M.S. Magnusson

Human Behavior Laboratory, University of Iceland, Reykjavik, Iceland

When measuring or analyzing continuous behavior records involving real-time - for example looking for cyclical organization, typical sequences and syntax, or behavioral profiles - it may be useful to consider the possible presence of repeated hidden structures which may be complex and of as yet unknown kinds. One basic type of hierarchical time pattern, called a T-pattern, and the corresponding detection and analysis software, Theme, is proposed for the detection of hidden behavior patterns.

T-patterns are probabilistic, repeated, recursive (self-similar and hierarchical) synchronic and/or sequential structures that may involve any number of individuals and behavioral, physiological and/or environmental event types. Other structural types are derived from the T-pattern type:

The search algorithm is described. Its main aspects are (1) detection of critical interval relations between pairs of time point series and (2) pattern construction, competition and selection. Examples of T-patterns detected in human interaction studies are presented including behavior coded with The Observer.

Paper presented at Measuring Behavior '96, International Workshop on Methods and Techniques in Behavioral Research, 16-18 October 1996, Utrecht, The Netherlands