Experiencing social acceptance and rejection through ‘likes’ and ‘dislikes’: Does sleep quality affect the processing of social rewards?

Experiencing social acceptance and rejection through ‘likes’ and ‘dislikes’: Does sleep quality affect the processing of social rewards?

Abigail Taylor-Spencer

2018

In adolescence, high importance is placed on peer evaluations and social rewards have increased salience during this developmental period. Sleep patterns also change in adolescence, as teenagers typically experience insufficient sleep. This research measured the pupil dilation of forty-four adolescents aged 16 to 18 using two tasks (audio and visual) to investigate whether sleep duration influenced the way social acceptance and rejection were processed. Sleep duration scores were obtained using the measure of sleep debt; this was calculated by subtracting sleep duration during the week from sleep duration at the weekend, plus weekday bedtime. It was expected that higher sleep debt would be linked to increased pupil reactivity towards social feedback and that there would be a greater pupil dilation in response to social rejection compared to social acceptance. In the visual task, it was found that sleep debt affected males and females differently when processing social rewards, as females with high sleep debt showed increased pupil dilation towards positive feedback compared to negative feedback, whereas males with low sleep debt showed a larger dilation towards positive feedback than females. It was also found that females with lower sleep debt gave more likes than dislikes when rating photos. This implies that sleep duration affects the social feedback adolescents provide. When a male voice was used in the audio task, more pupillary reactivity towards social acceptance was observed, however when a female voice was used, pupils dilated more in response to social rejection. Future research should further investigate these gender differences.

Adolescence
Pupil dilation
Social feedback
Reward
Rejection
Sleep debt.

Participants
Forty-four participants (N=44) were recruited from Haslingden High School and Sixth Form to participate in this research. The participants (35 female, 9 male) were all between the ages of 16 and 18 (Mage = 16.98, SDage = .63). Students in Psychology, Sociology and English classes were given the opportunity to participate in the research and contacted the researcher via email if they wished to participate. Each participant provided their informed consent before beginning the study.
Materials
Photo ratings. Firstly, the participants were shown a PowerPoint containing 40 photos, which had been previously collected by the researcher, and featured adolescents which the participants did not know. Each photo was displayed individually for four seconds, meaning that the presentation lasted two minutes and forty seconds in total. Participants were provided with a sheet of paper on which they had an option to tick either ‘like’ or ‘dislike’ for each photo on the PowerPoint (see Appendix A). The total number of likes was calculated for each participant.
Eye tracker. An eye-tribe desktop eye tracker with a 30Hz sampling rate was used to measure the pupil dilation of the participants in response to stimuli on two tasks - a visual task and an audio task. A chin rest was used to ensure the participants kept their heads still.
Visual task. The visual task involved showing the participants the same 40 photos which they had previously been shown in the photo rating task, however, each photo had either a ‘like’ symbol or ‘dislike’ symbol (see Figure 1) in the bottom right hand corner. Participants were informed prior to beginning the task that if a photo contained the ‘like’ symbol, it meant that the individual in the photo had liked the participant’s picture, however the ‘dislike’ symbol meant that the individual in the photo had disliked the participant’s picture. The presentation of photos was randomised across participants

Audio task. The audio task involved the participants listening to forty voice recordings, which each lasted between six and seven seconds in length. Twenty of these recordings were nice comments and twenty were nasty comments, which were found on online social media platforms. An example of a nice comment is; ‘You look unreal and your outfit is amazing. You are a true inspiration to everyone’ and an example of a nasty comment is; ‘You are so fake, and you are such a liar. Every single thing you say is a lie’ (see Appendix B for the complete list of comments). A male voice read out half of the nice and half of the nasty comments, and a female voice featured in the other half of the recordings. The nice comments were characterised as positive social feedback, and the nasty as negative social feedback. The presentation of nice and nasty comments was randomised across participants. The audio material was rated for emotional valence and arousal; the former being how positive or negative the recordings were, and the latter being the intensity of this positivity or negativity (Citron, Gray, Critchley, Weekes, & Ferstl, 2014). See Appendix C for the emotional valence and arousal scores, which were rated by six individuals using Qualtrics. Presentation of the nice and nasty comments was randomised across participants.
Questionnaires. Participants were asked to complete two questionnaires; one which was an adaptation of the MCTQ questionnaire (Munich ChronoType Questionnaire; Roenneberg, Wirz-Justice & Merrow, 2003), to identify the sleeping patterns of the participants (see Appendix D), and a questionnaire about their social media use (see Appendix E) which was used to maintain the ruse that the study was interested in the participants’ social media use.
This study received ethical approval from Lancaster University on 05/04/2018.
Design
Variables. The dependent variable in this study was pupil size, which was measured in arbitrary units, using an eye tribe eye tracker. An average pupil diameter was calculated for each trial; each participant had 40 average pupil size measurements in the visual task and 40 average pupil diameter measurements in the audio task. The dependent variables of median and area under the curve were used. The independent variables in the study were; feedback valence, sleep debt, gender voice and gender.
Feedback valence. The feedback was within subjects, as all forty-four participants experienced both positive and negative feedback in both tasks. In the visual task, all participants saw twenty people who had supposedly ‘liked’ their photo, and twenty people who had supposedly ‘disliked’ their photo. In the auditory task, all participants heard twenty positive comments and twenty negative comments. This was analysed to assess whether varying pupillary responses were elicited towards positive and negative social feedback.
Sleep debt. Sleep debt was determined by the MCTQ (Roenneberg et al., 2003); a value of sleep debt was calculated by subtracting sleep duration during the week from sleep duration at the weekend, plus weekday bedtime. Participants were split into two groups; high sleep debt and low sleep debt. Those with a high sleep debt had less weekday sleep and greater weekend sleep, which is a marker of poor sleep quality. This was a between subject factor, as half of the participants were in the high sleep debt group, and half in the low sleep debt group.
Voice Gender. In the audio task, half of the audio clips featured a male voice, and half featured a female voice, therefore this was a between subject factor. This was analysed to investigate whether the gender of the voice or pictured individual had an effect on the pupillary responses.
Gender. In the visual task, the gender of the participants was investigated as a between subjects factor, as nine of the participants were male, and thirty-five were female.
Audio task. The design of the audio task was a factorial design with a between subjects factor of sleep debt (which had two levels – low and high) and a within subjects factor of social feedback valence (two levels: positive and negative) and a second within subjects factor of voice gender (two levels: male and female).
Visual task. The design of the visual task was a factorial design with a between subjects factor of sleep debt (which had two levels – low and high) and within subjects factors of social feedback valence (two levels: positive and negative) and participant gender (two levels: male and female).
Procedure
Approximately two weeks prior to the beginning of data collection, students in Psychology, Sociology and English classes at Haslingden Sixth Form were contacted and given the opportunity to participate in this research. Those who were interested in participating, and would provide consent, were asked to send a picture containing only themselves (eg. a Facebook profile picture) to the researcher via email for use in the study. The participants were informed that the photo they sent would be liked or disliked by students at another school, and that that during the study, there would be an opportunity to like or dislike photos of the individuals who rated their picture. No other information about the other ‘students’ was provided. The participants were led to believe that the study was investigating whether social media use affects responses to being judged online, and whether the use of social media affects sleep patterns in adolescence.
All participants were tested in the same office in Haslingden High School and Sixth Form. Participants were invited into the office and invited to sit down a desk which featured an eye-tribe eye tracker, 24-inch iMac monitor and keyboard, and a chin rest was placed 50 cm away from the eye tracker. The computer had MatLab 2015 installed. Each participant was provided with an information sheet (see Appendix F), and was given the opportunity to ask any questions, before signing an informed consent form (see Appendix G) if they still wished to participate and consented to partake in the study.
Once the consent form had been signed, the photo rating task was explained. This task involved presenting forty photos to the participants using Microsoft PowerPoint. The photos were shown individually; each photo was on an individual slide, and each one was presented for four seconds. The participants were asked to mark whether they ‘liked’ or ‘disliked’ each photo on a sheet of paper (see Appendix A). The presentation was on an automatic timer however, the participants were informed that if a slide moved on too quickly, the left arrow key would take them back to the previous slide, and the timed presentation would continue by pressing the right arrow key. The participants were led to believe that the photographs they were rating were of the individuals who had rated their photos. The eye tracker was not used during this task.
Next, the participants were asked to place their head on the chin rest, and the eye tracker was calibrated. Participants were asked to keep their heads as still as possible, and to move their eyes towards the dots as they appeared on the screen. The calibration was accepted when three stars or above was achieved, and the eye tracker was used for both the visual and auditory tasks. The order in which the tasks were completed was counterbalanced, therefore half of the participants completed the visual task first, and half completed the auditory task first. The participants were informed what would happen during each task and were given the opportunity to ask any questions before the tasks began.
The participants were told that, in the auditory task, they would hear forty voice clips; twenty nasty and twenty nice. They were asked to look at a black cross that was located in the centre of the screen whilst the voice clips were playing. Ten of the ‘nice’ clips and ten of the ‘nasty’ clips were read aloud by a female, and the remaining were read by a male voice. The nice and nasty comments which featured in the voice clips were found on online social media platforms (see Appendix B for the full list of comments used), however the participants were asked to imagine that the comments had been directed towards themselves. Participants were told that, in the visual task, they would view the photographs which they had previously ‘liked’ or ‘disliked’ in the photo rating task. However, this time, the photos would either have a ‘like’ thumb or a ‘dislike’ thumb in the bottom right hand corner (see Figure 2 and Figure 3 for examples). If a photo had a ‘like’ thumb, it meant that person had supposedly liked the participant’s photo, whereas a ‘dislike’ thumb meant the individual in the photo had disliked the participant’s photo. Half of the participants completed the visual task first, and half of the participants completed the audio task first; the tasks were counterbalanced to determine whether the order in which they were presented influenced the outcome.
After finishing both the visual and auditory tasks, participants were asked to complete two questionnaires; the MCTQ (Roenneberg et al., 2003) to determine a sleep debt score and a questionnaire on social media use. After completing the questionnaires, participants were informed that their photo had not actually been seen or rated by pupils at another school, and that the ratings which they gave in the photo rating task wouldn’t be seen by the individuals in the photos. Participants were then provided with a debrief sheet (see Appendix H) and given the opportunity to ask any questions they may have had.
Analysis
Preliminary data analysis. In order to measure the magnitude of change in pupil dilation and compare across the conditions, each trial pupil size was baseline adjusted by subtraction of the mean pupil size in the 300ms prior to stimulus onset from each sampled value during the further 4 seconds of stimuli presentation. The area under the curve and median were then calculated from the trial level baseline adjusted data to provide the dependent variables in the analysis. These were used as dependent variables to show the magnitude and duration of the effects. The median was used as opposed to the mean because the median is less likely to be skewed by outliers.
Two multilevel mixed effects general linear mixed models (GLMM) were used to analyse the data for the two tasks with participant included as a random effect with intercept. An AR(1) heterogeneous first order autoregressive structure with homogenous variances was selected because it was expected that the error variance would become less correlated as the trials became further apart. The total number of likes each participant gave on the photo rating task was calculated and a 2 (gender: male vs. female) x 2 (sleep debt: low vs. high) between factor analysis of variance (ANOVA) was carried out.

Lancaster University

data/SPSS.sav

Taylor-Spencer2018

Ellie Ball

Open

None

English

Data

LA1 4YF

Judith Lunn

MSc

Cognitive Psychology
Developmental Psychology

44 Participants (9 male and 35 female)

ANOVA
Linear Mixed Effects Modelling