The Development of an Attentional Bias toward Body Size Stimuli: Performance on a
Novel Stroop Task

The Development of an Attentional Bias toward Body Size Stimuli: Performance on a
Novel Stroop Task

Raegan Bridget Cecilia Whitehead

2015

Distorted perceptions of body size have been identified and well-documented in eating disordered (ED) and eating-restricted populations, however, less is known about the development of this distortion. Research has employed Stroop food- and body-word tasks to investigate attentional biases towards semantically-related words and found a significant Stroop effect to such stimuli in ED, and sub-clinical, cohorts. The Size Congruity Effect (SiCE) has confirmed the perception of inanimate object size, however such an effect has not yet been studied in regards to body size specifically. This study recruited a novel Stroop size task to measure the perception of conceptual body size versus physical object size in four developmental age groups (Child, Adolescent, Young Adult and Adult). The Body Satisfaction Questionnaire (BSQ-34) was also taken as a measure of body dissatisfaction in participants over the age of 18. Findings indicate that a significant attentional bias towards body size is present across all age groups, but is most prevalent in adolescent and young adult participants. These findings imply that cognitive interference towards body size stimuli is not only present in the typical population, but is also present in children from aged 7. Body dissatisfaction, measured using the BSQ-34, did not have a significant effect on Stroop interference scores, suggesting that dissatisfaction with one’s own body does not implicate perception of others body size. The findings contribute to the fields understanding of body size misperception throughout typical development, the results also infer that body size perception is special, and not processed in the same way as inanimate size.

None

Participants
Eighty-eight participants (N = 88) were recruited to participate in this research. The participants (35 males, 53 females) were aged between 7 and 59 years (Mage = 23.38, SDage = 14.34). Participants were divided into one of four groups, dependent on their chronological age.
Child group. Child participants (N = 24, 8 male and 16 female), aged between 7 and 11 years (Mage = 10.04, SDage = 1.23), were recruited from St Boniface RC Primary School, Salford. A minimum participation age of 7 years was enforced for this experiment as previous research has not identified a consistent Stroop effect with younger children (Comalli et al., 1962). Parental consent was obtained prior to the research and participant assent was obtained on the day of testing. Five participants required glasses to correct their eyesight and were permitted to wear these throughout the testing period. Five participants reported having a specific learning difficulty (SLD); three participants had dyslexia, one participant had dyspraxia, one participant had attention deficit disorder (ADD) and one participant had attention deficit hyperactivity disorder (ADHD). One participant was on the autism spectrum (ASD). All participants with additional needs were performing well in mainstream school and were therefore considered able to participate in this research. Twelve participants had white British or white Irish ethnicity. Three participants had white European ethnicity. Three participants had black African ethnicity. Three participants had mixed or multiple ethnicities. One participant had Chinese Asian ethnicity. One participant had Irish traveller ethnicity. Two participants spoke English as a second language, however both were fluent English speakers. Each child received a reward sticker for their participation.
Adolescent group. Adolescent participants (N = 22, 9 males, 13 females), aged between 13 and 16 years (Mage = 14.73, SDage = 1.12), were recruited through opportunity sampling. Social media posts were used to advertise the study, as well as word of mouth. All participants were recruited from Greater Manchester. Parental consent was obtained prior to testing and participant assent was obtained on the day of testing. One participant was colour blind. Six participants required glasses to correct their eyesight and were permitted to wear these throughout the testing period. Four participants reported having a SLD; two participants had dyslexia, one participant had dyslexia and dyscalculia and one participant had dyslexia and ADHD. All participants with SLD’s were performing well in mainstream school and were therefore considered able to participate in the research. Eighteen participants had white British or white Irish ethnicity. Two participants had black African ethnicity. One participant had mixed or multiple ethnicities. One participant had British and Chinese ethnicity.
Young Adult group. Young Adult participants (N = 22, 7 male and 15 female), aged between 22 and 33 years (Mage = 25.86, SDage = 2.34), were recruited through opportunity sampling. The researcher utilised social media, approached classmates in Lancaster Univeristy’s Psychology Department and workplace colleagues to participate in the research. All participants were recruited from the North West of England. Each participant provided their informed consent prior to the research. Six participants required glasses to correct their eyesight and were permitted to wear these throughout the testing period. Two participants reported having a SLD; one participant had dyslexia and one participant had ADD. Both participants reported after testing that they were able to complete the task with no additional difficulty as a result of their SLD. Fifteen participants had white British ethnicity. Five participants had white European ethnicity. One participant had white American ethnicity. One participant had mixed or multiple ethnicities. Three participants spoke English as a second language, however both were fluent English speakers.
Adult group. Adult participants (N = 20, 11 male and 9 female), aged between 37 and 59 years (Mage = 45.75, SDage = 8.27), were recruited through opportunity sampling. Social media posts were used to advertise the study, as well as word of mouth. All participants were recruited from Greater Manchester. Each participant provided their informed consent prior to the research. Ten participants required glasses to correct their eyesight and were permitted to wear these throughout the testing period. One participant had dyslexia. This participant reported after testing that they were able to complete the task with no additional difficulty as a result of their SLD. Fourteen participants had white British ethnicity. Three participants had white European ethnicity. Two participants had black Caribbean ethnicity. One participant had mixed or multiple ethnicities.
Three participants were removed from the data sample due to a high number of errors. The responses from 85 participants were subsequently included in the data analyses.

This study received ethical approval from Lancaster University’s ethics committee on
1st May 2018.

Materials
Task. The novel Stroop task was created using Psychopy, an open source Python –based programme used to run psychological experiments (Peirce, 2007, 2009). In the task participants were presented with computer-generated images of female bodies. Each body was individually presented on the screen and remained there until the participant made their screen size selection. One hundred and eight images were presented in total; 54 in the congruent trial and the same 54 in the incongruent trial. Eighteen unique images were presented three times, each time the screen size of the image was varied in order to ensure all 18 images were presented in all 3 screen sizes. The 18 images consisted of three model types (See Figure 1) which were used to represent the polarities of body size (3 small body sizes, 3 large body sizes; see Figure 2).
Figure 2. An image to show the body ‘models’ used in the experiment. Row 1 Left – Right: Model 1, Model 2, Model 3, Model 4. Row 2 Left – Right: Model 5, Model 6, Model 7, Model 8.

The first testing phase of the Stroop task consisted of the individual presentation of 54 stimuli, these stimuli were presented with congruent screen and body sizes; all stimuli presented with a small screen size (10 x 4cm, 11 x 4.4cm, 12 x 4.8cm) contained a small body size, all stimuli presented with a large screen size (21 x 8.4cm, 22 x 8.8cm, 23 x 9.2cm) also contained a large body size. The second testing phase of the Stroop task consisted of the individual presentation of the same 54 stimuli as the first phase. These stimuli were presented with incongruent screen and body sizes; all stimuli presented with small screen size contained large body size, all stimuli presented with large screen size contained a small body size. See Appendix A for screenshots of the Stroop task, demonstrating the congruent and incongruent presentation of the stimuli as described here. The order of stimulus presentation was pseudo-randomised within Psychopy, so that each individual image was presented only once per participant. Randomising the order of stimulus presentation assured that participants were not subjected to order effects (Shaughnessy, Zechmeister, & Zechmeister, 2006).
Participants were instructed to determine the screen size of each stimuli and respond as quickly and accurately as possible using the keyboard keys indicated to them in the instruction phase. The relevant keyboard keys (A and L) were indicated with white stickers on the external keyboard. Key allocations (e.g. A = Small, L = Big) were also visible on screen throughout the task, see Appendix A for screenshots of the task. Participant response times were recorded within Psychopy and exported to Microsoft Excel. The task was presented on a Toshiba Satellite Pro laptop computer with a 15.6-inch HD non-reflective display with a 16:9 ratio and LED backlighting.

Body stimuli. Eighteen images of computer generated semi-nude female bodies, ranging in body size and physical appearance, were used in the current study. These were created and donated to the researcher by Dr Martin Tovee, body size perception researcher, for the purpose of the current experiment. The bodies ranged in size from ‘emaciated’ to ‘overweight’, the variations in body size were visually distinguishable (see Figure 1), Eight ‘models’ were created, each with variations in physical appearance including hair colour and style, skin tone, facial features and eye colour (see Figure 2). All bodies were presented in a forward facing 0o pose, in order to eliminate visual preference or difficulties in comparing stimuli. Image size was manipulated as a factor of the experiment; to reflect ‘small’ screen size all images were presented at 10 x 4cm, 11 x 4.4cm and 12 x 4.8cm. To reflect ‘big’ screen size all images were presented at 21 x 8.4cm, 22 x 8.8cm and 23 x 9.2cm. These sizes were chosen as they created incremental differences in screen size that were visually distinguishable, as can be seen in Appendix B.
Figure 2. An image to show the body size increments in the stimuli provided by Dr. Tovee. Model 3 is used to illustrate the size increments. Row 1 Left – Right: Size 1, Size 2, Size 3, Size 4, Size 5. Row 2 Left – Right: Size 6, Size 7, Size 8, Size 9, Size 10. For the purpose of the current experiment, sizes 1, 3, 4, 7, 8 and 10 were used as body size stimuli as these bodies had the largest size variation when visually scrutinised.

Questionnaires. All participants were required to complete a demographic questionnaire, see Appendix C. The parent/guardian of a participant under the age of 16 was required to complete this questionnaire on behalf of the participant. This questionnaire was used to ascertain factors which may affect a participants ability to successfully complete the Stroop task.
Participants over the age of 18 years were also required to complete the Body Shape Questionnaire (BSQ-34; Cooper, Taylor, Cooper & Fairburn, 1987). The BSQ-34 is a 34-item scale which measures participants feelings toward their own weight and body shape (Taylor, 1987). For example; ‘Have you been afraid that you might become fat (or fatter)?’ and ‘Has seeing your reflection (e.g. in a mirror or shop window) made you feel bad about your shape?’. Each item of this scale is scored on a six-point Likert scale, ranging from 1 (never) to 6 (always). BSQ-34 scores are totalled using Likert scale points; a score less than 80 indicates no concern with shape, a score between 80 and 110 points indicates a mild concern with shape, a score between 111 and 140 indicates moderate concern with shape and a score of 140 and above indicates a marked concern with shape (Cooper et al., 1987). The BSQ-34 was originally intended for use with female participants; the authors have since approved changes to items 9, 12 and 25 for use with male participants, this version was provided for male participants in the current experiment. The BSQ-34 was not considered suitable for participants under the age of 18 due to the explicit mention of clinically salient stimuli.
The BSQ-34, as well as participants consent forms and demographic questionnaires, were provided to participants on Adobe Fill & Sign using an Apple Ipad and touchscreen pen. All participants indicated daily or weekly use of a touchscreen and/or computer.

Design
Variables. The dependent variable in this study was task response time, recorded by Psychopy in milliseconds. Mean response times (MeanRT) were calculated for the congruent and incongruent trials, per participant. An interference score (incongruent MeanRT minus congruent MeanRT) was also calculated for each participant. The dependent variables of MeanRT and Interference Score were both used in the current data analyses. The independent variables in the study were; AgeGroup and Congruency.
AgeGroup. This was a between subjects factor. Participants were placed into one of four age groups, based solely on their chronological age.
Congruency. This was a within subjects factor. All eighty-eight participants completed the same novel Stroop task, containing both congruent and incongruent trials. The order of trial presentation were randomised for each participant.

Procedure
Three months prior to testing, the parents/guardians of children in years five and six of St Boniface RC Primary School, Salford were contacted and given the opportunity for their child to participate in this study. The children of parents/guardians who returned the consent form and completed questionnaire were able to participate. The research was also advertised, via social media and word of mouth, to potential participants. The parents/guardians of participants under the age of 16 years, and participants over 16 years, were provided with an information letter, consent form and demographic questionnaire (See Appendices C, D and E).. Those who responded with a complete consent form and questionnaire participated in the research.
Participants were individually invited to complete the procedure in a small quiet room. All participants were seated at a desk in front of the testing laptop and an external keyboard, see Figure 3 for the testing set-up. Participant consent, and child assent, was ascertained once the participants were seated. All participants were encouraged to ask any questions they had and child participants were reminded that they could return to their class at any time, without providing a reason. Once the preliminary period was completed participants were then asked to complete the computerised Stroop task.
The task was visible on the screen prior to each participant entering the room. Participants were aided through the initial instruction screens of the task and encouraged to stop and ask questions at this stage. The researcher read all instructions to participants under the age of 16, and to any participant who requested that the instructions be read to them. The task then contained two practice trials, in order to ensure that participants understood their role in the task. All participants were able to complete the two test trials without difficulty and were therefore permitted to complete the rest of the task. The researcher left the room and waited nearby for all participants over the age of 16 years, the researcher remained in the testing room for younger participants. Participants were informed that they should only take a break when they reached an instruction screen as their times were being recorded on all testing screens.
Participants were asked to alert the researcher once they had completed all stages of the computer task and reached the end screen. Child participants were given an envelope containing a parental debrief and escorted back to their classroom. Young adult and adult participants were asked to complete the BSQ-34 (Cooper et al., 1987) using Abobe Fill & Sign on an Apple Ipad. The BSQ-34 was provided after the task as Davison and Wright (2002) reported that this method reduced demand characteristics in a similar study. Upon completion of the testing period all participants were thanked for their time and provided with a debrief sheet as well as help and information pertaining to eating disorder or body anxiety concerns. Child participants were rewarded with a sticker for completing the task. Please see Appendix F for the participant debrief.
Each participants response times were recorded in an Excel document which was then encrypted and saved to the researchers password protected laptop. All data was also stored on an encrypted external hard-drive, this copy of the data will be securely destroyed upon completion of the data analyses.
Figure 3. A photograph to show the testing set-up used in the current study. Note, participants were encouraged to adjust their seat height to remain at a ninety-degree angle to the screen. The testing set-up was replicated for all eighty-eight participants to ensure continuity.

Lancaster University

Data/SPSS.sav

Whitehead2015

Rebecca James

Open

None

English

Data

LA1 4YF

Dr Michelle To

MSc

Cognitive, Developmental

Eighty-eight participants

ANOVA