There Is No Safe Word at That Depth

This is a two-part post on consent and power in psychedelic therapy. I take up the power differential and what it asks of us structurally in a companion piece on power dynamics in psychedelic therapy.

In a ketamine-assisted psychotherapy intake, I ask every client a version of the same question: if you want me to stop something during the session, how will you let me know? Most people have a good answer ready, a word or a gesture we agree on together. One client, weeks later in integration, told me there had been a moment she wanted to use ours and couldn't find it. Nothing in that session had gone wrong. The dose was moderate, she was comfortable, I was watching closely. The signal we'd rehearsed simply wasn't there when she reached for it.

That is not a story about a failure of technique. It is a story about what consent can and cannot do once a person is in a non-ordinary state. And it is the reason the safe word, which so much of our touch in psychedelic-assisted therapy rests on, is a weaker safeguard than we treat it as.

The writer and bioethicist Neşe Devenot has argued, in her Chemical Poetics essays, that there are no safe words in psychedelic therapy worth the name: a person under MDMA or ketamine, inside an amplified power differential, cannot reliably notice that a boundary is being crossed, locate the agreed signal, and produce it. I think she is right about the mechanism, and the empirical work is catching up to the concern.

A 2025 qualitative study in Therapeutic Advances in Psychopharmacology interviewed researchers and therapists about therapeutic touch in psychedelic trials (McHerron et al., 2025). Their findings are blunt. Participants under the influence have a reduced capacity for decision-making, may struggle to communicate preferences, and can experience the same touch agreed to in preparation as something entirely different under the drug. The authors note there is currently no consensus across trials about the conditions for consent at all. The safe word is not a small tool inside a solid system. In many settings it is most of the system.

This is where consent in psychedelic therapy has to become a practice rather than a moment. Informed consent in this work is ongoing by necessity, not as a nicety, and the practice has to be built before the medicine is ever administered.

Consent for touch has to be established before dosing, in a preparation session that does real work. The McHerron study describes this as a kind of preference-setting: an explicit conversation, documented, about which forms of touch are welcome, which are not, and under what conditions.

Concretely, that means distinguishing the types of touch rather than asking a single yes-or-no question:

• Instrumental touch (helping a client to the restroom, adjusting an eye mask) versus supportive touch (a hand held at the client's request) versus somatic or processing touch, which carries the most risk and the least clear evidence.
• The client's trauma history and cultural context, because a neutral gesture for one person is a threat cue for another.
• A predetermined way to communicate non-consent, while naming honestly that it may not be reliably available in the moment, which is exactly why the default has to be conservative.

The goal is not a longer form. It is a shared, specific understanding that you both walk into the session already holding.

If you offer this work, you should have a written touch policy, and it should be specific enough that a licensing board reading it would know exactly what you do and do not do. Devenot's sharpest point about the MAPS-lineage protocols is that their vagueness was the problem: under-specified touch language is a loophole, not a kindness.

A defensible policy names, at minimum:

1. What touch is permitted, by type, and what is categorically off the table in your practice.
2. That consent is set in preparation and documented.
3. That touch defaults to none unless it was explicitly pre-agreed.
4. How touch is documented after the session.
5. How supervision and consultation back the whole thing up.

Write it down before you need it. A policy you can hand to a board, a client, or a consultation group is worth more than a careful intention you can only describe after the fact.

If you take a single operational rule from the research, take this one: do not expand consent after dosing. The McHerron participants were clear that a client asking for more touch while in the acute state is not a basis for providing it, even though it feels, in the moment, like the most responsive thing to do. The default position is to hold to what was agreed in preparation and no further. Consent can contract during a session at any time. It cannot ethically expand.

That rule will sometimes mean declining something a client appears to want. That is the point. The heightened suggestibility and altered judgment of the state are exactly why in-the-moment enthusiasm cannot do the work that pre-dosing agreement does.

I take up that power differential and what it asks of us structurally in a companion piece on power dynamics in psychedelic therapy.

Somatic and processing touch is where I'd urge the most caution, because it carries real therapeutic appeal and the thinnest evidence base, and it is the easiest place for "intuitive" practice to substitute for an agreed plan. If you do this work, you need supervision for it, and ideally a consultation group where touch decisions can be examined out loud rather than rationalized privately.

None of this is an argument against touch. Used well and inside a real consent practice, it can be genuinely supportive. It is an argument against the comfortable fiction that a rehearsed safe word makes a non-ordinary state safe. The safeguard was never the word. It was everything you built before the session, and your willingness to hold the line during it.

Peter Addy, PhD, LPC, LMHC is a Portland-based licensed therapist and the founder of Psychedelic Affirming Education, an NBCC-approved continuing education provider for licensed mental health professionals and Oregon Psilocybin Services facilitators. His research background includes work at Yale School of Medicine on psychedelic substances.

Consent and touch are where good intentions meet real liability. Our CE course works through the ethical and documentation questions this raises, with the standards a board would expect.

For the broader screening and assessment groundwork that a sound consent practice sits on, the Comprehensive KAP Assessment Bundle (2 CEs) is a natural next step.