Projection Is a Form of Prediction
There is a moment in almost every therapeutic process when a person can see the pattern clearly and yet find themselves repeating it.
They know the story. They can name where it began. They recognize the dynamics while they are unfolding. And yet, days or weeks later, they find themselves caught in the same familiar cycle.
It is one of the most common experiences in the consulting room, and one of the least well explained by the idea that insight produces change. If insight alone were enough, understanding would be sufficient. It rarely is.
A recent account from contemporary neuroscience offers a compelling explanation for why. More surprisingly, it arrives at an understanding of the mind that echoes ideas depth psychology has been exploring for more than a century. It begins with a deceptively simple claim: the brain is not primarily a passive receiver of information. It is a prediction engine.
The Brain Is Not Waiting for Reality
Predictive processing has become one of the leading frameworks in contemporary neuroscience. Rather than imagining the brain as a passive receiver of information, it proposes something much more active. The brain is continuously generating predictions about what it expects to encounter, comparing those predictions against incoming experience, and then responding in one of two ways. It can revise its internal model, or it can act on the world in ways that make the world conform more closely to what it expected in the first place.
Imagine walking into your home at the end of the day and expecting someone to be angry with you. A neutral expression is read as irritation. A brief answer becomes evidence of withdrawal. Before long, you may respond defensively, creating the very tension you anticipated. The prediction has helped organize the experience.
Our brains are constantly performing this kind of work. Most of the time it serves us extraordinarily well. It allows us to move through an impossibly complex world without having to rediscover reality from scratch every moment. The difficulty comes when our predictions no longer fit the world we are actually inhabiting.
Projection as the Second Move
This is where a recent paper offers a fascinating bridge between predictive processing and classical psychoanalytic thought. Writing in Entropy in 2026, Erik Stänicke, Bendik Sparre Hovet and Line Indrevoll Stänicke of the University of Oslo argue that many of the mechanisms Freud described — including projection, transference, and wish fulfillment — can be understood through the lens of predictive processing. Their paper, Freud’s Model of the Mind Within a Predictive Processing Neuroscientific Paradigm, is open access.
Note: The Debrief has an accessible write-up for readers who would rather not read the paper itself.
When the brain cannot revise its model, it begins working on reality instead. Rather than changing the prediction, it changes the interpretation of experience. It attributes to others what it cannot yet recognize in itself. It recreates familiar relational dynamics long after they have ceased to serve any useful purpose. The world becomes organized around confirming what is already expected.
In psychoanalytic language we call this projection or transference. In predictive processing, it is the same movement described differently: acting on the world so that it continues to support the model already held by the mind. Different language, remarkably similar insight. I have written elsewhere about how projection shows up in ordinary life, and the mechanism described there is the one this research is now mapping.
Insight Is Not Information
The most interesting part of the paper is not its reinterpretation of Freud. It is what it suggests about change itself.
The authors describe insight as a form of prediction error, a moment in which reality differs enough from expectation that the existing model can no longer simply absorb the discrepancy. Something has to give.
This helps explain an experience almost every therapist knows well. A client can understand a pattern intellectually for years without fundamentally changing it. That understanding becomes incorporated into the existing model. It becomes one more thing the mind already knows. Real insight is different. It is not simply acquiring new information; it is encountering reality in a way that makes the old organization of experience no longer sustainable.
Understanding is often cumulative.
Insight is disruptive.
Why Safety Matters
If that is true, another question immediately follows. Why don’t we revise our models more often?
Karl Friston, whose work underlies much of predictive processing, offers a remarkably simple answer. When I spoke with him, he put it in almost ordinary terms. Not knowing what is going to happen next, and not knowing what you are going to do next, is the essence of unsafe. Safe is everything becoming predictable. In the language of the mathematics, expected surprise is minimized.
He took this directly into development. Good parenting, on this analysis, consists largely in providing predictable responses — enough consistency that a child can learn the world is secure, and therefore safe.
That definition reaches far beyond mathematics. If changing our internal models requires surprise, then the conditions in which those models are revised must also be safe enough to tolerate surprise.
This has profound implications for therapeutic work. The task is not simply convincing someone that they are mistaken, nor is it offering better interpretations. It is helping create a relationship stable enough that being wrong becomes survivable.
It also helps explain something that often puzzles people. Predictability is not the same as wellbeing. A nervous system organized around danger may continue returning to familiar forms of suffering because they are predictable. The unfamiliar, even if objectively healthier, may feel considerably more threatening than the known.
People do not simply cling to old patterns because they are irrational. Often they cling to them because those patterns remain the safest predictions available.
Change, then, is not only about seeing differently. It is about becoming safe enough to allow reality to reorganize what we thought we knew.
Two Languages Describing the Same Reality
One of the reasons I find this work so compelling is that it continues a pattern I have noticed throughout my career. Again and again, contemporary neuroscience, depth psychology, and the contemplative traditions seem to arrive at remarkably similar observations while speaking entirely different languages.
The cognitive neuroscientist Chris Niebauer, whose work I discussed with him in a recent conversation, argues in The Abstract Mind that we become so identified with our conceptual models that we stop being able to tell them apart from reality. The language differs from predictive processing. The observation is strikingly close. The model becomes the world.
This is where the conversation becomes genuinely interesting. Not because science validates the contemplative traditions. Not because the traditions somehow anticipated neuroscience. But because both are looking carefully at the same human reality and reporting back in different languages.
My own work has been an attempt to listen carefully to both.
An Invitation
This is the material we work with in the Transformational Coaching Certification I teach with Shilah Sarkissian. Friston's model comes up directly, including the argument that it undercuts polyvagal theory — the kind of question a practitioner has to be able to sit with rather than resolve by picking a side. The year-long training is for therapists, coaches, yoga teachers, bodyworkers, and contemplatives who have reached the edge of what their original discipline alone can offer.