Toward a Model of Visual Performance: Foundations and Data

Visual performance has been a central topic of research and discussion in illuminating engineering for many years. This interest is driven largely by practitioners who want to know how lighting affects the performance of workers in industrial and commercial environments. Data and theory that could quantitatively relate lighting and productivity for economic evaluations are, perhaps, the most desirable research goals for lighting practitioners.

The purpose of this paper is to present plausible and useful calculation procedures of suprathreshold visual performance for lighting practitioners. The plausibility, and therefore utility, of the proposed model is based upon two propositions. First, visual performance must be extracted from task performance. The latter is generally an unknown combination of both visual and nonvisual factors that contribute to the behavioral response. The confounding of visual and nonvisual contributions limits the utility of a calculation procedure based on such studies because one does not know how much of the predicted behavior is based upon visual responses under the control of the lighting practitioner and how much is based upon nonvisual responses evoked by the multitude of of other factors. Second, any model of visual performance must be consistent with the literature describing basic visual response. If visual performance can be extracted from task performance, then it should be possible to verify the extraction as an adequate representation of visual processing from data available in the vision literature. Although complete agreement may be difficult, well-established principles in visual sciences limit the number of possible descriptions of suprathreshold visual performance.

This paper describes in some detail, then, the foundations of the proposed model in visual sciences as well as the attempts to extract visual performance from task performance at a simulated realistic task. It is not argued that this visual performance model is complete. A variety of other parameters that are important to the complete specification of visual performance, as well as factors ultimately necessary for making economic evaluations of lighting for task performance, are not considered. Further, there are some assumptions in the model that require further testing. Therefore, this paper and the model serve as a milestone rather than the goal for lighting practitioners. The paper hopefully establishes an appropriate algorithm for a model of suprathreshold visual performance, but it does not provide a complete economic analysis relating lighting and productivity. A subsequent paper describes some of the requirements for future studies attempting to describe visual performance more accurately.