The premise.
A protocol is a scaffold built around a clinical insight. The insight is what matters; the scaffold is provisional. When the scaffold gets confused with the insight ~ when fidelity to the manual becomes a measure of competence ~ the work suffers and so do the clients.
And the components inside the scaffold ~ alternating bilateral stimulation, systematic desensitization, guided imagery, behavioral rehearsal, dual attention, imaginal and cognitive interweaves, somatic tracking ~ are basic clinical tools. Not Shapiro's intellectual property. Deconstructing EMDR is, in part, returning those tools to the clinicians who already use them.
EMDR's eight-phase protocol is one of the great procedural achievements in trauma therapy. It is also, like every manualized intervention, a snapshot of what a single clinician believed at a single moment about how memory works. That snapshot deserves respect. It does not deserve worship. The historical case for that claim ~ the six prior therapeutic traditions Shapiro synthesized ~ lives on the Origins page; this one picks up where that one ends.
The protocol is not a sacrament. It is a procedure with a research base, a set of indications, a set of contraindications, and a set of clinical situations where it has to bend or be set down.
Fidelity vs. flexibility ~ a false binary.
"Fidelity" in psychotherapy research means: did the clinician do the protocol the way the manual says? It's a useful construct for trials. It's a terrible proxy for clinical skill.
Two clinicians can score identically on fidelity and produce wildly different outcomes. A clinician who runs the protocol cleanly with a client who is dissociating is technically faithful and clinically wrong. A clinician who modifies the protocol to keep a complex-trauma client in the window of tolerance is technically infidelitous and clinically right.
What flexibility actually means
Modification is not the absence of structure. It's structured deviation, anchored to the same clinical objective as the original phase. When you skip the body scan because the client is overwhelmed, that's clinical judgment. When you skip it because you're behind schedule, that's negligence. The map matters; so does knowing why each square is on it.
The AIP model ~ on inspection.
Adaptive Information Processing is the theoretical scaffolding Shapiro built around the procedure she'd discovered worked. It posits that traumatic memories are stored in a maladaptive, state-dependent network, and that alternating bilateral stimulation enables the brain to reprocess them into the adaptive memory network.
As a clinical heuristic, AIP is useful ~ it gives both clinician and client a frame for what's happening. As a neuroscientific model, it is underspecified. The mechanism by which ABLS facilitates reconsolidation is still contested; the leading candidates (working memory taxation, orienting response, REM-like processing) all have empirical support and all have limitations.
The honest position is narrower than the marketing. EMDR has solid evidence for single-incident adult PTSD, weaker and more cautious evidence for complex trauma and dissociation, and an unresolved question about whether the alternating bilateral input is doing the work. That is the reason modifications exist ~ phased pacing, restricted processing, dual-attention prep, Flash, EMD/EMDr, the eight kinds of interweave ~ and the reason clinicians have to read the inclusion criteria, not just the headline. The protocol works on the population it was tested on. Your clients may not be that population.
The orthodoxy problem.
What gets called "EMDR-as-religion" in the field is a pattern that recurs in any community organized around a procedure: a slow drift in which adherence to the procedure becomes the marker of belonging, modifications are framed as deviance, and clients who don't fit the protocol are pathologized for not fitting it.
The cost falls hardest on the clients who most need our skill: complex trauma, dissociation, attachment injury, chronic medical illness, severe SUD comorbidity, late-life trauma. The protocol as written assumes a single-event target memory in a regulated client. That client exists. The other clients exist too, in greater numbers, and they need clinicians willing to think rather than recite.
When fidelity becomes harm
Running standard desensitization with a client who is dissociating is not "high-fidelity EMDR." It is poor clinical practice with an EMDR vocabulary. The protocol is not exempt from the broader ethics of dosing and titration that govern every other procedure in mental health.
What survives the critique.
Most of it. The eight phases describe a clinically sensible arc: orient, prepare, target, process, install, check, close, follow up. The components ~ alternating bilateral stimulation, the cognition pairing, the body scan, the calm/safe place, the container ~ are clinically useful tools regardless of whether you're doing "standard" EMDR or something that shares its lineage but doesn't share its name.
What doesn't survive is the idea that the protocol as written is the only legitimate way to use these tools. The rest of this site shows what the components do when you take them out of their assigned slot and put them where the clinical work needs them.