The free energy principle uses the following formula to explain free energy minimization in the context of Bayesian inference:
It might be helpful to think of accuracy in the context of "understanding". The better we understand a complex external state, the better we can reduce "surprise".
We utilize what is called "active inference" to engage with an external state through action. This external state is similar to what Woods and Cook refer to as states "below the line of reference" when looking at complex systems.
As complex states can never be fully known, we are continually acting and sensing in an Ecotone of the internal and external states.
Stress within an ecotone is essential for growth. The stress is the difference between the complexity of the external state and an understanding of it from our internal state - our mental models.
In order to survive, living organisms seek to reduce stress, that is, to minimize the free energy created by the difference between complexity and accuracy. When complexity increases, organisms must increase accuracy, that is, understanding.
Here is where the correlation with Piaget and Koestler becomes interesting.
Piaget talks about schemas, an individual's understanding of the world. In the learning process, this understanding is either extended or reconfigured by a learning experience.
Koestler talks about matrices of thought that are on intersecting planes. It is in the stress of dynamic orthogonal intersections that creative opportunity emerges.
He also explored the holarchical nature of organic complexity.
These theoretical models align with the context of Markov Blankets that make up Bayesian inference networks - those networks by which we understand. Those built on the relationship of parents, their children, _and_ their children's parents.
Stress, introduced by surprise as external states overlap with internal states, incite a process by which internal states must adapt. The speed of this adaptation is critical for survival.
Agile, as a learning mindset, introduced new approaches to accelerate Learning Cycles when engaging with the complex external states of computers. Here Simplest Experiments became valuable.
Through these engagements we increased the rate of active inference, allowing schemas of understanding - a Meaning Matrix - to more quickly evolve as we sought to reduce stress - that is, minimize free energy.
In our work in Dayton, we illuminated the potential of applying the learning cycles of the Agile Mindset to the core of learning experience in schools to accelerate the _rate of adaptation_ - that which is essential for survival.
DOT FROM preview-next-diagram