Cognitive models in the extreme

Imagine a scenario: A client comes in to your office. Let’s say they are rehabilitating from a drug addiction. You want to ascertain if impulsivity is a problem for this client so that you can tailor their treatment to effectively suit their needs. So you give them a quick game to play; the Balloon analogue response task (BART). You use a cognitive model to analyse the data and the results suggest a highly impulsive individual. With estimates of impulsivity much higher than we would normally expect, this individual is an extreme performer. So how should you interpret these results? Should you trust the results? The answer to these questions may not be as straight forward as we think. We present parameter recovery simulations for a popular model of behaviour on the BART and raise the question whether individuals can be effectively modelled in the extreme.