Activity: Talk or presentation types › Invited talk
Description
Neural representation and categories
Computational Logic workshop
The neural code maps states of the environment or states of 'mind' to
the electrical activity patterns of neurons. There is evidence that the
brain uses a sparse and explicit neural code, meaning that each state is
coded by the activity of a small fraction of available neurons, and that
neurons represent semantically interpretable aspects of the state. The
overlap of the codewords then imply semantic relationships between the
represented items. This structure within the code is well captured by
Formal Concept Analysis (FCA), where objects are the items to be
represented and the attributes are the neurons activated by the items.
FCA and its statistical extension are useful tools for representing
categories in a flexible way.