Why Your Brain Prefers Familiar Suffering to Unfamiliar Healing
Ask someone who has been depressed for a long time to describe who they'd be without the depression.
Most people can't do it. Not in any detail. They can name what they'd stop feeling. They struggle to say what they'd start feeling, or what they'd want, or how they'd spend a Tuesday. The unsuffering version of themselves isn't a person they can picture. It's an absence where a person would go.
I've come to think this is one of the more important facts about stuck patterns, and that it explains something that otherwise looks like a failure of will.
You can know exactly what's wrong. You can articulate the pattern, trace it to its origins, and describe what a better arrangement would look like in the abstract. And still not move. Not because you're weak or unmotivated or secretly attached to your suffering. Because the alternative isn't real to you yet, in the specific sense that your system has no working model of it.
There's a framework that makes sense of this, and it comes from the same research group behind much of the recent work on psychedelics and the brain.
What Canalization Means
In 2023, Robin Carhart-Harris, Karl Friston, and a dozen colleagues published a paper proposing a single underlying feature common to much of psychopathology. They call it canalization.
The word comes from the biologist Conrad Waddington, who pictured development as a ball rolling across a landscape of valleys. The deeper the valley, the more strongly the ball is channeled into it. Apply that image to mental life and the valleys become habitual states, the well-worn paths your system falls into.
Their argument is that what makes a pattern pathological isn't its content. It's its entrenchment. Specifically, its degree of influence and its resistance to revision when new contexts or new evidence arrive.
And here's the part I find most useful clinically. They describe canalization as usually beginning as an adaptive response to adversity, distress, or dysphoria. It made sense once. It becomes a problem when it grows too dominating of a person's future cognitive and behavioral state-space.
Read that phrase slowly, because it's the whole thing.
State-space means the range of states you can reach. Every self you could be, every way you could respond, every configuration available to you. Canalization doesn't just make one valley deep. It shrinks the map. Over time, the depressed self isn't one option among many. It becomes most of what's reachable.
Which is why the person can't describe who they'd be without it. That version isn't hidden. It's out of range.
The authors are careful about one thing, and I want to repeat it. Canalization isn't inherently bad. Consolidating well-trodden paths is how learning works at all. A person with no deep valleys would have no skills, no habits, no stable character. The question is never whether a system is canalized. It's whether the canalization has become too deep and too dominating for the life the person is now living.
This connects directly to the too much order end of the spectrum I've written about elsewhere. Canalization is what excessive order looks like when you zoom in on a single pattern.
Why Insight Doesn't Reach It
If the problem were missing information, insight would fix it. Learn the true thing, replace the false thing, feel better.
That isn't how it goes, and the canalization model explains why.
The trouble isn't that you hold a wrong belief. It's that the belief sits at the bottom of a deep valley, and the system keeps rolling back into it regardless of what you've understood. Knowing you're safe doesn't flatten the valley. Understanding where a pattern came from doesn't flatten it either. The predictions about who you are that organize your experience were built from years of input, and a paragraph of insight doesn't outweigh them.
This is also why contradicting evidence so often fails to register. When someone is kind to a person who expects unkindness, the kindness tends to get absorbed rather than counted. It was pity. They want something. They don't know me yet. The valley is deep enough that new information rolls back down into the existing shape.
There's a version of this I find worth naming plainly. A pattern that resists revision is doing what it was built to do. It got built because revision was once dangerous or expensive. Calling it self-sabotage misses that entirely.
The Part That Feels Like Going Backward
Something I've watched often enough to consider it ordinary: people can get shakier when life improves.
Anxiety arrives when a relationship becomes healthier. Discomfort shows up as the chronic weight starts lifting. Someone doing genuinely well finds themselves drawn back toward an old arrangement they had already left behind.
I want to be careful about how I explain this, because the tidy story is that the brain is seeking the familiar, and the honest answer is that I'm describing a clinical observation rather than a finding.
But the canalization picture makes it legible. Leaving a deep valley means spending time on unfamiliar ground where the system has no reliable model of what happens next. That's genuinely destabilizing, even when the direction of travel is good. The old pattern was painful and it was known. The new one is better and it is not yet mapped.
People often read that period as evidence they're failing. It's closer to the opposite. Feeling unmoored as things improve is roughly what you'd expect from a system leaving territory it knows.
What Actually Updates a Model
If entrenchment is the problem, then the work is whatever reduces entrenchment. In practice that seems to mean a few things.
Repetition over revelation. One contradicting experience gets absorbed. Many, over time, start to reshape the landscape. Safety has to be felt again and again before safety becomes predictable.
Relationships that don't behave as predicted. A steady therapeutic relationship, a partnership that doesn't repeat the original pattern, a friend who stays. These are hard to explain away precisely because they're sustained and embodied rather than argued.
Practices that loosen habitual prediction. Meditation is the most studied of these, and it works slowly and from the inside.
Tolerating the unmapped stretch. The period between the old model loosening and a new one forming is uncomfortable by nature. Support through that stretch is often what determines whether someone stays with the change or retreats to what they knew.
I'd call these clinically well-supported rather than proven by the canalization paper specifically. The paper describes the mechanism. What follows from it for practice is inference, mine and other people's.
Where Psychedelics Fit
The canalization model has a companion. The REBUS model, from the same authors, proposes that psychedelics reduce the confidence weighting on the brain's existing predictive models, which has the effect of flattening the landscape those models sit in.
Flatten the landscape and the ball stops rolling automatically into the same valley. Patterns that felt like facts can briefly be seen as patterns. Evidence that would normally be absorbed has a chance to land.
The authors themselves speculate that previously dominant attractors, including pathological ones formed in response to trauma-related uncertainty, may be relaxed in an enduring way after successful psychedelic therapy.
Note the hedging in that sentence. It's theirs, and it's appropriate. REBUS remains a proposal, and direct testing of its core claims is still early.
What the framework does clarify is why the medicine alone isn't the treatment. Flattening the landscape doesn't determine where the ball goes next. If nothing new happens during the window, the old valleys are still there when precision returns. The integration work is what puts something else on the map.
One narrowing worth making. The strongest trial evidence for psychedelic-assisted treatment of trauma specifically involves MDMA, which isn't a classic psychedelic and works through a different mechanism. For treatment-resistant depression the classic psychedelic evidence is stronger. I'd rather be precise about that than let the categories blur.
The Honest Part
There's something humbling in this framework and I don't want to smooth it over.
If the difficulty is structural, then healing isn't mainly a matter of understanding or trying harder. It's the slow business of building a model of a life you haven't lived yet, using experiences you have to actually have rather than merely consider.
For people who grew up in chaos, or survived something, or spent decades inside a particular shape of suffering, that can be the hardest work they ever do. Not from weakness. From the fact that they're working against a system that got good at keeping them safe by keeping the map small.
There's grief in it too, which surprises people. The old predictions organized a life. They were wrong and they were load-bearing. Letting them go is a real loss even when it's also a release.
A Closing Thought
The useful thing about canalization is that it reframes the question.
Not what's broken in me. What has narrowed, and how far, and what would widen it.
Valleys that deep took years to cut, usually for reasons that made sense at the time. They don't flatten because you've understood them. They flatten through repeated experience of something else, held long enough for the system to believe it.
Which brings it back to where I started. If you can't picture who you'd be without this, that isn't a failure of imagination. It's an accurate report on the size of the map right now.
Maps can be redrawn. Slowly, and usually not alone.