Mandler's (1984) model of emotion is summarized in this paper and is operatonalized analytically and statistically using Thom's (1975) catastrophe theory. Data were collected from (n=15) mathematics majors in a pilot study to test Mandler's model and the nonlinear effects of emotions in solving mathematical problems. The data were found to fit Thom's cusp model almost twice as well as the conventional linear model, thus strongly supporting Mandler's model. The implications of these findings are discussed in terms of mathematical problem solving and mathematics education. Contains 69 references. (Author/MKR)