=Paper=
{{Paper
|id=None
|storemode=property
|title=The Connection Graph Proof Procedure
as a Logical-Connectionist Model of the Mind
|pdfUrl=https://ceur-ws.org/Vol-764/paper01.pdf
|volume=Vol-764
|dblpUrl=https://dblp.org/rec/conf/nesy/Kowalski11
}}
==The Connection Graph Proof Procedure
as a Logical-Connectionist Model of the Mind==
<pdf width="1500px">https://ceur-ws.org/Vol-764/paper01.pdf</pdf>
<pre>
Invited Keynote Talk




The Connection Graph Proof Procedure as a Logical-Connectionist Model
of the Mind

Robert Kowalski, Department of Computing, Imperial College London, UK


In this talk, I present an agent architecture in which thinking is modeled as
activating links in a connection graph of goals and beliefs, represented in
abductive logic programming (ALP) form. In this ALP agent model, beliefs are
represented as logic programming clauses, and goals are represented as a
variant of range-restricted FOL clauses. This clausal representation facilitates a
connectionist (connection graph) implementation, in which forwards and
backwards reasoning are different ways of selecting and activating links. The
activation of links can also be determined in the manner of Patie Maes' activation
networks, by associating different strengths with different goals and beliefs and
different weights with different links, spreading the strength of activation
throughout the graph in proportion to the weights on the links.




                                        1

</pre>