A new performance-curve method for the detection of simulated cognitive impairment

To evaluate a new performance-curve method for detecting simulated cognitive impairment.

The method was based on quantitative analysis of choice reaction times in a series of eight matching-to-sample tasks of graded complexity, in which participants were required to identify which of two comparison stimuli was identical to a simultaneously presented sample stimulus. The stimuli consisted of arrays of letters, the number of letters ranging from 1 to 10 in the eight tasks.

Mean reaction times of a standardization group of 20 healthy young adults were used to construct regression equations to describe the behaviour of three test groups: 20 healthy individuals instructed to perform to the best of their ability (non-simulators), 20 healthy individuals instructed to simulate cognitive impairment incurred as a result of a head injury (simulators), and 25 patients with a history of closed head injury (patients).

There was a linear relationship between individual participants' latencies and the mean latencies of the standardization sample; the patient group showed steeper slopes, and the simulators higher intercepts, than the non-simulator group. A univariate measure of performance, based on the proportionate increase in latency corresponding to a standard increase in the standardization group's latency was able to identify 14/20 simulators while misclassifying 2/20 non-simulators and 2/25 patients. Other measures, including the number of errors made in the choice reaction time task and in recognition memory tasks, and a performance-curve measure based on Raven's Standard Progressive Matrices, were less successful in identifying simulators.

This new performance-curve method may prove to be a useful addition to tests currently used to diagnose simulated cognitive impairment in clinical and medicolegal settings.