Scientists and forensic experts have attempted to identify physiological responses and behaviors that are related only to deception, but there is no clear and unique sign of deception (DePaulo et al., 2003). Therefore, some indirect methods have been developed to detect deception. The Forced Choice Test (FCT) can be used as a crime memory detection test for this purpose. In a typical FCT, examinees are presented with questions about a crime, and each question is accompanied by two answer alternatives, where one alternative is always correct (i.e., a crime-related alternative), while the other one is always incorrect (i.e., a crime-unrelated alternative). Examinees must select one of two alternatives that they deem to be the correct alternatives. The FCT classifies responses according to the following logic: the total test score of unknowledgeable examinees is expected to fall within the level of chance because they have no knowledge of the crime and therefore truly guess each question. In contrast, the scores of knowledgeable examinees are expected to be biased because they have knowledge of correct answers and can calibrate their responses deliberately. Typically, examinees with knowledge of the crime purposely select incorrect alternatives, resulting in test scores so low that they are deemed to be too unlikely to have occurred through chance, known as underperformance. That is, avoidance of correct answers is considered the main sign of the possession of crime information. In addition, it has been reported that overperformance could also be a sign of the possession of crime information (although it occurs rarely, Jelicic et al., 2004; Orthey et al., 2017). In empirical studies with a typical FCT alone, the specificity of the test (i.e., correct classification of unknowledgeable examinees) has been reported to be 90–100%, and the sensitivity (i.e., correct classification of knowledgeable examinees) has been reported to be 40–70% (Giger et al., 2010; Jelicic et al., 2004; Meijer et al., 2007; Merckelbach et al., 2002; Orthey et al., 2022; Shaw et al., 2012). Hence, the FCT is relatively accurate at high rates of specificity. Moreover, FCT has two other advantages as a memory detection test. One advantage is that FCT can be combined with other crime memory detection tests that have different test mechanisms. In this regard, Meijer et al. (2007) demonstrated the effectiveness of the combined use of FCT and an autonomic-based concealed information test (ANS-based CIT). In the ANS-based CIT, selective bodily responses to crime-related questions are considered the main sign of the possession of crime information. Meijer and his colleagues argued that FCT can support ANS-based CIT because the detection principle of FCT differs from that of ANS-based CIT. In their experiment, they applied a two-step procedure for detection, where examinees are judged knowledgeable if they fail one of the two tests. As a result, it was demonstrated that the combined use of the two tests resulted in higher detection efficiency than the use of a single test. Another advantage is that FCT is a cheap, fast, and easy tool when used on its own, with a detection accuracy comparable to that of CIT.
Although the FCT may look promising as a tool for memory detection, there is room for improvement from the viewpoint of actual forensic investigations. Some concern is related to the number of questions that can be formulated with evidences. In the context of memory detection for a crime, many laboratory studies have used twelve to twenty questions in a typical FCT (Jelicic al., 2004; Meijer et al., 2007; Orthey et al., 2018), with larger test size being considered desirable (Denney et al.,1996). However, in real criminal investigations, it is difficult to find much evidence to use for memory detection tests of specific crimes. For example, the test is typically not immediately conducted after the crime has been committed. Hence, the culprit might forget some peripheral information about the crime by the time the test is conducted (Meixner & Rosenfeld, 2014), which means that even the culprit would guess on many questions due to genuine memory loss. In addition, if details about the crime are leaked and become public information, unknowledgeable examinees may become aware of these details and consequently demonstrate the same patterns as knowledgeable culprits when faced with a question about leaked information (Podlesney, 2003). As forensic experts who conduct memory detection tests cannot use leaked information, the available evidence that is useful for memory detection tests is limited. Finally, there are also cases where the information itself is so scarce that a minimum test length of 12 questions cannot be achieved. For any of these reasons, an FCT cannot be constructed in practice; its applicability is limited.
To address this problem, Orthey et al. (2022) developed repeated FCT in which individual pieces of evidence are repeated to achieve the necessary test length. In their FCT, the question specific to each piece of evidence (“What weapon was the offender carrying?” or “Which item did the offender steal?”) was changed to a more general question format (Which answer is more related to the crime?”), in which each crime-related answer alternative was combined with the incorrect answer alternative of all other questions. That is, each crime-related alternative was repeated; however, the combination of two answer alternatives was always unique in the FCT. The underlying idea of this approach is that FCT assumes that examinees genuinely judge which alternative they believe to be correct, which means that representing the same questions should always lead to the same answer. Consequently, the number of actual judgments, and therefore the real test length, does not increase. By providing a new comparison on each repetition, due to novel parings, the repeated FCT allows for variance in choices to occur when a piece of information is presented again. This should increase the real test length, and repeated FCT did not significantly differ in detection rate from typical FCT under the condition that all the same evidence was used (Orthey et al., 2022).
In this study, we proposed another version of the modified FCT to cope with the scarcity of information. In our modified FCT, multiple crime-unrelated answer alternatives were prepared for each question, and each crime-unrelated alternative was presented with the crime-related answer alternative as a unique pair. That is, a crime-related answer alternative (i.e., a correct answer alternative) had multiple crime-unrelated answer alternatives (i.e., incorrect answer alternatives). This test structure is inspired by ANS-based CIT, in which a question is composed of a crime-related item and three or four crime-unrelated items. For example, in the ANS-based CIT for a theft case where a watch was stolen, the question “Do you know if the stolen article was a personal computer?” would be followed by a series of answers, “A document?”, “A smartphone?”, “A watch?”, and “A purse?”. Introducing the test structure of ANS-based CIT to FCT contributes to increasing the test length without repeating the same pair of alternatives. Moreover, as two answer alternatives presented in each trial are within the same category, participants’ intention of choice for each piece of evidence would be more specific in each trial.
Recent FCT studies have also examined the theoretical background of typical FCTs. In particular, Orthey et al. (2017) demonstrated that participants’ test behavior is governed by specific strategies. In their study, participants’ behavior could be categorized into three different levels in accordance with cognitive hierarchy theory (Carmerer, Ho, & Chong, 2004). The level 0 strategy is carried out by a participant who complies with the test instructions, where the participants select the correct answer if she/he does not know the answer. The Level 1 strategy is carried out by a knowledgeable participant who assumed that typical FCT uses a level 0 strategy (i.e., compliance with test instructions) for detection. That is, the participant avoids correct answers, which leads to underperformance typically seen in a substantial proportion of knowledgeable participants with typical FCT. Finally, the level 2 strategy is carried out by knowledgeable participants who assumed typical FCT using a level 1 strategy (i.e., avoidance of correct answers) for detection. The participant attempts to calibrate performance at the chance level, which leads to the application of mixed selection with correct and incorrect answers. Orthey et al. (2018) altered the horizontal presentation of the correct answer alternative between each trial to countermeasure the Level 2 strategy, which therefore led to increased detection accuracy. Hence, it has been demonstrated that clarifying participants’ strategies during tests is important for increasing detection accuracy.
This study aimed to examine the applicability of the modified FCT as a memory detection tool in criminal investigations. We designed an experiment to compare the detection efficiencies of a typical FCT and a modified FCT. In our modified FCT, we prepared three incorrect answer alternatives and one correct answer alternative for each question. Each answer alternative was paired with every other answer alternative, resulting in three FCT trials that each provided a unique choice per question. In this study, the typical FCT had twelve questions, while the modified FCT had four questions with three trials each, resulting in two tests of equal test length. The same test length means that the number of choices that examinees make for crime-related answer alternatives is the same for both FCTs. We could not predict the superiority of either FCT with solid evidence from previous studies. However, if the amount of evidence is important for the efficacy of FCT, the typical FCT would be superior to the modified FCT. On the other hand, if the test length is important for efficacy, the efficacy would not differ between the two FCTs. Therefore, our modified FCT could be considered as a good memory detection test because it can be applied in a wider range of criminal investigations due to its smaller demand for available evidence.
In this study, we also aimed to examine the characteristics of modified FCT from the viewpoint of participants’ strategies. In recent FCT studies, the relationships between participants’ strategies and detection rates have been investigated in detail, and new methods of stimulus presentation have been developed to counteract specific intentional strategies (Orthey et al., 2018). Therefore, in this study, we aimed to clarify the specificity and sensitivity of the modified FCT based on each participant strategy.