[Node to be completed]
We define a *prediction* as
a statement that a certain process, referred to as a *test*
comes to a *successful end*, i.e.
to a certain, specified in advance, stage, after which
we simply do not care what happens to the process.
The prediction that a test `T` is successful will be denoted as `T!`.

We define formally a generalized model as anything
that produces one or more predictions.
When we speak of producing predictions, we have in mind, of course,
some objects that represent predictions, e.g. texts in a certain
language which enable us to reproduce the process that the prediction
is about. The objects representing processes are referred to
as their *objectifications* (see).

Formally, we can fit our general concept of prediction into the frame
of the modeling scheme, if we even further expand the range of possible
actions `a`, namely, allow for `a` being an arbitrary process which may
include both actions of the *subject* of the model,
and any other actions and processes.
Let the brain of the subject be always found in one of only two states,
let them have the names *True* and *False*. The representation
function `M_a(w)` will result in *True* if `w` is the end state of
the process `a` which succeeded, and *False* otherwise.
The modeling function `M(r)` will be universal and very simple:
it immediately produces the object True. Now the model we built makes
exactly one prediction : that the process `a` ends in success.

Copyright© 1991 Principia Cybernetica -
Referencing this page