• No results found

The effect of different sounds on selective attention at different stress levels

N/A
N/A
Protected

Academic year: 2021

Share "The effect of different sounds on selective attention at different stress levels"

Copied!
17
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

The effect of different sounds on selective attention at

different stress levels

Abstract

Every moment of the day the human senses are exposed to different stimuli with attention being key to filter out the most relevant stimuli from the background. Attention operates in two systems and in our study we will focus on the top-down system which is responsible for selective attention. Different forms of acute stress, including mild, moderate and severe stress have been shown to influence certain cognitive abilities. In addition to this, sounds and music have been shown to influence the performance and attention during cognitive tasks. The proposed study will investigate the effect of different sounds on selective

attention at different stress levels by measuring the theta, alpha, beta and gamma frequency waves of the brain, which are affected by selective attention. The performance and

theta/beta ratio of the Flanker task and Mouse Tracking task will be used as well to indicate an alteration in the electroencephalograms. It is expected that music will increase the selective attention, especially in test subjects experiencing acute severe stress. The results from this study can contribute to a better understanding of selective attention, which in turn may lead to a reduction of symptoms as seen in attention-deficit hyperactivity disorder and attention-deficit disorder.

Summary

In our research, we would like to investigate the possible effects of different sounds, such as music, nature and environmental sounds, on attention. The focus will be primarily on

selective attention, which is the ability to filter out the irrelevant stimuli in the environment in order to select the most relevant stimulus. In addition to this, the effects of the sounds will be analysed at different stress levels, which could give a better understanding of the role stress plays during selective attention and how these sounds could possibly affect it.

Keywords

Selective attention; Sound; Music; Stress (hypothalamus-pituitary-adrenal axis);

Electroencephalograms (EEG); Trier Social Stress Test (TSST); Flanker task; Mouse Tracking task

Arjun Persad (11713887)

Bachelor Thesis (research proposal)

(2)

2

Research topic

General background

The human brain is constantly exposed to sensory information in the environment. In order to process the most relevant information and filtering out the irrelevant information

attention is needed. Attention influences the perception and increases the awareness by shifting between the most relevant sensory information (Katsuki & Constantinidis, 2014). This cognitive process makes sure that the senses do not get overwhelmed by the sounds, sights and smells from the surroundings. The two mechanisms involved with attention are the bottom-up (exogenous) and the top-down (endogenous) systems which operate in a stimulus-driven or a goal-oriented way respectively. These two mechanisms are in a push-pull competition with each other for the attentional resources (Huang & Elhilali, 2020). The bottom-up system is often described as a competition between stimuli in the environment of which the most salient stimulus is selected and processed. Depending on the nature of the stimuli, different regions in the brain are activated or supressed. For example, visual information is sent through the dorsal and ventral visual cortical pathways and processed in the prefrontal cortex (PFC) and posterior parietal cortex (PPC) in both attention mechanisms. The top-down system interacts and works simultaneous with its counterpart, however there are certain differences in the processing which result in the ability to have selective

attention (Katsuki & Constantinidis, 2014).

In this research proposal, the focus will be on the goal oriented attention mechanism, since it will give more insights in the active control of attention by the brain. Selective attention will be indicated with the use of electroencephalography (EEG) during selective attention tasks, which will reveal the electrical activity in the brain of the top-down system. Changes in selective attention can be reflected by the neural oscillatory activity in the brain in the different frequency bands, such as the theta (4-7 Hz), alpha (8-12 Hz), beta (13-28 Hz) and gamma (29-70 Hz) bands. The theta/beta ratio (TBR) is often used to indicate attentional control because of its negative correlation, suggesting that as the TBR increases, attentional control decreases and vice versa (Putman et al., 2010; Angelidis et al., 2016). In the study of Putman et al. (2010) it was discovered that this ratio is more and stronger prevalent in the PFC than in the more posterior areas. Alpha (8-12 Hz) modulation also plays an important role in the top-down mechanism of attention. When the alpha frequency increases in power, information processing by the top-down system is reduced and the bottom-up system is primarily used. That is why the brain actively supresses the alpha band during tasks that require selective attention, such as mathematical, writing and reading tasks. The alpha suppression increases as the task becomes more complex, which is mostly found in the fronto-central and the parieto-occipital regions (Magosso et al., 2019). Another important predictor for selective attention is the gamma band (29-70 Hz), which is enhanced in the dorsal fronto-parietal regions during complex auditory, visual and somatosensory tasks (Huang & Elhilali, 2020).

There are many ways attention can be influenced positively or negatively with stress being a major factor. Although stress has multiple effects on the human body, for example increased blood pressure and heart rate (Larra et al., 2016), this study will be limited to the effects of acute stress on selective attention. Physical stress and psychological stress are two different types of stress, caused by physical or social threat respectively. Both of these stress types initiate the immediate release of dopamine and noradrenaline which will result in increased activity of the amygdala and a decrease in PFC activation, favouring the bottom-up system

(3)

3

over the top-down system (Sänger et al., 2014; Qi et al., 2018). In addition to this, there is a delayed response in the activation of the hypothalamus-pituitary-adrenal axis, which will result in the production and secretion of cortisol in the adrenal cortex. Once the cortisol crosses over the blood brain barrier it will also alter the PFC activity after binding to its glucocorticoid receptors (Larra et al., 2016). The severity of stress is often quantified by the concentration of cortisol in the blood or saliva, with the cortisol levels being less than 1.5 times in mild stress, 2-3 times higher in moderate stress and three or more times higher in severe stress, relative to the state of no stress (Shields et al., 2019). Previous research has already shown that acute moderate and acute severe stress impair the selective attention, however, it is yet unclear what the effect of mild stress is on selective attention. Several studies have concluded that acute mild stress improves the response speed during

attentional tasks, which could be attributed to the enhancement of the top-down system (Qi

et al., 2018; Shields et al., 2019). Response speed of a task can thus be used as an indicator

of selective attention at different stress levels, which will be implemented in the proposed study.

As mentioned before, attention can be influenced by different factors, however, it is not yet fully understood how sound can affect the top-down system. Music has been shown to enhance cognitive performance during attentional tasks and influence the emotional state depending on the category of music. Fernandez et al. (2019) suggested that attention on visual information is increased when a beat in music occurs, although the exact mechanism has yet to be discovered. It is still unclear whether the effects depend on the nature of the music, such as tempo, or on the changes in the emotional state when exposed to music. In their research they additionally concluded that high-arousing and joyful music was

associated with the enhancement of the top-down system (Fernandez et al., 2019). An example of such music is elevator music which will be used in the proposed study. The modulation of the beta wave is increased in musicians compared to non-musicians during a working memory task, suggesting that music could have the same effect during a selective attention task since musicians are exposed to music on a daily basis (Hsu et al., 2017). Based on this knowledge about the oscillatory changes during the top-down system and the possible effects of stress and music on selective attention, a pilot study was conducted in order to check the feasibility of a protocol and to investigate the different frequency waves and performances of test subjects during an attention task preceded by music, nature or environmental sound.

The “NS stationslab” pilot study

In 2019 under supervision of Dr. A.B. Mulder, a pilot study called “NS stationslab” was performed to investigate the effects of different kind of sounds, including music, nature sounds and environmental sounds, on attention with test subjects at different pre-existing stress levels. The experimental design and the data were used to explore the possibilities of indicating selective attention and to test and improve the method for the eventual proposed experiment. The pilot experiment was performed at a NS train station with 177 travellers, ranging from the age of 4 to 82 (mean= 36 and standard deviation= 19), who were asked to participate in the tests. First, the participants were given a unique number and were asked to fill in a survey, indicating their subject number, age, sex and the level of stress that was experienced at the time on a scale from 1 to 5 with one being the lowest amount of stress experienced. Thirty-nine out of 177 participants were used to record the EEG and to perform the attention task. The attention of the test subjects was indicated by using EEG with an

(4)

4

electrode on the forehead measuring the activity of the prefrontal cortex while performing an attention task. The use of only one active electrode was chosen instead of an EEG cap with multiple electrodes, because of the hygiene risks and the relative easy setup for at a train station. A ground electrode on the forehead (next to the active electrode) and a reference electrode on one of the ears were used besides the active electrode on the forehead.

At the start of the task, instructions were given to the test subjects to close their eyes and to listen to music, nature sound or the environmental sounds at the time, for 90 seconds. Elevator music was used as music and the nature sounds were samples of chirping birds and the blowing of the wind. When the sound ended they got instructed to open their eyes and to read the instructions for the different questions they had to complete, which included ten math questions (e.g. “800+366=”), five rhyming questions (e.g. “Rhyme three words with

waiting” ) and five series completion questions (e.g. “Complete the series: 1, -2, -5, -8,…”).

All the instructions and questions were given in Dutch in the program Stimulus-Presenter and the EEG recorder started with recording from the moment the instructions for the sound were given, as displayed in figure 1. Both software programs ran in a MATLAB environment. The instructions for the questions were then followed by 20 questions that the test subjects had to type the answer to with no time limit and no pause between the questions. The questions were in a randomised order, however no question of the same category

(math/rhyme/series) appeared twice in a row. Stimulus-Presenter saved all the answers and the duration of each question in excel files. The top-down mechanism of attention was mainly activated during the task, since the test subjects were instructed to only focus on answering the questions in a goal-oriented manner.

Figure 1. Flow chart of the attention task. The EEG of the whole task including the instructions and listening to the sound was recorded

(shown with the dotted arrow). The duration of the sound was 90 seconds (filled arrow) and the duration of the instructions and questions depended on the reaction/typing speed of the test subjects. An example of a mathematical question is displayed in the figure.

Preliminary analysis

The raw data from the “NS stationslab” was received from Dr. A. B. Mulder. For the performance and EEG data-analysis, only the test subjects were included who met the following criteria: EEG data in MATLAB, no duplication of data in different excel files,

complete information in the survey and task data in excel with the task answers. Twenty-two test subjects met the criteria and were divided into three main groups with the different sound categories. Elevator music was played for the music group (M), nature sounds were played for the nature group (N) and the environmental group (O) was exposed to the sound of the environment at the time. Each of these main-groups had sub-groups for the different levels of stress that was experienced at the start of the experiment on a scale from 1 to 5, with one experiencing no stress and five being the most severe. Group music contained seven test subjects, group nature nine test subjects and group environmental six test subjects. Table 1 shows the amount of test subjects in the subgroups within each main

(5)

5

group. The age of all test subjects ranged from 16-69 years old (mean= 32 and standard deviation= 15).

Stress

level Music group (M) Nature group (N) Environment group (O) Total

1 4 4 3 11

2 2 4 2 8

3 1 1 1 3

4 0 0 0 0

5 0 0 0 0

Table 1. The test subjects per subgroup. The main groups were merged in the last column of total amount of test subjects per stress level.

Stress level 1 was considered the lowest form of stress and stress level 5 the most severe form of stress experienced at the start of the survey.

Performance data-analysis

The performance of the test subjects, which included the duration and the score of the 20 questions, was analysed to investigate any differences in performance between the three main groups as well as to investigate any differences between the different stress levels. A choice was made regarding the difference between the stress levels, based on an insufficient amount of test subjects having a stress level of three or more, to merge the stress levels two and three to even out the number of test subjects within the two categories.

A Shapiro-Wilk normality test was used in R-studio to check if the data containing the duration in seconds of the 20 questions was normally distributed. All the data across the different groups was normally distributed. The one-way analysis of variance (one-way ANOVA) was executed to test for significance (p<0.05) for the duration in the three sound groups (n=22). The one-way ANOVA was used again to test for significance for the duration between the stress levels (stress level 1 versus 2-3), containing each 11 test subjects (n=22). The same groups were used again to test significance of the score in percentage of the twenty questions between the main groups and the different stress levels. The answer for each question had to be completely correct in order to be graded as a correct answer and for every correct answer five percent was assigned towards the test subjects score (100% divided by 20 questions = 5%/correct question). The data of the scores in percentage was not normally distributed in each group after using the Shapiro-Wilk normality test in R-studio. A Kruskal-Wallis test was applied to test for significance (p<0.05) for the score in the different sound groups (n=22) and for the duration between the stress levels (stress level 1 and 2-3), containing each 11 test subjects (n=22).

EEG analysis

In the first step of the EEG data-analysis, the EEG data was cut from the start of the first question to the end of the last question. A band-pass filter was used at 0.5 to 48 Hz to exclude the main power. In order to remove artefacts by visual detection, the data was segmented in trials of five seconds and all trials containing an amplitude lower than 100 µV and with a value within the mean plus or minus two standard deviations were kept. Test subjects with less than 15 trials left were considered as invalid to use for a power spectrum analysis. Twenty-one of the 22 test subjects had less than 15 trials left, resulting in an insufficient amount of trials to conduct a power analysis for the different groups. A time

(6)

6

frequency analysis was performed on one test subject containing 18 trials, using a frequency range of 0 Hz to 48 Hz. The power over the trials was averaged and the power of the

different frequency ranges corresponding with the theta (4-7 Hz), alpha (8-12 Hz), low beta (13-20 Hz) and high beta (21-28 Hz) waves were calculated. For the TBR, the average power over the beta wave was divided over the average power of the theta wave.

Preliminary results

Experienced stress levels from the surveys

The surveys at the start of the experiments showed that most of the test subjects (83 out of 177) experienced stress level 1, followed by stress level 2 (53 out of 177) and stress level 3 (35 out of 177) as displayed in figure 2. Only five test subjects had experienced stress level 4 and one had experienced stress level 5. None of the test subjects experiencing the most severe stress levels (stress level 4 and 5) met the inclusion criteria for the performance and EEG data-analysis (see figure 2: stress level 4 and 5).

Figure 2. The stress levels of the test subjects. The figure shows the experienced stress levels of the test subject as given in the survey

before the task. The blue bar illustrates the total amount of test subjects per stress level (1-5) and the red bar is stacked upon the blue bar illustrates the stress levels of the test subjects that met the inclusion criteria. The number of test subjects is displayed on the y-axis (total n= 177). The stress levels 4 and 5 had no test subjects after the exclusion.

The excluded test subjects

From the 39 test subjects who participated in the attention task, 17 were excluded before the performance and EEG data-analysis, with three test subjects not having excel data, five not having filled in the survey, five having incomplete EEG data (absent markers), two having duplicated data in different excel files and two having disturbed EEG data.

The duration of the task compared to the different sounds and stress levels

The one-way ANOVA test was used to discover any significant difference in duration (s) of the task between the three sound groups. Figure 3 shows the duration (s) of the task per sound group with the average duration of each group (n=22). No significant difference was found between the sound groups (F = 2.47, df = 2 and p-value = 0.118).

The same tests were used again to investigate any difference between the stress levels 1 and 2-3. In figure 4 the duration (s) of the task is displayed with n=22. The duration of the stress

(7)

7

levels were also not significantly different from each other (F = 1.78, df = 1 and p-value = 0.197).

Figure 3. The duration of the task per sound group. The music group contained seven test subjects, the nature group contained nine and

the environment group contained six test subjects (n=22). No significant difference was found, with F = 2.47 and p-value = 0.118 (df=2).

Figure 4. The duration of the task per stress level. Each stress level group contained 11 test subjects (n=22). No significant difference was

found, with F = 1.78 and p-value = 0.197 (df=1). Stress level two and three were merged because of the insufficient amount of test subjects with stress level three or higher.

The score of the task compared to the different sounds and stress levels

The Kruskal-Wallis test was used to discover any difference in score (%) of the task between the three main groups as seen in figure 5. The score is the percentage of correct questions (5% per correct answer). The means of the groups were between the 93-98 percent and one outlier was detected in the music group. No significant difference was found between the sound groups (p-value = 0.746; Chi-squared = 3.488).

(8)

8

There was also no significant difference between stress level 1 and stress level 2-3 (p-value = 0.698; Chi-squared = 2.073). Figure 6 shows the score (%) of the task with 11 test subjects in each group (n=22).

Figure 5. The score of the task per sound group. The score is the percentage of correct answers out of the 20 questions (5% per correct

answer). The y-axis has a broken line for visual presentation purposes. The small circle in the boxplot displays an outlier. The music group contained seven test subjects, the nature group contained nine and the environment group contained six test subjects (n=22). No significant difference was found, with p-value = 0.746 and chi-squared = 3.488.

Figure 6. The score of the task per stress level. The score is the percentage of correct answers out of the 20 questions (5% per correct

answer). Each stress level group contained 11 test subjects (n=22). The y-axis has a broken line for visual presentation purposes. The small circle in the boxplot displays an outlier. No significant difference was found, with p-value = 0.698 and chi-squared = 2.073. Stress level two and three were merged because of the insufficient amount of test subjects with stress level three.

Power analysis of the frequency bands

After the artefact removal of each test subject, 6 test subjects had 0 trials left, 11 had between 1-3 trials left and 5 had between 5-9 trials left. One test subject containing the most usable trials (18 trials) was used for measuring the theta (4-7 Hz), alpha (8-12 Hz), low beta (13-20 Hz) and high beta (21-28 Hz) waves. The average power over the frequencies, including the TBR, are displayed in table 2. The closer the number is to zero the higher the

(9)

9

power is, for example the theta frequency has with -11.55 the highest power of the tested frequencies. The average power for the beta wave is -12.31 and -11.83 for the alpha wave. The TBR of the average powers is 0.9386.

Test subject ID Theta (4-7 Hz) Alpha (8-12 Hz) Low beta (13-20 Hz) High beta (21-28Hz) Theta/beta ratio

PP33 -11.5511 -11.8262 -12.1471 -12.4655 0.938633

Table 2. The average power of one test subject. This test subject contained 18 trials with no artefacts and was used for the time frequency

analysis. The average power was calculated for the theta (4-7 Hz), alpha (8-12 Hz), low beta (13-20) and beta (21-28 Hz) waves. The theta/beta ratio was calculated by dividing the average theta power over the average beta power (-12.3063). A higher power correlates with a number closer to zero.

Conclusion and discussion

First of all, it stood out that the most common form of experienced stress amongst the test subjects was mild stress, because of the frequently indicated stress level 1 and 2 from the surveys. It also has to be mentioned that only 6 out of the 177 participants experienced a stress level of four or five when filling in the survey. There was also no difference in duration or score between the music group, nature group and sound group or between the stress levels 1 and 2-3. Despite the high amount of artefacts detected in the EEG data, it was still possible to measure the average power over the theta, alpha, low beta and high beta frequency waves of one of the test subjects.

The stress factor used in this pilot study was subjective to the participants because in the survey, they had to fill in the stress level on a scale from 1 to 5 (1=mild and 5=most severe) based on what was experienced at the moment. This was done without testing any

biomarkers for stress, such as cortisol concentration, hearth rate variability or blood pressure, which resulted in a highly biased indicator of stress. The lack of participants that experienced a stress level of 4 or 5 also made it impossible to compare the different stress levels with each other. In order to prevent this from happening again in the future

experiment, it will be a better option to induce stress at different levels on the test subjects and to measure the biomarkers to categorise them appropriately.

Another important factor to mention is the exposure to the different sounds before the task instead of during the task, meaning that every group was exposed to the same

environmental sounds while answering the questions. This could explain why there was no difference in duration or score between all the tested groups. Investigating the effects of the sounds during the task will be key in the proposed experiment to discover any changes in duration, score or brain activity.

When the EEG data was being analysed, an excessive amount of artefacts were detected which resulted in the removal of most of the test subjects for the time frequency analysis. The abundance of artefacts could be motor artefacts due to typing of the answers which could disturb the EEG recording, the use of only a single electrode on the PFC, excessive eye movements and the distracting environment. Although one test subject was still used for the analysis, the data was still not entirely clean or sufficient to make any conclusions regarding selective attention. There was also no baseline to compare the brain activity with.

The proposed study

With the knowledge from the pilot study and the general background the method can be expanded and improved to investigate the effects of different sounds

(music/nature/environmental sounds) on selective attention at different experienced stress levels. In order to investigate possible alterations in selective attention, multiple frequency bands including the theta, alpha, beta and also gamma bands will be analysed to determine

(10)

10

the effects of the different sounds. The average power of each frequency band will be measured, as well as calculating the TBR, the score and the duration of the task. All these parameters of selective attention will additionally be tested at induced acute mild, moderate and severe stress to investigate possible effects of the sounds on acute stress, which could cause an increase in selective attention. The hypothesis is, therefore, that music will improve selective attention during the task, especially in test subjects with a severe acute stress level. It is expected that the TBR will be lower in the brain regions while exposed to music during the task, suggesting that the beta wave increases and/or theta wave decreases as a result of enhanced selective attention. This is thought because as mentioned before, musicians have shown to have increased beta wave modulation during certain tasks while listening to music (Hsu et al., 2017) and the TBR decreases (primarily in the PFC) as the task becomes more complex (Angelidis et al., 2016). Alpha suppression in the fronto-central and the parieto-occipital regions as well as the gamma wave in the dorsal fronto-parietal will be more enhanced under the influence of music according to the expectations, both favouring the top-down over the bottom-up system during the processing of information. Finally, a higher score and a shorter duration of the task is expected in the music group, since certain

categories of music enhance the cognitive performance during attentional tasks (Fernandez

et al., 2019). As for the stress factor, a greater difference in selective attention as result of

music is predicted in the test subjects with severe stress compared to the moderate stress group and mostly the mild stress group. This is in line with the previous research stating that acute moderate and severe stress impair selective attention and mild stress improves the performance (duration) during attentional tasks (Qi et al., 2018; Shields et al., 2019). In the proposed experiment, the method will be improved and expanded upon based on the experience of the pilot study and other studies. First and foremost, the most important change in the proposed method will be the exposure of the sound throughout the attention tasks. The task used in the pilot study will also be replaced with the Flanker test and the Mouse Tracking task (Shields et al., 2019), which both require selective attention to perform and also enables the test subjects to answer without typing, resulting in less motor related artefacts in the EEG data. Another way to minimise the amount of artefacts will be the use of an EEG cap with multiple electrodes measuring the PFC, the dorsal fronto-parietal, the fronto-central and the parieto-occipital regions (F1-2, FC1-2, C1-2 and PO1-2). The main improvements regarding the stress factor will be implemented by using the Trier Social Stress Test (TSST)(Labuschagne et al., 2019) to induce stress in test subjects and by

measuring the cortisol concentration to give a more objective indication for the stress levels (mild/moderate/severe instead of level 1/2/3/4/5). The TSST has been proven to reliably induce moderate stress in 80% of test subjects with a two-to-three fold increase in cortisol concentration, meaning that the TSST will be slightly modified to induce other stress levels (mild/severe)(Allen et al., 2017). The survey will also be modified to screen for any potential musicians, which should be excluded from the analysis because of the known effects of music on beta modulation and cognitive performances in musicians versus non-musicians. At last, a baseline will be measured to compare the possible alterations in frequency waves in the different conditions of sound (music/nature/environment) and no sound. The Appendix displays the risk assessment of the possible complications that may occur in the procedure with the possible corresponding solutions.

(11)

11

Scientific and social impact

It is important to mention that, to our knowledge, this will be the first study combining these different methods (TSST, Flanker task and Mouse tracking task) to test the effect of sound on selective attention at different induced levels of stress, as well as using the theta, alpha, beta and gamma frequency waves to discover any changes in selective attention. By investigating the effects of different sounds (music/nature/environmental sounds) on selective attention at different experienced stress levels, knowledge will be gained of the possible effects of sound on the brain frequencies (theta, alpha, beta, gamma) and the TBR during selective attention tasks, how the top-down mechanism operates on a EEG level, the effects of sounds on selective attention when under different levels of stress (mild/moderate/severe) and how different methods (TSST, Flanker task and Mouse tracking task) can be effectively combined or modified to test the influence of stress on selective attention. This knowledge can be the foundation for further studies to get more insight in the process of attention, which in turn may lead to reduction of symptoms as seen in attention-deficit hyperactivity disorder (ADHD) and attention-deficit disorder (ADD). In addition to this, sound can possibly be used to increase selective attention when it is needed, for example during an exam.

Approach

Test subjects and survey

The test subjects will be recruited in cooperation with NS at a train station, where travellers will be asked to participate in the experiments. The pilot study has shown that there were 175 participants with a wide age range (5-85 years) and with an almost even distribution of male or female. The amount of test subjects and the heterogeneity of the groups is expected to be almost the same when gathering test subjects for the experiments. In order to create a baseline (no sound) for the possible effects of sound, all of the test subjects will do both the TSST (inducing the same stress level both times per test subject) and the selective attention tasks (2x Flanker task and 2x Mouse Tracking task) twice in a week with two days in

between. The sequence of the conditions no sound and sound will be switched for half of the participants to avoid the factors of already having experienced the task and fatigue of the test subjects. Figure 7 illustrates the procedure of a test subject on the first day.

Figure 7. The procedure a test subject will undergo on one day. The time (in minutes) is displayed under the different steps as well as the

moments the cortisol samples will be taken. The TSST will start from the moment of acclimatisation and end after the debriefing. The EEG of both the Flanker task and Mouse Tracker task will be recorded. Test subjects will get the same sound (M/N/E) in both tasks, unless in the no sound condition. * The survey will only have to be completed the first time and only the ID of the test subject is required **the second day of the experiment will start with the acclimatisation step and will follow the second condition (no sound or sound)

(12)

12

The experiment will start between 2pm-4pm to avoid any elevated cortisol levels as a result of a meal or drinks (Labuschagne et al., 2019). When the participants arrive, the

experimenter will assign an identification number (ID) to each participant and instruct them to fill in a survey indicating their age, ID, sex and experience with music (options are:

musician/ex-musician or non-musician: listen to music for more/less than an hour every day). The test subjects will be given ten minutes after arrival to complete the survey and to acclimatise, which could reduce pre-existing acute stress.

The Trier Social Stress Test

After the surveys are turned in (10 minutes after arrival), the participants will be given a glucose drink, to factor in the inter-individual variation in baseline blood glucose levels, followed by a glass of water (100 mL) to rinse their mouth. Saliva samples from passive drool will be taken before the TSST and stored in a freezer (at -20℃) to measure a baseline for the cortisol concentration. The experimenter will then lead the participants into a room (panel room) with a panel of two people where instructions will be given (read aloud from a paper by the panel around 5 minutes) to prepare a speech for a mock job interview and that it will be filmed for further review and comparison with other candidates. They will get five

minutes of preparation time in the waiting room, which would be referred to as anticipatory stress period. One man and one woman wearing professional clothing and formal shoes will be the members of the panel. Throughout the tasks, the panel will remain neutral and non-responsive towards the test subjects. The experimenter will lead the test subject back into the panel room where the researcher starts the video recording while in view of the test subject, to create a social evaluative component. The five minute speech will then start and the experimenter will leave the panel room. At the end of the five minutes, one of the panel members will present the test subject to a surprise arithmetic task (unexpected stress component) which will have to be completed within five minutes. The arithmetic task will exist of relative easy to difficult math questions (e.g. 45/9 or 1166+37). After the arithmetic task, the test subject is returned to the waiting room where the experimenter will be giving a debriefing and answer any question the test subject might have (debriefing and recovery period of 10 minutes).

Saliva samples from passive drool will be taken again to measure the cortisol concentrations ten and 15 minutes after the surprise arithmetic task, after which these are immediately stored in a freezer (-20℃). The cortisol concentration (nmol/L) of the samples before and after the TSST will be analysed by conducting cortisol assays with the use of ELISA kits. In order to induce different levels of acute stress (mild/moderate/severe/control) the TSST will be slightly modified in the following ways: (1) To create mild stress, the test subjects will also be introduced to the arithmetic task when the experimenter instruct them for the first time and the panel will give more positive feedback (e.g. giving complements with correct answers), which will eliminate the surprise factor and reduce the social evaluation stress; (2) To create severe stress, the test subjects will be told that the video recording of the tasks is livestreamed in front of a judges panel (in addition to the panel on location) and the panel will give some negative feedback (e.g. comparing with other participants who performed better or saying when an answer is wrong), which will increase the social evaluation and uncontrollability threat; (3) The arithmetic task will have more difficult questions for the severe stress group and more easier questions for the mild stress group; (4) The test subjects for the control group will follow the same procedure as the other groups, but with the

(13)

13

uncontrollability threat. Instead of an arithmetic test, the test subjects will be asked to stay in the panel room for five minutes.

According to the concentration of cortisol and variant of the TSST (initial label), the test subjects will be labelled with mild stress (<1.5x the baseline cortisol concentration), moderate stress (2-3x the baseline), severe stress (>3x the baseline) and control group (no stressor and no change relative to the baseline). The test subjects will be randomly assigned to a sound group based on their initial stress label (variant of the TSST), so that the music group, nature group and environment group will contain an approximately equal amount of test subjects with different stress levels (mild/moderate/severe/control).

The selective attention tasks

Following the last cortisol sample, the test subject will be asked to sit in an isolated room in front of a computer screen where the experimenter will place an EEG cap with electrodes on the F1, F2, FC1, FC2, C1, C2 PO1 and PO2 positons, on the head of the subject. A conductive gel will be used to optimise the measurements of the EEG. On the screen of the computer will be the instructions (in Dutch) that the test subject has to sit as still as possible, minimise blinking of the eyes and to only speak in case of an emergency. After pressing “enter” on the keyboard, the instructions for the Flanker task (presented in Stimulus Presenter) will be given, stating that the participants have to press the right or the left arrow key on the keyboard depending on the direction of the arrow in the centre of the screen and that the centre arrow can be flanked on both sides with two arrows pointing in the same direction (congruent), opposite direction (incongruent) or flanked with nothing (neutral). Further instructions were given to ignore the flanker arrows and to respond with the same arrow key as the centre arrow as accurately and quickly as possible (Figure 8 (a)). The EEG recorder in

MATLAB will be started by the experimenter after the test subject presses “enter” to start 20

practice trials. In addition to this, elevator music, nature sounds (samples of chirping birds and blowing wind) or no sound (E) will be played from a speaker depending on the condition (sound or no sound). The speaker will be placed directly behind (3 meters) the test subject and the sounds will be played at the same volume level across all sound groups.

The Flanker task will be divided into four blocks containing each 130 trials (520 total) with 52 congruent trials (208 total), 52 incongruent trials (208 total) and 26 neutral trials (104 total). There will be an even number of trials with the centre arrow pointing left or right (260 left and 260 right) and after each completed block the test subjects will be presented with their progress. The participants can pause at the score screen for as long as they want, after which they will have to press “enter” to continue with the next block of trials. A single trial consist of 500 ms display of a fixation cross in the centre of the screen, followed by the centre arrow and the flanking arrows (dependent on in-/congruent or neutral trial) until an answer is submitted or the time limit of 2000 ms is reached.

After completion of the Flanker task, the experimenter will set up the Mouse Tracker task in

Stimulus Presenter. The test subject can press “enter” when ready and the instructions for

the Mouse Tracker task will be presented on screen. The instructions will state that the participants have to click the same letter (A/B) in the centre as fast and as accurate as possible in the upper-left or upper-right corner from a string of five letters, which could be a congruent (“AAAAA” or “BBBBB”) or a incongruent (“BBABB” or “AABAA”) trial (Figure 8 (b)). When the box at the bottom (centre) of the screen is clicked eight practice trials will be presented, followed by 48 trials (56 trials total) with 24 congruent and 24 incongruent trials.

(14)

14

(a) (b)

Figure 8. Examples of the Flanker task (a) and the Mouse Tracker task trials. The flanker task trial (a) displays an incongruent trial where the

right arrow key has to be pressed to be correct. The Mouse Tracker task trial (b) displays a congruent trial where the box in upper-left corner has to be clicked with the mouse to be correct.

Furthermore, 12 of the congruent and incongruent trials will have the letter “A” in the centre and the other half of the trials will have the letter “B” in the centre. The letter “A” will be in the upper-left corner and the letter “B” in the upper-right corner in half of all test subjects, after which the positions are switched for the second half. A time limit for a single trial will be set at 1500 ms (after target onset) and all trials after this limit will be terminated followed by a warning to “Please respond faster”. The incorrect trials will be followed by a red “X” for 500 ms and the correct trials followed by a black fixation cross for 500 ms. The experimenter will stop the sound played by the speaker and the EEG recorder upon completion of both tasks and save the data with the correct labels (“ID” and “Sound”) in excel files (performance data) and MATLAB files (EEG data).

Performance analysis

The performance of the condition with music will be analysed by comparing the score (in percentage: correct answers divided by total trials) and average response time (RT in seconds) of the Flanker task and Mouse Tracking task to the baseline for the sound groups (M/N/E) and for the different stress levels (mild/moderate/severe/control). The practice trials of both tasks will not be included in the test. In the case of sufficient test subjects with different stress levels in each sound group, the difference in performance of sound in combination with a stress level can also be tested (e.g. music + mild stress group versus music + severe stress group). All of the performance excel data will be analysed in R-studio. The score and RT of both tasks will first be tested for normality by using the Shapiro-Wilks test. If the scores or RTs are normally distributed, then one-way ANOVA tests (p<0.05) will be used, otherwise the Kruskal-Wallis tests (p<0.05) will be used. The one-way ANOVA test will be followed by a Tukey’s HSD-test and the Kruskal-Wallis test will be followed by a Paired Samples Wilcoxon test, on the condition that a significant difference (p<0.05) is found. A significant difference between the conditions (baseline no sound versus sound) will indicate an effect of sound on a group and a significant difference between groups will indicate which sound (nature/music) the most effect has on which stress groups.

EEG analysis

First, the EEG data of the test subjects will be cut from the start of the first trial (excluding the practice trials) to the end of the last trial (done for Flanker task and Mouse Tracker task).

(15)

15

A band-stop filter of 48-51 will be applied. The data will then be cut in segments of five seconds in order to visually detect artefacts (amplitude higher than 100 µV and/or SD=2) which will be removed from the data set. Test subjects with more than 15 segments left are used to conduct a power analysis. A time frequency analysis will be performed to calculate the average power of the theta (4-7 Hz), alpha (8-12 Hz), beta (13-28 Hz) and gamma (28-70 Hz) waves for each test subject.

The one-way ANOVA or Kruskal-Wallis tests (depending on normally distributed data) will be used in R to investigate any difference (p<0.05) between the average frequency powers of the sound condition and the baseline power (no sound condition). The same will also be done for the TBR average power, which is calculated by dividing the average power of the theta wave over the average power of the beta wave. Again in the case of sufficient test subjects with different stress levels in each sound group, the difference in the frequency waves and TBR of sound in combination with a stress level can also be tested (e.g. music + mild stress group versus music + severe stress group). Significant outcomes from the baseline comparison will indicate the effect of sound versus no sound and the significant outcomes from between the groups will indicate the sound group with the most effect of on a stress group.

(16)

16

Literature

1. Allen, A. P., Kennedy, P. J., Dockray, S., Cryan, J. F., Dinan, T. G., & Clarke, G. (2017). The trier social stress test: principles and practice. Neurobiology of stress, 6, 113-126.

2. Angelidis, A., van der Does, W., Schakel, L., & Putman, P. (2016). Frontal EEG theta/beta ratio as an electrophysiological marker for attentional control and its test-retest reliability. Biological psychology, 121, 49-52.

3. Fernandez, N. B., Trost, W. J., & Vuilleumier, P. (2019). Brain networks mediating the influence of background music on selective attention. Social cognitive and affective neuroscience, 14(12), 1441-1452.

4. Hsu, C. C., Cheng, C. W., & Chiu, Y. S. (2017). Analyze the beta waves of

electroencephalogram signals from young musicians and non-musicians in major scale working memory task. Neuroscience letters, 640, 42-46.

5. Huang, N., & Elhilali, M. (2020). Push-pull competition between bottom-up and top-down auditory attention to natural soundscapes. eLife, 9.

6. Katsuki, F., & Constantinidis, C. (2014). Bottom-Up and Top-Down Attention: Different Processes and Overlapping Neural Systems. The Neuroscientist, 20(5), 509–521.

7. Labuschagne, I., Grace, C., Rendell, P., Terrett, G., & Heinrichs, M. (2019). An introductory guide to conducting the Trier Social Stress Test. Neuroscience & Biobehavioral Reviews. 8. Larra, M. F., Pramme, L., Schächinger, H., & Frings, C. (2016). Stress and selective

attention: Immediate and delayed stress effects on inhibition of return. Brain and cognition, 108, 66-72.

9. Magosso, E., Ricci, G., & Ursino, M. (2019). Modulation of brain alpha rhythm and heart rate variability by attention-related mechanisms. AIMS Neuroscience, 6(1), 1.

10. Putman, P., van Peer, J., Maimari, I., & van der Werff, S. (2010). EEG theta/beta ratio in relation to fear-modulated response-inhibition, attentional control, and affective traits. Biological psychology, 83(2), 73-78.

11. Qi, M., Gao, H., & Liu, G. (2018). The effect of mild acute psychological stress on attention processing: an ERP study. Experimental brain research, 236(7), 2061-2071.

12. Sänger, J., Bechtold, L., Schoofs, D., Blaszkewicz, M., & Wascher, E. (2014). The influence of acute stress on attention mechanisms and its electrophysiological correlates. Frontiers in behavioral neuroscience, 8, 353.

13. Shields, G. S., Rivers, A. M., Ramey, M. M., Trainor, B. C., & Yonelinas, A. P. (2019). Mild acute stress improves response speed without impairing accuracy or interference control in two selective attention tasks: implications for theories of stress and cognition.

(17)

17

Appendix

Risk assessment

Although the methods that will be used are well established, some small complications may occur. For instance, during the Flanker task or the Mouse Tracking task a test subject might use speed for a strategy at the cost of accuracy, resulting in a lower score and a higher RT for the tasks. This problem can be solved by excluding the test subjects whose score and RT are significant lower than the average score of their group (indicating a possible speed versus accuracy trade-off). Another factor that may influence the results is the fact that the TSST will be performed on two separate occasions (two times in a week), possibly resulting in a decreased stress cortisol concentration compared to the first time of induced stress due to habituation and experience. The reverse situation can also be a possibility, since the test subjects may experience more stress due to anticipation of the task. It will be key to use the test subjects with the same stress level (mild/moderate/severe) of both TSSTs (day one and two) based on their cortisol concentration in order to negate these effects.

Referenties

GERELATEERDE DOCUMENTEN

To investigate the effects of the social stress context and the cortisol responses (CR) on the selective attention to angry and neutral faces we conducted a two-way ANOVA rm for

Binnen de bebouwde kom van de gemeente Enschede is een studie naar de veiligheid van fiets verkeer op een viertal wegvakken uitgevoerd door Bureau Goudappel Coffeng bv,

Note that a considerable number of clusters are tissue-specific (i.e., they contain genes differentially expressed in one or two tissues), reflecting the aims of the experimental

Although presumed environmental-friendly and safe, we observed that the nanocellulose materials induced oxidative stress in aquatic organisms of three different trophic levels

This will allow personality psychologists to declare upon the connections between the levels of the model (Eysenck, 1993a, 1993b), and to declare upon the differential genetic

Because BRB enhances the learning ability for boundary features and further guides the detection of corner The International Archives of the Photogrammetry, Remote Sensing and

Based on the research results of relevant scholars, this thesis divides corporate social responsibility into three specific pillars: environmental, social and governance, and

Voor bedrijven die 's nachts wel en overdag niet schermen bij extreme koude, valt de piek in het gasverbruik meestal overdag... Actuele informatie over land-