ABSTRACT
This chapter introduces a generalized procedure for detecting second-order nonlinear effects in behavioral data sets. A linear autoregression equation contains the weighted sum of previous time-series observations without any multiplicative terms involving two or more observations. The analytic procedure of choice is the General Linear Model (GLM), which predicts the dependent variable using categorical independent variables obtained from both main effects and their interactions. Consequently, an inconceivable amount of deterministic evidence from over a century of published psychological research may be concealed in the error terms of ANOVA models. Linear systems would then emerge as a special case under stable conditions when the interactions between two or more entities are negligible, as is the case in Newtonian mechanics. Hemispheric effects were evaluated by presenting different information processing tasks to the left and right visual fields. The lexical task used words compiled in a study of Australian semantic norms by Casey and Heath.
