|
|
|
Shared Features
|
Connectionist models of disorders share a common
set of assumptions which guide their construction, and a common method
of investigating disorder. Common assumptions, generally termed
the "connectionist approach" or "connectionist framework" have evolved
over the last four decades from the desire to represent behaviors without
a central "program" or mechanism controlling them, so as to reflect the
lack of central control in neurological processes. A common approach
to investigating disorder, described below, follows naturally from these
guiding assumptions, and involves systematic perturbations of models of
non-pathological functioning. |
|
|
|
 |
There is a great deal of room for variability in the
models which may be created and the ways in which they may be tested and
validated, within the framework presented above. Of particular interest
in this review are what aspects of pathology may be modeled with
neural networks, what approaches may be used to model disorder,
what types of claims may be made from successful neural network models,
and how these claims are validated. These dimensions give rise to formal
codings for network models which are cataloged in the Aggregate Analyses
below. |
|
|
|
|
