Exposure Therapy, in VR
Designing for Relapse Prevention: A VR Exposure Module for Substance Use Disorder
[Wundrsight Health]
Methods: Questionnaires (100+), secondary research, contextual interviews, IDIs, focus groups, usability testing (50+ users) Frameworks: ERP (Exposure and Response Prevention), CBT
Role
User Research
Timeline
3 months
Problem Space
Treatment for substance use disorder is fragmented. Most interventions rely on verbal recall and imagination, asking someone to describe a craving rather than confront it. Relapse rates stay high partly because therapy rarely rehearses the moment that actually breaks someone: the drink placed in front of them, the friend who orders one anyway, the specific room a relapse happens in.
Addiction is psychosomatic. The body reacts before the mind catches up. Any intervention that skips the physical sensation of craving is only treating half the problem.
Research
I led a mixed-methods study to understand where existing treatment fell short and where VR could realistically add value without sacrificing therapeutic utility.
100+ questionnaire responses surfaced gaps in current treatment access and comfort with using an emerging technology like VR, which shaped our accessibility and onboarding decisions early.
Secondary research into existing digital therapeutics and clinical literature on ERP helped define the product's therapeutic framework and feature set before a single scene was built.
Generative research (contextual interviews, IDIs, focus groups) with people in recovery and with treating clinicians surfaced what a "trigger moment" actually feels like from the inside, and what a therapist needs to see in order to intervene in real time.
This became the design thesis: environments built to provoke genuine physiological and emotional reactions, always under a clinician's watch, with that reaction itself forming the basis of treatment.
Who we designed for
Arjun, 29, six months post-detox, in outpatient care
Arjun relapsed twice before, both times in situations he hadn't rehearsed for: a colleague's farewell drinks, a friend topping up his glass without asking. In therapy, he can describe these moments calmly. In the moment itself, he can't.
He's motivated but distrustful of anything that feels clinical or performative. He's also new to VR and a little self-conscious wearing the headset, so the first session couldn't ask anything emotionally difficult of him before he'd even settled into the environment. He's sensitive to overstimulation, both from the disorder itself and from unrelated sensory sensitivities, so a chaotic environment would pull him out of the exercise before the exercise could do anything useful.
His outpatient sessions already use motivational interviewing, so he's articulate about his own triggers and genuinely wants to change. That awareness just doesn't hold up under real conditions. He can talk through a craving in a therapist's office and mean every word. What he hadn't done was practice staying in the room when it actually happened.
Design decisions
ERP as the backbone. Systematic desensitization meant sessions had to escalate gradually. A user's first exposure couldn't be a crowded bar; it had to build from a quiet, low-stimulus room toward higher-risk social settings, with a therapist adjusting pace live.
Real agency, not passive viewing. Users could pick up objects, decline a drink, walk away. Passive exposure teaches recognition; interactive exposure builds the physical rehearsal of refusal, which is what actually needs to hold up outside the headset.
Sensory load as a clinical variable, not a preference setting. Usability testing with 50+ users showed that visual noise and text-heavy UI pulled attention away from the exposure itself, working against the therapy rather than supporting it. We cut on-screen text significantly and gave users direct control over voice volume, brightness, and color saturation. We defaulted to darker, lower-saturation palettes rather than bright ones, since overstimulation was consistently reported as a barrier, particularly for neurodivergent users.
One input at a time. Every scene was built around a single primary sensory or decision-making task, so users weren't processing a visual, an audio cue, and an interaction prompt simultaneously. For a population where sensory regulation is often already compromised by substance use, this wasn't accessibility as an add-on. It was a therapeutic requirement.
The therapist stays in the loop. No module was designed to run unsupervised. Real-time feedback from the user, visible to the clinician, let the therapist adjust exposure intensity mid-session, turning the module into a live extension of the therapy room rather than a standalone tool.
Why this mattered
Relapse prevention depends on rehearsal under real conditions. By combining evidence-based exposure therapy with an interactive, sensorily-considered environment, the module gave clinicians something talk therapy alone couldn't: a way to observe and shape a client's physiological response to a trigger, in the room, before it happens in the world.