Brain scientists keep landing on an idea Freud had 130 years ago about why bad habits stick

The idea that gets compared to Freud here is projection: the mind reads new situations through expectations built from old ones. Three psychology researchers at the University of Oslo argue in a 2026 paper that this looks a lot like the “prediction” at the center of modern brain science, and that the match may help explain why some harmful patterns of thought and relating are so hard to dislodge. Freud began working out such ideas about 130 years ago.

AI, explained every day: The Deep View is the go-to daily newsletter for 600,000+ founders, knowledge workers and professionals who want to stay up to date with artificial intelligence.

Get The Deep View →

We may earn a commission if you sign up.

The paper is a conceptual argument, not an experiment. It reports no data from patients or brain scans, and the authors themselves call the link an analogy. ScienceDaily’s repost frames the story as a surprising idea Freud had 130 years ago, and that framing is the repost’s, not a measurement. That distinction runs through everything below, because the popular version, that neuroscience has vindicated Freud, claims more than the paper does.

Projection and repetition: the Freudian ideas on the table

The paper, “Freud’s Model of the Mind Within a Predictive Processing Neuroscientific Paradigm,” appeared in the journal Entropy on March 12, 2026 (volume 28, issue 3, article 318; DOI 10.3390/e28030318). Its authors are Erik Stänicke, the corresponding author, Bendik Sparre Hovet and Line Indrevoll Stänicke, all in the University of Oslo’s psychology department, with Line Stänicke also affiliated with the Nic Waals Institute at Lovisenberg Hospital.

The central comparison, which the University of Oslo team presents as a conceptual proposal, according to the paper’s full text, is between Freud’s concept of projection, in which feelings or expectations are attributed to other people, and prediction in neuroscience, in which the brain interprets new experience using learned expectations. For the bad-habit angle, the authors use repetition compulsion, Freud’s term for the tendency to recreate familiar patterns, as an everyday example of repetition that serves stability. They also tie perceptual identity to Freud’s 1895 Project for a Scientific Psychology, the early model that gives the “130 years” its anchor, while noting that he never fully developed it.

Other Freudian concepts appear in the paper as well: wish-fulfillment, from The Interpretation of Dreams in 1900; transference, meaning reactions to a therapist based on past relationships; introjection; projective identification as developed by Melanie Klein and Wilfred Bion; and Freud’s pleasure, reality and Nirvana principles. The authors treat projection as the cleanest fit, and note that the concept has changed considerably across psychoanalytic writers, so the parallel depends on which version is used.

Predictive processing and active inference on the neuroscience side

The modern mechanism is predictive processing. In this account the brain constantly predicts incoming sensory input and works to shrink the gap between prediction and experience, called prediction error. It has two routes, as the authors describe them: perceptual inference, which updates the prediction in light of the input, and active inference, which acts on the world so that experience matches the prediction. Both belong to the free energy principle, a mathematical framework for how living systems stay stable.

That is where the habit argument enters. Because stability is tied to homeostasis, a familiar model can persist even when it is inaccurate,, which is the authors’ explanation for why a harmful pattern can feel stable from the inside. In the release quoted by ScienceDaily, which is based on the University of Oslo release, the researchers say psychoanalysts see people “recreate familiar relational patterns, even when these are poorly adapted,” and add that rigid symptoms “may be stable but not very flexible prediction models.”

Stänicke is quoted saying that psychoanalysis “has developed psychological theories about how predictions take place at a subjective level.” In the authors’ framing, neuroscience supplies a mechanism and psychoanalysis supplies the subjective experience of living inside a prediction. The release also floats a clinical implication: if symptoms are rigid predictions, therapy may sometimes have to work through the relationship with the therapist, offering new experiences, and not only through conscious reasoning.

What is analogy, and what is evidence

The list of limits comes from the authors. They say the projection and prediction parallel is an analogy and that they do not ground projection or introjection in the mathematical formalism of the free energy principle. They say the predictive paradigm is mostly descriptive while psychoanalysis is normative, aimed at treating illness. And they flag that predictive processing may blur the line between beliefs and desires, a point of dispute. The evidence for the free energy principle itself is described in the paper as circumstantial.

The release adds that the similarities do not establish that Freud’s broader theories were correct, and the examples given, such as a person who expects rejection, are illustrations and not reported findings. The paper’s publication fee was paid by the Norwegian Psychoanalytic Society and a university department, a fact the authors disclose.

The honest summary is narrow, and it applies to habits of thought and relating, not to habits like smoking or snacking, which the sources read for this article do not discuss. A neuroscience framework that treats the brain as a prediction machine happens to have a conceptual neighbor in one of Freud’s oldest ideas, and that neighbor offers a vocabulary for why expectations can lock in. Whether the mechanism operates the way the analogy suggests is a question for experiments the paper does not contain.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.

AI, explained every day: Get The Deep View →

We may earn a commission if you sign up.