Precision-Weighting · Active Inference · Niche Construction
2023-present
Adaptive
An original framework translating the predictive processing framework into a rigorous methodology.
Client
Ongoing · independent
scope
Precision-Weighting · Active Inference · Niche Construction
year
2023-present
duration
3+ years, ongoing
Overview
The question.
Predictive processing gives cognitive science a formal account of the mind as a hierarchical prediction engine: a system that generates models of the world, weights its confidence in those models against incoming evidence, and revises them accordingly. Much of the field's explanatory power sits in that mechanism — precision-weighting, the process by which the brain assigns confidence to competing predictions and thereby determines which errors drive belief revision and which are discounted as noise. This mechanism is well established theoretically and has been applied productively in clinical and computational contexts, from models of psychosis to accounts of depression, but its application has remained largely diagnostic and researcher- or clinician-mediated. Comparatively little work has asked what it would mean to apply this same mechanism deliberately, outside a clinical frame, to the models a person holds of themselves and their own capacities.
The Challenge
Translating a clinical mechanism into a self-directed practice. |
Adaptive is my attempt to work out an answer to that question. Developed over three years, it draws directly on the mechanics of precision-weighting and on active inference's account of niche construction — the finding that an agent's environment is not a passive backdrop but something the agent actively structures, and which in turn stabilises or destabilises the agent's own predictive models. Adaptive takes these established mechanisms and asks how a person might work with them intentionally: how the confidence assigned to a self-model might be deliberately raised or lowered, and how the environment surrounding a person might be restructured to support that revision rather than work against it. Where the underlying theory is silent or where I extend beyond it — for instance, in treating mental imagery as a route to adjusting the precision of a self-model, or in applying evidentiary-scaffolding principles from hierarchical belief-updating to how a person builds confidence in a new self-belief over time — I try to be explicit about what is established, what is a reasonable extension, and what is closer to hypothesis. |
the solution
A disciplined, non-clinical operationalisation. |
This is deliberately not a therapeutic or self-help framework. It does not diagnose, and it makes no claims about mechanism that the literature does not support. Its contribution, more modestly stated, is translational: taking a rigorous but largely clinical or theoretical body of work and asking what a disciplined, self-directed, non-clinical operationalisation of it would actually look like — and being honest, at each step, about where that operationalisation is grounded in existing theory and where it is proposing something new. I'm now formalising this work academically, situating it properly within the predictive processing and active inference literature it draws from, ahead of postgraduate study. |
