Facial affect (or emotion) recognition is a central issue for many VMC and naturalistic computing applications. Most computational models assume "categorical perception" of facial affect, in which a benign illusion promotes robust recognition of emotional expressions even under severe degradation conditions, including temporal compression. However, this applied interest in human facial affect perception is coming at a time when the evidence for categorical perception is being challenged in the basic research literature, largely on methodological grounds. The research presented here systematically addresses the classic evidence for categorical perception of facial affect, using high-quality digital imaging and display technologies and improved research methods. In doing so, it illustrates a fruitful convergence of basic and applied research. The evidence does NOT support categorical perception of facial affect, which in turn suggests the importance of preserving high-fidelity...
Diane J. Schiano, Sheryl M. Ehrlich, Kyle Sheridan