I find it useful to see a person’s belief system as a system of domain-specific models. Each of us builds them unnoticed over years to survive and thrive: one for money, one for work, one for strangers, one for illness. Each is a set of rules of thumb.
The models differ. Some are simple and some complex. Some run all day, some hardly ever.
Accuracy matters less than we think. A model survives if it satisfies its owner’s passions and seems to get results. Herbert Simon called that satisficing: stopping at good enough. Gerd Gigerenzer adds that a rule of thumb works when it fits the setting it grew up in. Move it to another setting and it fails.
Psychology has older names for the parts. George Kelly held in the 1950s that every person is a scientist, building models of how the world works. My label borrows from cognitive science, where domain specificity means the mind comes with specialized, inborn learning devices, a view Pinker and Chomsky hold. The models start from inborn parts, and a lifetime of learning shapes them.
Two things make a model hard to change. It sticks to identity like velcro, which is the ground cultural cognition covers. And it comforts. Few of us step outside the pull of our own.
The worst models are often installed in childhood, like shame about sex or a politics of enemies. Parents who mean no harm pass on what their values require, a cause of unintended consequences Robert Merton named. The carrier is more often a victim of the process than wicked.
The common failure is misapplication. A rigid mind reaches for the model it knows and uses it on the wrong problem. Psychologists call this the Einstellung effect: solving a problem the familiar way even though a better way exists. It is the old line about the hammer and the nail, the law of the instrument. A flexible mind asks first which model fits.
The way out is reflexivity, seeing a model as a model. I do not think this sorts people into two kinds. Everyone has sealed models in some domains, and I have mine. The less I know a field, the less likely I am to notice. The work is gradual: find one sealed model, test it against the world, and repair it. Then help someone else do the same.