Current Research

Ernest Davis, Department of Computer Science, New York University

Unless otherwise indicated, Ernest Davis is the sole author.

Integrating scientific and commonsense reasoning

Understanding science is not merely a matter of knowing mathematical formulas and fitting them to measurements. Equally important is knowing how scientific theory connect to physical reality as manifested at the human level and in particular to the realia of scientific experiments. I have been studying how symbolic reasoning can be used to connect commonsense with formal science.

A particular emphasis in this work has been the formalization of a number of ontologies for matter. Matter and its properties can be viewed in a number of different ways: in terms of particles; of fields; of the quantity of stuff in a region; of moving stuff; or of individuated things. I have been developing representations and theories that will allow an AI system to use these, separately or in combination, in automated reasoning.

Recent Publications



Containers

Containers are ubiquitous both in everyday life and in specialized contexts. In collaboration with Gary Marcus I have carried out an extensive study of symbolic methods for commonsense physical reasoning about containers in human cognition and in automated reasoning. The web page for the project is here.

Recent Publications



Qualitative Spatial Reasoning

Developing languages for expressing qualitative spatial knowledge and analyzing the expressive power of those languages. Developing algorithms for qualitative spatial reasoning. Studying the problems of spatial reasoning that arise in applications.

Recent publications



Commonsense Reasoning and Text Comprehension

What kinds of commonsense knowledge are involved in comprehending natural text?

The TACIT project, aims toward building a corpus of short texts that are hand-annotated with all the commonsense knowledge and reasoning needed to interpret them, and developing a systematic analysis of the issues involved. The project is currently on a back-burner, but may be resumed at some point, if I can find a meaningful way to move it forward. The project is currently on a back-burner, but may be resumed at some point, if I can find a meaningful way to move it forward.

In collaboration with Hector Levesque and Leora Morgenstern, I have assembled a collection of Winograd schemas.

Recent Publications



Commonsense Reasoning and Probability Theory

Analysis of various issues that arise in using probability theory for commonsense reasoning.`

Recent Publications





Limits of Simulation Theory

There is a common theory that human physical reasoning does and that AI automated reasoners should work using simulation; that is, methods more or less akin to the physics engines used both in video games and in much scientific computation. Working with cognitive psychologists Gary Marcus, Todd Gureckis, Ethan Ludwin-Peery, and Neil Bramley, I have argued, first that there are many situations in which simulation is an inappropriate method for reasoning; and second, that there are situations in which it can be shown empirically that human beings are not using simulation for reasoning.

Recent Publications





Limits of Bayesian and Optimalist Cognitive Theories

Many cognitive psychologists have argued recently that the human mind carries out reasoning optimally, or as well as possible given fundamental resource limitations, and in particular that reasoning can largely be characterized in terms of Bayesian inference. Collaborating with Gary Marcus, I have argued against this point of view.

Recent Publications

There is an interesting discussion of the meanings of "common sense" in English and Latin literature. in C.S. Lewis' Studies in Words.