As pointed out elsewhere (e.g. [1,2]), analogies exist between the real-time structure of behavior and the structure of DNA molecules and proteins. One such analogy concerns the existence of t-patterns. These are repeated patterns of elements occurring in a fixed order, with distances between consecutive elements being significantly similar each time the whole pattern occurs. They are also characterized by a hierarchical/recursive structure, i.e. there are typically patterns of patterns of the same kind [1]. The behavioral data referred to are time-based records, i.e. real-time streams of various event-types, where any (large) number of types and instances may occur, both sequentially and/or concurrently. Because the quality of data for detecting t-patterns in such records must be controlled, a new measure - called 't-kappa' - has been developed, and will be illustrated.

Vast amounts of genetic (molecular) data are becoming available, most of which are strictly sequential strings of symbols. There are typically four different symbols (bases) in DNA, but twenty (amino acids) in proteins. DNA (genomes) may, however, be transcribed in terms of 'codons': base triplets that may correspond to amino acids. The number of symbols may vary from hundreds or thousands (proteins) to millions (DNA, genomes). In a collaborative study between this laboratory, www.patternvision.com and www.decode.com, behavioral t-pattern methodology is now being adapted to molecular analysis in a specialized version of Theme, called GeneTheme, as will be briefly shown.

In both the behavioral and molecular cases, the search for t-patterns requires the setting of two search parameters: (a) the significance level of each critical interval relationship, and (b) the minimum number of occurrences of a pattern for it to be retained (other parameters are optional). A new automatic procedure for the objective setting of these search parameters for individual behavioral records has now been developed, allowing more optimal detection of patterns. It also allows better comparisons between the structure of behavioral streams (data, records) of varying length, which may be essential for detecting the effects of independent (experimental) variables. Some new illustrative results will be presented.

References

1. Magnusson, M. S. (2000). Discovering hidden time patterns in behavior: T-patterns and their detection. Behavior Research Methods, Instruments and Computers, 32(1), 93-110.
2. Magnusson, M.S. (accepted/in print) Repeated patterns in behavior and some other biological phenomena: The t-system and its corresponding detection algorithms. In: Evolution of Communication Systems (working title). Konrad Lorenz series in theoretical biology. Oeller, K. et al. (eds). MIT press.

Paper presented at Measuring Behavior 2002 , 4^th International Conference on Methods and Techniques in Behavioral Research, 27-30 August 2002, Amsterdam, The Netherlands

© 2002 Noldus Information Technology bv