top of page
Writer's pictureMichael DeBellis

Using OWL Models to Bring Rigor to the "Soft" Sciences

The first paper I published on the Semantic Web was about a model I developed called the Universal Moral Grammar (UMG) Ontology. My goal was to explore ideas from Evolutionary Psychology about the theory of morality. In the last few years I've taken a detour into more applied work but the idea of using OWL (and other software tools) to bring rigor to what are sometimes described as the "soft" sciences is still a goal I want to pursue. My view of science and philosophy is the same as Noam Chomsky's (also ironically the same as Sam Harris, probably the only thing the two of them agree on). That is that the traditional division between science and philosophy or science and the humanities is artificial. I think any topic can be studied scientifically. Of course it will be difficult to do controlled experiments in certain fields but that doesn't preclude use of the scientific method. I see the scientific method as not just one rigid method as is often described in text books but rather a collection of methods and philosophical approaches. Experiments, falsifiability, evaluation based on reason and data rather than on reputation and books with special status, etc. As I see it we use as much of the scientific method as we can on any given problem. The problems we call philosophical are (to the extent they are coherent problems that can be studied at all) are problems where our theories are still in their infancy.


In the social sciences, the definition of the problem is often as important as forming a theory to answer the problem. I have seen so many debates in these fields go in circles because researchers would use terms like "morality", "justice", "intention", "belief", etc. without a rigorous definition of what these terms mean. They think they are debating substantive issues when in reality they are not because they don't have a shared vocabulary. I think OWL can help address this. The eventual goal of a mature science is to have mathematical models that make testable hypotheses. However, before we can develop such models we need to have rigorous definitions of our terminology. Human language is inherently ambiguous. In every day speech "speed" and "velocity" are synonyms. In physics they aren't. Speed is a scalar and velocity is a vector. My goal with the UMG Ontology was to bring rigor to the scientific study of morality. As I look back at that paper, I realized that I had left out a critical issue: separating what evolutionary psychologists would call the cognitive module components of morality that are innate in our genome from the socially defined components. To do that I realized it was necessary to do a higher level analysis of all the various faculties that evolutionary psychologists call cognitive modules. There has been significant and impressive work in this area in the last few decades and yet no-one until now has taken a look at all of it and attempted to create a formal model based on this research. That was my goal in a paper that I recently published in the CAOS VII Workshop on Cognition and Ontologies: Modeling Cognitive Modules with the Web Ontology Language: A Functional Architecture of the Mind.

To say the least, this is an ambitious attempt. I'm reminded of an Introduction to Cognitive Neuroscience class that I audited at UC Berkeley. The first day of class the professor (a brilliant researcher named Jack Gallant) started by saying: "Everything you will hear in this class is wrong. But some parts are less wrong than others." He was acknowledging the immaturity of our understanding of neuroscience compared for example to mature sciences such as physics. It is with the same caveat that I present this paper. It is certainly wrong but I hope some of it is less wrong. More importantly, for a complete theory of psychology I think some model such as the one I describe in this paper is a necessity and before we can get the right model we need to start with a model. That was my intention with this paper. The graphic above from the AllegroGraph Gruff visualization tool shows an excerpt from the model where I model a scenario among hunter gatherers described in the book Moral Origins by anthropologist Christopher Boehm.

Recently, I worked with a colleague who shares my vision of using OWL with the social sciences. He is a PhD candidate in Turkey named Samer Sharani. Samer has developed (with some help from me) an OWL model of a social theory about refugee home return. I.e., how likely a refugee is to return to their home nation of their own free will. This model is completely different from my work in Evolutionary Psychology. However, Samer and I share the goal of using OWL to create rigorous models from research in the social sciences. In his ontology, Samer has rigorous definitions for such nebulous concepts as home, refugee, status, and agency. The figure above shows some example definitions in the ontology modeled with UML. The actual ontology was developed in Protégé and populated with data into a knowledge graph in the AllegroGraph triplestore. Samer presented our work at the same workshop CAOS VII Workshop on Cognition and Ontologies: Refugee Ontology v1: Ontology of Refugee Home Return.







Comentários


bottom of page