COMMENT ⊗   VALID 00003 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	8. Machine Intelligence Workshop 9
C00005 00003	EPISTEMOLOGICAL PROBLEMS OF COGNOLOGY
C00007 ENDMK
C⊗;
8. Machine Intelligence Workshop 9

9. Yes.

10. University of Edinburgh, Edinburgh Scotland
and (this time) USSR Academy of Sciences

11. Repino, USSR (near Leningrad)

12. April 17, 1977 to April 23, 1977

13. April 15 or 16

14. 13 days

15. San Francisco to Leningrad (via London - no stopover);
return via Western Europe with stopover in Sweden and/or UK for which
arrangements have not yet been made.  One day stopover in Pittsburgh,
April 29.

16. Requesting roundtrip air fare S.F.-Leningrad less roundtrip S.F. - Pittsburgh.
This is x - y = z.

17. Yes, ARPA ...

18. No.

19 IJCAI (spell out) Sept. 1975, Tblisi USSR.  Stanford University support.

	3 month sabbatical Spring 1975, Kyoto University, Japanese government
support.

	Working group 2.2, International Federation of Information Processing
Societies, Helsinki, Finland, ARPA support.

20. The Machine Intelligence Workshops have been major forums for the
exchange of experience and the presentation of papers to be published
in Artificial Intelligence.  The early workshops were held near Edinburgh,
but the latest was held here in California at the University of Santa Cruz.
Holding one in the Soviet Union provides an opportunity for substantial
contact with the leading Soviet workers in the field.

21. No.


EPISTEMOLOGICAL PROBLEMS OF COGNOLOGY


(Cognology is a new name for what has been called artificial intelligence,
with some extension to include studies of natural intelligence).

Intelligent  behavior  requires  an  organism  or  machine  to   represent
internally facts about  the particular situation  and general facts  about
the world and to use  these facts to decide  what action will advance  its
goals.  The  epistemological  problem is  to  decide what  information  is
available in given circumstances, how it  can be represented and what  are
the rules by which new information can be inferred or conjectured.

This paper presents some new ideas  on (i) general criteria for  ascribing
particular   mental   qualities   to   computer   programs   under   given
circumstances, (ii)  a scheme  for representing  intensional objects  like
propositions and  individual concepts  in  unmodified first  order  logic,
(iii)  some  formalisms  for  expressing  beliefs  and  knowledge  and   a
discussion of  when  non-knowledge can  be  proved,  (iv) a  new  mode  of
conjectural  reasoning  called  minimal   inference  that  expresses   the
conjecture that the known facts are all the relevant facts for a problem.