Developed by Kieron O’Connor and Frederick Aardema in Montreal, inference-based CBT starts from the observation that the obsessional doubt in OCD isn’t a normal doubt. It isn’t triggered by anything in the senses. You aren’t looking at dirt; you’re considering that dirt might be there.
The model calls this inferential confusion: confusing an imagined possibility with an actual perception, and then acting on the imagined one as though it were evidence. The person is standing in front of a clean-looking counter and reacting to a story about the counter.
The reasoning that builds the doubt
I-CBT is interested in how the obsessional story is constructed, and identifies recurring devices. Personal irrelevant facts — “contamination is real, it happens.” True, and unconnected to this counter. Hearsay and out-of-context rules. Abstract possibility — “it could happen, therefore.” Discrediting your own senses: “I know it looks clean, but looks can deceive.” Sequences of apparently reasonable steps that carry you a long way from what is actually in front of you.
The work involves learning to spot your own particular version of this narrative, and then coming back to what your senses actually report — which the model calls trusting the self, and treats as the central restoration.
How it differs from ERP
ERP accepts the doubt as given and trains you to act without resolving it. I-CBT argues the doubt should never have been accepted as a legitimate question, and works on dismantling the reasoning that produced it.
It uses little or no formal exposure, which is either its great advantage or its central weakness depending on who you ask. Trials to date — several randomised, mostly out of Canada — suggest comparable outcomes to ERP with lower dropout, which is a meaningful finding given how many people refuse or leave exposure treatment.
The honest state of the evidence
ERP has decades of trials and remains the first-line recommendation in every major guideline. I-CBT is newer, has a smaller literature, and has been studied by fewer independent groups. It’s promising rather than established.
What that means practically: if you haven’t tried ERP, try ERP. If you’ve tried it properly and it did not work, or you couldn’t tolerate it, I-CBT is a genuine alternative worth asking about rather than a fringe option — and some clinicians combine elements of both.
Who it seems to suit
People with strong rumination and little visible ritual. People with real event and false memory themes, where the entire question is about what actually happened. And people who found exposure work intolerable — which is a large group, and one the field hasn’t served well.