The European Association of Psychology and Law (EAPL) annually organises a conference to bring together researchers and practitioners operating in a forensic context. Combining different disciplines, such as psychology, criminology and law leads to a multidisciplinary conference with presentations on topics like detecting deception, false memories, presenting forensic evidence in court, investigative interviewing, risk assessment, offenders, victims and eyewitness identification (see program). This year’s conference took place during the 24-27th of June in St. Petersburg and I summarised a selection of talks given during this conference.
Tuesday the 24th of June, 2014 Symposium 16.30-18.00 – Allegation: True or false
Van Koppen: I don’t know why I did it: motives for filing false allegations of rape. The first in a series of three talks (see Horselenberg & de Zutter). Explained the basis of the false allegations of rape in a Dutch & Belgian research project. Their conclusions are that the existing data (Viclas) and models are insufficient. Researchers on the current project went through rape cases between 1997-2011 and found more than 50 false allegations. Subsequently, they investigated the reasons why these people made false allegations and found in addition to the already known factors (especially emotional reasons and the alibi factor were often present; on the other hand, mental issues and vigilance were not) that in a substantial amount of cases it was unknown why this person made a false allegation of sexual abuse. Some people reported not knowing why they made the false allegation (even when pressured by the interviewer to provide a reason), and in other cases the researchers couldn’t find out because the police hadn’t asked or did not write down the reasons down, so it wasn’t in the case file. In conclusion, false allegations of rape happen, they cause problems, and it is not always clear why people make these false allegations.
Horselenberg: Detecting false allegations: A trait, skill or a gamble? How good are we at detecting false allegations of rape? And does training help? To answer these questions, they tested 3 groups of people and measured if some people are better at distinguishing true and false allegations of rape: lay, police officers and vice officers. All groups scored around chance level; experience did not increase performance. However, it did increase confidence level. Tool is needed.
De Zutter: A valid tool to detect the true nature of allegations of rape. False allegations will be similar to fabrication lies, including: Behave as liars do, based on own sexual experiences and based on beliefs. Beliefs are partly formed by information presented by the media (more extreme than “regular rape” and not very detailed). Ran a study to test a model they made based on 27 true and false allegations of rape. Decision matrix: N = 75. 30/32 correct detection, false positive: 1/23 and false negative: 1/9. Prediction matrix: 19/24 correct, 3/14 false positive, 2/11 false negative, Error rate is 7/75 (9.9%). Although the false positive and negative rates are worrying, the results are better than current decision-making strategies based on instinct.
Discussion:– 1) How did you decide which criteria to use to select the cases to build the model on? Answer: We needed to know the ground truth (i.e., that the false allegations were really false) to make sure we were building the model on real true and false allegations, so our criteria were quite strict. 2) Do vice officers get specific training to detect false allegations? Or how do you decide that training doesn’t work? Answer: The police officers just finished their training on how to detect false allegations, so they did receive training. However, to make sure that training didn’t help, one should test the group before and after training, rather than the different categories used now. 3) Do you really believe some of the false allegation people did not know why they did it? Answer: I couldn’t interview them in person, so I had to rely on the existing police interviews. But police officers really didn’t like it when people said they didn’t know, and they experienced lengthy interviews about the false allegation, and they stuck to: Spur of the moment, but I don’t know why. 4) Is unknown due to the police interview really unknown or undocumented? Answer: Can be both. 5) Because police officers are trained to discriminate between true and fake criminals, but those characteristics are different than true and false rape allegations. Answer: We found chance level, but we expected a truth or lie bias. Police officers were behaving similarly to lay people, so they could not rely on their training. Although training did make people more confident. So maybe they need help from an objective tool to make these decisions. The tool is not to replace the police, but to aid them, to help them to improve their detection skills.
Korkman: The prevalence of founded and unfounded Child Sexual Abuse (CSA) suspicions in a representative sample of Finish adolescents. CSA police reports have been increasing in the past decade (OPTULA crime statistics Finland, 2012). This may be due to an increase in reporting rather than an increase in cases. However, not all reported CSA cases are true, and forensic investigators have a truth bias regarding CSA allegations. They collected data on 12 & 15 year old Finish adolescents, n = 11364, response rate 75%. CSA was defined as sexual experiences with someone >5 years (so they include consensual sex between a girl with an older boyfriend). 2.4% reported CSA experience (3.6 for girls and 1.0% for boys). 13.3% of CSA has been reported to authorities (no difference in age and gender). A question about false allegations (has any adult at any time suspected you had been sexually abused although this had not happened) led to 1.5% of false allegations. 14.5% of those false allegations were investigated by authorities. In total, according to the children, of all cases reported to authorities, about 41% were unfounded, and 59% were founded. So false suspicions exist, and get reported to the police. The media and the option to google CSA has played a role in this increased awareness and the existence of false suspicions à availability bias. On the bright side, the authorities are more aware of CSA and more cases get reported and crime rates go down (in US).
Questions: Why was the response rate so low (25%) if study conducted in high school? Answer: The questionnaire was quite long, they may got bored, thought it didn’t apply to them or didn’t understand the question. You didn’t include CSA by people who were between 0-5 years older (or younger) than the alleged victim. Do you have any insights? Answer: There were one or two cases, but we didn’t include them here.
Tadei: A Bayesian model to assess child sexual abuse reports and identify the offender. He created a model to predict if a child was sexually abused based on specific factors. The example he presented was a model predicting if a child was sexually abused given that he is anxious. In general, the CSA prevalence in Europe in: 11-16.5% for girls and 3.8-8.4% for boys. In Finland, 8% for girls and 2-3% for boys. The variability in these percentages is thought to depend on country (both prevalence and reporting) and the age of the sample. However, there are problems with CSA reports: 50% of questions are suggestive and in 70% of the cases there was no physical evidence. So how to use this information Using Bayesian statistics for CSA reports: Prevalence of CSA = 5%. 60% of abused children have anxiety. 30% of non-abused children have anxiety. The likelihood of a child being abused knowing he/she is anxious, is: 60%*5%/28.5%+3%=9.5%. This model will become better when more variables are included. The authors created software that other researchers can use as well. They tested the model in a pilot and large data file of 24482 Finnish 9 & 15-year old children; it included 742 variables. The software/model answered the question: Given a certain case with several known variables, what is the likelihood that this girl was abused?
Wednesday the 25th of June 14.00-15.30 – Symposium Lie detection: Research, theory and practice
Watkins: Sampling variations in deception detection experiments: The consequences of misspecification. Main message: Start using a general linear mixed effects model instead of the standard ANOVAs.
Watkins: Lie hard with a vengeance: Getting specific about cognitive manipulations. There is a recent trend towards a cognitive approach to the detection of deception, increasing detection rates with imposed cognitive load on the interviewee. Do these lie imposed cognitive processes disrupt executive control? This is based on the idea that we have a limited amount of resources. And is it true the other way around, is deception impaired by executive control processes? And if so, does this occur specifically when lying? Study: Senders: N = 114, participants tell a true statement, relax for 15 minutes to remove any depletion caused by the first task. Between-subjects false-opinion design including 2 control conditions. Judges: N=762, each judged 6 truths and 6 lies. The results showed that when cognitively loaded, senders were more often correctly judged (overall truths and lies) compared to participants in the low cognitive load condition. This effect was particularly large for the liars. Question: How does these results translate to judges and lawyers, to the real world? And will training help? Answer: It will not translate to high-stake lie situations, but the judges were not just students but a more general MTurk population, so it should be generalizable to other low-stake lies.
Warmelink: Evaluative priming task as a lie detection tool: More data, better statistics. This study is based on the concept of intention related positivity. They used a reaction time task, using a prime followed by an adjective that should be classified. People are faster on classifying positive words (but not negative) when primed on their true intentions. This method can be used as an intention lie detection task, because true intention tellers will respond quicker to positive words, whilst there is no difference between positive and negative words for false intentions. N = 130 participants, 60 truth tellers and 64 liars; reaction time experiment on true and false intentions including 3 time groups. Ran 2 types of analyses: Both ANOVA and General Linear Mixed Effects Model. The ANOVA should a three-way effect of valence, veracity and time: In the medium (1000 millisecond) group, truth tellers responded quicker to positive words than the liars. However, this effect was not found in the small and large time group. Similar results were found using a general linear mixed effects model in R. They created a model using several predictors/factors to predict reaction time. The model shows that although most people are faster on positive words, but some people are faster on negative words, and who will perform in which way cannot be predicted by the model. So in conclusion, people respond differently to adjective valence and the results that truthful people respond quicker to positive words than negative words were replicated, but only in the 1000ms condition.
Question: It seems like you did the second model to try to explain why the effect wasn’t found in the slow and quick time condition. What did you find? Answer: Nothing. I suspect that the effect might be driven by a particular type of people that don’t show the quicker respond to positive words. And that those people were accidentally grouped in the same time condition. Or maybe my participants had difficulty understanding the prime words. Or 1 second is “the magic spot”. It gives them time to process information and respond.
15.45-17.15 Wagenaar Symposium – Lie detection & Memory
Brackmann: The impact of testing on the development of false memories. When witnessing a crime, often you will immediately talk about this event with others, including the police. What effect does this recall have on memory later on? And what effect does post-event misinformation have on the development of false memories? In the current literature, there are 2 opposing findings. Pansky et al found that testing leads to less misinformation. However, Chan et al found that people who were tested had more false memories. Her study, N=182 (children and adults). 2 day study, day 1: Participants witnessed a theft and were tested or not, and verbatim or gist. On day 2 there was an eyewitness account with or without misinformation. Subsequently, participants did a memory test. The results showed that children were more susceptible than adults. Testing helps to preserve eye-witness memory. However, findings did not support the work of Pansky et al or Chan et al.
Shaw: False memories, real consequences? What is the implication of implanting false memories of crime on real life? If past behavior is the best predictor for future behavior, what will be the implications of thinking you did something bad, although you didn’t (due to the implantation of a false memory). Methods: Familial informant false narrative procedure; implanting a memory about a participant’s own life, which involves contacting their parents for information and making sure that the target events hadn’t really occurred. Implanted memories in experimental condition were: Theft, assault and assault with a weapon. In the control condition: Animal attack, losing a large sum of money and injury. Cover story: Emotional memories, which is why their parents were contacted. Participants had experiences 1 target event (truth; asked parents) and not experienced another (false memory). During the first interview, false information was fed to the participant to enhance the false memory (including a close your eyes and imagine what happened tactics), and the experimenter pretended that their parents told them these details. 70% formed false memories of criminal events, and 77% for non-criminal events. Importantly, the false memories looked much like real memories. For example, complex multisensory memory realism occurred, including mentions of smell and observers could not distinguish true from false memories. If these false memories seemed so real, will that negatively affect future behavior? To measure this, self-report questionnaires about unethical and criminal behavior were administered before and after the false memory manipulation. Changes were found in the self-appraisal questionnaire & the pro-social tendencies measure (less likely to help a stranger), but not in the criminal thinking scale. In an Implicit Association Test pairing crime & me found that people were faster at combining me & crime when a criminal false memory was implanted compared to people who had a non-criminal false memory. So we can generate false memories of both emotional and criminal crimes and this affects their implicit associations and self-reports.
Question: Why would future behavior be changed by memory rather than by being an anti-social person; Personality is quite stable and might be the main cause that past behavior is the best predictor of future behavior, rather than having a (false) memory of bad behavior. Good question, I don’t know.
Li: Remembering what I didn’t do: An experimental investigation of the impact of lying on memory for actions We assume that lying impacts on memory, partly because you miss out on an opportunity to rehearse the truth, and lying is a form of self-generated misinformation. So does lying impair the memory for the truth? And do liars incorporate the lies in their memory? Method: Perform activities including handling several items, followed by getting instruction for the retrieval, 4 conditions: Truth, lie, opposite and no-rehearsal (control condition). A week later participants came back for an item recognition task: 48 out of 60 items were items they came across during the retrieval practice phase. Of the not-used items, 6/12 items were old and 6/12 were new. If people lied about an item, they had worse recognition memory for the action statement. Saying the opposite led to a marginally significant decrease in memory compared to the truth as well. Also, people were more likely to falsely remember items they had used for their lies than other items they did not include in their lies. So lying impairs the memory for the truth and it creates false memories.
Hagsand: Stumbling down memory lane: When to interview alcohol intoxicated eyewitnesses – immediate or delayed interview. Alcohol causes problems: 50-70% of all violent crimes in Sweden are alcohol related, and alcohol consumption is increasing. However, little is known about alcohol-intoxicated eyewitnesses. Alcohol interferes with the formation of long-term memories, which can make it difficult to remember things the next morning (blackout vs. grayout; no or little information is encoded into long-term memory). Participants consumed the equivalent of 3-4 glasses of wine in 15 minutes (vodka with juice). RQ: Should we question intoxicated witnesses immediately or when sober? If no alcohol is involved, immediate recall is better than delayed. Method: 2 (alcohol or no alcohol) x 2 (immediate and delayed or only delayed interview) between subjects design. No main effect of alcohol on amount of information (both sober and intoxicated witnesses revealed similar amounts of information). Alcohol also did not affect new details (both accuracy and amount of info) given in the second interview compared to the first. However, time was important; the immediate recall session led to more information than the recall session. In conclusion, immediate interview was better than delayed, both for sober and intoxicated witnesses and alcohol did not really impair eyewitness memory. Question: How can you distinguish between the effect of 2 vs 1 recall moment and the immediate vs. delayed recall? Because you only had 2 conditions: immediate and delayed or only delayed recall. Answer: So there were recall condition: 1 recall which included only delayed recall, and a 2 recall session which included both an immediate and a delayed recall. To answer the ‘amount of recall sessions’ hypothesis, we compared the 1 recall session with the 2 recall session. To answer the ‘immediate vs. delayed recall session’ hypothesis, we compared the delayed recall from the 1 recall session with the immediate recall from the 2 recall sessions. This way we did not need a condition in which participants only did the immediate recall.
Szpitalak: Social contagion and the misinformation effect: A comparison. The general question in the misinformation research area is that if people are provided with false information after an event, does that influence their memory on the final test? (this is traditionally tested by watching a video that includes a green car, followed by misinformation about a red car). And, how is this affected by the way misinformation is presented? Often, in misinformation research, the information is delivered via a written narrative. In the current study it was tested what the effect was of presenting misinformation through social interaction, rather than in writing. How close is this to the concept of social contagion? N = 108, 3 conditions (computer written up front, handwritten by other participant during the study and through social interaction). The results revealed that it did not matter which of 3 conditions people participated in on the amount of misremembered items, so it does not matter how misinformation is delivered. Although it is interesting to know that it does not matter how misinformation is presented to still have an effect, this study was not able to shed a light on the role of social contagion in the misinformation effect.
Thursday the 26th of June Symposium 10.00-11.30 – Interviewing tactics
Bull: The effects of different ways within interviews of revealing information to suspects. Overview talk on investigative interviewing: The early days of investigative interviewing started in the eighties, when the police became interested in using psychology. The first workshops were organized, and luckily the police listened to us, partly thanks to talks given by Shepherd, who was a former marine and someone the police admired. The audience was expecting him to “talk tough”, but instead he advocated the use of rapport building rather than the use of threat and violence, which helped warming the police up to the idea of investigative interviewing. Important research in the interviewing area: Holmberg et al, 2002 describing the importance of letting the suspect talk. The need to new strategies was shown by Baldwin (1993), who analyzed 600 police interviews (audio or video). His conclusions were that interviewers did not challenge statements, and interviewers often were badly prepared. Similarly, Moston, Stephenson and Williamson (1992) realized that in the majority of the interviews, interviewers did not try to obtain the account of what happened. Instead, they accused the suspect, and any evidence the police had was often presented early during the interview. This type of research on interviewing techniques was difficult back then, because accessing police interviews was very rare and normally restricted to the government, rather than researchers. The field then became interested in the elicitation of false confessions, and how they are induced by accusatory interviewing strategies. Soukara (2005; Soukara & Bull, 2010) managed to access 40 interviews and analyzed those interviews (divided in bits of 5 minutes). She showed which interview tactics (e.g., information disclosure) were leading up to a shift from denial towards confession.
Ewens: Drop the small talk when establishing baseline behavior in interviews. Police officers are advised to examine a suspect’s natural and truthful behavior at the beginning of an interview through small-talk, especially focusing on expressiveness, word-use and tone of voice (Inbau, Reid, Buckley & Jayne, 2013). This behavior is used as a baseline, and behavior during later parts of the interview that differ from this baseline behavior are seen as “suspicious”. However, small-talk is low-stakes, while the actual interview, especially when met with suspicion by the interviewer, is a high-stakes situation that is likely to elicit different behavior regardless of guilt, which makes this a bad comparison. Research has showed that context influences how people behave, such as discussing embarrassing vs. neutral topics, and discussing a topic that interests you vs. a topic you don’t care about, and being challenged vs. not being challenged during the interaction. Seen as behavior is topic related and can change over time, looking at differences between baseline behavior during low-stakes small-talk and the suspect’s response to critical interview parts may not be a good way to decide if someone was lying. In the present study, N=243, interviewees’ videoed behavior was rated (7-point Likert Scale) on emotions, cognitive load and behavioral control. A 2 x Veracity (truth vs. lie) 3 Period (baseline vs. Target period 1 at the middle of the interview vs. Target period 2 at the end of interview) between subjects design was used. Participants told a truth or lie about their occupation and had a few days to prepare their interview. Importantly, the study suffered from problems with low inter-rated reliability (emotion was entirely removed, and the other 2 categories were .6-.7). The results showed that all participants’ behavior as rated by the two independent raters, differed between the baseline and target periods regardless of veracity (i.e., also for the truth tellers). Therefore the current use of baseline behavior does not work. And if one really wants to use baseline behavior, more high-stake baseline (truthful) behavior should be established.
Question: How did you rate the videos? Answer: It was rated rather than coded, so the outcome is subjective. But we did use 2 raters so we could measure the inter-rater reliability. This analysis showed that we had to remove the emotional (nervousness) category due to low inter-rater reliability, so the current conclusions are drawn based on cognitive load and behavioral control.
Jo: Effectiveness of interrogation tactics and rapport building in obtaining changes in the suspect’s account from denial to admission. This study investigates how often information-gathering (open questions, rapport building) and accusatory interview (minimization, maximalization) strategies are used in Korea at the moment, and which tactics are effective in eliciting a confession. Nowadays, a confession is (still) very important evidence in Korea, but the pressure used to obtain a confession can be lead to false confessions. In Korea, a written statement from the suspect is always submitted to the court (written by the prosecutor and in cooperation wit the police and this can be used as evidence in court). There is also an electronic recording system that can be used (if there are any discrepancies) to rectify the written statement of the suspect. The current study that is investigating what type of interviewing tactics are used during Korean police interviews and if those tactics differed between interviews in which the suspect kept denying vs. interviews in which the suspect switched from denial to confession, is based on a study by Soukara, Bull, Vrij, Turner and Cherryman (2009). They found a total of 17 tactics used in police interviews by interviewers trained in the PEACE model (investigative interviewing). The current study involved transcribing and coding (2 coders; inter-rater reliability of .67 for interview tactics and .95 for suspect response types) 48 videos, 8 homicides and 37 sexual assault cases. The most common tactic was leading questions (31.4%), followed by interruptions (3.8%) and challenging (3.1%; notice the large difference between leading questions and any other type of interview tactic). As a comparison, open questions were only asked 1% of the time. Worrying, even more leading questions were used in the group of suspects that changed from denial to confessing. Rapport building was only used in 41% of the interviews, and did not differ between the change and continuous denial group. In conclusion, 9/10 cases were successful in obtaining a confession, mostly through accusatory techniques, which is worrying with regard to false confessions.
Richardson: Interview tactics: Using mimicry to encourage cooperation. Little research has been conducted on simple language use in interviews; rather, the focus so far mainly has been on interview tactics. Can changes in simple language use influence cooperation? There is a positive relationship between mimicry and cooperation, and this is shown through a similar the use of function words (9 categories: adverbs, articles, auxiliary verbs, conjunctions, indefinite pronouns, negations, personal pronouns, prepositions and quantifiers). Function words are about how something is said, rather than what is said, which means this type of analysis can be used across situations. In a first study using transcriptions of real police-interviews, a correlation was found between confession and language style matching (i.e., verbal mimicry; measured using Pennebaker’s Linguistic Inquiry and Word Count); interviews ending in confessions progressed to a state where the suspect gradually mimicked the language of the interrogator, whilst the opposite was found for interviews that did not lead up to a confession (this finding was ‘type of function word’ specific, so did not occur for all 9 categories of function words). In the next study the authors investigated the link between cooperation and confession in a more experimental setting. Instead of looking at the natural occurrence of verbal mimicry, can we train people to mimic in order to increase cooperation? And can the interviewee be primed to use certain categories of function words? Results: Yes, changing the interviewer’s language can be used as a prime, as it increases the desired type of word use by the interviewee. But does this change in function word use also increase cooperation? To answer this study, a second experimental study (third in total) was conducted, n = 64, in which a confederate was trained to match the type of function word used by the interviewee and this was implemented during a negotiation task. Results: When the interviewer was mimicking the interviewee, he gained more points on the negotiation task than when he was not mimicking the interviewee. So when the interviewer deliberately mimics, he benefits more from the interview than the interviewee does, and this effect is covert (i.e., not clear to the interviewee).
Symposium 11.45-13.15 – Lies in the investigative process
Akehurst: Creating a false identity: Investigating cognitive load for, and perceived credibility of, native and non-native speakers. The author used to investigate detecting deception in children, but is now looking at another vulnerable group; non-native language speakers, which is especially useful in border control situations, where people speaking lots of different languages continuously meet. How does second language use affect deception? Little is done on second language research so far. Lying behavior is likely to be affected by second language use because both lying and second language use is cognitively demanding. This increase in cognitive load may lead to increased detection rates. However, there is also the risk of false-positives, as non-native speakers might be thought to be lying just because of the behavioral changes caused by speaking in their second language. In a first study by Cheng & Broadhurst (2005), this hypothesis was not supported as they did not find a difference in detection rates between first and second language speakers. But the author argues that this may be due to methodological issues, seen as they used a false opinion paradigm, which is troublesome because you don’t know the ground truth. Also, participants in their study were allowed to switch between both languages (code switching), and participants were put in different conditions based on self-reported language proficiency. The current study was designed to overcome the methodological flaws of the Cheng & Broadhurst (2005) study and test the effect of second language use (French native and English as second language) on deception detection rates. 2 (veracity) x 2 (language) design in which truth tellers studied their own passport, while liars studied a false identity card and were then interviewed in an accusatory/confronting manner about “themselves”. Ground truth could be checked because all real and fake passports were present at the study, so the answers participants gave could be checked against the real and fake identity cards. The interviewer then judged deception. Results: The authors found a strong truth bias, especially for native speakers. Regarding truths, they found a 91% correct detection rates for native speakers, and 88% for non-native speakers. Regarding lies, only a 27% correct detection rate was found for native speakers (truth bias; the interviewer categorized more stories as true than false, leading to a high detection rates for truths and a low detection rates for lies) and a 60% correct detection rate for non-native speakers. So in general, the native speakers were rated as more credible than non-native speakers. Also, non-native speakers were rated as more nervous when lying compared to when telling the truth, although there was no nervousness effect caused by lying in native speakers. Last, native speakers were monitoring their own behavior more than non-native speakers. In a follow-up study 120 people will look at these videos (rather than using this one interviewer who was categorizing truths and lies in the current study), listen to the audio or read the transcripts to see if the results can be generalized and to measure if accents matter when judging deception. Possible explanations for differences between native and non-native speakers: Limited communication, blunted emotion and increased cognitive load.
Solodukhin: Lying in a non-native language: Effects on cognitive load and emotions. What effect does language proficiency level has on detecting deception and the suspicion they are met with during interviews? A 2 (veracity) x 4 (language proficiency; native, beginner, intermediate and advanced level) design was used, with 16 participants per group. Participants watched a video about a normal office situation, or a suspicious office situation (including maps, blueprints and bomb-making manuals) and were subsequently interviewed about this office, followed by a set of questionnaires measuring emotional and cognitive cues. Results showed there was no effect of veracity and proficiency on nervousness (i.e., emotion), thinking hard (cognitive load) and monitoring. However, for the truth tellers, native speakers thought they were less likely to be believed by the interviewer compared to non-native speakers. And regardless of proficiency, people preferred to be interviewed in their native language. Question: It didn’t seem like a difficult interview, there were a lot of closed questions. The easy questions might have influenced your lack of results. Answer: It might be, but I interviewed them, and non-native speakers did have difficulty answering the questions and did not know the right words for some items, so I personally don’t think the interview was too easy.
Strofer: Deceiving – More than lying. Mere the intention to deceive increases cognitive load. Lying can be more cognitively demanding than truth telling, due to suppressing the truth (Spence et al 2001; Walczyk et al 2004), saying something not true (Spence et al 2001; Walczyk et al 2004) and self-monitoring / impression management (Kassin 2005). Most cognitive load research in this area is done on actual lying, although in some situations the intention of lying is enough to create a false belief rather than an actual lie. Therefore the author experimentally investigated the effect of cognitive load on the intention to lie. Participants had to answer questions truthfully, deceitful or with the intention to lie in an emotional or cognitive load setting. The dependent variable was measured through skin conductance. Skin conductance was higher for lies than truths and intentions. In the cognitive load setting, the highest skin conductance occurred during lying, followed by intentions to lie and then the truth. So having an intention to lie also increases skin conductance regarding cognitive load, but not emotional responses.
Van Der Zee: Restoring the balance: How feelings of injustice affect lie tendency. Deception research is currently focused on improving detection rates, rather than looking at the deterrents of deception. However, a better understanding of the underlying factors that cause people to lie creates the possibility to decrease the amount of lies people tell in the first place. In an online study, it was found that people’s perception of the fairness of a request was the most consistent and significant determinant of the amount of personal information they would disclose (Malheiros, Preibusch, & Sasse, 2013). If people are more likely to be truthful when they feel the situation is fair, does that mean they will be more deceitful when a situation is unfair? The current study investigates this question by analyzing the effect of feelings of unfairness on lie tendency. The cover story of this first online experiment was a language proficiency study consisting of a grammar and semantics test. First, participants were told that they would receive a monetary bonus if they would answer all grammar questions correctly. Secondly, participants were told they would receive a monetary bonus for each correctly provided definition during the semantics test. At the end of the grammar task, participants received true (n = 75) or false feedback (n = 81), causing participants in the false feedback condition to believe that they would not receive the bonus despite having answered all questions correctly. It was then measured how many people cheated during the semantics test in order to increase their monetary bonus. Results revealed that the manipulation worked; participants in the unfair situation felt less happy and more anxious and frustrated after the feedback manipulation, whilst participants in the control condition did not. Consequently, participants who received false feedback cheated more often than people who did not receive such feedback, implying that being treated unfairly affects both emotions and lie tendency. However, in the first study, we did not distinguish between the effect of being rejected in general and effect of the perceived fairness of this rejection. To test this question in a practically relevant context, a second study was conducted in which participants were asked to fill in a travel insurance claim form based on a scenario and the main insurance guidelines. Participants received random feedback; or their claim was accepted, or it was fairly rejected (based on a violation of the main insurance guidelines), or it was unfairly rejected (the wide interpretation of a vague concept). The results revealed that the negative emotional response was caused by being rejected in general. And although rejection led to an increase in lying, this effect was especially large when rejected unfairly. So although the fairness of a rejection does not impact people’s feelings, it does affect their lie tendency. This knowledge has a high practical relevance, for example in the context of rejected insurance claims. A lack of transparency when rejecting applications may create feelings of unfairness in clients, increasing the chances that those clients will lie in order to “restore the balance”.
Vihkman: Cognitive and emotional characteristics of liars. The author promoted the research center in Russia he works at where they study different aspects of psychology and law. For example, they studied the personality of liars measuring SJT (based on information manipulation theory by McCormack), Machiavellianism and other personality measures. N=210 (3 conditions based on participants’ scores at personality scales: Liars, norm(al) group and honest respondents). The results showed that Liars scored higher in psychotism, and lower in moral control, volition and emotional stability. Liars differed from the norm group in personality traits such as self-control and affect. Honest participants differed from the norm group as well, mainly in temperament. We then wondered, how are the tendency to lie and our thinking system connected? N=70, police investigators and students had to decide if 5 (real) criminals were convicted for a violent murder or a non-violent fraud crime. Information about criminals was presented on a website, incl. info and picture and participants could navigate their own way through the website. Results: Success of identification was not affected by professional work experience (no difference between students and police officers). However, the time participants spend on making a decision did make a difference, and longer, deliberate decisions were better than quick decisions. That made the authors wonder, what is the link between suggestibility to cognitive biases and detecting deception? To this end, they are developing a psychometric test to measure the susceptibility to cognitive biases.
Question: Was this an experimental design, i.e., did you assess participants to the true, norm and lie condition, or was this division based on personality? Answer: Based on personality. We used 2 tests, a psycholinguistic case test and a Machiavellianism scale. Whenever people scored high on both, we categorized them as liars, when they scored low they were categorized as honest and middle group served as the norm? Criticism: If you categorize people based on their personality as liars, normal and honest people, then how much does it mean that you find personality differences between liars and truth tellers?
Symposium 13.30-15.00 – Lie detection tools
Tekin: Strategic Interviewing to elicit new information: Making liars more forthcoming. Can we use a strategic evidence disclosure (SUE technique) to improve police interviews? (i.e., get more information/make suspects more forthcoming). Study: N=120, 3 (interview technique: SUE, early disclosure and control interview) x 2 (veracity: Truth and lie) design. Each interview started with a free recall section, followed by questioning. Dependent variables were the amount of details disclosed and the suspect’s perception of the evidence. Suspects did a pre-interview task and were subsequently interviewed about this experience. The interviewer had some information (especially what happened before and after the crime), but not all crime related information, and when interviewers revealed this evidence was one of two independent variables. Results: A main effect for veracity was found; truth tellers revealed more new information than liars. Interview technique did not affect information disclosure. However, interview technique did affect the opinions of the suspect. Also, the more information liars thought the interviewer had about the crime, the more details they gave. This did not affect truth tellers. So the perception suspects had regarding how much evidence and information the interviewer knew about the crime was more important for the amount of information disclosed by the suspect, rather than when this information was disclosed (i.e., interview technique). In a next study however, it was shown that the SUE technique (withholding the knowledge the interviewer has about the evidence until later during the interview) did have a positive effect on information disclosure, so investigating when evidence should be presented during interviews seems an interesting avenue.
Hillman: How much does it hurt? Detecting deceptive motion using point light displays. Can we detect deceptive motion? In medical insurance cases, the impact an injury has on someone’s life will depict how much money they will get. Some people exaggerate or lie during these medical assessments, leading to the cost of insurance fraud of 2.1 billion a year. Can behavior show if people are lying? However, often behavior is measured per isolated body part rather than looking at the entire body. Therefore, in this study the author used point-light display with 9 cameras to measure the full-body behavior of people with “genuine” (the “injury” was constructed by attaching a Lego wait to the leg) and fake injuries. In this study, light-point videos were created existing of 10 women who, entirely faked an injured limp, exaggerated an injured leg or had an actual impaired limb (existing of a healthy leg with a Lego weight attached to it to create an impairment). Observers n=85, rated each video clip for pain and strangeness, not realizing that some of the videos were by the same person due to the light-points. Exaggerated limbs were rated are more painful than fake limbs, and both exaggerated and faked limbs were rated as more painful than the genuine limbs. Importantly, exaggerated limbs were rated are stranger than fake limbs, and both exaggerated and faked limbs were rated as stranger than the genuine limbs. In addition, taking a biomedical perspective, the authors showed evidence that when faking a leg injury, liars tried to control the opposite side of their injury (mainly arms). These results indicate that we can use full-body analysis of human behavior as an indication of deceit in faked injuries situations.
Nahari: Inter-personal reality monitoring: Does judgment framework matter? Does the purpose of the judgment affect the criteria assessments? N=120. Participants were trained in using Reality Monitoring (RM). Then the judgment context was manipulated; relevant judgment context (the criteria predict veracity), non-relevant judgment context (criteria predict literary attractiveness) and no judgment context (criteria test textual features). Each participant rated 6 life stories (3 genuine and 3 false). In all 3 conditions, participants perceived the truthful stories as having more details than false stories. Importantly, accuracy rates for both truths and lies were higher when the context was relevant, compared to not relevant, so context matters.
Aldert Vrij: Lie detection in intelligence interviews
1. Undercover interviewing: Big difference in intention to lie vs. lying about past events, and in intelligence interviews the intention to lie is mainly important. First topic is undercover interviewing. This happens if need to be decided if someone is dangerous or not. A suspect will be followed for several days to gather intelligence, and undercover interviews (that are supposed to not blow the interviewer’s cover) will occur. Those interviews need to be very brief, but lie detection is normally easier if a suspect talks for a longer period of time. However, spatial questions, that are useful to detect deceit, can still be asked. Beneficial, these short undercover interviews are ideal to measure cooperation and if the suspect is being forthcoming. We know that liars prefer preparing (stereotypical) alibis, and they prefer not to be linked to incriminating evidence, which can decrease their cooperation and make them less forthcoming. In Study 1 (Vrij, Mann, Jundi et al., 2012), n=86 (half were genuine tourists and half were instructed terrorists). Terrorist participants had to prepare a story on how they would visit the Isle of Wight. A casual, undercover student interviewer (pretended to do a tourist survey for the university) asked each participant some questions: What are you going to do? (events vs. activities) At what time will you return? (time) Can you show the places you are going to visit on the map? (spatial) Do you mind if I get a quick snap? (measures cooperation). Results showed that the liars reported more events, but less activities and time mentioned. Liars also made more mistakes at identifying on a map where they were going, and had more hedges. Truth tellers were more cooperative and allowed a picture to be taken. In a second study, 53 participants were send to take some photographs at a square with a cover story of watching the 2012 sport event on large screens at that square. Before the participant left the square, the mime artist asked if they took a picture of him, and if he could check. There was no difference in the amount of people that took a picture of the mime player between truth tellers and liars. However, 60% of the truth tellers allowed to check the photo, while only 20% of the liars were okay with this request. So again, truth tellers were more forthcoming and cooperative than liars. In conclusion, undercover interviewing can be used to measure if people are forthcoming or not. Specifically, will they later in a follow-up interview mention the undercover interview or not? If not, it is more likely that they are lying. This claim was studied and confirmed in a collective undercover interview experiment about a visit to the park where participants were undercover interviewed, followed by a “more traditional interview”. They found that the number of times that participants mentioned the undercover interview in a first yes/no question was dependent on being a truth tellers=, 0.44 or a liar 0.09. During the rest of the interview a similar pattern was found, truth tellers acknowledged the undercover interview 2.22 times vs. liars who only mentioned the interviewer 1.63 times. In conclusion, undercover interview elicit diagnostic cues to deceit for several reasons, including that liars are less cooperative and forthcoming than truth tellers.
2. Collective interviewing: Suspects often commit crimes in groups, yet deception research is often focused on the individual. This is of especial use in practical applications at the street (e.g., border control, groups of youth/people in the street and interviews for obtaining citizenship through marriage). An additional benefit of collective interviews is that you have an extra set of information; how the interviewees communicate with each other. We know that recalling shared experiences lead to many interactions (Rajaram, 2011), but liars try to keep their story simple, so there is little to add and little need for interruptions and corrections. So we expect liars to interact less with each other than truth tellers. In a study, n = 43 pairs of friends, truth tellers had lunch together and liars had to steal money and pretend they had lunch instead. Results showed that truth tellers interrupted each other more (large effect), corrected more, added more information (large effect) and gaze more at each other, but less at the interviewer than liars. These effects can be use strategically by the interview by encouraging turn-taking; decide who starts talking and ask the other person to continue after 20 seconds. This was tested experimentally, n = 46 pairs of real couples (dating at least 1 year an cohabiting) and fake couples (friends pretending to be a couple and cohabiting). Pairs were send out for coffee to prepare the interview and we subsequently interviewed together. Dependent variables were: Waiting after response latency, waiting to start taking over; Repeating what the other had already said, Continuing with what the other was saying. Liars paused more per swap, repeated more what was already said and continued less then truth tellers. A next study investigated the effect of turn taking on detecting deceit. 90 observers read transcripts and decided on truthfulness of the statements; no-turn taking 47%, turn-taking control 50% and turn-taking cues, 77%. So this study showed that when interviewing using the forces turn-taking strategy, observers are better at detecting deceit.
3. Analyzing the veracity of verbal and written statements: CBCA (criteria based content analysis; weak theoretical basis and difficult to use), RM (reality monitoring; useful when investigating children, memory based lie detection tool and easy to use) and the verifiable detail approach (memory based lie detection tool; can be used with adults and within-subjects; is more resistant to countermeasures and is easy to use; Nahari, Vrij & Fisher, 2012). The verifiable detail approach is based on the idea that liars want their lies to be detailed to be convincing, but they don’t want to reveal too much evidence, so they give details that can’t be checked. Truth tellers and liars had to write down what they had just been doing and the amount of verifiable details was measured. Truth tellers provided more verifiable details than liars. In a next study, half of the participants were told that the interviewer would ask details that can be verified. This prime led to a high increase in verifiable details in truth tellers, but not in liars (no effect of prime in liars; interaction effect). So warning interviewees that the interviewer wants details that they can check will led truthful people to give you much more verifiable details than liars will, so this will benefit the interviewer rather than the interviewee. In real life, 75% of all alibis are corroborated by a witness, and still 60-80% of people are willing to lie for their friend or partner by providing them with a false alibi. 60% feels that faking an alibi is easy to do. How to prove that alibis are false? Study by Nahari & Vrij (2014) in which pairs were or kept together (truth) or separated during which one of the two committed a crime (lie), followed by the request to provide a written statement together about what happened. Truth tellers mentioned more verifiable details, 39.14 than liar, 13.06. So asking for verifiable details will magnify difference in providing those details between truth tellers and liars. Importantly, these details do not need to be checked, just mentioning they should be checkable is enough.
Cheryl Thomas (women’s human rights attorney): Psychological abuse and domestic violence law. We are working with activist groups and translating their wishes to the judicial system. Domestic violence is defined within this talk as violence between intimate partners. What happens after reporting a case of domestic violence? Are the responses effective? And would till report lead to a prosecution? Or would the victim end up being blamed? We found 25 reports on the judicial response to domestic violence. Shockingly, 20 years ago, most countries did not have any laws that dealt with domestic violence, but luckily now most countries in the West do have such laws. However, domestic violence is still a problem: The World Health Organization report (2013) showed that 30% of women worldwide suffered from intimate partner, physical violence. It’s the most prevalent form of violence against women. Similarly, a large European Union survey (28 member states) found that 1 in 3 women had suffered from sexual or physical violence since the age of 15. And these statistics do not include any psychological and emotional abuse, only physical, even though victims consistently identify psychological violence (coercive, shaming, intimidating and threatening behaviors) as a devastating part of the abuse. But this psychological violence is harder to proof. She gave an overview of current laws in different countries, such as Croatia, Tonga and Latvia. Although domestic violence laws are improving, the real effectiveness is dependent on how the judicial system is implementing these laws. Removing the threat of being prosecuted as a victim is an important step to make victims come forward and report these crimes.
Friday the 27th of June Symposium 10.00-11.30 – Forensic Expertise and Eyewitness identification and line-ups
De Keijser: The expert factor in DNA evidence: A source of substantive variation. Really interesting talk during which the author pointed out that there are differences between how DNA forensic reports are written and how they are subsequently interpreted in the courtroom. He ran a study in which the same description of a criminal case including 3 types of DNA evidence was sent to 30 forensic labs in Europe. A full report, each one competed by a different forensic DNA expert, was returned by 19 labs. First of all, a comparison of these reports showed that the reports were very different. For example, some reports were only 2 pages and only described the conclusion, whilst other reports were up to 13 pages long and much more informative. Subsequently, the conclusions that experts had drawn based on the presented DNA evidence were compared, and this comparison showed that although all reports were based on the same DNA evidence, experts did not agree with each other. Shockingly, statements went from: It is 5.7 million times more likely that the DNA belongs to the suspect than to someone else, which is highly incriminating for the suspect, to statements that it was unclear if the DNA belonged to the suspects, or even probably did not belong to the suspect. A follow-up study was conducted to measure how people who may participate as a jury one day interpreted these reports. Participants were invited to make a judgment based on three of these forensic reports, on the likelihood of the suspect’s guilt. Results showed that people’s perception of the suspect’s guilt was mainly dependent on which forensic report they received. These studies show that forensic reports are subjective and need to be more standardized and harmonized to limit damage, but it is also clear that the problem is bigger than just that.