Studying the effects of digital arts-based interventions on teenagers’ social media usage, brain connectivity, and mental health: study protocol of the SMART project

ABSTRACT

Background: Social media (SM) has become ubiquitous among youth. However, which SM activities are beneficial or detrimental for the wellness of children and adolescents is still under debate. While some reports highlight positive outcomes of SM in learning, social interaction, and wellbeing, other investigations suggest that the overuse of SM induces decreased attention, cognitive, and emotional control, and increases mental-health related disorders (e.g. depression and anxiety). Interestingly, the cognitive and emotional functions negatively affected by the intense use of SM, as well as some of its neural underpinnings, have been previously and consistently reported to benefit from music and arts-based interventions.

Objectives: The protocol for the ‘Social Media Artistic tRaining in Teenagers (SMART)’ project (ClinicalTrial: NCT06402253) is presented here: digital art-based interventions will be used to teach adolescents how to use SM in more goal-oriented and stimulating ways, in the context of learning music or photography composition/editing through specific open-source software.

Methods: Participants (aged 13–16) will be evaluated before and after completing a 3-month music or photography composition/editing intervention programme. Participants will also provide weekly measures of SM usage and mood. A matched passive control group will also be recruited, evaluated, and followed for 3 months. Evaluations will include cognitive (attention), mood, and mental-health (depression, stress, anxiety, self-esteem) measures, as well as functional and structural connectivity and morphological biomarkers obtained via MRI and MEG techniques.

Discussion: We expect observable changes in self-reported use and attitudes towards SM, and benefits in attention, mood, and mental-health measures, as well as in the neural substrates supporting these processes. The data we plan to collect will confirm or challenge these expectations, aiming to improve our understanding of the impact of SM overuse on brain function, cognitive state, and mental health. Our findings could also inform potential strategies to mitigate SM negative effects.

Trial registration: ClinicalTrials.gov identifier: NCT06402253..

HIGHLIGHTS

• In today's highly digital world, where smartphones and social media are ubiquitous among children and adolescents, debate persists over which online activities benefit youth and which may be harmful.

• Mental health, attention, cognitive and emotional control, as well as the neural underpinnings of these brain functions, seem to be negatively affected by social media but positively influenced by music, arts, and creative training.

• Our project –SMART– aims to train 13–16 year-old teenagers in composition and editing of either music or photography materials, using this artistic experience to promote critical thinking around the use of social media, in an attempt to overcome the detrimental mental health and cognitive effects associated with the passive use of these platforms.

1. Background

The use of digital devices (i.e. smartphones, tablets) with access to online activities and social media platforms has exponentially increased in the last decades. Currently, more than 60% of the global population uses the internet –with almost 95% of users connecting through their smartphones– and around 5 billion people have a social media (SM) account (Meltwater Global Digital Report, 2024). A growing social concern is the recognition that this widespread digital connectedness is not just present among adults but is perhaps even more prevalent among youth. For the past two decades, children have increasingly been introduced to smartphones and social media as early as age two, and previous reports suggest that around 10% of 1 year-old infants are already being regularly exposed to these devices/sites nowadays (Chang et al., 2018; Durham et al., 2021; Kim et al., 2020; Zimmerman et al., 2007). Regarding SM usage, 40% of children 8–12 years old report using these sites despite being under the legal age to access these platforms, although this may be changing with modifications that SM companies are starting to implement, such as the Instagram Teen Accounts (Meta News, 2024), or with public-health-based governmental initiatives such as Australia’s Online Safety Amendment (Social Media Minimum Age) Act 2024 (Australian Government eSafety Commissioner, 2024). Nevertheless, more than a third of adolescents aged 13–17 years report using SM ‘almost constantly’ – up to 7 h per day, on average (Meltwater Global Digital Report, 2024; Office of the US Surgeon General, 2023).

Research has found evidence that excessive smartphone use during childhood and adolescence is associated with negative effects on cognitive and emotional control. Specifically, cognitive and mental health issues linked to overuse of smartphones/SM include impairments on impulse control, executive functions, cognitive flexibility and emotional regulation, as well as low self-esteem and poor academic performance (Kim et al., 2020; Wacks & Weinstein, 2021). Importantly, some investigations have shown that these effects can evolve into serious mental health issues at mid- and long-term levels, such as anxiety and depression disorders that can persist into adolescence and adulthood (Conte et al., 2025; Elhai et al., 2019; Ivie et al., 2020; Kim et al., 2020; Wacks & Weinstein, 2021). However, SM effects on youth are complex and their overall impact can vary depending on individual differences (e.g. gender) and how social media is used (Orben et al., 2022; Vidal et al., 2024).

Interestingly, SM has also been linked to positive outcomes: such as promoting healthy habits, boosting creativity, building support communities, and helping in new-skill acquisition (Haddock et al., 2022; Lucero, 2017; Vaingankar et al., 2022). According to previous research, prolonged used of smartphones and SM can elicit cognitive and affective benefits associated with empathic abilities in adolescents, and a greater engagement in health promotion activities across the lifespan (Gabarron et al., 2021; Manago et al., 2012; O'Reilly et al., 2019; Uhls et al., 2017).

At the neural level, there is limited evidence regarding the structural and functional brain underpinnings of SM use/overuse. Neuroimaging reports with adult participants have mainly shown a relationship between SM usage and brain regions and networks involved in reward processing and executive functions. Specifically, excessive use or SM overuse is linked to reduced grey matter volume in the amygdala, ventral striatum, and the nucleus accumbens (He et al., 2017), to enhanced connectivity between reward pathways (e.g. connections to ventral striatum and ventromedial prefrontal cortex; Wilmer et al., 2019), and to deficient interhemispheric connectivity (e.g. corpus callosum, He et al., 2018). Lower or regular SM usage, on the other hand, has been reported to be associated with enhanced structural and functional connectivity in networks connecting reward-related areas with regions related to executive functions (e.g. ventral striatum and the dorsolateral prefrontal cortex; Wilmer et al., 2019), as well as with the self-referential cognition network (consisting of the medial prefrontal cortex and posterior cingulate cortex/precuneus, and involved in self-reflection and comparisons with others), the salience network (covering the anterior insula and dorsal part of the anterior cingulate cortex), or the default mode network (see Wadsley & Ihssen, 2023 for a review). Most of these previous reports focused on investigating SM or internet addiction, and their brain structural and functional findings support well-known theories of reward deficiency, enhanced impulsivity, and sensitivity for immediate rewards, among others (He et al., 2017, 2018; Montag et al., 2017; Turel et al., 2018).

These counterintuitive findings suggest that the negative outcomes associated with SM use may not directly result from excessive use itself (Montag et al., 2021). Instead, the relationship between well-being and SM use appears to be more complex, with individual differences playing a crucial role in how people experience the impact of SM. Based on previous reports (Verduyn et al., 2015, 2017), we hypothesise that (i) the specific activities performed, (ii) nature of content watched online, and (iii) whether the use of SM is passive or active, may be crucial factors determining the effect of SM use on youth’s cognition and mental health. This is supported by research showing that the beneficial or detrimental effects of SM on young users depend on a widespread array of factors, including baseline mental health status, number of SM accounts and concrete platforms used, purpose and type of usage of these online sites, and individual differences in demographic variables (Nesi, 2020; Orben et al., 2022; Rajamohan et al., 2019; Vidal et al., 2024).

Interestingly, the brain networks previously linked to the negative impacts of SM have been shown to undergo positive changes following art-based interventions. Cognitive, creative, and artistically focused interventions in children and teenagers positively influence wellness and induce neuroplasticity (Luis-Ruiz et al., 2020; Roman et al., 2017; Vander Linden et al., 2019; Zatorre, 2013). Engaging in visual arts, aesthetic perception activities, and music-related practices – such as listening, composition, and performance – has been shown to enhance the connectivity of both structural and functional brain networks across the lifespan. These activities strengthen critical cortical and subcortical structures, as well as key white matter pathways, including the cingulum, corpus callosum, and the inferior longitudinal fasciculus, among others. Additionally, they modulate brain plasticity functional systems such as the Default Mode Network, auditory and visual networks, and Executive Function/Cognitive Control networks (Bashwiner et al., 2016, 2020; Cheng et al., 2023; De Pisapia et al., 2016; Habibi et al., 2018; Hennessy et al., 2019; Hong et al., 2023; Hudziak et al., 2014; Jünemann et al., 2023; Li et al., 2018; Wang et al., 2022). These plastic changes seem to be associated with better cognitive performance across domains, including verbal memory, attention, creativity, error prediction, and emotional control (Chatterjee, 2011; François et al., 2015; Moumdjian et al., 2017; Särkämö & Soto, 2012; Sihvonen et al., 2017). Based on these reports, we hypothesise that art-based interventions could be a useful approach to overcome the detrimental effects of an excessive passive SM use in children and adolescents. We hypothesise that the generalised brain stimulation and enriched environment that art processes provide (Särkämö et al., 2008; Zatorre, 2013), paired with their inherent rewarding component (Mas-Herrero et al., 2023), will elicit a boost in different cognitive and emotional functions, that may transfer to other domains beyond those explicitly practiced in visual arts and musical interventions (although these have been less consistently replicated, Alain et al., 2019; Moreno & Bidelman, 2014). It is important to note, however, that previous reports suggest that when music and visual arts interventions are compared, music seems to be the experience associated with greater plastic changes or transfer effects to other cognitive domains (Flaugnacco et al., 2015; François et al., 2013; Frischen et al., 2021; Jaschke et al., 2018; Lu et al., 2022).

Here we present and define a protocol for the SMART (Social Media Artistic tRaining in Teenagers: https://smartproject.blog/) intervention. The main research question behind SMART is to assess whether art appreciation and practice can change the way teenagers interact with and use SM; and whether and how digital arts-based interventions can help to overcome the negative influence of smart devices and social media platforms on mental health.

1.1. Goals of the current study

SMART aims to promote changes in the way teenagers use SM by implementing an arts-based intervention. The protocol combines training in accessible, open-source digital tools for the creation of music or visual art with group discussions that stimulate critical thinking about SM. We hypothesise that the stimulation provided by the art-based training together with the SM resources and reflections will foster a more mindful and balanced approach to social media use, aiding the participants, for instance, to be critical about the content shared online, to seek healthier and more respectful online interactions, and/or to look for more stimulating activities to do online. In other words, the main goal of this project is to assess how the SMART intervention changes SM behaviour in teenagers.

SMART’s secondary goal is to elucidate the behavioural and brain mechanisms behind the effects that the art-based training (in music or visual-arts) might have on cognition and mental health. Based on previous work (Barnett & Vasiu, 2024; Fong et al., 2021; Masika et al., 2020; Särkämö et al., 2014; Sihvonen et al., 2017, 2020, 2021), we hypothesise that improvements in mental health variables will correlate with plastic changes in emotional and reward-related centres (e.g. ventral striatum, amygdala, ventromedial prefrontal cortex), while changes in cognitive function will be related to plastic changes in prefrontal regions related to attention and executive functions.

1.2. Target population

Teenagers aged 13–16 years (9th–10th graders in the USA educational system) attending participating public high schools in the New York City area. Target was placed on public high schools because we believe the population attending these institutions are more representative of the general population of New York City.

2. Methods

2.1. Summary of intervention study design

Two active groups and a passive control group will be included in the study, with the two active groups receiving one of the interventions (either music or visual-arts). Interventions will last for 3 months (12 weeks), consisting of two 45-min arts-based classes per week, plus a 15-minute session per week focused on discussions, reflections, and activities around social media. Before and after this intervention period, participants will complete two evaluation phases where measures of mental-health, behaviour, and cognition will be collected. We will also collect weekly measures of mood and SM usage from all participants through online surveys. See Figure 1 for a schematic representation of study design.

Figure 1.

Schematic representation of the SMART project design: 2 evaluations (Baseline, Post-Intervention), with a 3-month (i.e. 12 weeks) intervention or control period. Open-source software that will be used during the interventions is depicted by their logo in the corresponding column: Soundtrap for the Music composition programme, and PixelR for the Visual-Arts (i.e. photography) intervention programme. Abbreviations: MUS: music composition/edition intervention group; VIS: photography/visual-arts composition/edition intervention group; CON: control group with no intervention; min: minutes.

The study aims to acquire brain structural, functional, and neurophysiological data through magnetic resonance imaging (MRI) and magnetoencephalograpy (MEG) techniques. This part of the protocol will be optional due to restrictions imposed by the New York State Department of Education during their ethical review of the protocol. MRI and MEG data will be collected at the Center for Brain Imaging at New York University (New York City, USA).

Assessments (behavioural, neuroimaging, and neurophysiological data) will be done 1–3 weeks prior to the beginning of the intervention (baseline evaluation), and 1–3 weeks after the end of the intervention (post-intervention evaluation). Behavioural assessments will occur during the same class periods assigned for the intervention sessions (i.e. during regular class hours or after-school hours, as agreed with the participating high schools), while neuroimaging and neurophysiological measurements will be conducted outside of scheduled class times.

2.2. Study setting

Data collection is planned to be conducted in the New York City area, through collaborative partnerships with public high schools in Manhattan and Brooklyn. Regular classrooms or auditorium spaces will be used for SMART sessions, depending on the availability in each high school and the size of the student group at the time of intervention. Students in both intervention groups will use their smartphones to take pictures (VIS), record sounds (MUS), and do some level of editing through built-in and available editing software in their personal devices (VIS, MUS), and/or will use the available laptops in the schools – e.g. Macbook Pro A2159– (MUS). Expert instructors and teaching and counselling personnel from the participating highschools will be present at all intervention sessions (see details about personnel below).

2.3. Participants

2.3.1. Inclusion & exclusion criteria

As criteria for inclusion, students will be eligible irrespective of their country of birth, native language, race, or gender. Furthermore, we will not exclude subjects based on their neurodevelopmental, neurological, or psychiatric disorders, or based on their background or previous training in music, visual/plastic arts, or film/multimedia. Thus, any student attending the participating high schools, and aged 13–16 years is a potential participant.

In addition, we intend to recruit the parents of child participants to answer questions about their child. For these participants, we have no inclusion criteria except for being the parent/guardian of a child participating in the study.

2.3.2. Expected demographics

Participants will be recruited from public high schools from our city area with which agreements have been set in place. Based on data from the from New York State Education Department (NYSED) Student Information Repository System (SIRS) regarding both general city-wide data (Zimmerman, 2023) and average data from our current participating research sites (NYC Department of Education’s 2023–24 School Quality Snapshot High School corresponding to our participating high schools), the demographics of our sample are expected to be: 47–56% female vs. 44–53% male, and 0.01% non-binary students (please note that the most recent official information found on non-binary / gender non-conformant / gender fluid / gender expansive students in the city of implementation is from summer 2023 and is based on documentation that students themselves had to fill out under the approval of their parents, which may lead to some students deciding not to mark their preferential gender category); 29–30% Asian, 3–8% Black, 26–31% Hispanic or Latinx, < 1% Native American/Native Hawaiian/Pacific Islander, and 33–35% White; 16–18% students with Individualised Education Programs (IEPs); and 2–12% English-as-a-second-language learners. In the US education system, an IEP is a legal document developed for each child enrolled in a public school who has an identified disability that requires special education. This document is reviewed every year to ensure the appropriate progress in the child’s education (Mueller & Moriarty Vick, 2019; IEP definition from the NYSED official website).

Demographics corresponding to parents of participants are expected to be similar to those of the student participants. According to information obtained from participating schools, the students that will be enrolled in SMART will be coming mostly from Brooklyn, with a few of them coming from Queens or the Bronx boroughs of NYC. Families with teenagers in these boroughs are expected to show significant economic diversity, which may be connected to academic performance disparities among our high-school student-participants. Although specific socio-economic status (SES) information from our participants and their families would have to be directly asked, we expect our cohort to reflect the average SES that has been described for our participating high schools, as well as from the NYC boroughs where they and their families live. To mention a couple of specific SES proxies, it is expected that between 56% and 73% of our students would be eligible for free lunch (which indicates a medium-low family income, 2023–24 NYSED Student and Educator Report corresponding to our participating high schools); and that their families present a diverse range of maternal education, from communities where only around 20% of adults have college degrees (the Bronx: 21–23%) to areas with bachelor's degree attainment rates approaching 50% (Brooklyn: 46%, Queens: 34–39%) (Neches et al., 2024). This data suggests that our participant cohort would likely encompass both economically disadvantaged and more privileged family contexts typical of NYC's outer boroughs and the underserved public high schools they host.

2.3.3. Sample size / power analysis

A pre–post longitudinal experimental design will be implemented. Recruited participants (whose parents have given their written permission and who have assented to participate) will be divided in three groups: (a) Music composition/edition intervention (MUS group), consisting in a training in using music creation and edition by learning to use open-source digital tools combined with an introduction to music appreciation and structural musical concepts; (b) Photography composition/edition intervention (VIS group), comprising image edition training using open-source digital tools combined with an introduction to photography/visual art appreciation and basic photography concepts; (c) Control group, with no specific intervention (CON group). Based on this design, a sample size analysis using MorePower (Campbell & Thompson, 2012) determined that, to have 80% of power to detect a large effect in a 2 (Baseline/Post) × 3 (Group: MUS, VIS, CON) mixed between-within repeated measures ANOVA, 18 participants per group are required. We intend to recruit 30 participants per group to account for attrition.

2.3.4. Recruitment strategy

Recruitment will take place adapting to the communication channels already in place at the participating high schools. This process will be completed by using flyers posted in the high school or sent to the families physically and/or via the high-school newsletter. Researchers will also attend school orientation, open-house events, after-school fairs and/or specific Q&A sessions at the schools or online to provide information about the study to parents and students. If students are interested in participating, researchers will have consent, assent, and parent permission forms available for students and their parents/guardians to sign. Students will be asked to contact the researcher if they are interested in participating. Once they have contacted the researcher, they will be given a physical parent consent form to bring home for their parent or guardian to sign. After their parents have reviewed and signed their consent form granting them permission to participate, students will be provided with an assent form to read and agree to participate.

Parents will receive an extra consent form and will be asked to complete questionnaires at the beginning and end of their child’s intervention. In that consent form, parents will be asked whether they would like to receive information about additional optional procedures within this same study. If they decide to receive this information, the research team will explain the optional neuroimaging sessions their child could complete, and will provide the parents with both parental consent and child assent forms for both MRI and MEG sessions.

While participants for the MUS and VIS groups will be recruited from the participating high schools, two complementary strategies will be employed to recruit the control group: (a) utilising an internal lab database, and (b) extending recruitment efforts within the same participating high schools. In the first case, a database of former research participants from our labs and institutions who agreed to be recontacted will be used to find a CON group matched in age, gender, and socioeconomic status to the active-intervention groups. In both cases, participants from the database who agreed to participate in SMART will complete a baseline and a 3-month follow-up evaluation and will be contacted weekly to complete the mood and SM usage surveys and will not receive any training. In the second case, participants in the CON group recruited from participating high schools, will be offered to complete the training of their choice in the following semester after they have completed their participation as a control participant in the programme.

Student participants will be compensated for their time in each of the two behavioural assessments with a $25 Visa gift card. If any student withdraws from the study earlier than expected, they will be compensated with Visa gift cards for their first assessment session only. Parents of students participating in SMART will be also compensated for each of their assessments with a $25 Visa gift card, being only compensated for their first assessment if they withdraw from the study early. Compensation of students and parents will be handled independently, following each participant’s assessment session(s) completion. The amount allocated for compensating participants for their time invested in the behavioural assessment part of our protocol (i.e. $25) was chosen to match the maximum amount possible as per the NYC Department of Education guidelines for studies conducted in public high school populations.

If students choose and are given permission to participate in the optional neuroimaging part of the study, they will be compensated for their time with a $40 Visa gift card for each of the two scanning sessions (MEG, MRI) at each assessment point (baseline, post-intervention). They will also receive a $10 Visa gift card to reimburse their travels to NYU facilities, where the Center for Brain Imaging is located. If they choose to only complete one of the neuroimaging sessions or only one of the assessments, participants will be paid consequently, based on the sessions they complete. Amounts for compensating participation in the neuroimaging part of SMART (i.e. $40 per scan, $10 for transportation) were decided based on usual practices at NYU’s Center for Brain Imaging.

2.3.5. Group assignment procedure

Group assignments will be performed based on the participating high school demands and curriculum (during regular or after-school classes), as well as on teaching artists’ availability, while attempting to take students’ preferences into account as much as possible. Thus, full randomisation will not be implemented. Variables to control for school of origin will be included in the statistical analyses to compensate.

2.4. Study personnel

2.4.1. Teaching artists

Teaching artists (TAs) in charge of SMART’s arts-based interventions have been selected to: (1) have extensive artistic experience in the disciplines included in our interventions, and (2) have experience in teaching and in developing and leading workshops with an artistic curriculum.

For the music composition/edition intervention, all TAs are part of the American Composers Orchestra (ACO) education team. ACO is a New York-based organisation dedicated to the creation, celebration, performance and promotion of orchestral music by American composers. ACO maintains a robust educational programme for teenagers and young adults that leverages music composition and improvisation as a gateway to creative thinking.

Regarding the photography composition/edition intervention, agreements have been placed with freelancing TAs, who have been selected with the following criteria: (1) professional artists showing specific higher education in visual-arts/filmmaking matters, (2) extensive experience creating and participating in visual arts/filmmaking projects with an important emphasis on photography, (3) proven experience in the use of open-source digital tools, and (4) experience teaching visual-arts / photography / filmmaking content for youth audiences.

2.4.2. High school personnel

In order to adapt to actual demands of the participating high schools, and to disrupt as little as possible their regular activities, the SMART project will be implemented either (a) as part of the regular advisory–arts’ lesson curriculum or any other open-curriculum subject, or (b) as part of the after–school activities programme.

In the first case, the regular high school teacher in charge of the assigned subject will be present at all times during the intervention sessions/periods. These teachers will usually have a background in performing arts or science-related domains. In the second implementation alternative, personnel from the counselling team or other personnel assigned by the principal to supervise the after-school activities will be present during the intervention sessions.

2.5. Description of the interventions

The curriculum / syllabus for both interventions was planned and developed with teaching artists and education directors of ACO –MUS intervention–, and the Downtown Community Television Center (DCTV, a non-profit organisation that aims to use filmmaking to inform and empower communities, with accessible visual-arts educative programmes) –VIS intervention. To create parallel content between the music and the photography syllabi, the intervention materials were created simultaneously for both types of intervention-groups (MUS, VIS). ACO’s Sonic Spark Lab’s lesson programme (https://www.americancomposers.org/education/sonic-spark-lab) served as the basis for our novel curriculum. ACO Sonic Spark Lab is an educational arts-integrated programme addressed to middle and high/school students that teaches how to engage creativity and imagination to transform an idea or ‘spark’ into an artistic project, such as a full song. Several aspects of ACO Sonic Spark Lab’s programme were tailored to fit the scientific demands, timeline, and ideas of the SMART project. The materials for the VIS intervention were created by DCTV specialists by matching the structure of the MUS ones, with input from the freelancing visual-arts TAs. The materials for the MUS and VIS intervention programmes, including the social media component of the interventions can be found in the Supplementary Materials.

2.5.1. Music composition / edition (MUS group)

Lessons in the MUS intervention will review basic structural musical concepts, making the students practice active listening and music appreciation skills, while teaching them to use specific open-source music-composition software (i.e. Soundtrap). Specific software was selected to be accessible (free to use) and compatible with the usual digital devices the students have access to (e.g. their smartphones, tablets, and/or class laptops). Overall, our intervention programme will give the students tools to express themselves and express their identity through music, while learning about musical storytelling through the creation of their own musical pieces from scratch.

2.5.2. Visual-arts composition / edition (VIS group)

Lessons in the VIS intervention will review basic structural photography/visual-arts concepts as well as history of photography to provide a linear understanding of how photography as a media has evolved, teaching the students active observation and visual-arts appreciation skills, and the use of specific open-source photography edition software (i.e. built-in photo editors in students’ smartphones and the free software Pixlr). Specific software for this intervention was selected to be accessible to the students (free to use) and usable on their class laptops or other digital devices they may have access to. All in all, this intervention programme will give the students resources to express themselves and their identity through photography / visual-arts, while learning about visual storytelling through the creation of their own photography projects from scratch.

2.5.3. Social media component of the interventions

During the intervention period, activities involving the direct use of social media or reflections around it will be included, aligning with the goals and activities planned for every week as part of the arts-based sessions. This will be implemented via guided discussions with the students or activities implemented in class, under the guidance of the SMART research team, for 15 min a week. Among other topics, during this allocated time students will participate in activities including: looking for artistic inspirations, commenting on their favourite artists, discussing popularity online, reflecting on what artists share online, converse about the expression of opinions online / hate culture, and giving and managing constructive criticisms. The social media component of the interventions can be found in the Supplementary Materials.

2.6. Outcome measures

2.6.1. Social media usage

We have developed a self-report questionnaire, the ‘Usage Attitudes & Opinions about Social Media Questionnaire’ (UAOSMQ) to measure different aspects of social media usage. The questionnaire will be completed at baseline and post-intervention. The full questionnaire can be found in the Supplementary Materials. In addition to the UAOSMQ, participants will provide weekly measures of social media use. These measures will be collected for all groups, including the CON. The following sections specify the different SM-related measures that will be collected.

2.6.1.1. Changes in attitudes and importance given to SM

Changes in attitudes and importance given to SM will be measured using the following items of the UAOSMQ: 6 (open/ended question about their opinion of SM) and 7 (multiple forced choice or MFC with 5 levels, ranging from ‘Not important at all’ to ‘I can not imagine a day without using social media’).

2.6.1.2. Changes in SM profiles followed in SM

Changes in SM profiles followed will be measured using the following items of the UAOSMQ: 3 (list where they can mark and add as many options as needed), 4 (MFC to mark whether the profiles followed are mostly female, male, mixed, LGBTIQ+ or neutral-led or if they do not know), and 5 (list to mark the options that apply to what they expect from the accounts they follow, i.e. entertainment, news, etc.). Furthermore, the post-intervention version of the UAOSMQ includes an item for the participants to self-report on their perceived change in profiles they follow: item 16 (open-ended question about whether they have changed the profiles they follow).

2.6.1.3. Changes in mood and emotional reaction while being online

Changes in Mood and Emotional Reaction while being online will be measured using the following items of the UAOSMQ: 8, 10.1, and 10.2 (these three items consist of lists of emotions to mark and add as many as apply for them in each situation).

2.6.1.4. Changes in detrimental behaviours while being online

Changes in several negative online behaviours will be measured using UAOSMQ items: 9.1 (MFC with 5 levels to report whether they compare themselves with the people they see online, ranging from ‘Yes, always’ to ‘No, never’) and 9.2 (MFC with 3 levels to report how they feel when they compare themselves with people online, ranging from ‘Mostly happy’ to ‘Mostly unhappy’).

2.6.1.5. Changes in the use of smartphones during in-person social contexts

Changes in the use of smartphone / social media while being with friends will be registered with UAOSMQ items: 13.1 (MFC with 5 levels to report whether they use their smartphone while being with friends, ranging from ‘Yes, always’ to ‘No, never’), 13.2 (MFC with 5 levels to report whether they use SM specifically, when using their smartphones in this social gatherings, ranging from ‘Yes, always’ to ‘Non, never’), and 14 (open-ended question for them to explain what they think the use of smartphones and SM in social contexts adds to the situation).

2.6.1.6. Changes in self-reported overall screen-time

Changes in self-reported overall screen-time will be measured using item 1 of the UAOSMQ (MFC with 5 levels, ranging from ‘Less than 1 hour/day’ to ‘More than 10 hours/day’).

2.6.1.7. Changes in SM platforms and overall perceived usage

Using UAOSMQ item 2 (participants can mark and add as many options as needed), we will measure changes in the SM platforms used. In addition, during the post-intervention evaluation, there are 3 items that will help us measure the changes the participants have perceived in their usage of SM: 15.1 (MFC with 4 levels to report whether they think they have changed in their way of using SM, ranging from ‘Not really, I use them in the same way’ to ‘Yes, I have completely change the way I use social media’), 15.2 and 16 (both of them open-ended questions to report how has that change been and what have been the changes in the profiles they follow).

2.6.1.8. Changes in types of posts shared online

Changes in types of posts shared online will be measured using the following items of the UAOSMQ: 11.1 (open-ended question about what they share online), 11.2 (MFC with 6 levels to report frequency of posting, ranging from ‘Once a month’ to ‘Several posts per day’), and 12 (subdivided in 6 sub-items showing different potential situations and then asking how likely they would be to share a post in that situation using a MFC that reflects a range of ‘taking a couple of pictures or video and sharing on your favorite platform’ to ‘taking a picture or a video for your records only’).

2.6.1.9. Weekly evolution of overall screen-time

Number of hours of screen-time use will be provided by each participant via a self-report questionnaire to be filled by using the screen-time built-in app of their smartphone once a week.

2.6.1.10. Weekly evolution of apps most used during weekday

Participants will provide the list of actual most-used apps and for how long has each of them been used during a given weekday as listed in the built-in screen-time app of their smartphones. Specifically, we will ask the participants during a weekday (Tue-Fri, depending on the days the SMART team will be in the school in each participating site) to fill in the information corresponding to the previous day.

2.6.1.11. Weekly evaluation of apps most used during weekend day

As recorded for weekdays, participants will consult and annotate every week the list of their most-used apps and the amount of time invested in each of them during a weekend day. On the day of registering these weekly measures, we will ask the participants to fill the information corresponding to the previous Saturday.

2.6.1.12. Output measures from parental reports

Parents/guardians will fill the parent/guardian version of the UAOSMQ as well, in order to measure parents’ attitudes and importance towards SM, but mostly, their perceptions of their children’s attitudes, usage and importance of SM, structured in a parallel way as the UAOSMQ filled by the teen participants, and differing on items: 1.2 (open-ended question about their opinion on the amount of time their child spends online), 6 (open-ended question about parent’s opinion on SM –pre-intervention–, or whether their opinion has changed in the last 3 months –post-intervention–), 7 (MFC with 5 levels, ranging from ‘less than an 1 hour/day’ to ‘more than 10 hours a day’ to measure the time the parents spend on SM), 8 (lists of emotions to mark and add as many as apply to report the feelings parents have while using SM), and 14 (open-ended question to report whether the parents have ever shared their personal opinion about SM with their child). The parental version of the UAOSMQ can be found as Supplementary Material.

2.6.2. Mental health markers

2.6.2.1. Evolution and changes in mood

The ‘Positive and Negative Affective Scale’ (PANAS – Watson et al., 1988; primary outcome) will be collected at baseline and post-intervention and also on a weekly basis during the intervention period (or 3-month participation, in the case of the CON group). The PANAS consists of a list of 10 positive emotion-related adjectives and a list of 10 negative emotion-related adjectives, presented in an intercalated order. Its score ranges from 10 to 50 for Positive Affect, with greater values meaning greater presence of positive mood; and from 10 to 50 for Negative Affect, with greater values meaning greater presence of negative mood.

In addition, we will also measure the baseline vs post-intervention potential differences in Total Mood Disturbance index (secondary outcome), a score obtained through the Profile of Mood and States Scale (POMS, McNair et al., 1992). The Total Mood Disturbance measure is calculated out of summing up the negative-mood subscales (Anger-Hostility: 0–20; Confusion-Bewilderment: 0–20; Depression-Dejection: 0–28; Fatigue-Inertia: 0–20; Tension-Anxiety: 0–24) and subtract the positive mood subscale (Vigor-Activity: 0–24). The higher the scores the greater the probability of having a mood-related issue.

2.6.2.2. Changes in self-Esteem

We will use the Rosenberg Self-Esteem Questionnaire (Rosenberg, 1965) at both evaluation timepoints to understand potential post – baseline differences in self esteem (primary outcome). This test ranges from 0 to 30, with scores between 15 and 25 indicating a normal range and scores below 15 suggesting low self esteem.

2.6.2.3. Changes in depression

The Children Depression Inventory (CDI, Kovacs, 1981, 1992) will be used to measure post – baseline differences in depressive state (primary outcome) in our participants. The CDI is a 27-item scale ranging from 0 to 54: the higher the score, the higher the depressive state.

In addition, we will also collect the ‘Depression, Anxiety, and Stress Scale 21’ or DASS-21 (Lovibond & Lovibond, 1995). Using the DASS-21 – Depression subscale we will measure differences between baseline and 3-month/post-intervention in depression states (secondary outcome). 7 items are dedicated to measuring depression within the DASS-21 scale. Scores range from 0 to 42 points, with higher scores indicating more severe depressive symptoms: scores 14–20 suggest moderate depression, scores greater than 28 suggest extremely severe depression.

2.6.2.4. Changes in anxiety

Anxiety will be mainly measured via the ‘Screen for Child Anxiety Related Disorders’ (SCARED, Birmaher et al., 1997) questionnaire, whose scores range from 0 to 82. Values greater than 25 in the total score indicate the presence of anxiety disorder. Further, it can be subdivided in subscales for Panic Disorder (cut-off: 7), Generalized Anxiety Disorder (cut-off: 9), Separation Anxiety Disorder (cut-off: 5), Social Anxiety Disorder (cut off: 8), and School Avoidance (cut off: 3). Hence, post – baseline differences in Anxiety as measured by SCARED (primary outcome) will be explored.

Moreover, we will measure changes in anxiety before and after the intervention period by applying the DASS-21 – Anxiety subscale (secondary outcome). DASS-21 dedicates 7 items to measure anxiety. Scores range from 0 to 42 points, with higher scores indicating more severe depressive symptoms: scores 10–14 suggest moderate anxiety, scores greater than 20 suggest extremely severe anxiety.

Lastly, we will also measure post-intervention – baseline changes in social anxiety (secondary outcome) via the ‘Social Anxiety Scale for Adolescents’ questionnaire (SAS-A, La Greca & Lopez, 1998). The SAS-A is an 18-item scale whose total scores range from 18 to 90 points; it also counts with subscales in Fear of Negative Evaluation (8–40), Social Avoidance and Distress–New (6–30), and Social Avoidance and Distress–General (4–20). The higher the scores, the greater the social anxiety symptoms (the usual recommended cut-off for interpreting this test’s results is 50; Ranta et al., 2012).

2.6.2.5. Changes in stress

Post-intervention – baseline differences in stress (secondary outcome) will be measured via the DASS-21 – Stress subscale. 7 items within the DASS-21 are dedicated to evaluating stress states. Scores range from 0 to 42 points, with higher scores indicating more severe stress symptoms: scores 19–25 suggest moderate stress, scores greater than 34 suggest extremely severe stress.

2.6.2.6. Output measures from parental reports

Participants’ parents who decide to participate in the project will be asked to fill a baseline and a post-intervention evaluation concerning some aspects of their child’s mental health. Specifically, participating parents will fill the parent/guardian report from: BASC3 (a questionnaire with 29 MFC items identifying behavioural and emotional risk of the child, evaluating the relationships between the student and their parents, conformed by 4 subscales with specific ranges of normality, Reynolds & Kramphaus, 2015), and SCARED (a test with 41 MFC items measuring social anxiety in their kids, with scores above 25 indicating potential presence of social anxiety in their child; Birmaher et al., 1997).

2.6.3. Cognitive markers

Based on previous findings of art-based interventions’ influence on cognitive functions (Chatterjee et al., 2011; Flaugnacco et al., 2015; François et al., 2013; Frischen et al., 2021; Jaschke et al., 2018; Hennessy et al., 2019; Särkämö & Soto, 2012), we decided to assess the effects of the intervention in attention, verbal learning and memory, and working memory.

2.6.3.1. Changes in selective attention

Selective attention will be measured via a computerised attentional flanker task (Eriksen & Eriksen, 1974; Sanders & Lamers, 2002) previously validated elsewhere (Orpella et al., 2025). In brief, in each trial, a single line formed with 5 arrows is displayed on a screen. The surrounding arrows in the row could: point in the same direction as the one located at the centre (congruent condition), point in the opposite direction (incongruent condition), be changed by squares (neutral condition), or be changed by ‘X’ symbols in which case the participant will have to inhibit their response (‘No-Go’ inhibition condition; see Figure 2(A,B)). Participants will be asked to respond as quickly as possible using the computer keyboard arrow keys whether the arrow in the centre of the row is pointing to the left or the right, except in the case of the No-Go trials, where the correct response is not to press any button. Each trial is preceded by a fixation cross displayed in the centre of the screen for 2 s; then, the flanker stimulus appears for a maximum of 2 s and participants’ responses will be encoded during a window of 3 s maximum after the flanker stimulus appears on the screen.

Figure 2.

Schematic representation of the cognitive tasks. Flanker Task: (A) Depiction of the Familiarisation Phase, with fixation cross between trials, a possible response in the arrow-buttons of the computer keyboard, followed by the corresponding feedback in each case. From top to bottom, congruent, incongruent, neutral, and No-Go conditions. (B) Test Phase, with the same organisation and type of presentation of stimuli and conditions, but without feedback given to participants. FOIB task: (C) Target Familiarisation Phase in the two possible memory set sizes (top, high or MMS7: 7 different targets; bottom, low or MMS2: 2 different targets). (D) Recognition test, where participants are presented with individual stimuli (faces) and they have to answer whether they are targets or not. Feedback is provided to reinforce memorising the correct targets. (E) Task Familiarisation & Test Phases are presented in the same manner: moving stimulus (i.e. faces) on a grey background, with a score counter and a Next button at the centre of the screen. The difference is in the first part (Task Familiarisation) participants have to achieve 50 points, while in the actual task (Test Phase) they have to get up to 200 points by recognising and tapping on the correct targets. Abbreviations: MMS: memory set size; Y: Yes; N: No.

The task will start by presenting 24 training trials to familiarise the participants with the task. After each of these trials, feedback is provided by displaying on the screen the words ‘Correct!’ or ‘Incorrect!’ immediately after response or after the 3-second after-stimulus onset window if no response is provided. After the training trials, 72 test trials will have to be completed, with the same structure as the training trials but without any feedback displayed on the screen. Testing trials will be organised as follows: 18 trials per condition, with half of the presentations showing the centre arrow pointing towards the right, and the other half pointing towards the left. Furthermore, the task has been programmed so that trial duration is standardised in order to maintain the overall duration of the task at 6 min. Trial standardisation is provided by making the test trials appear exactly 5 s apart by introducing variable duration intervals after participants’ responses in a way that reaction time plus the introduced interval is always equal to 3 s; adding up the two seconds allocated for the pre-stimulus fixation cross presentations, results in the 5 s window of maximum total trial duration.

Measures of response accuracy and reaction time (measured from the onset of stimulus presentation) for congruent, incongruent, neutral trials will be obtained. For No-Go trials only accuracy will be used as the correct answer is not to press any button and, therefore, reaction times cannot be collected. Variables from the Flanker task will allow us to measure post-intervention – baseline differences in attentional and inhibition/cognitive control functions (primary outcome).

2.6.3.2. Changes in focused attention and executive functions

A previously developed gamified task called ‘Foraging Inattentional Blindness Task’ (FOIB) or ‘FORAGEKID’ (Gil Gomez de Liaño & Wolfe, 2022) will be used for exploring focused attention and executive functions at both evaluation timepoints. In collaboration with these authors, we modified the FORAGEKID task to adapt the stimuli to something more attractive to adolescent participants (pictures of celebrities, instead of pictures of stuffed animals). The FOIB task allows the simultaneous evaluation of executive functions and attention development in an engaging and fun way, operating as a videogame. In this game, participants are presented with visual targets and distractors that move around the screen, and they have to find some preselected targets as quickly as possible until reaching a pre-agreed score.

The task has been programmed in Matlab 2018b (The Mathworks, Natick, MA, USA) and the Psychophysics Toolbox version 3. The task will be presented on Dell Inc. Inspiron 15 3515 laptops (with Microsoft Windows 11 as OS) and participants will have to respond by using the laptop’s trackpad. Monitor resolution will be 1366 × 768 pixels. The stimuli (targets and distractors) will be randomly moving at a constant speed of 44 pixels/sec, changing directions at pseudo-random intervals. The stimuli are images of well-known celebrities (singers, actors, fashion models, politicians, etc.) presented on a homogeneous grey background. In the centre of the screen, a score will be displayed and updated, with the amount of points the participant is gaining or losing, depending on their performance on the task (correctly identified targets add 1 point, non-target stimuli selected as targets will subtract points to the final score).

Stimuli images will be drawn from an image pool of 150 different ‘celebrities’. Each image will be a circle placed on an invisible rectangle that would subtend a visual angle of 1.33° height, 1.00° width at a 60 cm viewing distance (although viewing distance will not be controlled). In the different screens, image presentation will change with possible set sizes being 40, 80 or 120 images. Targets will constitute 20–30% of the presented items in each display. Participants will be not informed about the number of targets or set sizes, and targets will be randomly selected for them in each display. Two runs of the task will be completed by each participant to allow for two memory load conditions: Low-Load Memory Set Size with 2 targets to look for (MMS2), and High-Load Memory Set Size with 7 different targets to search (MMS7). The memory set size indicates the number of potential targets that have to be held in memory by the participants, which is different from the visual set size of the number of image items displayed in each presentation screen of the game. The order of presentation of MMS2 and MMS7 will be randomised across participants.

The task contains a short target familiarisation phase, in which the whole MSS is presented, followed by individual presentation of each image in the MSS, and ending with a recognition/memory test in which participants are presented with images adding up to double the size of the MSS (4 in the MSS2 and 14 in the MSS7, 50% of them will be targets), and asked one by one whether the image is a target or not. Participants will have to get at least 80% of correct responses to proceed with the game part of the FOIB task; otherwise, they will repeat the recognition/memory test until reaching the 80% of correctly recognised targets. Participants will have to answer whether the presented images are targets (blue) or not (red) using two keys of the laptop keyboard marked with a sticker in blue and red colours. Responses will be followed by immediate feedback: the words ‘Correct!’ or ‘Incorrect!’ will be displayed in the centre of the screen. After this, a task familiarisation phase will be completed, where the participants will have to recognise their targets among the moving stimuli, and select them as quickly as possible, completing a total score of 50 points. Participants can always, at any point during the task, change displays by clicking on a ‘Next’ button located on the centre of the screen, right above the score counter. Travel time between displays after the ‘Next’ button is touched is set to 2 s. After completion of the task familiarisation phase, a reminder with all targets appears on the screen. Then, the test phase will be completed, which will proceed as the familiarisation phase, but participants will have to reach 200 points this time. This procedure will be repeated for both MSS2 and MSS7 in the corresponding order for each participant. Total completion of the two runs will take around 20–30 min. See Figure 2(C,D,E) for a schematic representation of this task.

Diverse variables elicited from response times and search strategy can be computed from the FOIB task results. First, percentage of correct responses from the total tapping times (the times the participant used the trackpad to select a stimuli item to respond) will be computed. Then, regarding response patterns, we will study runs, defined as repetitions of target response (i.e. tapping the image of Billie Eilish when the previous tapped target was also Billie Eilish), and switches (i.e. tapping the image of Brad Pitt after tapping on Billie Eilish). Runs and switch trials will equal 100% of correct responses. Since participants can change display at any point during the task, targets left on the previous display can be counted as misses. Then, when distractor items are tapped on, selected as targets, these will be considered as false alarm (FA) errors, which will grant us the computation of FA-Error Rates. Then, to study search quitting behaviour, we will calculate the average rate of leaving one screen for the next display (calculated as the total number of points collected at the point of clicking the ‘Next’ button, divided by the time spent in the whole task), and the instantaneous rate of leaving for the next display (calculating the average reaction times for each of the last 10 items tapped upon right before clicking the ‘Next’ button, counting in reverse order from the display switch moment).

All these variables will give us information about focus attention, visual memory and visual search strategy, allowing us thus for the study of executive functions. Hence, post-intervention – baseline differences in focused attention, visual search patterns and executive functions (secondary outcome) will be explored.

2.6.3.3. Changes in verbal learning and memory

Measures of verbal learning and memory will be gathered via the Rey’s Auditory Verbal Learning test (RAVLT, Schmidt, 1996; Vakil & Blachstein, 1993), a standard test consisting in the presentation and immediate recollection of a 15-word list for five times, the presentation and recollection of a distractor list, and a final recollection of the repeated list without presenting it again. The higher the number in each trial or on the memory trial, the better the participant’s verbal learning and memory functions are assumed. Thus, measures from the RAVLT will grant us the possibility of studying potential post-intervention – baseline changes in verbal learning and memory (primary outcome).

2.6.3.4. Changes in auditory working memory

Auditory working memory will be tested at both evaluation timepoints through the WAIS-V (Weschler Adult Intelligence Scale, Pearson) Letter Number Sequencing, a subtest of the IQ battery consisting of the presentation of letters and numbers in a mixed order, followed by the participant’s recollection and manipulation of the stimuli so that they mentally organise and say out loud first the numbers in increasing order, and then the letters in alphabetical order. Scores range from 0 to 21, with higher scores indicating higher auditory working memory abilities (Egeland, 2015: Wang et al., 2023). This test will allow us to explore potential post-intervention – baseline differences in auditory working memory (primary outcome).

2.6.3.5. Changes and insights from parental reports

Parents will also fill the BRIEF2 questionnaire (Gioia et al., 2015), a 63 MFC-item tool to discard learning disabilities, attention and executive function disorders, and other neuro and neuropsychological developmental issues.

2.6.4. Neuroimaging markers – MRI

Using a 3.0T Siemens Prisma scan located at the Center for Brain Imaging at New York University, participants who assent and whose parents consent of their participation in the neuroimaging section of the protocol will complete an MRI session at baseline and post-intervention. This session will serve to obtain anatomical and structural connectivity. The following sequences will be collected: T1-weighted, as previously used to explore grey matter alterations after arts-based interventions (Bashwiner et al., 2016; Habibi et al., 2018; Hudziak et al., 2014); diffusion-weighted MRI, to explore plastic effects on white matter after arts-based training (Cheng et al., 2023; Hong et al., 2023; Li et al., 2018); resting state data, to study differences in (MRI) functional connectivity at rest after arts-based interventions (Bashwiner et al., 2020; De Pisapia et al., 2016; Hennessy et al., 2019; Jünemann et al., 2023; Li et al., 2018); and a pathology-discarding T2-weighted structural image.

Before entering the scanner, participants will have to answer a scanning contraindication form to make sure it is safe for them to undergo this procedure; everything metallic that participants may be carrying on their outfits / bodies will be removed and a metal detector will be used to guarantee participant’s safety to complete the MRI session.

2.6.4.1. Changes in functional connectivity at rest

Participants will complete a 6 min-long MRI resting state sequence with the following parameters: voxel size = 2.0 × 2.0 × 2.0 mm, TR = 1000 ms, TE = 37.40 ms, 66 interleaved slices, FoV = 208 mm. Participants will have to maintain their gaze looking at the screen, where a fixation cross will be placed in the centre for the total duration of this sequence (i.e. 6 min). They will be asked to keep their eyes open looking at the screen, and be relaxed but as still as possible, trying not to fall asleep. Resting-state information will grant the study of potential differences in functional connectivity at rest between baseline and post-intervention evaluations (primary outcome).

2.6.4.2. Changes in grey matter morphology / brain anatomy

Participants will complete a T1-weighted MRI sequence at both evaluations, that will allow for the study of participants’ cortical and subcortical grey matter. MPRAGE T1-weighted data will be acquired with the following parameters: voxel size = 0.8 × 0.8 × 0.8mm, TR = 2400 ms, TE = 2.24 ms, 256 interleaved slices acquired in a single shot, FoV = 256 mm. During the anatomical acquisition sequences, a neutral-relaxing video with slow-motion nature pictures and no sound will be displayed to entertain participants, and to reduce potential anxiety for being inside the scanner. Participants will be asked to be as relaxed but as still as possible and either look at the video or close their eyes. T1-weighted data will grant us the possibility of studying post-intervention – baseline differences in grey matter brain morphology (primary outcome).

2.6.4.3. Changes in brain structural connectivity

A diffusion-weighted MRI sequence will enable the study of participants’ white matter structural connectivity. The following parameters will be used to acquire this data: voxel size = 1.5 × 1.5 × 1.5 mm, TR = 4150 ms, TE = 85.20 ms, 81 interleaved slices, FoV = 225 mm, 128 diffusion-weighted volumes with 9 interleaved non-diffusion-weighted images; b-value of 1500 s/mm2. As described for the T1-weighted sequence, participants will be asked to be as still as possible and relaxed, and to either look at a neutral nature video without sound or to just close their eyes.

Diffusion MRI data will allow for exploring the baseline vs. post-intervention differences in brain structural connectivity (primary outcome).

2.6.5. Neurophysiological markers – MEG

Participants who assent and whose parents consent to their participation in the neurophysiological section of the protocol will complete an MEG session. Data will be collected using a 157-channel whole-head MEG biomagnetometer with 5-cm baseline axial gradiometers and max inter-sensor spacing of 2.5 cm (Kanazawa Institute of Technology, Kanazawa, Japan) scan located at the Center for Brain Imaging at New York University (NYU)’s premises.

Before entering the MEG scanner, all participants will have to fill a form with some demographic information, and will have to answer again a scanning contraindication form to make sure it is safe for them to undergo this procedure. In order to localise their head position during the recordings, five electromagnetic coils will be attached to the forehead of each participant. The location of these coils will be registered with respect to the MEG sensors at the beginning and end of the experimental session. Participants’ head shape will be digitised immediately before entering the MEG scanner using a Polhemus digitiser in combination with the Source Signal Imaging 3D digitiser software, taking into account the anatomical landmarks where the coils will be placed plus two additional points: the nasion and bilateral tragus. This will grant the proper alignment of MEG data with an anatomical magnetic resonance template or participants’ own MRI anatomical data, if available for the whole cohort. After this preparatory phase, participants will enter the magnetically shielded room, and lay supine for the whole MEG recording session. Visual and auditory stimuli (including the fixation cross used during the resting-state task) will be presented to participants by using Presentation software in a PC connected to the MEG display system.

The experimental session will consist of a resting state sequence, an auditory function localisation task, and an auditory / melodic pre-attentive change detection paradigm. We decided to focus on auditory-related tasks (auditory localisation + auditory violation-detection) because, based on previous reports (Flaugnacco et al., 2015; François et al., 2013; Frischen et al., 2021; James et al., 2024; Jaschke et al., 2018; Lu et al., 2022; Wang et al., 2022), we hypothesise that the music composition intervention will have the greatest neuroplastic effect – when compared to the passive control or the photography intervention groups.

The auditory function localisation task performed in the middle of our session will be only used as a confirmation of the spatial location of auditory processing in our participants, but no post-intervention vs. baseline differences are expected or will be checked in this data. In this task, participants will be presented with 2 tones (250 and 1000 Hz) in an alternating way with a jitter at 900, 1000, 1110, 1200 and 1300 ms. Each tone will be presented 100 times, with a duration of 0.40 ms each time. Stimulus presentations will be done through the MEG-compatible headphones, while we ask participants to fixate their gaze on a fixation cross displayed at the centre of the screen. The duration of this task will be 5 min.

2.6.5.1. Changes in neurophysiological correlates of pre-attentive auditory detection

A previously validated melodic multi-feature MMN paradigm (Putkinen et al., 2014; Vaquero et al., 2021) will be implemented. MMN responses are consistently evoked and detected in healthy participants when presenting them with a sequence of repeated auditory stimuli –standard– that gets suddenly altered by a less probable stimulus –deviant– that changes particular sound features (duration, frequency, or intensity of a tone, the interstimulus interval, etc.). These sound features suppose a combination of different characteristics or the presentation of more complex patterns, or that imply the absence of an expected trait in the pattern with respect to the previously presented stimuli (Grimm & Escera, 2012; Näätänen et al., 2005). Interestingly, previous works in the field have reported observable modifications in MMN responses after musical training or long-term practice (Paraskevopoulos et al., 2012; Putkinen et al., 2014; Virtala et al., 2014).

Thus, during this paradigm, participants will be presented with a silent movie (an excerpt from ‘Modern Times’ –Charles Chaplin, 1936– prepared explicitly for this experiment) and asked to pay attention to it. At the same time, participants will be passively presented via MEG compatible headphones with a continuous melodic stream where standard melodies get altered in melodic or rhythmic features. The MMN responses to these auditory violations have been suggested to reflect pre-attentive detections of discrepancies between the input sound and what was predicted based on the previous auditory presentation, implying that there is a memory trace serving as an internal template against which sensory inputs get compared to in a sort of automatic fashion (Koelsch et al., 2019; Näätänen et al., 2007; Winkler et al., 2009). The visual stimuli (silent film, in this case) ensures thus that participants’ attention is on the visual and not the auditory stimuli. Participants were instructed to pay attention to the action of the movie, and asked after the experiment to respond to a 10-item questionnaire about the movie. The score in this questionnaire will be used as a proxy of the level of attention participants were paying to the visual stimuli.

The auditory stimuli consists of 360 brief melodies previously composed specifically for this paradigm (full details about these melodies are described in Putkinen et al., 2014). In brief, they are melodies played on acoustic piano tones respecting the Western tonal rules and recursively repeated. Melodies start with a triad (300 ms) followed by four tones varying in length, with a 50-ms gap between each tone, and followed by a final tone whose duration is 575 ms. Each melody lasts 2100 ms, and they will be presented with a 125 ms gap between them. The whole experiment lasts for 15 min, where participants will be presented with these melodies in a looped manner.

Six different deviants can be included in the melodies: three low-level changes (i.e. not affecting the melodic contour) and three high-level changes (i.e. modifying the melodic contour). Melodies can contain more than one deviant or change. Low-level changes are: (i) Mistuning: moving upwards the pitch of a tone half a semitone, 3% of the fundamental frequency of the original sound (appearing in 20% of the total set of melodies); (ii) Timbre Deviant: the timbre of one tone is change for that of a flute (instead of the original piano played for the rest of the tones, happening in 16% of the melodies); (iii) Rhythm Mistake or Timing Delay: including a 100 ms silence gap (appearing in 17% of the melodies). High-level changes are: (i) Melody Modulation: altering the third or four tone of a melody in pitch (happening in 22% of the melodies, this slightly alters de presented melody and continues until a new Melody Modulation appears); (ii) Rhythm Modulation: affecting the second or third tone, in this deviant a long tone gets replaced by a short tone (shortening) or a short tone gets replaced by a long one (lengthening) by introducing the reversal of the duration of two sequential tones (appearing in 22% of the melodies); (iii) Transposition or Key Change: the whole melody gets transposed by one semitone up or down its original key (happening in 26% of the melodies) –after a transposition, the following melodies stay in the transposed key until a new Transposition deviant appears (Tervaniemi et al., 2014). All deviants were musically plausible, and the high-level variants become the new repeated form of the melody in the following presentations in a roving-standard fashion (Cowan et al., 1993). Only one deviant of the same kind can occur within each melody, and each single tone can be altered only by one type of deviant. After a high-level change, the ‘new standard’ melodic / rhythmic pattern or key will be repeated at least once (on average, three repetitions happen) before the next change of the same kind appears.

The MMN paradigm will be repeated in the MEG protocol at both evaluation times. This way, we will explore whether there are any differences after the intervention in these neurophysiological correlates of pre-attentive musical/auditory detection (primary outcome).

2.6.5.2. Changes in functional connectivity at rest within different frequency bands

A resting state sequence with eyes open will be collected at both evaluation timepoints. Participants will have to fix their gaze on a fixation cross displayed on the screen in front of them while laying on the MEG bed, without doing any particular task. This sequence will last for 5 min. Potential baseline vs. post-intervention differences in brain connectivity at rest between brain sources of interest will be calculated within different frequency bands (secondary outcome) and compared between groups.

2.6.6. Intervention observation measures

In addition to the TAs, a team member will attend every lesson of each intervention group (MUS, VIS) and will collect information about treatment fidelity and student engagement.

2.6.6.1. Treatment fidelity

Accuracy and fit between the actual lessons carried out and the original planned lesson programmed for every week and session of the intervention will be annotated. These annotations will take the form of a percentage of accuracy and fit to the original plan –in other words, treatment fidelity (secondary outcome).

2.6.6.2. Students’ engagement during interventions

During each intervention session, participants’ attendance and engagement (secondary outcome) will be recorded and rated using a 1–7 Likert scale.

2.6.6.3. Students’ feedback about the intervention

At the end of the intervention, students participating in the active groups (MUS, VIS) will fill a questionnaire in order to rate how much they have liked the intervention programme, and express their opinions about the type of intervention programme in which they have participated. We will use a brief questionnaire specifically developed for this purpose (see Supplementary Materials).

2.7. Data management and storage

Subjects will be identified by a code for data files that will be used for analysis within the study team. The correspondence between identifiable data and participant codes will be in an encrypted file in a RAID server to which only the principal investigator (PI) has access to. Physical copies of informed consents will be stored in a secured locked cabinet in the PI’s office at New York University. Anonymized behavioural, neuroimaging and neurophysiological data will be stored in a local computer within NYU premises for data analysis. The complete anonymized dataset and workflow, with specific instructions for potential reanalysis and replication, will be made publicly available (shared with NYU and non-NYU affiliated researchers) at OpenAIRE, Zenodo, or similar public repositories. Identifiable data will be deleted after 5 years of data collection. For this protocol paper, all data and materials are available in the supplementary materials.

2.8. Statistical analysis

All behavioural analysis and analysis performed on neuroanatomical, structural or functional connectivity values extracted from neuroimaging or neurophysiological data will be performed using JASP computer software (JASP Team (2024); Version 0.19.3; https://jasp-stats.org/), R (R Core Team (2021); version 4.1.2; 2021-11-01; https://www.R-project.org/), MNE-python (Gramfort et al., 2013), the CONN toolbox (Whitfield-Gabrieli & Nieto-Castanon, 2012), Statistical Parametric Mapping software package v.12 (SPM12, Wellcome Department of Imaging Neuroscience, London, UK, available at: https://www.fil.ion.ucl.ac.uk/spm/), FMRIB Software Library (FSL www.fmrib.ox.ac.uk/fsl/), and the Automatic Fiber Quantification software (AFQ, Stanford University, described in Yeatman et al., 2012). For details see Supplementary Materials, section Statistical Analysis).

2.9. Monitoring, safety & ethics

The study was designed and is being carried out in accordance with the declaration of Helsinki (World Medical Association, 2013). The initial proposal for this project was reviewed, specifically regarding ethical concerns, by the European Commission in Research and Innovation. The final protocol has been reviewed and approved by the Institutional Review Board of New York University (NYU Protocol Record IRB-FY2023-7419), and the Institutional Review Board of the New York City Department of Education (NYC DOE IRB protocol number 4952). All study personnel engaging with students in the high schools –including TAs– have or will complete general ethics education applied to behavioural cognitive studies (CITI training), as well as appropriate federal background checks and fingerprinting (for child abuse and any other criminal activity clearance) to work within the public school system.

Furthermore, this study protocol has also been reviewed and registered as a Clinical Trial within the US National Institutes of Health system (ClinicalTrials.gov, trial number: NCT06402253).

Recruitment processes will be carried out and closely monitored with personnel from the participating high schools. Supervisors from all the collaborating institutions will monitor and ensure the appropriate development and completion of the protocol.

Treatment fidelity and students’ engagement will be monitored at every intervention session, and a field diary will be implemented to note as many details as possible of both positive and negative outcomes, feedback, or maladjustment that may occur during the development of the interventions, but also the project as a whole.

2.9.1. Mental-health issues response protocol

Since the beginning of the partnership with the different research sites, the scientific team of SMART has worked closely with the counselling team of the participating high schools to guarantee that any mental-health or safety issue detected in any of the student participants during SMART evaluations, weekly survey completion, or intervention sessions is properly and timely addressed. To that end, a mental-health issue response protocol tailored to the resources of each participating high school, as well as a mental-health resource document, were created in collaboration with the counselling department of each participating high school. These documents and protocols were originally inspired by materials from the Hartley Lab at NYU and are attached to this manuscript in Supplementary Materials. All involved personnel in SMART will follow the Mental Health Response Protocol adjusted to each research site’s resources, when and if needed.

2.9.2. Consultation and communication with direct stakeholders

Collaborating high school personnel participated in the development of the intervention programme, and are going to be consulted at least weekly to best adapt the implementation and individual sessions and activities to the student group’s general work-load, mood, and specific requirements during the implementation period. Further, each intervention session contains activities that can be easily adapted to specific requirements or preferences of the student group. As such, participating students will be consulted for preferences and interests at least weekly, if not in every session, to maintain an appropriate level of motivation throughout the sessions, while at the same time keeping the rigour and main structure of the intervention so that it is still replicable.

Lastly, in addition to the communication efforts planned for this project targeting scientifically and general audiences (frequent updates on the project’s website, outreach activities and articles, formal articles in scientific journals), two specific communication initiatives will be implemented to ensure that participating families, students and high-school personnel get to know the results of the data collected and the implications and conclusions that our project may have. First of all, at the end of the data collection period for the active groups, a SMART Exhibition will take place to share basic information about our participating cohort, personnel and institutions, as well as to showcase the students’ artistic works. This event will invite all enrolled students and their families, the involved and directive personnel from the participating high schools, as well as the SMART research and teaching-artist team and other potential stakeholders. Lastly, an end-of-the-project symposium event is planned. Its objective will be to communicate to all involved personnel and stakeholders the results, conclusions, and potential implications of our project. A summary of our findings, as well as any official publications whenever they are ready will be made available and shared with all participants, families, high school personnel, research and teaching artists team, and any other potential identified stakeholder.

3. Discussion

The main motivation and research question behind SMART is whether art appreciation and practice can change the way teenagers interact with SM, overcoming the negative influence of smart devices and SM platforms, and ultimately leading to beneficial cognitive and emotional effects.

We expect that the designed interventions, due to their artistic content, emphasis on critical thinking and promotion of a more stimulating use of SM, will be able to modify the attitudes and opinions that teenagers have of SM, as well as modify the activities they perform online. Importantly, we expect our interventions to positively affect mental-health markers, leading to a post-intervention improvement in depression, anxiety, stress and self-esteem indicators. At the cognitive level, better executive functions and attentional performance is expected after the 3-month interventions.

Further, we expect this cognitive and mental-health improvement to be paired with brain structural and functional / neurophysiological changes (see Figure 3). Specifically, we hypothesise that improvements in mental health will correlate with plastic changes in GMV of emotional and reward-related centres such as ventral striatum and ventromedial prefrontal cortex (Cheng et al., 2023; Särkämö et al., 2014; Vaquero et al., 2016). We expect cognitive improvements to be related to plastic changes in prefrontal regions and networks related to attention and executive functions (as measured by structural and functional MRI) (Hennessy et al., 2019; Jaschke et al., 2018).

Figure 3.

Schematic representation of the brain regions and networks that have been described to be negatively affected by social media and positively affected by artistic interventions/practice. SMART interventions are thus expected to have some plastic effects in these regions/networks and/or the cognitive domains supported by them. Laterality is not taken into consideration in this visualisation. Abbreviations: ACC: Anterior Cingulate Cortex; PCC: Posterior Cingulate Cortex; Inf: Inferior.

Functional connectivity measured by fMRI at rest is also expected to show more efficient patterns post-intervention in the active groups (MUS, VIS) when compared to the passive control groups, and especially significant effects are hypothesised for the MUS group in comparison with the other two groups (Bashwiner et al., 2020; Hennessy et al., 2019; Jünemann et al., 2023; Li et al., 2018). Structural MRI data is expected to show alterations in audio-motor and visual regions and structural networks, but also in limbic and cognitive control related regions/pathways (Bashwiner et al., 2016; Cheng et al., 2023; Habibi et al., 2018; Hong et al., 2023; Hudziak et al., 2014; Li et al., 2018).

In addition, the neurophysiological pre-attentive responses to musical alterations within a melodic stream (MMN responses) are expected to be faster and/or greater in amplitude post-intervention, specifically in the MUS group (as measured by MEG; Putkinen et al., 2014, 2021; Tervaniemi et al., 2014). Regarding MEG-measured functional connectivity at rest, there are not that many descriptions in the literature that can help us draw specific expectations. Nevertheless, based on previous reports (Lu et al., 2022; Wang et al., 2022), we hypothesise that effects on theta bands from regions related to executive functions will show post-intervention changes in the active groups (especially, in MUS).

We also expect the students involved in the active interventions to show increased engagement during the classes led by our teaching artists. Further, in SMART, we are targeting public high schools from the NYC area in order to provide artistic resources to underserved youth communities that may have difficulties accessing these kinds of resources (both artistically and mental-health related) or education elsewhere. Our interventions may also serve as a potential career or hobby discovery for the participants, which can promote long-term benefits related to their well-being and motivation.

We expect SMART to require adjustments during its implementation. For example, recruitment strategies and contact with prospective participants and their families are going to depend greatly on the participating high schools’ communications strategy and infrastructure. Resources such as the laptops to use during the interventions will also depend on the participating high schools’ hardware and other resources’ availability. Even though the TAs selected for this project have extensive teaching and arts-based experience and have been trained specifically to implement the SMART project according to our research and scientific-team guidelines, the fact that different TAs will be in charge of implementing the intervention sessions may add some variability in the results that could be difficult to control for (if not post-hoc, during the statistical analyses). Finally, adjustments in the intervention programme due to unforeseen circumstances, environmental issues related to the high schools’ context, difficulties or specific characteristics from the particular cohort of students, etc., may be necessary, potentially diverging from our original plan and intervention curriculum. Other miscellaneous issues we foresee could be related to data acquisition, such as loss of data due to technical or human errors. The team will be trained to follow the established protocol and avoid to the maximum extent possible these kinds of mistakes.

If the predicted benefits afforded by digital artistic interventions are confirmed, the SMART project could break new ground in the treatment of the adverse effects of SM in teenagers and might have an impact in policy-making regarding adolescents’ mental health and education. Projects like SMART have the potential to guide and promote public health ventures that could advise youth to use the arts paired with a creative and critical thinking approach to use digital devices, perform online activities and overall use SM in a healthier way. Moreover, the conclusions and materials developed within this project have the potential to be applied in clinical contexts (e.g. health education, therapeutic purposes) in a wide array of physical and mental health conditions, even beyond adolescents. For instance, arts-based interventions like the ones proposed here could be developed for the management of stress, depression, and anxiety symptoms associated with chronic health conditions, both in the physical and mental health realms (such as cancer, major depression, Parkinson’s disease, long COVID-19, children treated with growth hormone therapy, or ADHD, among others; de-Graft Aikins et al., 2024; Mas-Herrero et al., 2023; Saville et al., 2025; Sihvonen et al., 2017), and could be beneficial both for patients of those conditions or their main caregivers (Bourne et al., 2021; Pérez-Núñez et al., 2024).

Finally, as an overall goal and expectation, we hope SMART conclusions help us raise awareness around the beneficial impacts on well-being that arts-related activities can have during adolescent development.

Funding Statement

In addition, LV holds a Horizon Europe funded Marie Sklodowska-Curie Action Global Postdoctoral Fellowship (Grant Agreement ID# 101063319). PR, Prof. Catherine Hartley, and LV were recipients of a 2023 Grant for Music Research from the GRAMMY Museum Foundation. ELMV is a recipient of the Steinhardt Doctoral Fellowship Program. JLM is a 2024 awardee of the ‘Washington National Opera/The Kennedy Center American Opera Initiative' and an ‘American Opera Project Fellow’ (2023-2025). LO is a recipient of the 2025 Media Art and Tech's Marco Cosio fellowship (Brown Institute for Media Innovation, Columbia Journalism School & Stanford Engineering – Columbia University).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplemental Material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/20008066.2025.2550079.