A new method of assessing psychological processes of cognition is developed. It provides large amounts of highly diagnostic data on individuals in very short periods of testing. This method is demonstrated in a study of individuals' use of attention, but the primary goal of the thesis is showing the utility of the method rather than producing new insights about attention. The experimental task uses rapid and continuous appearance and disappearance of targets and foils in three positions on a computer monitor. The participant uses a 'mouse' to control a cursor and attempts to move the cursor toward and onto targets and not toward foils. The cursor movement data is analyzed with machine learning methods (based on a Hidden Markov Model) to produce a most likely position of attention every ten ms. of testing. Just a few minutes of testing coupled with such analyses allows valid inferences about an individual's use of attention. This is illustrated with a variety of statistics comparing the sequence of presented stimuli to the inferred movement of attention, allowing comparison within an individual and across the group tested. This method is particularly well suited for clinical assessment, for testing of individuals from very large groups, and for any settings in which extended testing is not feasible and in which large amounts of data from many conditions for each individual is required.