I’ve been reading a fascinating article by Steven Hayes et al. (2026), published in Clinical Psychology Review. As many of you know, I love ACT (Acceptance and Commitment Therapy) and process-based therapy, and I have a self-confessed confirmation bias for much of what Steve talks and writes about.
Here’s a quick paraphrased summary of the article for your perusal! As far as I can tell, the findings turn much of the current evidence-based science assumptions on their head! The info below comes straight from the paper, and isn’t my opinion about it (just to make that clear!).
For thirty years, clinical psychology has been running in place. Despite a mountain of new “evidence-based” protocols and enough funding to launch a small space program, patient outcomes have essentially flatlined. Effect sizes for Cognitive Behavioural Therapy (CBT) are no better today than they were in the 1990s.
The crisis isn’t a lack of effort; it is a foundational error in the math of human nature. Our diagnostic manual, the DSM-5, is a monument to this failure. It offers a dizzying 10 million ways to qualify for a diagnosis; add in “specifiers,” and that number explodes to 161 septillion combinations. To put that in perspective: there are 636,120 unique ways to meet the criteria for PTSD alone! We have built a system of staggering complexity that attempts to categorize “latent diseases” that have never actually been found. At the heart of this stagnation is the “Average Person”—a mathematical ghost that literally does not exist. We are currently witnessing the birth of the “Idionomic Revolution”: a shift from a “one-size-fits-none” science to a rigorous science of the particular.
The Ergodic Error: You Are Not a Statistic
Clinical science is haunted by what is called the Ergodic Error. In physics, an ergodic process is one where the average of a group at a single moment matches the average of one individual over time. Psychology assumes human behaviour is ergodic. It assumes that if a treatment helps a group of 1,000 people on average, it will work for YOU this Tuesday.
This is a serious mathematical fallacy. You cannot model the dynamic trajectory of a flying bird by endlessly subdividing a high-speed photograph of a flock. A snapshot of the group (the ensemble) tells you absolutely NOTHING about the path of the individual bird over time. People are non-ergodic: we are historically embedded, context-sensitive, and path-dependent. Even the most “sophisticated” modern psychological tools, like Multilevel Modeling (MLM), act as a reality distortion field. MLM uses a process called “shrinkage,” which pulls unique individual data points toward the group mean to stabilize estimates. While this helps statisticians find a clean population average, it routinely erases—or even reverses—the direction of true person-specific effects. It effectively “shrinks” your unique life out of the data to satisfy the math of the aggregate.
“The average of an ensemble is never a description of any member of it.”
Here’s an example from Steve Hayes’ blog (before I get back to the paper!):
In 1950, the U.S. Air Force was troubled by a series of crashes, and so researchers took it upon themselves to measure thousands of pilots on over a hundred physical aspects, to update the design of their cockpits. After all, a cockpit built to the middle of the range should fit the greatest number of pilots and thus cut down on preventable accidents. But when a young lieutenant compared the average measurements with the individual records of pilots, he discovered that virtually none of them even came close to matching.
A cockpit built for the average, it turned out, fit almost no one. There simply was no such thing as an “average pilot.” The fix was to make the plane adjust to the person, with adjustable seats, pedals, and straps; simple things we now take for granted in every car.
It’s the same in mental health: many experts make the very same mistake. What works for the “average” person is supposed to work for you.
A Troubled Legacy: Statistics’ Dark Origins in Eugenics
Back to the paper…. Why did we adopt an analytic system so poorly suited to human healing? The answer is chilling. The tools we treat as scientific bedrock—the standard deviation, correlation, and regression—were never intended to help individuals grow. They were designed by Francis Galton and Karl Pearson to rank, classify, and control “unfit” populations (i.e. eugenics). We are essentially trying to map the architecture of the human soul using tools originally forged to measure barley plots and manure treatments! Galton and his disciples, though, weren’t interested in the “seed”; they were interested in the “yield” of the aggregate to support a eugenic agenda of “racial fitness.” This historical amnesia has left us using tools of classification and control to attempt acts of healing.
The end point of this “normative” thinking was not just bad therapy, but human catastrophe. These statistical rankings fuelled the 1924 Immigration Act and provided the “scientific” veneer for Aktion T4, the Nazi program that murdered hundreds of thousands of disabled people. Facility staff used personal details—like “the wart on your mother’s foot became cancerous”—gathered during nationwide professional screenings to lie to families about the deaths.
“Three generations of imbeciles are enough.”
This infamous quote from the U.S. Supreme Court’s Buck v. Bell decision wasn’t just a legal opinion; it was the ultimate application of group-level “norms” used to erase an individual’s right to exist.
The Equisyncratic Trap: When “Proven” Methods Backfire
The danger of the average becomes life-threatening when we encounter what they call equisyncratic relationships. These occur when a process helps SOME people but harms others, resulting in a “zero” effect in a group study (in which case it will not be recommended).
Consider “Valued Action”—the idea of doing what matters regardless of mood (absolutely a core component in Acceptance and Commitment Therapy). In a typical ensemble study, acting on values ideally brings sadness down. But when you look at individuals, a different story emerges. Researchers identified a subgroup of people whom valued action was unrelated or even negatively related to joy; it made them sadder! They persist through challenges, but it doesn’t make them “feel better” in the moment. For most other people, though, the link is strong.
If a therapist follows the “proven” clinical finding and tells a client to “act on your values to increase your happiness,” the advice is not just useless—it is empirically false for that person’s dynamic if they don’t respond “as per normal”. This can lead to harm, where the “cure” makes the patient feel like a failure because they aren’t achieving the “average” result. Common blanket advice like “just be more mindful” can actually predict increases in anhedonia and sadness for certain individuals.
The Idionomic Revolution: A Science of the Particular
The way forward is what the authors describe as the Idionomic Workflow. We are moving away from “Syndromal Diagnosis” (labels like MDD or GAD, based on the DSM-5) toward Process-Based Therapy. This approach doesn’t ask what category you fit into, but what unique processes drive change in your life.
This new toolbox uses modern technology to turn “measurement” into a form of “deep listening” Just to give you a few -highly technical- examples:
- High-Frequency Data: Using wearables and daily check-ins (EMA) to capture the “rhythm” of a life as it happens, rather than a single, biased memory from a clinic chair.
- tsBoruta Algorithms: Traditional machine learning often mistakes the “noise” of yesterday’s mood for a real predictor. tsBoruta is a “conservative” algorithm that first removes these temporal dependencies, allowing it to identify the true drivers of change. It is designed for when clinical time is scarce and chasing false positives is too costly.
- i-ARIMAX: This tool finds the specific, bivariate links in your data while controlling for trends and cycles, ensuring the “map” of your progress is actually yours
Beyond the Procrustean Bed
The article describes an old myth. In Greek mythology, Procrustes invited travellers to spend the night in his bed, but he would stretch them or amputate their limbs to ensure they fit its fixed length. For a century, psychology has forced individuals onto a Procrustean bed of statistics, amputating the unique context of human lives to fit the “average” protocol.
A “science of the particular” recognises that clients’ path through time is not a rank on a curve. It is a unique, path-dependent system. The field is finally moving away from the “sugar high” of group-average data toward an approach that treats every individual as their own valid universe of study.
As you look at your own path to growth, ask yourself: Are you a point on a bell curve, or a unique system moving through time? The future of science suggests the only evidence that matters is the evidence of what works for you.
I’m sure this paper will cop some flak from the ‘establishment’. I saw a LinkedIn post by Steve Hayes already gathering a large number of (sometimes very) critical comments.
Critical Reflection?
What are your thoughts? How do you feel about ‘evidence-based science’ as the thermometer of ‘what works’? How do you use established protocols, treatment plans, suggestions and ‘what the research says’ in your practice (if at all)? Or are you purely guided by what the person in front of you tells you works (and doesn’t work) for them, so that the both of you, together, can cobble together a plan that actually works for the client? This is the premise of John McLeod’s and Mick Cooper’s Pluralistic Approach (this approach makes complete sense to me).
Let me know what you think!
Here’s a direct link to the article: https://www.sciencedirect.com/science/article/pii/S0272735826001091
Marc de Bruin is an ACA Level 4 counsellor, clinical supervisor (for both PACFA and ACA), university tutor, and former barrister-solicitor with over 20 years’ experience in counselling, supervision, and professional development. He is the founder of Counselling Supervision and Simplifying Life.
