Using computational methods and models, philosophers, epistemologists, logicians, and cognitive scientists have access to a broader range of methods of conceptual analysis that can provide new points of view and new interpretations of the standard topics. The investigation involves some very interesting ideas from the epistemological and philosophical traditions and aims at providing a computational interpretation of them. The aim of CPL is to explore basic concepts in the following three areas:
Theory Formation and Hypothesis Generation
The computational problem of discovering new
concepts and theories (hypothesis generation and theory formation)
is closely related to the classical theme of the epistemology of scientific
discovery and conceptual change (see L. Magnani, Epistemologia applicata, Marcos y Marcos, Milan, 1991).
Many programs are available and some of them belong to the fundamental
history of AI. The problem of analogical reasoning can be related to this
topic (see L. Magnani and N.J. Nersessian (eds.), Model-Bases Reasoning. Scientific Discovery, Technological Innovation, Values., Kluwer Academic/Plenum Publishers, New York, 2002).
Abduction and Nonmonotonic Reasoning
Abduction is becoming a very popular
term in AI and seems to be computationally involved in several intelligent
tasks, such as diagnostic reasoning, planning, natural language understanding,
learning, and image recognition (see L. Magnani, Abduction, Reason, and Science. Processes of Discovery and
Explanation, Kluwer Academic/Plenum
Publishers, New York, 2001).
This suspicion has prompted many efforts to understand the logical structure
of abduction as a form of nonmonotonic reasoning, i.e. reasoning drawing
defeasible conclusions from incomplete information. This kind of reasoning
can be modeled by means of a non-classical logic and represented at the computational
level for instance with the help of different methods. Philosophy regards
abduction as a creative process of generating new hypotheses in science:
consequently the problem of abduction is also related to the previous topic.
Explanation
The problem of explanation applies
to the acceptance and rejection of hypotheses in both scientific theories
and reasoning in everyday life. The inference to the best explanation, that
evaluates different hypotheses (or theories) involves a dynamic set of criteria
having a multi-dimensional character. For instance if a scientific theory
is simpler and explains more significant data than its competitors, it can
be acceptable as the best explanation. This inference is also involved in
diagnostic reasoning, where the task is to find out the best explanation
(the best diagnosis) among a pre-stored encyclopedia of diagnostic hypotheses.
A very interesting theory of scientific change in terms of a computational
system that performs so-called explanatory coherence is given in
Paul Thagard’s book Conceptual Revolutions
(Princeton University Press, Princeton,
NJ, 1992, Italian edition edited by L. Magnani, Rivoluzioni concettuali. Le teorie scientifiche alla prova
dell'intelligenza artificiale, Guerini & Asociati, Milan,
1994).
We consider CPL of strategic importance for these reasons:
- Importance of the research topics: discovering new concepts, the role of explanation, abduction are all specific and well-known topics of AI. The analysis of their complexity from both the computational and philosophical points of view and the elaboration of a new general theory of hypothetical reasoning seems to be of great interest for many areas of congnitive science; moreover, this analysis will help us to renovate the neglected rationalistic philosophical tradition in Italy.
- Cultural importance of AI methods and tools: these are not yet considered as new and unavoidable models for analyzing and clarifying many indefinite topics coming from the immense unexploited tradition of philosophy, logic, and human sciences. Finally, the results of the research will be clearly relevant for the analysis of the cognitive status of the different kinds of reasoning studied.
- CPL research has excellent links to the area of AI and exact sciences but also to the wider area of philosophy and human sciences. These links have to be regarded in the context of an international perspective trying to identify an established and relatively autonomous field of studies. A general aim is also to place some AI methods on firmer epistemological ground, by eliminating theoretical ambiguities.
References about Philosophy of Science and Computational Methods
Go back to the other areas of research