of 300-ms with the word stimulus shown in the middle placeholder. After this, only the placeholders were visible for 200-ms with the word stimulus. The fixation point (a black plus-sign) then appeared in the center
placeholder for the duration of 300-ms, replacing the word stimulus. Another empty display with only the placeholders and word stimulus visible was followed by this for 150-ms. On two of the practice trials and 24 of the experimental trials no target was visible (i.e. catch trials) and only the placeholders and word were visible for 1000ms. On half of the remaining four practice trials and 72 experimental trials the target (a black square) was shown on the left side for half of the trials whereas on the other remaining half of the trials it was shown on the right side for 1000ms.
The word stimuli consisted of 96 words that had either a negative, neutral, or positive valence. The 96 words were selected from a database that was used by Zeelenberg, Wagenmakers, and
Rotteveel (2006). The present experiment did not use all the words Zeelenberg, Wagenmaker, and Rotteveel (2006) used in their experiment, so a selection was drawn from their database. The words that were used in the present experiment were selected from the database based on their frequency in the Dutch language, for the positive, negative, and neutral words the frequencies per million were 19.16, 21.52, and 23.01 respectively. Also the words were selected on their length, the average word length for the positive, negative, and neutral words were 5.67, 5.47, and 5.44 respectively. The evaluative ratings of the words were tested by Zeelenberg, Wagenmakers, and Rotteveel (2006). The selection used in the present experiment had an average evaluation, on a 7-point scale, of 5.67, 2.26, and 4.15 for the positive, negative, and neutral words respectively. This was as close as possible to the averages of the words that Zeelenberg, Wagenmakers, and Rotteveel (2006) used for their experiment. The cue that drew the attention away from the center was a black circle. The target stimulus was a black square and the fixation point was a black plus-sign.
Each of the three types of words (positive, neutral, or negative) were used the same number of times, with each word being used only once. The distractor cue was displayed randomly on the left side on 50% of the trials and the remaining 50% on the right side. Because of a programming error the randomization was not completely successful. The cue and target did appear equally on the same side and on the opposite side of each other, but not equally within each valence condition (positive, negative, and neutral).
For the affective evaluation of the word stimuli, a slider was used that either went from positive to negative for 50% of the participants, and from negative to positive for the other half of the participants. The participants could put the slider anywhere between the two opposite poles. Data analysis and programming were done using a Windows 7 laptop.
Procedure
Participants were asked to sit in the room in front of the computer, were informed, and signed informed consent. The participants were instructed to react as fast and accurately as possible to the targets by pressing the spacebar, while trying to remember the words that were shown in the middle placeholder. The participants were also instructed that the best way to do the task is by trying to keep looking at the middle square. At the start of the trial only the three placeholder squares were visible. Participants were than shown the placeholders with the word stimulus shown in the middle
placeholder. After that, the distractor cue was shown in either the left or right placeholder with the word stimulus shown in the middle placeholder. After this, only the placeholders were visible with the word stimulus. The three placeholders followed this with the fixation cue in the center placeholder, replacing the word stimulus. Then, another empty display with only the placeholders and word stimulus visible followed. After that the participant saw either a target (either on the left side or the right side) or only the three empty placeholders (i.e. catch-trials).
After reacting to the target participants were asked to evaluate affectively the word stimulus from positive to negative by sliding a slider from positive to negative, or from negative to positive depending on which condition they were in. After completion of these trials participants were asked to fill in the PANAS (Watson & Clark, 1988) to measure to what extend the participants were feeling positive and negative affect, and an emotional-contagion questionnaire (Doherty, 1997) to measure how sensitive the participants were to other’s emotions. After filling in the PANAS and the Emotional Contagion questionnaire the participants were asked to fill in their age, gender, if they were a student, and whether or not they knew the experiment. At the end of the experiment an exit-interview was conducted with follow-up questions to check if the manipulation was successful and to ask if they knew what was being studied within this experiment.
Results Participants
For this experiment 40 participants were acquired. Of the 40 participants, 12 (30%) were male and 28(70%) were female. In total, 37 participants (92,5%) were student and 3 (7,5%) were not a student. The average age was 20.35 years, with an SD of 3.34, and a range from 18 to 23, and one person that was 38 years old.
Two participants were excluded from further analysis because of a programming error; these were participant one and two. The 38 participants that remained gave on average in 22,89 (SD = 1,71) of the 24 catch-trials a correct response, meaning they did not press the spacebar. One participant was removed from further analysis because of responding incorrectly too many (i.e. more than 2.5 SD’s above average, > 6.4) times on the catch-trials (participants number 7). Six participants’ data were excluded, from which one was already excluded based on the catch-trials, because of responding more than 2,5SD slower or faster than the rest of the participants in at least one of the 12 conditions (3 (positive, negative, and neutral) x 2 (cue on the left of the right side) x 2 (target on the same side as the cue or the opposite side)). These were participants seven, eight, 13, 16, 17, and 29.
Datasets can be checked for outliers using different methods. The method used by Spalek (2005) was used for this paper. The data was also analyzed with an alternative outlier method, to check the robustness of our results. This is not shown in the results section of this paper, but is summarized in Appendix I. As can be seen in Graph 1 in Appendix I most of the responses were given within the 100ms and 900ms time window. When a participant responded faster than 100ms on a target it was considered an outlier, as this can be seen as an anticipated respond. Also observations above 900ms were considered outliers, as this indicates that a participant was not paying attention and therefore might affect the effect that is being studied in this experiment. In total 63 (2.30%) of the 2736 observations were excluded from the analysis using the 100 and 900ms cut-off points outlier method. Throughout the results section footnotes will highlight the results of the analysis using the outlier method with cut-off points 100ms and 900ms.
Reaction Times
Response latencies were analyzed using a within-subject repeated measures ANOVA with as variables valence of the words (positive vs negative vs neutral), side of the cue (left vs right), and side of the target in contrast to the cue, here and forward known as cue-target relationship (same vs opposite). Participants responded faster when the target appeared on the opposite side of the cue than when the target appeared at the same side as the cue (see Table 1), F(1, 31) = 39,63, p<0.0001, ηp2= .56. This means that there was an inhibition of return effect present in the latencies because participants needed more time to respond when the target and cue were shown on the same location1. This is in line with previous research from other researchers (i.e. Spalek & Hammad, 2004, Posner, 1980, and Van Weert, 2009).
Next to the main effect of the cue-target relationship, also a marginal main effect of valence was found. Contrasts revealed a marginal difference between the reaction times on targets that were shown with negative words and targets that were shown with neutral words (see Table 2),
F(1,31)=3.86, p=0.059, ηp 2
= .11. This indicates that participants reacted faster to targets that were shown with negative words than with neutral words2. Post-hoc analysis with a paired-sample t-test revealed that participants also reacted marginally faster to targets that were shown with positive words than with neutral words3, t(31)=-1.80, p=0.082. There was no difference between the reaction times to targets that were shown with positive or negative words. This indicates that participants reacted faster to targets that were shown with words with an emotional meaning, either positive or negative, than to words that have no emotional meaning such as neutral words. No interaction effect was found between valence, side of the cue, and cue-target relationship. Also no main effect of the side of the cue on the reaction times was found.
_________________________________________________________________________________
1 With cut-off points 100ms and 900ms an IOR-effect is also found, F(1, 37) = 42.58, p<0.0001, η
p2= .54.
2 With cut-off points 100ms and 900ms a similar effect is found, F(1,37)=3.76, p=0.060, η
p2= .09.
Table 1. Means and SD’s of Reaction Times (ms) for Side Cue (Left vs Right) vs Cue-Target Relationship (Same vs Opposite).
Side Cue-Target Relationship
Side Cue Same Opposite
M SD M SD
Left 323.22 61.21 287.47 59.06
Right 324.70 74.17 286.47 56.44
Average 323.75 66.23 286.62 55.97
Table 2. Means and SD’s of Reaction Times (ms) per Valence (Positive vs Negative vs Neutral) per Side Cue (Left vs Right) vs Cue-Target Relationship (Same vs Opposite).
Valence
Positive Negative Neutral
Side Cue Side Cue-Target relationship
Same Opposite Same Opposite Same Opposite Mean Left M=316.61 SD=75.31 M=283.59 SD=62.13 M=315.48 SD=67.05 M=281.35 SD=57.87 M=337.56 SD=78.08 M=297.46 SD=72.82 M=305.34 SD=57.62 Right M=318.41 SD=75.39 M=285.54 SD=66.15 M=330.45 SD=91.86 M=281.59 SD=74.90 M=325.23 SD=80.99 M=292.29 SD=57.88 M=305.59 SD=60.74 Average M=317.51 SD=67.84 M=284.56 SD=58.59 M=322.96 SD=69.90 M=281.47 SD=62.41 M=331.40 SD=72.45 M=294.88 SD=58.29 M=301.04 SD=58.91 M=302.22 SD=62.10 M=313.14 SD=61.68 Evaluative Ratings
To analyze the evaluative ratings of the words, the same Repeated Measures ANOVA was conducted as was used with response latencies. The within-subject variables were affective valence of the words (positive vs negative vs neutral), side of the cue (left vs right), and cue-target relationship (same vs
opposite). A significant main effect of valencewas found4 (see Table 3), F(2, 62)=454.35, p<0.0001, ηp
2
= .95. Contrasts reveal that positive words were rated more positive than negative words5, F(1, 31)=547.57, p<0.0001, ηp
2
= .95. Negative words were rated more negative than neutral words6,
F(1,31)=310.94, p<.0001, ηp2= .91. Post-hoc analysis with a paired sample t-test reveal that positive words were rated more positive than neutral words7, t(31)=21.12, p<0.0001. These results were as expected and show that the three valences were indeed different from each other.
Also a marginal interaction effect between valence and cue was found8, F(2, 62)=2.82, p=0.67, ηp
2
= .08. This interaction effect was specifically associated with positive and negative words. The influence of the side of the cue was not the same for the positive and negative words9, F(1, 31)=3.824,
p=0.11. Contrasts reveal that the participants rated the positive words more positive when the cue was
presented on the right side than when presented on the left side. In contrast, the negative words were rated more positive when the cue was presented on the left side than when the cue was presented on the right side. This effect is visualized in Graph 1. Interestingly, the evaluations of the neutral words seem to be affected in the same way as negative words, as the neutral words are also rated more positive when the cue is presented on the left side. To our knowledge this effect has not been found previously and will be further explored upon in the discussion.
As the means in Table 3 show, there seems to be a difference in evaluative ratings between the attention shifts for the neutral words. When the cue is presented on the left side and the target on the right side, so when attention shifts from left-to-right, the participants rated the words more positively than when the cue was presented on the right side and the target on the left, so when the attention shifts from right-to-left. Even though no interaction effect was found between valence, cue, and cue-target relationship a paired-sample t-test was conducted to see if there might be a directional effect of attention shift. What was found is that when attention shifts from left to right the neutral words __________________________________________________________________________________
4 With cut-off points 100ms and 900ms a valence main effect is also found, F(1, 62) = 36.00, p<0.0001, η
p2= .94.
5 This effect is also found with cut-off points 100ms and 900ms, F(1,37)=538.89, p<0.0001, η
p2= .94.
6 With cut-off points 100ms and 900ms this is also found, F(1, 37) = 340.28, p<0.0001, η
p 2
= .90.
7 Analysis with cut-off points 100ms and 900ms shows the same effect, t(1,37)=20.62, p<0.0001.
8 The same effect is found with cut-off points 100ms and 900ms, F(2, 62) = 2.66, p=0.08, η
p2= .07.
9 The results of the analysis with 100ms and 900ms indicate a similar effect, F(1, 37) = 2.94, p=0.10, η
(M=57.09, SD=7.17) were rated marginally more positive than when the attention shifts from right to left10, M=53.94, SD=6.77, t(31)=1.98, p=0.057. This effect only seems to apply to neutral words, and not to positive or negative words. Maybe this is due to a lack of emotional valence of neutral words, as they are somewhere between positive and negative evaluations, which might make them more
susceptible to the influence of the IOR. Positive and negative words have a certain emotional valence and might therefore be less likely to be changed because of the small effect of an attentional shift.
Table 3. Means and SD’s of Evaluative Ratings for Valence (Positive vs Negative vs Neutral) per Side Cue (Left vs Right).
Valence
Side Cue Positive Negative Neutral
M SD M SD M SD
Left 78.34 8.94 22.04 7.73 56.89 7.39
Right 79.62 7.69 20.07 7.80 54.98 7.10
Average 78.98 7.89 21.05 7.12 55.93 6.53
Table 4. Means and SD’s of Evaluative Ratings per Valence (Positive vs Negative vs Neutral) per Side Cue (Left vs Right) vs Cue-Target Relationship (Opposite vs Same).
Valence
Positive Negative Neutral
Side Cue Side Cue-Target relationship
Same Opposite Same Opposite Same Opposite
Left M=78.51 SD=10.61 M=78.17 SD=8.50 M=22.20 SD=10.07 M=21.87 SD=7.32 M=56.70 SD=9.63 M=57.09 SD=7.17 Right M=80.16 SD=8.55 M=79.07 SD=8.45 M=19.69 SD=8.74 M=20.45 SD=8.51 M=56.01 SD=9.77 M=53.94 SD=6.77 Average M=79.34 SD=9.02 M=78.62 SD=7.58 M=20.94 SD=8.23 M=21.16 SD=7.26 M=56.35 SD=8.85 M=55.52 SD=5.33 __________________________________________________________________________________
Graph 1. Interaction Effect of Valence (Positive vs Negative) * Cue (Left vs Right Side). Evaluative Rating on a Scale from 0 (Negative) to 100 (Positive). Primary Y-axis is the Evaluative Ratings of the Positive Words and Secondary Y-axis is the Evaluative Ratings of the Negative Words.
Correlation Between Inhibition of Return and Evaluative Ratings
To analyze the correlation between an IOR effect in response latencies and associated evaluative ratings of the words, a bivariate correlation was conducted. Inhibition of Return is computed as reaction times or ratings when the target is presented on the same spatial position as the cue minus the reaction times or ratings when the target is presented on the opposite spatial position as the cue. A marginal positive correlation between the IOR on all the reaction times trials and the IOR of the ratings of all the words is found11, r(30)=.32, p=.07. This indicates that when the difference in reaction times between presenting the target and the cue on the same spatial position and presenting the target on the opposite spatial position of the cue increases, the difference in ratings between presenting the target on the same versus the opposite spatial position of the cue also increases. This means that when the reaction time increases when the target is shown on the same spatial position as the cue, which is the IOR condition, in contrast to when the target is presented on the opposite spatial position as the cue, the not-IOR condition, the participants evaluate the words more positive when the target was presented at the same spatial position as the cue than when the target was presented on the opposite spatial position as the cue. This indicates that an IOR effect increases the evaluation of
__________________________________________________________________________________
11 With cut-off points 100ms and 900ms this is also found, r(36) = 0.31, p=.06.
20 20,5 21 21,5 22 22,5 77,5 78 78,5 79 79,5 80 left right Eval ua tiv e Ra tin g N eg at iv e W or ds Ev al ua tiv e Ra tin g Po si tiv e W or ds
Side of the Cue
Interaction Effect Valence * Cue
Positive Negative
the words, which is in contrast to what was expected, as it was expected that IOR would lead to a more negative evaluation of the words.
The positive correlation between IOR on reaction times and IOR on the ratings is specifically strong for the IOR in the reaction times with neutral words. This correlation is significant12, r(30)=.43,
p<.05. When the difference in reaction times between the target being presented on the same spatial
position versus the opposite spatial position of the cue increases in trials with a neutral word, the difference between the ratings of the words between presenting the target on the same versus the opposite spatial position of the cue also increases. Ergo, the neutral words were rated more positive when the IOR effect was stronger.
Table 5. Correlation Coefficients Between the IOR Response Latencies of All the Trials, Trials with Neutral, Positive, or Negative Words and the IOR Evaluative Ratings of All the Trials (Total Evaluation IOR), and Trials with Positive, Negative, and Neutral Words.
Correlation Coefficients
IOR Affective Evaluation IOR Latency
Total Neutral Words Positive Words Negative Words
Total .32* .43** 0.07 .18 Positive Words .10 .14 -.16 .24 Negative Words -.02 -.05 .12 -.03 Neutral Words .05 .21 -.06 .04 *p<.1 **p<.05 Catch
Catch trials were trials where no target was presented. In catch trials participants were supposed to restrain from a response. Veiling et al. (2008) showed that when participants were required to restrain from pressing a key in a Go/No-Go task they rated the stimuli more negative, although this effect was only found on positive stimuli. In a way restraining from a response can be seen as an inhibition, as __________________________________________________________________________________
participants were required to inhibit a response in contrast to the other trials where they were asked to react as quickly as possible to a target. To analyze the evaluative ratings on the catch trials in relation to the non-catch trials a 3x2 Repeated Measures ANOVA was conducted, with valence (positive vs negative vs neutral) and trial (catch vs non-catch) as variables. A marginal main effect of catch was found13 (see Table 6), F(1, 31)=3.96, p=0.055. Participants rated the words in the non-catch trials more positive than words in the catch-trials. This indicates that a possible inhibition experienced from restraining from a reaction might negatively influence the evaluation of the words. There was no interaction effect between the valence of the words and the catch vs non-catch, this indicates that the effect of the trials being catch or not did not differ between the different types of valences14.
Table 6. Means and SD’s of Evaluative Ratings for Valence (Positive vs Negative vs Neutral) per Trial (Catch vs Non-Catch).
Valence
Trial Positive Negative Neutral Average
M SD M SD M SD M SD
Catch 78.01 8.08 19.41 6.90 54.98 5.19 50.80 3.46
Non-Catch 78.98 7.89 21.05 7.12 55.93 6.53 51.99 3.34
Emotional Contagion
The emotional contagion questionnaire measured to what extent a participant was sensitive to other’s emotions and was influenced by other’s emotions. The participants were divided into two equal groups; those with an average score above 3.17 and those with a score lower than 3.17. The group with an average score higher than 3.17 can be seen as ‘high sensitive to other’s emotions’ and the group with an average score lower than 3.17 can be seen as ‘low sensitive to other’s emotions’. To check what kind of influence the emotional contagion of participants had on their ratings of the words, an Independent Sample T-test was conducted with the evaluative ratings of the different
__________________________________________________________________________________
13 With cut-off points 100ms and 900ms a significant main effect of catch is found, F(1, 37) = 4.53, p<0.05, η
p2= .11
valences (positive, negative, and neutral) as test variables and Emotional Contagion as grouping variable with 3.17 as cut-off point. The results of the analysis revealed that Emotional Contagion did not lead to a significant difference in the evaluative ratings of the words. Nevertheless, it is interesting to see that the effect of emotional contagion did not influence every valence equally, as can be seen in Table 7 the difference between high-sensitive people and low sensitive people was bigger for positive and negative words and smallest for neutral words. This might be explained by the lack of emotional information of neutral words, and are therefore less contagious than positive or negative words15. Further analysis on the correlation between emotional contagion and the evaluative ratings of the words indicated that there was no correlation between the emotional contagion of the participants and the evaluative ratings of the words.
Table 7. Means and SD’s of evaluative ratings for valence (positive vs negative vs neutral) per EC score (<3.17 vs >3.17).
Valence
Mean EC Positive Negative Neutral
M SD M SD M SD
>3.17 81.10 5.65 19.68 5.82 56.48 4.37 <3.17 76.86 9.33 22.42 8.19 55.39 8.27
PANAS
The PANAS questionnaire is designed to measure participants’ current affect. The questionnaire was divided into two different affects, positive, and negative. To analyze the influence of participants’ current affect on the affective evaluation of the words two Independent-Sample T-tests were
conducted, one to analyze the influence of the positive affect and one to analyze the influence of the negative affect on the evaluative ratings of the three different word valences (positive, negative, and neutral).
__________________________________________________________________________________
For each analysis the participants were divided into two, if possible, equal groups. The cut-off point for the positive affect was 3.75, with higher than 3.75 being the ‘high-positive’ group and lower than 3.75 being the ‘low-positive’ group. Because of equal mean scores, the ‘high-positive’ group had two more participants than the ‘low-positive’ group, 17 and 15 respectively. The t-test revealed that the positive affect of the participants during the experiment did not influence the evaluative ratings of the words. Also the positive affect of the participants did not influence the ratings on the independent valences (positive, negative, and neutral)16.
For the second t-test, to analyze the influence of the participant’s negative affect on the affective evaluation of the words during the experiment, the cut-off point was 1.81. The participants with an average score higher than 1.81 formed the group ‘high-negative’ and those with an average score lower than 1.81 formed the ‘low-negative’ group. Both groups had 16 participants. The independent-sample t-test revealed that also the participant’s negative affect did not influence the evaluative ratings of the words. Further correlational analysis to see if there is a correlation between the positive or negative affect and the evaluative ratings of the words showed that there was no correlation between the participants’ current affect and their evaluation of the words17.
Table 8. Means and SD’s of Evaluative Ratings for Valence (Positive vs Negative vs Neutral) per PANAS Positive Affect Score (<3.75 vs >3.75).
Valence
Mean PANAS Positive Negative Neutral Average
M SD M SD M SD M SD
>3.75 77.67 7.03 21.74 6.36 56.10 5.73 51.93 3.59
<3.75 80.46 8.78 20.28 8.06 55.75 7.54 52.35 3.66
__________________________________________________________________________________
16 This is also found with cut-off points 100ms and 900ms.
Table 9. Means and SD’s of Evaluative Ratings for Valence (Positive vs Negative vs Neutral) per PANAS Negative Affect Score (<1.81 vs >1.81).
Valence
Mean PANAS Positive Negative Neutral Average
M SD M SD M SD M SD
>1.81 79.85 8.44 20.64 7.60 55.72 7.25 52.21 3.24
<1.81 78.10 7.48 21.47 6.83 56.15 5.95 52.04 3.98
Table 10. Correlation Coefficients Between the PANAS Positive Affect, PANAS Negative Affect, and Emotional Contagion and the Evaluative Ratings of All the Words (Total), Positive Words, Negative Words and Neutral Words.
Evaluative Ratings
Total Positive Words Negative Words Neutral Words
PANAS Positive .16 -.08 0.07 .25
PANAS Negative -.15 -.01 -.05 -.14
Emotional Contagion -.04 .17 -.18 -.02
Three x Two design
The observations per cell in the 3 (positive vs negative vs neutral) x 2 (cue left vs cue right) x 2 (cue-target-relationship: same vs different) varied from one to 11 (M=6.00, SD=1.88), meaning some cells only had one observation which makes the results somewhat questionable. To solve this issue the design was changed to a 3 (positive vs negative vs neutral) x 2 (cue-target-relationship: same vs opposite) design. The observations per cell in the 3x2 design varied between seven and 18 (M=11.98, SD=2.35) which makes the results more trustworthy.
Reaction times 3x2
Again, the reaction times of the participants on the target were analyzed with a within-subject
Repeated Measures ANOVA, but this time with valence of the words (positive vs negative vs neutral), and cue-target-relationship (same vs different) – as variables. Just as with the 3x2x2 design, a main
effect on the cue-target relationship was found18. Participants responded faster when the target appeared on the opposite spatial position of the cue (M=286.97, SD=55.13) than when the target appeared at the same spatial position as the cue, M=323.96, SD=65.22, F(1, 31) = 39,63, p<0.0001, ηp2= .56.
In line with the results from the 3x2x2 design, a marginal effect of valence on reaction times was found19. The contrasts revealed that participants reacted faster on targets in trials with negative words (M=302.22, SD=62.10) than in trials with neutral words, M=313.14, SD=61.68, F(1,31)=3.86,
p=0.059, ηp 2
=.11. A post-hoc analysis using a paired sample t-test indicated that participants reacted marginally faster to a target in a trial with a positive word than in a trial with a neutral word, t(31)=-1.80, p=0.082. No difference was found between the reaction times on targets in trials with a positive word and trials with a negative word. Also no interaction effect on reactions times is found between valence and the cue-target relationship20.
Table 11. Means and SD’s of Response Latencies (Ms) per Valence (Positive vs Negative vs Neutral) per Side Cue-Target-relationship (Same vs Opposite).
Valence
Positive Negative Neutral
Cue-Target-Relationship
Mean SD Mean SD Mean SD
Same 317.51 67.84 322.96 69.90 331.40 72.45
Opposite 284.56 58.59 281.47 62.41 294.88 58.22
_________________________________________________________________________________
18 A main effect of cue-target is also found with cut-off points 100ms and 900ms, F(1, 37) = 40.88, p<0.0001, η
p2= .53.
19
There was no marginal effect of valence, but contrasts do reveal a marginal difference on the reaction times between
negative and neutral, F(1, 37) = 3.68, p=0.06, ηp2= .09.
20
Marginal interaction effect between valence and Cue-Target relationship is found with cut-off points 100ms and 900ms,
F(1,37)=3.27, p=0.079, ηp
2
Evaluative Ratings 3x2
A 3x2 repeated measures ANOVA analysis with valence (positive vs negative vs neutral) and cue-target-relationship (same vs opposite) showed similar results as the 3x2x2 Repeated Measures ANOVA. There was a significant main effect of valence21, F(2, 62)=454.35, p<0.0001, ηp2=.94. The contrasts revealed that positive words were rated more positive than negative words, F(1, 31)=547.57,
p<0.0001, ηp2=.95. Negative words were rated more negative than neutral words, F(1,31)=310.94, p<.0001, ηp2=.91. Post-hoc analysis with a paired sample t-test revealed that positive words were rated higher than neutral words, t(31)=21.12, p<0.0001.
The results of the repeated measures ANOVA did not show an interaction effect between valence and the cue-targetrelationship22. Also no difference is found between trials where the cue and target were presented on the same spatial position and trials where the cue and the target were presented on opposite spatial positions.
Table 12. Means and SD’s of Evaluative Ratings per Valence (Positive vs Negative vs Neutral) per Side Cue-Target Relationship (Same vs Opposite).
Valence
Positive Negative Neutral
Cue-Target-Relationship Side Cue-Target relationship
Mean SD Mean SD Mean SD
Same 79.34 9.02 20.94 8.23 56.35 8.85
Opposite 78.62 7.58 21.16 7.26 55.52 5.33
Mean 78.98 7.89 21.05 7.12 55.93 6.53
__________________________________________________________________________________
21 A significant main effect of valence is found with cut-off points 100ms and 900ms, F(1, 37) = 461.26, p<0.0001, η
p2= .93.
22
No interaction is found between valence and cue-target relationship with cut-off points 100ms and 900ms. *Analysis with
cut-off points 100ms and 900ms did show a significant interaction effect between positive vs negative and same vs
Discussion
The results indicate that an Inhibition of Return effect exists in our data, when a target is presented at another location than the previously cued location, people respond faster to that target than when a target is presented at the same location as the previously cued location. This is in line with what previous research has shown and we were able to replicate this pattern of results. However, in contrast to what was expected, no influence of Inhibition of Return was found on the evaluative ratings of the words. This indicates that the inhibition experienced during the reaction time task does not affect people’s evaluation of the words.
Results do show an effect of valence on reaction times. People respond faster to targets coupled with positive and negative words than to targets coupled with neutral words. This can possibly be explained by the fact that neutral words do not have an emotional meaning, like positive and negative words. The emotionality of a stimulus can affect the time it takes for attention to capture it (Carretié, Hinojosa, Martín-Loeches, Mercado, & Tapia, 2004). Attention captures negative stimuli fastest, than positive stimuli, and attention captures neutral (or non-emotional) stimuli slowest. When people need to respond to a target after seeing a stimulus, they probably respond faster to that target when it is coupled with an emotional stimulus than with a non-emotional stimulus, as people will have captured the emotional stimulus faster than the non-emotional stimulus. This can possibly lead to a shorter reaction time when the target is presented with a positive or negative word, than in a situation with a neutral word, as attention captures emotional words faster than non-emotional words.
Furthermore a directional effect, as was shown by Spalek and Hammad (2005) was not found on the reactions times. However, the results did indicate a directional effect on the evaluative ratings of the neutral words in line with the findings of Phaf and Rottelveel (2009). When attention shifts from left to right people seem to evaluate neutral words more positive than when their attention shifts from right to left.
There is a correlation between IOR on reaction times and IOR on the ratings; this effect was specifically strong for reaction times on neutral words and the total ratings of the words. Inhibition of Return is computed as the difference between same, which is when the cue and target are presented at
the same spatial position, and opposite, where the cue and target are presented on opposite spatial positions. This means that the correlation shows that when the difference between reaction times with target on the same spatial position of the cue and the target on the opposite spatial position of the cue increases, the same difference occurs on the ratings. This can mean that when people react faster to a target that is presented on the opposite spatial position, in contrast to the same spatial position as the cue, people evaluate the words more positive in a situation where a target is presented on the same spatial position of the cue, in contrast to the opposite spatial position. This is in contrast with what was expected, as it was expected that a bigger difference in reaction times between the target being
presented on the same spatial position in contrast to the opposite spatial position of the cue, would lead to a more negative rating of the words.
An effect of not-responding, i.e. catchtrials, on the evaluative ratings of words was found. The results indicate that when people are not supposed to respond they have a tendency to evaluate words more negatively. This was also found in previous research of Veling et al. (2008), although they only found the effect on positive stimuli, and not on negative and neutral stimuli. The method used in the present experiment was somewhat different, as it did not show a to-be-ignored stimulus but rather the absence of a target that signaled the participant to restrain from responding.
Both Emotional Contagion, the extent to which someone is influenced by other’s emotions, and the PANAS, to what extent someone is experiencing positive or negative emotions, did not influence the evaluative ratings of the words. Results did show a bigger, but not significant, difference between ‘high-sensitive’ people and ‘low-sensitive’ people on the evaluation of positive and negative words, in contrast to neutral words. This might be due to the fact that neutral words do not have a distinctive valence and are therefore less contagious than words with a positive or negative valence.
Interestingly, no effect of IOR on the evaluative ratings of the words was found. Tim van Lent, conducting the exact same experiment as the present experiment except using face-stimuli instead of word-stimuli, did find an effect of IOR on the evaluative ratings of the face-stimuli. When the target was presented on the opposite spatial position of the cue the participants rated the face-stimuli more positive than when the target was presented on the same spatial position as the cue. It is possible that
attention affects face-stimuli in a different way than word-stimuli. Words have a certain meaning and every person has seen the words that were used in this experiment. In contrast, the participants have never seen the faces that were used in Tim van Lent’s experiment and may therefore be more
susceptible to interpretation and might be easier to be influenced by an inhibition of return effect. Van Weert (2009), using a similar experiment with face- and with word-stimuli, did find an effect of IOR on the evaluative ratings of the negative words and the face-stimuli. This makes it even more
interesting that in the present study no IOR effect on the evaluations of the words was found. The difference between Van Weert’s (2009) experiment and this experiment is that the evaluations at Van Weert’s (2009) were given at the end of all the trials, approximately 20 minutes after the IOR had occurred during the trials. In this experiment the evaluations were given directly after the IOR had occurred, which was expected to result in a stronger effect on the evaluations. A possible explanation for the different findings of Van Weert’s (2009) study and the present study is that there might be an incubation time before IOR can affect the evaluations of words. This would explain why no effect of IOR on the evaluative ratings of words was found in this experiment.
Another possible explanation for the discrepancy between Van Weert’s (2009) results the current experiment’s result might be that in this experiment a 100-point scale was used to measure the evaluations, in contrast to the 7-point scale Van Weert (2009) used. This may have led to different response behavior. The reasoning behind the use of a 100-point scale in contrast to a 7-point scale is to measure a more sensitively affective evaluation. However, it may be more difficult for people to discriminate between for example a 70 and 74 on a 100-point scale, in contrast to for example a 4 and 5 on a 7-point scale. This may have led to people evaluating the words less seriously and more randomly; which would explain the different effects on the evaluative ratings between valences. Research on the impact of scale-size does show that different scale-sizes can give different means on the same question (Dawes, 2008); a 10-point scale can give lower means than 5- or 7-point scales. Even though there has, to the writer’s knowledge, not been any research conducted on the difference between a 7-point scale and a 100-point scale, it can be assumed that scale-size can influence the data. It is possible that because the scale-size varied greatly from Van Weert’s (2009) experiment the results
differed as well. This has to be studied further before any conclusions can be drawn on the exact effect of scale size. A suggestion for further research may be to replicate the present experiment but instead of using the 100-point scale, using the 7-point scale Van Weert (2009) was using.
The current research did find a directional effect of IOR on the evaluative ratings of the words, but not on the reaction times. Previous research from Spalek & Hammad (2004) did find that the direction of the IOR, either from left-to-right or right-to-left, influenced the reaction times of
participants. Western participants, who commonly read from left-to-right, reacted faster when the IOR went from left-to-right than from right-to-left. Right-to-left readers showed the same effect, but opposite. In the current research all the participants were Dutch, meaning they read from left-to-right, so a faster response to left-to-right IOR would be expected. Yet there was not an effect found on the reaction times, but there was an effect of direction on the evaluative ratings. This is in line with previous research from Phaf and Rotteveel (2009), using an experiment with arrows either pointing from left-to-right or from right-to-left. Dutch participants, who read from left-to-right, rated the arrows pointing from left-to-rate more positively than arrows pointing from right-to-left. In the present study it is found that people have a tendency to rate words more positively in a left-to-right attention shift than in a right-to-left attention shift. This might be because it may not necessarily slow people down when their attention shifts from right-to-left instead of left-to-right, but it may feel unusual and
therefore be evaluated more negatively. To see if this effect is influenced by a person’s reading habit it is necessary to conduct further research with people with different reading habits, such as from right-to-left.
The present research did not fulfill all the expectancies, as no influence of IOR on the affective evaluation of words is found, and possibly indicate that more research on this topic is needed to get further insight. The effect of IOR on people’s reaction time is yet again confirmed, people do respond slower when a target is presented on the same spatial position as a cue after a delay, in contrast to a target that is presented at the opposite spatial position of the cue after a delay. However an effect of the IOR effect on the evaluation of word-stimuli has yet to be conclusively found. This research indicates that such an effect of inhibition of return on the evaluation of words is difficult to find and
maybe even absent when measured directly. But further research needs to be conducted to support this conclusion, or discard it. For now this research takes the academic world one step further in exploring the effects of the inhibition of return.
References
Bower, G.H. (1981). Mood and Memory. American Psychologist, 36, 129-148
Britton, B.K., & Tesser, A. (1982). Effects of Prior Knowledge on Use of Cognitive Capacity in Three Complex Cognitive Tasks. Journal of Verbal Learning and Verbal Behavior, 21, 421-436 Carretié, L., Hinojosa, J.A., Martín-Loeches, M., Mercado, F., & Tapia, M. (2004). Automatic
Attention to Emotional Stimuli: Neural Correlates. Human Brain Mapping, 22, 290-299 Dawes, J. (2008). Do data characteristics change according to the number of scale points used? An
experiment using 5-point, 7-point and 10-point scales. International Journal of Market
Research Vol. 50 Issue 1, 61-77
Dittrich, K., & Klauer, K.C. (2012). Does ignoring lead to worse evaluations? A new explanation of the stimulus devaluation effect. Cognition & Emotion, 26, 193-208
Doherty, R. W. (1997). The emotional contagion scale: A measure of individual differences. Journal
of nonverbal Behavior, 21(2), 131-154
Eder, A.B., & Rothermund, K. (2008). When Do Motor Behaviors (Mis)Match Affective Stimuli? An Evaluative Coding View of Approach and Avoidance Reactions. Journal of Experimental
Psychology, 137, 262-281
Fenske, M.J., Raymond, J.E., Kessler, K., Westoby, N., & Tipper, S.P. (2005). Attentional Inhibition Has Social-Emotional Consequences for Unfamiliar Faces. American Psychological Society,
16, 753-758
Folk, C.L., Remington, R.W., & Johnston, J.C. (1992). Involuntary Covert Orienting Is Contingent on Attentional Control Settings. Journal of Experimental Psychology, 18, 1030-1044
Forgas, J.P., Bower, G.H., & Krantz, S.E. (1984). The Influence of Mood on Perceptions of Social Interactions. Journal of Experimental Social Psychology, 20, 497-513
Fredrickson, B. L., & Branigan, C. (2005). Positive emotions broaden the scope of attention and thought‐action repertoires. Cognition & emotion, 19(3), 313-332
Kastner S, & Ungerleider L.G. (2000). Mechanisms of visual attention in the human cortex. Annu Rev
Klein, R.M. (2000). Inhibition of return. Trends in Cognitive Sciences, 4, 138– 147
Lazarus, R.S. (1982). Thoughts on the Relations Between Emotion and Cognition. American
Psychologist,37, 1019-1024
Posner, M.I. & Cohen, Y. (1984). Components of visual orienting. In Attention and Performance Vol.
X (Bouma, H. and Bouwhuis, D., eds), 531–556, Erlbaum
Phaf, R.H., & Rotteveel, M. (2009). Looking at the Bright Side: The Affective Monitoring of Direction. Emtion, 9, 729-733
Raymond, J.E., Fenske, M.J., & Tavassoli, N.T. (2003). Selective Attention Determines Emotional Responses to Novel Visual Stimuli. Psychological Science, 14, 537-542
Spalek, T.M., & Hammad, S. (2004). Supporting the attentional momentum view of IOR: Is attention biased to go right?. Perception & Psychophysics, 66, 219-233
Spalek, T.M., & Hammad, S. (2005). The Left-to-Right Bias in Inhibition of Return is Due to the Direction of Reading. Psychological Science, 16, 15-18
Tipper, S.P., Grison, S., & Kessler, K. (2003). Long-Term Inhibition of Return of Attention.
Psychological Science, 14, 19-25
Veling, H., Holland, R.W., & Van Knippenberg, A. (2008). When approach motivation and behavioral inhibition collide: Behavior regulation through stimulus devaluation. Journal of Experimental
Social Psychology, 44, 1013-1019
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: the PANAS scales. Journal of personality and social psychology,
54(6), 1063
Zajonc, R.B. (1980). Feeling and Thinking: Preferences Need No Inferences. American Psychologist,
35, 151-175
Zeelenberg, R., Wagenmakers, E., & Rotteveel, M. (2006). The Impact of Emotion on Perception: Bias or Enhanced Processing? Psychological Science, Vol. 17, No. 4, pp. 287-291
APPENDIX I: Analysis with Different Outlier Detection Observations lower than 100ms and higher than 900ms were excluded
Reaction Times
Table 1. Means of Reaction Times (Ms) per Valence (Positive vs Negative vs Neutral) per Side Cue (Left vs Right) vs Side Cue-Target Relationship (Opposite vs Same).
Valence
Positive Negative Neutral
Side Cue Side Cue-Target relationship
Same Opposite Same Opposite Same Opposite Left M=309.66 SD=56.64 M=294.83 SD=66.18 M=321.63 SD=71.99 M=287.76 SD=54.74 M=333.55 SD=64.79 M=305.40 SD=85.45 Right M=324.57 SD=75.06 M=281.71 SD=64.66 M=330.68 SD=73.73 M=281.46 SD=57.16 M=327.22 SD=74.60 M=289.92 SD=63.19 Average M=317.13 SD=60.24 M=288.27 SD=53.51 M=326.15 SD=63.79 M=284.61 SD=52.21 M=330.38 SD=65.03 M=297.66 SD=68.19 M=302.69 SD=52.85 M=305.38 SD=54.79 M=314.02 SD=61.65
Table 2. Means of Reaction Times (Ms) for Side Cue (Left vs Right) vs Side Cue-Target Relationship (Same vs Opposite).
Side Cue-Target Relationship
Side Cue Same Opposite
M SD M SD
Left 321.61 55.95 296.00 59.38
Right 327.49 68.17 284.37 54.28
Evaluative Ratings
Table 3. Means of Evaluative Ratings per Valence (Positive vs Negative vs Neutral) per Side Cue (Left vs Right) vs Side Cue-Target Relationship (Opposite vs Same).
Valence
Positive Negative Neutral
Side Cue Side Cue-Target relationship
Same Opposite Same Opposite Same Opposite Left M=80.24 SD=11.42 M=79.08 SD=9.41 M=20.31 SD=10.69 M=20.01 SD=8.94 M=55.85 SD=10.20 M=57.51 SD=8.03 Right M=80.36 SD=8.77 M=80.40 SD=9.24 M=17.91 SD=9.76 M=18.80 SD=10.04 M=55.35 SD=10.21 M=53.57 SD=6.70 Average M=80.30 SD=9.40 M=79.74 SD=8.40 M=19.12 SD=9.28 M=19.40 SD=9.01 M=55.60 SD=9.33 M=55.54 SD=5.67
Table 4. Means and SD’s of Evaluative Ratings for Valence (Positive vs Negative vs Neutral) per Side Cue (Left vs Right).
Valence
Side Cue Positive Negative Neutral
M SD M SD M SD
Left 79.66 9.57 20.16 9.06 56.68 7.93
Right 80.38 8.14 18.35 9.28 54.46 7.79
Average 80.02 8.41 19.26 8.68 55.57 6.94
