This two-day workshop will bring behavioral scientists and ontologists together to discuss the current state of, and future directions for, organizing knowledge of behavioral theory. We anticipate inviting approximately 30 experts across the behavioral science and text processing communities. See the menu to the right for a tentative agenda and list of participants.
The workshop will be held at the historic Stanley Hotel in Estes Park, Colorado. Famous for its old world charm, The Stanley Hotel boasts spectacular views in every direction and is less than six miles away from Rocky Mountain National Park. Multiple renovations have restored this 140-guest room hotel to its original grandeur. Listed on the National Register of Historic Places and a member of Historic Hotels of America, The Stanley offers over 14,000 square feet of unique meeting and event space equipped with modern day amenities.
Human decision-making and behaviors determine outcomes in all areas of society, including health, education, and finance. Research into behavior is expressed through hundreds of theories in more than a dozen disciplines—each theory often focusing on a given behavior and functioning as a vertical stovepipe for findings. Theories encapsulate and represent current knowledge in terms of key constructs and the relationships between them. It is not unusual for a theory to be tested and later extended in thousands of follow-up studies wherein findings are believed to build incrementally towards a greater understanding of a set of behavioral phenomena.
However, recent findings cast shadows on this belief, as even individual theories have expanded beyond comprehension. Theory review studies are unable to find and integrate more than a small percent of existing findings. Further, dozens of theories exist that address the same phenomenon in almost identical ways. Davis et al. (2014) found 82 theories of behavior and behavior change that had substantial overlap, only three of which were integrative. Yet as new and possibly redundant theories are proposed and as they gain adherents, old theories continue to thrive. Without an ontology of behavioral knowledge-embeddedness and -integration, researchers remain largely unaware of related findings, especially outside their own discipline, but also within a discipline or even within narrow research areas. This lack of awareness prevents the behavioral disciplines from becoming a front-player in the new sciences of behavioral big data. While studies building on big data are becoming more frequent—from examination of millions of Facebook users, for example—no single existing behavioral theory is capable of serving an integrative role.
We propose that analysis of extant behavioral theories may provide a much-needed ontology of behavioral knowledge-embeddedness and -integration, itself the necessary requirement to clarify the structure of knowledge within the large set of behavioral theories in existence. Without such an ontology, clarifying the conceptualizations and vocabulary that underlie knowledge is not possible.
Our initial focus will be on defining a grounded approach to discovery of key elements embedded inside behavioral articles, categorizing and prioritizing those elements, and finally, developing strategies for transdisciplinary knowledge extraction and dissemination efforts. Beyond developing an ontology, a consistent set of transdisciplinary knowledge bases will enable researchers to better search, integrate, and understand past research.
The major goals of this proposed workshop are to 1) evaluate the current state of disparate ontological foundations for behavioral theory efforts, 2) facilitate the development of shared understanding of ontology development and learning from experts in the fields of natural language processing, visualization, and computational ontologies, and 3) advance the rigor of theory-integrative work towards ontology-worthy efforts.
- Kai Larsen, Leeds School of Business, University of Colorado Boulder
- Michael Paul, College of Media, Communication and Information, University of Colorado Boulder
This workshop is supported by the National Science Foundation through award IIS-1612580.