Skip to main content
ARS Home » Southeast Area » Little Rock, Arkansas » Microbiome and Metabolism Research Unit » Research » Publications at this Location » Publication #164369
Title: AN APPROACH TO SCALING EVENT-RELATED POTENTIALS FOR COMPARISONS OF TOPOGRAPHIC SHAPES USING REGRESSION ANALYSIS

Author
item JING, H - UAMS
item PIVIK, R - UALR
item DYKMAN, R - ACH

Submitted to: Journal of Neuroscience Methods
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/2/2005
Publication Date: 3/15/2006
Citation: Jing, H., Pivik, R.T., Dykman, R.A. 2006. An approach to scaling event-related potentials for comparisons of topographic shapes using regression analysis. Journal of Neuroscience Methods. 151(2):239-249.

Interpretive Summary: Event-related potentials (ERPs) are brain responses to stimuli such as sounds or images. These responses are related to behavior by studying when they occur after a stimulus, as well as by their shape and size. This paper shows a new way for finding out how parts of the brain react to stimuli. This will help us better understand how brain events relate to behavior.

Technical Abstract: Recent studies have demonstrated that comparisons of scalp topographical distributions of event-related potentials (ERPs) between experiment conditions may not correctly indicate underlying changes in neural sources if the signals are not scaled prior to the comparisons. This important issue was re-evaluated in this paper using both simulated and experimental data. Simulated data were generated according to 16 different brain models containing 2-4 dipole sources varying in strength, orientation, origin, and number. The changes made in the strength, orientation, and origin included relative changes between the sources or symmetrical changes in the sources. Experimental data were ERPs collected from 45 infants at 3 months of age. Influences of linked-ear and average references were examined. A scaling method based on relations of signal amplitudes between conditions was devised and compared with the vector method. While real topographic differences generated by complex changes in underlying sources were preserved, interactions between condition and electrode site due to mere strength changes were successfully identified by the new method, irrespective of reference method used. However, the vector method was not always reliable because failure to differentiate or mistakenly indicate changes in sources may occur when a linked-mastoid reference was used. The method presented in this paper is reliable and recommended prior to topographic comparisons to distinguish different types of changes in underlying neural sources.