The research reported is concerned with the effects of alcohol on a maze-based task which had been designed to incorporate independently each of the four behavioral processes described by Berliner, Angell, and Shearer in 1968. Such a design allows study of the relative effects of alcohol on various behavioral domains. This type of . comparison had not been previously accomplished in a single study. Further, if a task can be characterized according to the behavioral dimensions of which it is comprised and alcohol levels at which performance of the task is likely to occur can be postulated, regression equations might be of use in estimating performance decrements on the task under alcohol versus no-alcohol conditions without experimental manipulation. The development of such regression equations is a second aim of this experiment.

Thirty-two subjects (16 of each gender) were given four different alcohol doses (0.00, 0.05, 0.07, and 0.09% BAC) coupled with four levels of maze difficulty. A Latin Square strategy was used to assign the BAC/Maze combinations. Each combination was repeated under speed and accuracy instructions.

Analyses of variance showed that alcohol impaired performance on (most independent variables in each of the behavioral domains. However, comparisons of estimated percent differences in performance across the dimensions revealed that the cognitive processes were most impaired by alcohol while the perceptual processes were most resistant to alcohol effects. Analyses of variance also indicated that there were no performance effects attributable to gender but that maze difficulty and instruction generally affected performance in the expected directions.

Regression equations which incorporated alcohol, instruction, and ratings of the contribution of each behavioral process were developed to predict task completion time. Gender did not enter into these equations. The predictions yielded by these equations are in agreement with the results found in the literature. Hence, they are satisfactory for use in estimating performance decrements due to alcohol on a task the behavioral components of which are known or can be measured.