Proceedings of the KI 2015 Workshop on Formal and Cognitive Reasoning




       Human Reasoning, Logic Programs and
             Connectionist Systems


                             Steffen Hölldobler
                 International Center for Computational Logic
                        Technische Universität Dresden
                                01062 Dresden
                                   Germany



Summary

    The suppression task, the selection task, the belief bias effect, spatial rea-
soning and reasoning about conditionals are just some examples of human rea-
soning tasks which have received a lot of attention in the field of cognitive sci-
ence and which cannot be adequately modeled using classical two-valued logic.
I will present an approach using logic programs, weak completion, three-valued
L
� ukasiewicz logic, abduction and revision to model these tasks. In this setting,
logic programs admit a least model and reasoning is performed with respect to
these least models. For a given program, the least model can be computed as
the least fixed point of an appropriate semantic operator and, by adapting the
Core-method, can be computed by a recurrent connectionist network with a
feed-forward core.




                                       4