Cognitive Approach to Fraud Detection

Fraud detection researchers have spent a great deal of effort looking for information cues (often termed ‘red flags’) that signal the presence of fraud (Albrecht and Romney 1986). This research has been motivated by the desire to improve on auditors’ accuracy at detecting fraud, in particular financial statements fraud. Despite the intuitive appeal of the red flag approach, studies have shown that this search for cues has not been entirely successful. Red flags have frequently been found to be ineffective, sometimes even hindering the ability of an auditor to detect fraud (e. g. , Pinkus 1989; Johnson, Jamal and Berryman 1989).

With striking similarity, psychology researchers have been searching for “Pinocchio cues,” which are cues or patterns of cues that are reliably present when one lies or attempts to otherwise deceive another. Not only have these cues proven to be elusive to discover, studies have also shown that reliance on such cues may be detrimental to accurate detection, as judges of deceptiveness often pay attention to cues that are not correlated with actual deception (DePaulo, Lindsay & Malone 2003). To overcome this impasse, researchers in both psychology and auditing have proposed shifting the focus of their research away from the search for red flags (Pinocchio cues), and towards a deeper understanding of the cognitive processes by which individuals identify, interpret, and evaluate information that may be indicative of a misrepresentation (e. g. , O’Sullivan 2003). We believe that this research strategy is powerful and fruitful. For over a decade we have been studying in great detail the problem solving behavior of external auditors who are and are not successful at detecting financial statement fraud (e. g. , Johnson et al. 1989;11992; 1993; 2001).

This paper reviews our approach to understanding an auditor’s cognitive processes, describes the theory that we have developed to explain what auditors do, and reviews key findings from our experiments. Our studies show that detecting fraud based on an analytical review of a company’s financial statements is difficult, but not impossible. On one hand, even Big 4 firm audit partners often fail to detect fraud using standard analytical procedures (the recorded accuracy in controlled laboratory settings is 50-75 percent - Johnson et al. 1992; 2001). On the other hand, these same studies also demonstrate the possibility of success: approximately ten percent of auditors are able to consistently detect the frauds in the cases submitted to them for review.