Neural Network Simulations Support Heuristic Processing Model of Cultivation Effects

Abstract

Many studies have shown that heavy TV viewers make social reality judgments more in line with televised reality. Shrum's (2001) heuristic model of cultivation effects predicted and found that biases in first-order cultivation judgments resemble heuristic processing. Systematic processing eliminated the effect. This study presents a series of computational simulations to examine whether a simple feed-forward neural network model exhibits learning behavior in accordance with Shrum's model. The data from the model are tested in contrast to data from human participants. Results closely fit human data. Simulations show that increased television exposure increases construct accessibility; television exemplars are not discounted when exhibiting the cultivation effect; and systematic processing reduces or eliminates the cultivation effect.

ACKNOWLEDGMENTS

The author thanks L. J. Shrum, University of Texas at San Antonio; Robert F. Potter, Indiana University; and Edward T. Palazzolo, Arizona State University, for helpful comments on earlier versions of this manuscript.