On the gap your brain actually learns from, and why insight never opens it.
You know exactly what happens to you.
You could draw it on a whiteboard. You know what age it comes from, what it does to you in relationships, which situation switches it on. You explain it better than anyone who has ever tried to explain it to you.
And it still happens.
Notice that the body says it before the mind does. There is a moment, half a second after the message arrives or after he answers curtly, when the shoulders lift slightly, the breath climbs higher, something in the belly tightens. You have not yet had a single thought. The system has already decided.
Most people I meet conclude from this that they have not worked hard enough, or that something in them is too deep to reach. I want to offer a different reading, and a mechanism that supports it.
In 1997 Wolfram Schultz, Peter Dayan and Read Montague published a paper in Science that became one of the foundations of modern neuroscience. They recorded from dopamine neurons, the cells popularly associated with pleasure and reward, and found something that contradicted the expectation.
Those neurons do not fire for reward. They fire for the difference between the reward the system predicted and the reward it received. When something better than expected happens, there is a burst. When something turns out exactly as expected, there is almost no signal at all. When something worse than expected happens, activity dips below baseline.
That difference is called prediction error, and it is the unit the system learns from.
On the source and its limits: the work was done in monkeys, on a simple brief reward such as a drop of juice, not on love, abandonment or relationships. Extending it to the emotional lives of human beings is a widely accepted interpretation in the field, not a directly measured finding. I offer it as a framework that explains well, not as proof. Schultz, Dayan & Montague, Science 275(5306):1593-9, 1997.
In 2013 the philosopher Andy Clark generalised this into a broader theory of how the brain works: a hierarchical system that predicts at every moment what is about to happen, and passes upward only what surprised it. Everything that matched the prediction is cancelled on the way and never reaches awareness. Clark, Behavioral and Brain Sciences 36(3):181-204, 2013.
And there sits the answer to the question this article opened with. Insight is not a gap. Insight is confirmation. It tells the system precisely what the system already held, and so it stops on the way up and is written nowhere.
You can understand your pattern from a hundred angles and never generate a single prediction error.
There is a second trap here, and it is sharper than the first.
Suppose your prediction says: if I do not hold on tightly, this person will disappear. So you hold on. One more message, one more explanation, one more small thing given up. And he stays.
From the outside it looks like success. Inside, the exact opposite of what needed to happen has happened: the prediction was not disproven, it was confirmed. The system logged one more piece of evidence that holding on worked. Ten years of holding on are ten years of evidence.
The pattern did not survive despite your effort. It survived because of it.
The same is true in the other direction. Someone who predicts that closeness always ends in injury, and therefore leaves first, never arrives at the point where it turns out it did not end in injury. Both patterns, the pursuer and the withdrawer, are preserved by exactly the same mechanism: the behaviour prevents the prediction from ever being tested.
This is the most practical distinction I know here, and almost nobody makes it explicitly. Both situations are real, and the response to each is entirely different.
First situation: the prediction has never been tested. The signs: you know exactly what you will do in a situation because you have always done it, you feel tension when someone suggests trying it differently, and you explain why in this particular case it is different. If this is you, what is missing is not more depth. What is missing is one small experiment. The practice below was written for you.
Second situation: the prediction was tested, but you were not present for it. The signs: good things have happened to you that never managed to land, someone stayed and you felt nothing, you describe corrective experiences in the language of a report rather than the language of the body, and while good things are happening you are analysing them. If this is you, more experiments are not the answer, because there have already been enough. What is needed is presence in the body at the moment it happens, and that is usually somatic work with support rather than alone.
This distinction matters, because people in the second situation keep running more and more experiments and despairing, when another experiment is not what they were missing.
There the picture is different, and it is worth saying so plainly before the practice.
A study published in Nature Communications in 2022 examined this mechanism in people with post-traumatic symptoms, and found that they lean excessively on prior beliefs to control memory, and respond more weakly to prediction error when an intrusive memory nonetheless breaks through. The weaker that response to the gap, the more severe the avoidance symptoms. Leone et al., Nature Communications 13:3300, 2022.
In plain language: after trauma the expectation is calibrated to danger at high confidence, and a signal of safety is filtered out as noise before it reaches awareness. This is why people say "I know I am safe and I do not feel safe". They are not imagining it. The system genuinely is not receiving the information.
A framework describing how this works computationally was published in 2017 in Frontiers in Psychology. It is theoretical in nature rather than experimental. Wilkinson, Dodgson & Meares, Frontiers in Psychology 8:1840, 2017.
The prediction line · one minute a day, for one week
Choose one situation that recurs for you. Not the most painful, the most recurrent. One notebook or one note on your phone, no more.
When to stop: if real flooding arrives during step two, shaking, dissociation or panic, stop immediately. A gap created inside a flooded system updates nothing, it only floods again. That is the moment this practice is done with support rather than alone.
This practice is not designed to make you understand more. It is designed to do the opposite: to turn what was automatic and invisible into an explicit prediction that can be tested, and then to have you present enough to register what actually happened.
You do not need to understand yourself more. You need to be surprised once, and to stay in your body while it happens.
In 2015 Richard Lane and colleagues proposed a framework arguing that every effective therapy, from cognitive behavioural to psychodynamic, works through the same bottleneck: activating an old emotional memory, and a new emotional experience that contradicts it while it is active. Lane, Ryan, Nadel & Greenberg, Behavioral and Brain Sciences 38:e1, 2015.
Here I particularly want to state the limitation, because this idea is sold as certainty in my field. The behavioural protocols that tried to translate the mechanism into a precise method in humans do not replicate consistently. More than twenty studies have tried, and at least fourteen reported non-replication. A direct replication of the best known protocol showed the original result depended on decisions about which participants to exclude. Chalkia, Van Oudenhove & Beckers, Cortex 129:510-525, 2020. So: the principle that the brain learns from a gap stands well. The promise that an emotional memory can be erased with a ten-minute protocol does not.
What I take from this into my own work is one simple rule. Before trying to understand another layer, it is worth stating exactly what we are predicting, and then giving reality one real opportunity to be different, under conditions where we can be present when it is.
Not every pattern needs support. The sign that yours does: when the very attempt to do it differently produces flooding, or when you do it differently again and again and nothing lands. There, another presence in the room does what is hard to do alone, not because you are weak, but because your system is also the instrument doing the measuring.
I work over Zoom from anywhere in the world, and in person around Lisbon and Sintra. My work is somatic and focused: not only talking about the pattern but catching it live in the room, in the body, as it happens. Full details on the coaching page, and for couples on the couples coaching page.
Insight is not useless, it is simply not sufficient on its own. Its job is to identify the prediction that drives you and to choose where it is worth testing. The change itself is written only when an experience contradicts the prediction while that prediction is active. Lane and colleagues propose in their 2015 framework that this is exactly what every effective therapy does, each in its own way.
It is the gap between what your brain expected to happen and what actually happened. The brain predicts constantly, and only what fails the prediction travels upward and gets registered. Anything that turned out exactly as expected produces almost no neural signal at all. In other words, the system learns from surprise, not from confirmation.
There is no single number, and anyone giving you one is selling you something. What can be said is that the measure changes: not how much you understood, but how many times you gave the prediction a real opportunity to be wrong. People usually report a change in body sensation before they report a change in behaviour.
With care, and not alone. The 2022 study found that in PTSD the response to prediction error is weakened, and that this tracks avoidance symptoms. A gap created inside a flooded system updates nothing, it only floods again. In that case the work is done with support, at the pace the body sets.
or book a free 30-minute intro call, no pressure.
Keep reading: Why you have been having the same fight for two years · All articles