How social media algorithms may be undoing the work of therapy, and what to swap in instead

You sit down to rest for ten minutes. An hour later you are still scrolling. In that hour, you have absorbed dozens of small comparisons against other people's bodies, careers, holidays, and relationships. You have seen distressing news from three continents. You have watched short clips about a parenting trend you now feel bad about not following. Your shoulders are tense. You feel a little flatter than when you started.

If this is familiar, you are not failing at willpower. The platforms are doing what they were designed to do.

This post is about how social media algorithms can quietly work against the gains people make in therapy and on medication, and what to swap in their place. I am not going to suggest deleting everything. For many of my clients, particularly those who are isolated or part of communities that meet primarily online, social media is genuinely useful. The goal is right-sized engagement, with the worst-fitting parts swapped out for activities that serve the same function more kindly.

What the algorithm is actually doing

The recommendation systems that drive your feed work on a principle that psychologists know well. Variable ratio reinforcement, the same schedule that makes poker machines so persistent, delivers unpredictable rewards spaced between neutral or negative content. The brain releases dopamine in response to unpredictable reward, more strongly than in response to the reward itself (Schultz, 2016). The scroll keeps you reaching for the next maybe.

Layer onto this the personalisation. Every pause, click, dwell-time, and scroll-speed feeds the system information about what holds your attention. A 2026 editorial in the British Journal of Psychiatry describes this as an “algorithmic dopamine economy”, an environment that does not so much reflect your preferences as actively sculpt them through repeated reinforcement (Putica et al., 2026). If you stop on anxiety content, you get more anxiety content. If you watch one clip about a medical symptom, you get a feed shaped by health worry. Over time, your feed becomes a mirror of your most pulled-on places.

This matters clinically because what is being reinforced may be the very pattern you are working hard to change.

How this shows up in depression, anxiety, and OCD

For depression, two mechanisms are well-established. Social comparison is one. Looking at curated images of other people's lives reliably produces lower mood and lower self-evaluation, with the strongest effects in people already struggling (Hunt et al., 2018, Journal of Social and Clinical Psychology). Passive consumption is the second. Hours of scrolling crowd out the activities that actually protect mood. A recent Australian meta-analysis of ten randomised controlled trials by May, Malouff, and Meynadier (2025), based at the University of New England and published in the European Journal of Investigation in Health, Psychology and Education, found that limiting or reducing social media use produced a small but significant reduction in depressive symptoms across 1,491 participants.

For anxiety, the feed becomes a vector for catastrophic content. The constant news cycle, doomscrolling, and the algorithm's preference for emotionally intense material combine to keep the nervous system in a state of low-grade alarm. A 2024 meta-analysis by Du and colleagues in BMC Psychology found consistent associations between problematic social networking use and anxiety symptoms.

For OCD, the picture is sharper. OCD latches onto whatever material is most available, and the feed offers an endless supply. Guazzini and colleagues (2022) studied 660 participants and found that people with high OCD symptoms reported greater mood reactivity to social media and gave it more subjective importance than non-OCD comparators. In clinical settings, three things tend to combine. Endless triggers, since the algorithm learns and serves whatever catches your attention. Frictionless reassurance-seeking, in the form of searching symptoms, watching OCD-content creators describe their rituals, or checking other people's reactions. And digital compulsions in their own right, like repeated checking of notifications, messages, or post engagement. Van Bennekom and colleagues (2018), publishing two case reports in BMJ Case Reports, have proposed adapting the Yale-Brown Obsessive-Compulsive Scale symptom checklist to capture social media and smartphone content as obsessions and compulsions, recognising them as a current and clinically relevant manifestation of OCD.

For neurodivergent clients, particularly those with ADHD or autistic traits, the feed often also serves genuine functions. Sensory regulation, novelty-seeking, and stimulation management. The swap suggestions further down are written with that in mind.

Why this matters for the therapy you are doing

This is the part I think gets discussed too little. Therapy and medication work against the load of your daily environment. They do not work in isolation.

If you are doing behavioural activation for depression, the model assumes that scheduled activities are competing with avoidance. Several hours a day of passive scrolling is, functionally, avoidance. The behavioural activation is being given a smaller share of your day than the avoidance behaviour it is meant to displace.

If you are doing exposure and response prevention for OCD, the work depends on you not engaging the compulsion during and between exposures. Reassurance-seeking is a compulsion, and the feed makes digital reassurance-seeking constant and almost effortless. A client doing structured exposures with me on a Tuesday, then searching symptoms or watching OCD content on Wednesday, is reinforcing the very loop we are trying to interrupt.

If you are doing Acceptance and Commitment Therapy, the work depends on making contact with valued action and learning to be with uncomfortable internal experience rather than escape it. Where the scroll is functioning as experiential avoidance, it is practising the opposite of what ACT is teaching, often for several hours each day.

For SSRIs and SNRIs, the medication reduces symptom intensity, though it does not change the environment that drives the symptoms. If your environment is steadily delivering comparison triggers, catastrophic content, and unpredictable reinforcement, the medication is doing remedial work against a sustained load. Sleep, often disrupted by late-night use, compounds this further. Disrupted sleep is itself a known predictor of poorer antidepressant response.

Swapping rather than cutting off

What tends to work better than abstinence, for most of my clients, is substitution by function. Ask what the scroll was doing for you in that moment, then swap in something that meets the same need at a lower cost.

When the function was variable reward and novelty, a podcast queue, audiobook rotation, or fiction reading can preserve the unpredictability without the comparison and trigger content. Activities with built-in variability work similarly, things like cooking new recipes, gardening, op-shopping, birding, or fishing.

When the function was soothing or emotional regulation, slow walking outdoors has good evidence behind it. Tyrväinen and colleagues (2014) and the broader green-space literature consistently show mood and stress benefits from time in natural settings. Warm-water immersion, weighted blankets, pet contact, and breathwork also fit here. For clients I see who are working with somatic approaches, orienting and pendulation practices are useful in-the-moment swaps.

When the function was connection, direct contact outperforms passive viewing. Hunt's 2018 trial found that limiting passive use actually reduced loneliness, which surprises people. A short phone call to one person, a weekly in-person catch-up, a hobby group, or a volunteer role provides the social nutrition that scrolling promises and does not deliver. For neurodivergent clients who find unstructured social contact depleting, parallel activities like reading or working alongside another person in a café can offer connection at a manageable intensity.

When the function was distraction from internal experience, a small piece of valued action tends to shift mood more reliably than further distraction does. Even five or ten minutes of something aligned with what matters to you. The behavioural activation literature suggests this is where mood actually changes.

When the function was information-seeking or reassurance, particularly for OCD, swapping to a different information source rarely helps. The skill that does help is response prevention combined with tolerable distress practice. A written script for the obsession, a delay timer, a grounding practice, and a planned alternative action are the tools that break the loop.

When the function was cognitive stimulation or filling a gap, content with an endpoint serves better than content without one. Long-form journalism, a print magazine, a structured language app, a short course, or a single podcast episode. Anything that finishes.

Redesigning the environment, not just the behaviour

The single most useful intervention I see, across diagnoses, involves changing the phone rather than relying on willpower.

Move social media apps off the home screen and into a folder on the second or third page. Turn off all notifications except calls and messages from named contacts. Switch the phone to greyscale, since the colour saturation is part of what makes feeds visually compelling. Charge the phone in a room other than the bedroom. Keep a paper book or e-reader on the bedside table.

Each of these small changes alters the friction of using social media without removing access. Most clients I work with find their scrolling time drops substantially within a week, with no real sense of deprivation.

A final note on social media and therapy

Social media is not the villain in anyone's mental health story. For many people it is also a source of community, education, identity, and humour. My point in this post is more specific. For clients doing the hard work of therapy, the feed often delivers exactly the content most likely to undo that work, and the design of the feed makes this difficult to notice while it is happening.

If you are working with me on depression, anxiety, or OCD, and you suspect your social media use is part of the picture, we can build a swap plan together that protects the gains we are making in session. The therapy is doing its job. We just need to stop the environment from working against it.

Dr Sarah Fischer is the Principal Psychologist and Director of Behavioural Edge Psychology, a specialist private practice with locations in Caulfield South and St Kilda, Victoria. She holds a PhD, MPsych, and AHPRA endorsement in organisational psychology. To book an appointment, visit behavioural-edge-psychology.au4.cliniko.com/bookings.

Frequently Asked Questions

Can social media make OCD worse?

Yes. Research by Guazzini and colleagues at the University of Florence (2022) found that people with high OCD symptoms are more mood-reactive to social media than non-OCD comparators in a sample of 660 participants. Algorithmic feeds also create a continuous supply of triggers, frictionless reassurance-seeking, and digital compulsions, all of which can reinforce the obsession-compulsion cycle.

Does social media affect how well antidepressants work?

Direct trial evidence is limited. Antidepressants reduce symptom intensity but do not change the environment that drives symptoms. Where social media use is generating continuous comparison triggers, catastrophic content, and disrupted sleep, the medication is doing remedial work against a sustained environmental load. This is best treated as a clinically reasoned concern rather than a confirmed empirical finding.

Does reducing social media use improve mental health?

A 2025 Australian meta-analysis of ten randomised controlled trials (N=1,491) by May, Malouff, and Meynadier at the University of New England found that limiting or reducing social media use produced a small but statistically significant reduction in depressive symptoms. The 2018 Hunt et al. trial also found reductions in loneliness and depression among university students who limited use to approximately 30 minutes per day.

What are healthy alternatives to social media scrolling?

Healthy substitutes are most effective when matched to the function the scrolling was serving. Variable reward can be met with podcasts, audiobooks, or fiction. Emotional regulation can be met with walking outdoors, warm-water immersion, weighted blankets, or breathwork. Connection is best met with direct in-person or phone contact. Cognitive stimulation is met with content that has a defined endpoint, such as a podcast episode, magazine, or short course.

Should I quit social media if I have depression, anxiety, or OCD?

Total abstinence is not necessary for most people. For many clients, including those who are isolated, neurodivergent, or part of online communities, social media provides genuine value. The recommended approach is right-sized engagement, with the worst-fitting parts swapped for activities that serve the same function more kindly. Reducing app accessibility, turning off notifications, and switching the phone to greyscale also reduce use without removing access.

Can social media interfere with CBT or ACT?

Mechanistically, yes. Behavioural activation for depression depends on scheduled activities competing with avoidance, and passive scrolling functions as avoidance. Exposure and response prevention for OCD depends on not engaging compulsions, including digital reassurance-seeking. Acceptance and Commitment Therapy depends on valued action and acceptance of internal experience, both of which are undermined when scrolling functions as experiential avoidance. There are no direct randomised trials testing this question, so the position is clinically reasoned rather than empirically settled.

What is doomscrolling and why is it harmful?

Doomscrolling is the prolonged consumption of negative news and content, often driven by algorithmic feeds that prioritise emotionally intense material. It is associated with elevated anxiety symptoms and a sustained state of low-grade nervous system alarm. Du and colleagues (2024) found consistent associations between problematic social networking use and anxiety in their meta-analysis published in BMC Psychology.

References

• Du, M. et al. (2024). Association between problematic social networking use and anxiety symptoms, a systematic review and meta-analysis. BMC Psychology, 12, 263.

• Guazzini, A., Gursesli, M.C., Serritella, E., Tani, M., and Duradoni, M. (2022). Obsessive-Compulsive Disorder (OCD) types and social media, are social media important and impactful for OCD people? European Journal of Investigation in Health, Psychology and Education, 12(8), 1108–1120.

• Hunt, M.G., Marx, R., Lipson, C., and Young, J. (2018). No more FOMO, limiting social media decreases loneliness and depression. Journal of Social and Clinical Psychology, 37(10), 751–768.

• May, W., Malouff, J.M., and Meynadier, J. (2025). Reducing social media use decreases depression symptoms, a meta-analysis of randomised controlled trials. European Journal of Investigation in Health, Psychology and Education, 15(11), 222.

• Putica, A. et al. (2026). Algorithmic dopamine economies, emerging psychiatric challenges in an artificial-intelligence-mediated attention ecology. The British Journal of Psychiatry.

• Schultz, W. (2016). Dopamine reward prediction-error signalling, a two-component response. Nature Reviews Neuroscience, 17, 183–195.

• Tyrväinen, L., Ojala, A., Korpela, K., Lanki, T., Tsunetsugu, Y., and Kagawa, T. (2014). The influence of urban green environments on stress relief measures, a field experiment. Journal of Environmental Psychology, 38, 1–9.

• Van Bennekom, M.J., de Koning, P.P., and Denys, D. (2018). Social media and smartphone technology in the symptomatology of OCD. BMJ Case Reports, 2018, bcr-2017-223662.