text
string | predicted_class
string | confidence
float16 |
|---|---|---|
The pathogenicity of IL-17+ CD8+ T cells has been mainly investigated using transgenic mouse models. Using the OT-I transgenic model, antigen-specific IL-17+ CD8+ T cells treated with IL-23 were found to be diabetogenic when adoptively transferred into RIP-mOVA mice. The pathogenicity was diminished upon treatment with anti-IL-17A and anti-IL-17F antibodies, indicating that IL-17 is essential for disease development . Furthermore, in EAE mice myelin oligodendrocyte glycoprotein-specific CD8+ T cells isolated from the lymph nodes and central nervous system at the peak of disease were found to express IL-17 ex vivo (after PMA and ionomycin stimulation). These cells did not express granzyme B, indicating their potential pathogenicity was not dependent on a cytotoxic mechanism and may be related to pro-inflammatory cytokine production . As previously mentioned, healthy individuals carrying the protective R381Q variant in the IL23R locus had a lower IL-17+ CD8+ T cell frequency in peripheral blood compared to those who did not carry the variant . Since carriage of this variant (or others in high linkage disequilibrium with R381Q) confers decreased susceptibility to immune mediated diseases such as inflammatory bowel disease, ankylosing spondylitis and psoriasis , , , this indicates a potential unidentified role of IL-17+ CD8+ T cells in the pathogenesis of these diseases. In the context of psoriatic arthritis, CD8+ IL-17+ T cells frequencies were increased in the synovial fluid compared to peripheral blood, correlated with several clinical parameters of disease and were associated with erosive disease, suggesting these cells may play a role in the pathogenesis of this disease .
|
study
| 99.94 |
A growing evidence base indicates the presence of IL-17+ CD8+ T cells at sites of inflammation in humans. These cells bear resemblance to their IL-17+ CD4+ T cell counterparts in terms of phenotypic markers and cytokine co-expression typically associated with Th17 cells. The exact requirements for differentiation and/or polarisation of IL-17+ CD8+ T cells are less well-defined, particularly in humans. Future in-depth phenotypic, molecular and functional characterisation of these cells will help determine how IL-17+ CD8+ T cells may contribute to human inflammatory disease.
|
study
| 97.9 |
Offenders in homicide and sex offense cases often claim crime-related amnesia (Cima et al., 2002, 2004; Pyszora et al., 2003, 2014; Bourget and Whitehurst, 2007). For instance, even though it is hard to determine to which degree defendants may intentionally feign amnesia following a crime, Pyszora et al. (2003) found that 29% of a 1-year cohort of individuals sentenced to life imprisonment claimed memory loss for their deeds (31.4% of those convicted of homicide). While it might be that the intense emotional arousal that some perpetrators experience during the crime might impair memory (e.g., Kopelman, 1995), there is also the distinct possibility that perpetrators feign memory loss (Centor, 1982; Marshall et al., 2005). Although majority of jurisdictions are reluctant to equate amnesia with incompetency, claiming crime-related amnesia in court raises the question whether the defendant’s ability to understand the trial proceedings or his capacity to consult with his attorney are impaired (e.g., Cima et al., 2002; Tysse, 2005; Tysse and Hafemeister, 2006). For that reason, some individuals who are charged with serious crimes pretend to have memory loss for their offense (Christianson and Merckelbach, 2004; Smith and Resnick, 2007; van Oorsouw and Merckelbach, 2010).
|
review
| 93.25 |
Furthermore, when offenders adopt that strategy relevant information might be forgotten as it has been demonstrated that feigning amnesia has a detrimental effect on the genuine memory reported by feigners for those target events (e.g., Christianson and Bylin, 1999; van Oorsouw and Merckelbach, 2004, 2006). Although what perpetrators truly remember about the crime may differ from what they actually select to report or claim to remember, because of the risk of undisclosed information in high-stake cases it is crucial for the legal context to ascertain how people remember remarkable information over time despite having previously feigned amnesia (Porter et al., 2001; Bourget and Whitehurst, 2007).
|
other
| 99.75 |
Several studies have shown that feigning amnesia can undermine actual memory for a crime (Christianson and Bylin, 1999; Bylin, 2002; Bylin and Christianson, 2002; van Oorsouw and Merckelbach, 2004, 2006; Sun et al., 2009; Mangiulli et al., unpublished). The typical procedure to investigate the memory-undermining effect of simulating crime-related amnesia is as follows. First, participants are usually exposed to a written/narrative story about a crime and instructed to identify themselves with the perpetrator or asked to commit a mock crime. Next, participants are assigned to one of the two conditions: Some participants are asked to comply with the police by reporting as accurately as possible all information they remember about the event (i.e., further referred to as confessors); some others are instructed to minimize or evade their responsibility for the crime by feigning memory loss for the offense (i.e., referred to as simulators; Christianson and Bylin, 1999; Bylin, 2002; van Oorsouw and Merckelbach, 2006). Sometimes, a third condition is included (Bylin and Christianson, 2002; van Oorsouw and Merckelbach, 2004; Sun et al., 2009; Mangiulli et al., unpublished) consisting of participants who do not receive any instruction, and who serve as a delayed test-only control condition (i.e., further referred to as controls). In these studies, during the first memory phase, participants in the first two conditions (confessors vs. simulators) were given a free and cued recall test pertaining to the crime. One week later, during the second memory phase, all participants of the two (or three) conditions were requested to genuinely report about the crime event, through the same free and cued recall test. Typically, participants initially instructed to feign amnesia show a poorer memory for their crime than those who were asked to confess it (Christianson and Bylin, 1999; Bylin, 2002; Bylin and Christianson, 2002; van Oorsouw and Merckelbach, 2004; Sun et al., 2009; Mangiulli et al., unpublished). Thus, research on feigning amnesia suggests that simulating memory loss impairs actual memory for a crime (e.g., Christianson and Bylin, 1999; Bylin, 2002; van Oorsouw and Merckelbach, 2004). Furthermore, when a delayed test-only control condition was included in the experimental design, no significant differences in memory performance were observed between controls and simulators 1 week after either being exposed to or committing the crime (e.g., Bylin and Christianson, 2002; van Oorsouw and Merckelbach, 2004). In contrast to confessors, simulators and controls did not have to provide details of the crime just after the crime stimulus, namely they did not engage in rehearsing the crime. This lack of rehearsal, therefore, might explain the memory-undermining effect of feigning amnesia (see van Oorsouw and Merckelbach, 2004).
|
review
| 97.2 |
Relatedly, it is commonly observed that feigning participants comply with their instructions by withholding, distorting, and introducing new information (i.e., commission errors) on the initial memory test (Bylin, 2002; Bylin and Christianson, 2002; van Oorsouw and Merckelbach, 2004, 2006). Clearly, simulating participants use laypeople’s ideas about how feign amnesia works (Bylin, 2002). Thus, even though feigning amnesia might mostly lead to omissions (e.g., Christianson and Bylin, 1999; van Oorsouw and Merckelbach, 2004), it is not surprising that van Oorsouw and Giesbrecht (2008) found that participants initially instructed to minimize culpability for a mock crime increased commission errors over time, as compared with genuinely responding controls. Accordingly, the act of coming up with a personal, self-generated story of the crime (i.e., simulated version of the crime) could enhance errors, but may not affect the number of correct details provided (Chrobak and Zaragoza, 2008, 2012; van Oorsouw and Giesbrecht, 2008; Ackil and Zaragoza, 2011).
|
study
| 99.94 |
More recently, Mangiulli et al. (unpublished) explored whether confronting simulators with visual and verbal cues about a crime – by trying to induce rehearsal of details of the offenses – would prevent impairments in simulators’ memory. With this set-up, simulators performed on a similar level as confessors (when they were prompted with verbal cues), and interestingly they outperformed controls on a memory test for the crime regardless of a rehearsal induction. This indicates that lack of rehearsal indeed does not fully account for feigners’ memory detriments. Unlike previous research in this field (e.g., Bylin and Christianson, 2002; Sun et al., 2009), Mangiulli et al. (unpublished) used a video clip instead of a narrative story as a mock crime. They suggested that compared to a narrative story, a mock crime video was better encoded by participants engaged in role playing, such as feigners and confessors, leading to a better memory performance over time. It is well known, indeed, that visual stimuli are generally remembered better than verbal stimuli (i.e., words, narrative stories) since images are encoded into both verbal and image codes, while words are primarily coded verbally (Paivio, 1976, 1986). Moreover, images are more distinctive in their features and better evoked than words (Nelson et al., 1976; Mintzer and Snodgrass, 1999). Thus, even though controls were exposed to the same crime material, it seems that both confessor and simulator groups actively elaborated upon the crime video so as to provide specific statements concerning their instructions, contributing to a more solid memory trace of the crime event compared with controls. The Mangiulli et al. (unpublished), suggests that the memory-undermining potential of feigning amnesia is more modest and fragile than it has previously been assumed. The results of that study seem to indicate that when using a mock crime video as crime material, the phenomenon is limited to the comparison between confessors and simulators.
|
study
| 99.5 |
Following this line, the main purpose of the present study was to further investigate lack of rehearsal as the best explanation for the memory-undermining effect of simulating amnesia. We replicated the standard procedure to study feigning amnesia effects (e.g., van Oorsouw and Merckelbach, 2004) by using the same mock crime video employed by Mangiulli et al. (unpublished) instead of a narrative mock crime story (e.g., Bylin and Christianson, 2002; Sun et al., 2009). We showed participants a video clip pertaining to a violent crime and asked them to either feign amnesia (simulators group) or confess the crime (confessors group) during the first memory phase. We also included a delayed test-only control group consisting of participants who did not receive any instruction. After 1 week, we requested all three groups to genuinely report all the information they could remember about the offense. We expected that simulators would recollect fewer correct details of the crime than confessors (hypothesis 1). However, we anticipated that both confessors and simulators would perform better than the delayed-test only control group (hypothesis 2). Finally, we predicted simulators to report more distortion and commission errors (i.e., introduction of new information) than confessors on the subsequent memory recall (hypothesis 3 and 4, respectively).
|
study
| 100.0 |
Moreover, we attempted to extend the study by Mangiulli et al. (unpublished) by investigating whether the memory-undermining effect of feigning amnesia is modulated by inner speech activity. Inner speech refers to the subvocal rehearsing of personal events (Alderson-Day and Fernyhough, 2015) and includes various characteristics such as dialogicality and condensation, the presence of other people voice and evaluative/motivational inner speech (McCarthy-Jones and Fernyhough, 2011; Alderson-Day et al., 2014; Alderson-Day and Fernyhough, 2015). For instance, the use of evaluative/motivational inner speech such as “I should do this,” might be linked to the feigners’ inclination in being consistent with their own simulated version of the crime in distinct circumstances (e.g., during preliminary investigations). Yet, perpetrators might estimate their deeds by engaging themselves in a self-evaluative-talk. Accordingly, common contents of inner speech refer to self- addressed evaluations and emotional states, in which continued inner speaking would regularly refresh experiences and maintain the corresponding memory traces in an “inner loop” (Alderson-Day and Fernyhough, 2015). Thus, simulators might internally think of the offense they perpetrated like entailing consequences for the event. By doing so, they might feed the actual memory of the crime. If this was the case, we would expect a significant correlation between the individual inner speech traits and the memory undermining effect of feigning.
|
study
| 100.0 |
The present study was approved by the standing Ethical Committee of the Faculty of Psychology and Neuroscience, Maastricht University (ERCPN application - 167 06 05 2016). Using a snowballing sampling technique (Goodman, 1961), we tested 111 individuals who volunteered to take part in the study (range 18–58, Mage = 22.60, SD = 9.64; 70% women). Participants were randomly assigned to one of the three conditions – simulators (N = 37), confessors (N = 37), and controls (N = 37). The study used a 3 × 2 mixed model design with condition (simulators vs. confessors vs. controls) as between subjects variable, and memory test–retest (T1 vs. T2) as a within subjects repeated measure variable. The dependent variable was the proportion of correctly recollected information in a free and cued recall test. Furthermore, we calculated distortion and commission errors generated during memory tests.
|
study
| 100.0 |
Participants were tested in a quiet room. The study consisted of two phases. During the pre-experimental phase, each participant was invited to complete the Structured Inventory of Malingered Symptomatology (SIMS; Smith and Burger, 1997), and the Varieties of Inner Speech Questionnaire (VISQ; McCarthy-Jones and Fernyhough, 2011). This last instrument was administered to explore the relation between inner speech and the memory undermining effect of simulating. In order to guarantee homogeneity in our sample before the experimental phase, the SIMS was assessed to check for possible differences among groups with regard to their feigning tendency.
|
study
| 100.0 |
The SIMS1 is a two option self-report measure to screen for over-reporting of mental symptoms and consists of a 75 items which are divided into five subscales (affective disorders; amnestic disorders; low intelligence; neurological impairment; psychosis). It includes items asking for atypical symptoms (e.g., “Walking is difficult for me because of my problems with balance”). Answers indicative of over-reporting are summed to obtain a total SIMS score (α = 0.73).
|
other
| 97.9 |
The VISQ is an 18 item self-report instrument measuring the phenomenological proprieties of inner speech along four dimensions: Condensed (α = 0.72; “I think to myself in words using brief phrases and single words rather than full sentences”); Dialogic (α = 0.85; “I talk back and forward to myself in my mind about things”); Other People (α = 0.85; “I experience the voices of other people asking questions in my head”); Evaluative/Motivational (α = 0.76; “I experience the voices of other people asking questions in my head”). Participants have to rate a 6-point Likert scale anchoring from “certainly does not apply to me” (1) to “certainly applies to me” (6).
|
other
| 85.6 |
After the pre-experimental phase, all participants were requested to pay attention to the mock crime video and were instructed to identify themselves with the character that appeared on the scene first (i.e., offender). The crime contained a violent scene between two armed men (2.30 min): A man entering a restroom was attacked by another man. After a severe fight, the attacker strangled the victim with his belt leaving him lifeless on the ground. After the exposure to the mock crime video, all participants were given a 10-min distractor task (i.e., computer game). This task was administered to avoid the possible ceiling effect in the following memory tests (Bylin, 2002).
|
study
| 100.0 |
Next, participants belonging to the simulator and confessor groups received the following instruction: “Imagine being the offender. Imagine that you have been arrested because you are the prime suspect of the murder. That day, a witness saw you there and all the evidence points to you. Right now, a policeman is asking you to tell what happened.” Following previous studies (e.g., Bylin and Christianson, 2002; van Oorsouw and Merckelbach, 2004; Sun et al., 2009; Mangiulli et al., unpublished) free and cued recall tests were employed as memory measures. Through a free recall, participants were invited to report their statements in accordance with the condition to which they were assigned. That is, simulators were instructed to report the crime as if they could not properly remember what happened. To evade punishment, simulating participants were free to omit, distort or report other information. Confessors, on the other hand, were asked to honestly report details about the crime in order to collaborate with the police. After this free recall, participants in both groups were given 14 cued recall questions concerning the mock crime video and were instructed to answer them by adhering to the instruction previously given (i.e., simulating or confessing). In line with previous research (e.g., Bylin and Christianson, 2002; van Oorsouw and Merckelbach, 2004; Sun et al., 2009), and in contrast to simulators and confessors, controls were not given a memory test. Although they were asked to identify themselves with the offender, participants in the control condition did not receive any instruction after the mock crime viewing and they were directly scheduled for the second session.
|
study
| 100.0 |
After 1 week, all participants – including controls – were specifically requested to be as specific as possible while providing all the information regarding the mock crime, as if they had decided to collaborate with the police. Contrary to the instruction received during the first memory phase, this time simulators were instructed to give up their role as feigner and recollect all they could remember about the target event. Confessors again received the instruction to comply with the police by reporting each and every detail about the mock crime video. Similarly, controls were now asked to recollect as much as they could remember about the criminal act. In the cued recall task, all participants were told to honestly answer the 14 questions. Finally, participants were thanked and debriefed.
|
study
| 99.94 |
Following Mangiulli et al. (unpublished), a scoring system was established to assess participants’ free recall. We first classified the mock crime video into 50 critical information units. Critical information was defined as a relevant piece of the video. For each correct unit of information reported (e.g., “I assaulted the victim from the back”), participants scored 1 point (maximum = 50). Moreover, participants earned a half point for each partially correct unit of information given (e.g., “I assaulted the victim”). In line with previous studies (i.e., van Oorsouw and Merckelbach, 2004; Sun et al., 2009), the entire correct score was transformed into proportions (range = 0–1) by dividing the number correct units reported by the maximum obtainable score. Additionally, we identified the number of distorted units (e.g., “I killed the victim by shooting him”) and commissions (i.e., introduction of new information that was not displayed in the video: “The victim was wearing a ski mask”). The first author and two assistants, who were blind to the study conditions, scored participants’ free recall. The interclass correlation coefficient (ICC) average measure for the number of correct free recall scores was 0.93 (p < 0.001); the ICC’s for distortions and commissions were 0.86 and 0.83, respectively (both ps < 0.001).
|
study
| 100.0 |
The cued recall test consisted of fourteen questions regarding both central (seven questions; i.e., weapon or blood), and peripheral details (seven questions; i.e., characters’ clothing, details about the location) of the mock crime video. Participants earned 1 point for each correct answer given (e.g., Question: “Where did the murder take place?,” “In the parking lot’s toilet”). Again, a half point was awarded for a partial correct answer (e.g., “In the toilet”). No penalty was given when participants did not provide any answer (e.g., “I do not remember”). The maximum obtainable score was 14. Similar to the free recall, the total cued recall score was transformed into proportions (range = 0–1). Furthermore, the number of distorted details and commissions were identified (e.g., “The murder took place near a fire extinguisher in the parking lot,” and “The murder took place in a bar,” respectively). The ICC average measure for the number of correct cued recall information was 0.96 (p < 0.001); the ICC average measure of distortions and commissions for cued recall was 0.88 and 0.79, respectively (both ps < 0.001).
|
study
| 100.0 |
A one way ANOVA was conducted on SIMS total score to exclude possible individual differences in the feigning tendency among three groups2 before the experimental phase. The main effect of condition was found not significant, F(2,106) = 0.03, p = 0.971, so that participants did not differ as to their simulating predisposition.
|
study
| 100.0 |
A 2 × 2 repeated measures ANOVA with condition (simulators vs. confessors) as a between subjects factor and memory test–retest (T1 vs. T2) as a within subjects factor was run on the free recall correctness score. The main effects of condition and memory test–retest were found significant, F(1,71) = 60.62, p < 0.001, ηp2 = 0.46, and F(1,71) = 8.86, p = 0.004, ηp2 = 0.11, respectively. The significant condition by test–retest interaction, F(1,71) = 24.66, p < 0.001, ηp2 = 0.26, showed that simulating participants reported more correct information at T2 than T1, t(35) = 7.83, p < 0.001, d = 2.00. This indicates that participants in the simulation condition properly followed their instruction. No difference was found for confessors in the proportion of correct information reported at the two memory phases, t(36) = 1.16, p = 0.25.
|
study
| 100.0 |
In order to observe differences in the correctness score among all conditions of the design (simulators vs. confessors vs. controls), a one-way ANOVA was conducted only on the retest memory phase (T2). The main effect of condition reached significance, F(2,107) = 17.74, p < 0.001, ηp2 = 0.25. Post hoc test with Bonferroni correction indicated that simulators reported less correct information than confessors, p = 0.014, 95% CI [-4.22 -0.36], d = 0.61. As expected, and in line with our hypothesis (Hp. 1), feigning amnesia undermined memory for the mock crime video. Furthermore, in line with our prediction (Hp. 2), both simulators and confessors were able to recall significantly more correct information than controls, p = 0.009, 95% CI [0.47 0.43], d = 0.83, and p < 0.001, 95% CI [2.77 6.60] d = 1.36, respectively. Proportions of free recall correctness3 are shown in Table 1.
|
study
| 100.0 |
Two 2 × 2 repeated measures ANOVAs with condition (simulators vs. confessors) as a between subjects factor and test–retest (T1 vs. T2) as a within subjects factor were separately performed on distortion and commission errors. Regarding the distortions rate, the main effect test–retest, and the interaction effect condition by test–retest reached significance, F(1,71) = 6.46, p = 0.013, ηp2 = 0.08, and F(1,71) = 12.06, p = 0.001, ηp2 = 0.14, respectively. By contrast, the main effect of condition was not found to be significant, F(1,71) = 1.68, p = 0.20. Surprisingly, simulators provided more distorted details at T2 than T1, while confessors did not over time, t(35) = 5.50, p < 0.001, d = 1.10, and t(36) = -0.561, p = 0.58, respectively. Unexpectedly, and contrary to our hypothesis (Hp. 3), no significant differences were found between simulators and confessors at T2 on the number of distorted details provided, t(71) = 1.16, p = 0.25, d = 0.27.
|
study
| 100.0 |
With regard to the commission errors, the main effects of condition and memory test–retest were found to be significant, F(1,71) = 16.30, p < 0.001, ηp2 = 0.19, and F(1,71) = 14.55, p < 0.001, ηp2 = 0.17, respectively. Moreover, the condition by memory test–retest interaction effect was analyzed, F(1,71) = 9.42, p = 0.003, ηp2 = 0.12. Simulators significantly provided fewer commissions at T2 compared to T1, while no differences were found in confessors between the two memory phases, t(35) = 3.60, p = 0.001, d = 0.81, and t(36) = -1.14, p = 0.26, respectively. Against our assumption (Hp. 4), simulators did not significantly differ from confessors at T2 on the number of commissions reported, t(71) = 1.49, p = 0.14, d = 0.34.
|
study
| 100.0 |
Finally, two one-way ANOVAs were independently run on distortion and commission errors to investigate differences among groups (simulators vs. confessors vs. control) during T2. The main effect of condition did not reach significance with respect to both distortions and commissions, F(2,107) = 0.73, p = 0.482, and F(2,107) = 1.96, p = 0.145, respectively. Absolute numbers for both distorted details and commission errors are displayed in Table 1.
|
study
| 100.0 |
In line with the free recall analyses, an identical pattern of ANOVAs was conducted on the cued recall correctness scores. The interaction effect condition by memory test–retest, F(1,72) = 27.91, p < 0.001, ηp2 = 0.28, showed that simulators reported more correct information units at T2 than at T1, which was in line with their instruction, t(36) = 6.23, p < 0.001, d = 1.09. However, confessors did not differ in the amount of correct information reported from T1 to T2, t(36) = -0.31, p = 0.76. Interestingly, and in contrast to our hypothesis (Hp. 1), a post hoc test with Bonferroni correction showed no significant difference between simulators and confessors with respect to the number of correct details during T2, p = 0.128, 95% CI [-1.98.16], d = 0.44. Thus, feigning memory loss did not impair genuine memory for a mock crime when participants were requested to honestly recollect the target event through a cued recall test.
|
study
| 100.0 |
Moreover, partially supporting our prediction (Hp. 2), only confessors reported more correct details than controls since no significant differences were found between simulators and controls at T2, p = 0.03, 95% CI [0.07 2.22], d = 0.64, and p = 1.00, 95% CI [-0.83 1.31], d = 0.13, respectively. See Table 1 for cued recall correctness4 proportions.
|
study
| 100.0 |
By running the same set of ANOVAs, a pattern comparable to the free recall was observed on the cued recall distortion and commission errors. In contrast with our hypothesis (Hp. 3), no significant differences were found between simulators and confessors with respect to distortions at T2 t(72) = 0.34, p = 0.73, d = 0.08.
|
study
| 100.0 |
Yet, the main effect of time was found significant for the commission errors, F(1,72) = 6.20, p = 0.015, ηp2 = 0.08, showing that all participants (i.e., simulators and confessors) reduced the number of commission errors from T1 to T2, t(73) = 2.48, p = 0.01, d = 0.53. No other main or interaction effects were found to be significant on commissions, Fs(1,72) < 1.35, p > 0.25, ηp2 < 0.02, meaning that simulators did not differ from confessors on commission errors at T2 against our expectation (Hp. 4).
|
study
| 100.0 |
With respect to differences among groups at T2, the main effect of condition was significant for distortion rates, F(2,108) = 5.02, p = 0.008, ηp2 = 0.08. Bonferroni corrected post hoc test revealed that controls reported more distorted details than both simulators and confessors at T2, p = 0.01, 95% CI [0.18 2.08], d = 0.70, and p = 0.03, 95% CI [0.05 1.95], d = 0.58, respectively. Finally, no significant main or interaction effects were found on commission errors among groups at T2, Fs(2,108) < 1.73, p > 0.18. See Table 1 for cued recall distorted information and commission errors.
|
study
| 100.0 |
In line with van Oorsouw and Merckelbach (2004), we first computed the simulating amnesia effect for the feigning participants’ free and cued recall performance by calculating difference scores for the memory variables (▲ Free Recall = Free Recall T2 – Free Recall T1; ▲ Cued Recall = Cued Recall T2 – Cued Recall T1). Next, we correlated the simulating amnesia effect with VISQ (MCondensed = 14.11, SD = 5.24; MDialogic = 11.38, SD = 5.25; MOtherPeople = 8.31, SD = 4.46; MEvaluative/Motivational = 13.02; SD = 4.71). No significant correlations were found between the memory-undermining effects of feigning amnesia and these individual difference traits with respect to both free and cued recall performances, rs < 0.12, p > 0.47, and rs < -0.05, p > 0.11, respectively.
|
study
| 100.0 |
The present study aimed to replicate and extend previous research on the feigning amnesia for a mock crime paradigm (e.g., van Oorsouw and Merckelbach, 2004; Mangiulli et al., unpublished) to further study decrements in rehearsal as an explanation for the memory-undermining effect of simulating amnesia. With respect to our first hypothesis, feigning amnesia undermined the actual memory for the criminal event since simulators provided less correct information than confessors on the free recall. However, the memory detrimental effect following feigning of amnesia took place only during the free memory test, since previous simulators and confessors did not differ on the final cued recall test. The same pattern of results was observed on the accuracy score. Hence, feigning amnesia for a mock crime in this study did not lead to the strong memory-undermining effect as shown in previous research (e.g., Christianson and Bylin, 1999; van Oorsouw and Merckelbach, 2004, 2006). Moreover, it seems that the lack of the simulating amnesia effect is related to the memory test that was used, as evidenced by better performance on the cued recall test than on the free recollection test.
|
study
| 100.0 |
Next, according to our second hypothesis, simulating participants did report more correct crime-related information than those who were not interviewed in the first place (i.e., controls) on the free recall test – although no significant differences were found on the accuracy rates between both groups, in which distortion and commissions errors were taken into account. Note that this prediction is particularly important since, based on the absence of significant differences between feigners and controls on the final recall tests, “lack of rehearsal” has been pointed out as the best explanation for the memory-undermining effect of feigning amnesia (e.g., Christianson and Bylin, 1999; van Oorsouw and Merckelbach, 2004). On the one hand, our findings might be related to the crime stimulus that was used (i.e., mock crime video). Indeed, a considerable body of research has demonstrated that visual stimuli are typically remembered better than verbal material (Nelson et al., 1976; Paivio, 1976, 1986; Weldon et al., 1989; Mintzer and Snodgrass, 1999). Hence, the use of a video could have led to a more solid memory trace for the mock crime information. Relatedly, on the other hand, perhaps simulators might have processed crime-related information more elaborately than controls – who did not receive any instructions after the mock crime viewing – in order to come up with a personal simulated version of the offense. In fact, although one could expect that feigning amnesia mainly leads to omitting information, previous studies suggested that when participants were asked to recall a mock crime in such a way they had great difficulties in remembering what happened, they were even likely to provide an alternative self-generated story (e.g., van Oorsouw and Merckelbach, 2004, 2006; van Oorsouw and Giesbrecht, 2008). Indeed, van Oorsouw and Merckelbach (2006) found that one-third of their entire sample used an alternative story in an attempt to feign amnesia. Yet, this result appears to be in line with the idea that by enhancing an active elaboration of information during memory encoding (McWilliams et al., 2014), being tested in itself could promote correct recollection (Chan, 2010; Fazio et al., 2010). However, even though feigners were more accurate than controls, no significant differences were found between groups on the number of correct responses provided during the cued recall test. This may suggest that individuals who did not receive any instruction during the first memory phase might find it easier to report correct information when prompted by open-ended cued recall questions rather than through a free recollection (Craik and McDowd, 1987; Padilla-Walker and Poole, 2002).
|
study
| 99.94 |
Finally, regarding our third and fourth hypothesis, no significant differences were found between simulators and confessors with respect to both distortion and commission errors during free and cued recall tests. This result might indicate that feigners recovered up to the level of confessors when it concerned distorted or self-generated information. Of interest, simulators increased distortions from T1 to T2 on the free recollection test. Conceivably, when participants instructed to feign amnesia come up with a self-generated version of the crime, which is strongly related to the original event, distortions are more likely to occur (Chrobak and Zaragoza, 2008; van Oorsouw and Giesbrecht, 2008; Otgaar and Baker, 2017).
|
study
| 100.0 |
A subsidiary aim of the present study was to explore whether or not inner speech (Alderson-Day and Fernyhough, 2015) might work as a buffer against the memory-undermining effect of feigning amnesia. One could argue that the more simulators tend to think of their crime, the less simulating amnesia affects the actual memory of the offense. However, our analyses suggest that inner speech might not be involved in preserving the genuine memory for the crime.
|
study
| 99.94 |
In sum, we suggest that feigning amnesia in the first place might be seen as a way to preserve and perhaps enhance memory for the target event over time, as compared to not being initially interviewed. However, possible memory decrements for feigners might depend on simulating amnesia in itself rather than a mere lack of rehearsal (e.g., Christianson and Bylin, 1999; van Oorsouw and Merckelbach, 2004). That is, drawing on the Memory and Deception (MAD) framework (Otgaar and Baker, 2017), feigning amnesia is inserted in a lying-continuum from false denial to fabrication of alternative scenarios. According to the MAD, the amount of cognitive resources required by individuals is directly proportional to the type of lie exerted, and the memory outcome for the actual target event is strictly affected by the different lie adopted. More precisely, whereas false denial implies less cognitive resources leading to omission, fabricating entire stories ex novo requires more cognitive resources leading to commission errors (Otgaar and Baker, 2017). Therefore, located somewhere in the middle of this framework, one could expect that feigning amnesia might cause a more distinct memory-undermining effect on the original experience by mostly withholding information than feigning amnesia by distorting or self-generating new information might do. In this latter case, perhaps due to a cognitive re-elaboration of the event (McWilliams et al., 2014), simulating amnesia might partially affect genuine memory for the crime even though participants might potentially report distortion and/or commission errors (e.g., Chrobak and Zaragoza, 2008; van Oorsouw and Giesbrecht, 2008; Otgaar and Baker, 2017).
|
study
| 99.94 |
Also, it could be argued that the instructions given to participants to feign amnesia for a mock crime might play a role in the actual memory for those events. As rightly noticed by Otgaar and Baker (2017), when those asked to feign amnesia account for the target experience, both attempts at simulating memory loss and fabricating may be occurring. Therefore, future studies might consider to bypass the “feigning” instruction by instead instructing participants to avoid thinking of the crime as a consequence of an emotional distress. Such deliberate avoidance, which would produce forgetting in itself as some research has pointed out (e.g., Anderson and Green, 2001; Anderson and Hanslmayr, 2014), might better reflect memory processes in real offenders.
|
study
| 99.94 |
Several caveats of the present study need to be mentioned. Although we tested a large number of participants, our sample mainly consisted of women. Even though we asked participants to identify themselves with the main character in the video, it might be hard for women to identify themselves with a male offender. To avoid this limitation in future research, it might be better to use a mock crime video recorded in point of view (pov) in which the perpetrator’s gender is indistinguishable. Moreover, we did not specifically assess participants’ ability to identify with the offender, which represents a limitation. Secondly, we did not assess the emotional impact of the mock crime video on participants. Specifically, we do not know whether or not the material used had contributed to maintain the participants’ memory performance over time per se regardless of the instruction given at T1. In future studies, therefore, it would be wise to assess the emotional impact of the crime material. Moreover, in future research, it might be interesting comparing two or more different types of crime materials to identify the most suitable stimulus to use in the mock crime paradigm (e.g., video vs. narrative story). It could be the case that the memory-undermining effect of simulating amnesia, that already seems to be less solid by adopting a mock crime video rather than a narrative, would perhaps be even weaker with a more ecologically valid set-up (e.g., mock crime through virtual reality). Thirdly, we did not ask participants what type of strategy they adopted to come up with a simulated version of the crime. Certainly, this information would be interesting since it is not fully clear to which degree this detrimental effect on the genuine memory for a crime is due to the feigning amnesia in itself or to the act of self-generating an alternative scenario for the same target event (Otgaar and Baker, 2017). Finally, another limitation has to do with the VISQ inner speech scales and the lack of correlation between these latter and the memory-undermining effect of feigning amnesia. Arguably, for this specific measure our sample size (i.e., 37 feigners) may not have been large enough to detect the predicted correlation. Future research in this direction should involve a larger sample size to better investigate a potential relation between the simulating amnesia effect and the inner speech traits. Moreover, even though this instrument has satisfactory psychometric reliability, the VISQ in its present form does not tap traits such as cognitive functions (McCarthy-Jones and Fernyhough, 2011). These functions – mnemonic and attentional uses of inner speech – are, however, assessed by other instruments which were not used in the present study (e.g., Self-Verbalization Questionnaire; Duncan and Cheyne, 1999). In future studies it would be wise to tap participants’ inner speech activities through self-report measures instead of mainly assessing their inner speech predisposition.
|
study
| 99.94 |
Although it is always difficult to generalize experimental findings to real life cases (Schacter, 1986), research using laboratory mock crime scenarios are fundamental to increase our knowledge about crime-related amnesia (e.g., McWilliams et al., 2014). As a matter of fact, knowing that genuine memory of a crime might be largely uncompromised despite having previously feigned amnesia appears to be informative to forensic practitioners who are asked to provide an opinion concerning crime-related amnesia cases. Relatedly, police investigators might find it interesting that suspects might actually preserve memory for the crime and contribute to disclose specific crime-related details. Oftentimes, indeed, crucial information of crimes remain undisclosed when a report of amnesia emerges (van Oorsouw and Merckelbach, 2004, 2006), regardless of the fact that some perpetrators admit their guilt (Porter et al., 2001). Thus, at least to some degree, our findings suggest that perpetrators are more likely to recall a larger amount of information when prompted by cues rather than being asked to freely recall the crime (see also Meissner et al., 2012; Mangiulli et al., unpublished, Study 2), particularly when they might be persuaded to collaborate with the justice department (e.g., plea bargaining situation).
|
study
| 99.94 |
In closing, by using a mock crime video instead of a mere narrative, our findings suggest that the memory-undermining effect of simulating amnesia occurs to a lesser extent than that observed in previous research (e.g., Christianson and Bylin, 1999; van Oorsouw and Merckelbach, 2004, 2006). The present study, indeed, indicates that simulating amnesia partially undermines actual memory for a crime and that, apart from a tendency to distort some details, offenders might still have relatively intact memory for the target experience.
|
study
| 99.94 |
IM conceived of the presented idea supported by HM, designed and directed the study in collaboration with KO, processed the experimental data and performed the analysis assisted by AC, and wrote the manuscript with input from all co-authors. MJ supervised the project.
|
other
| 99.94 |
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer JS declared a past co-authorship with one of the authors HM to the handling Editor.
|
other
| 99.94 |
In humans and other mammals, meiosis in females and males differs both in terms of timing and in the number of gametes produced. As an essential outcome of meiosis, recombination also differs substantially between the sexes. Genome wide, human female crossover rates are ∼1.6-fold greater than that of males12345. Sexual dimorphism in human recombination is not limited to simple changes to the overall recombination rate, but also includes differences in the distribution of recombination throughout the genome. Specifically, male autosomal recombination rates tend to be lower than female rates over the majority of the genome, but tend to be elevated towards to the telomeres125. While these broad-scale differences have been known for decades, the fine-scale differences in recombination between the sexes remain to be fully characterized.
|
study
| 99.25 |
Nonetheless, previous large-scales studies of recombination have characterized differences between females and males24, and identified a number of variants associated with sex-specific recombination rates2678910. These findings imply that the genetic regulation of meiotic recombination has evolved to be different in females and males. Yet, how recombination rates are regulated along chromosomes in a sex-specific manner is largely unknown.
|
study
| 99.94 |
In this paper, we build new sex-specific genetic maps based on crossovers from more than 100,000 meioses obtained from published pedigree studies. These maps reveal the variation of recombination rates in females and males throughout the genome at an unprecedented resolution. We characterize this variation over a range of scales using wavelet analysis and investigate the genomic and epigenomic features associated with the sex differences. We show that female and male differences in recombination rate can be mainly attributable to the fine scale, and find sexually dimorphic patterns of recombination in THE1B elements and gene promoter regions.
|
study
| 100.0 |
To study recombination differences between the sexes, we assembled a collection of recombination events from six recent studies of human pedigrees26711121314, pertaining to a total of 104,246 informative meioses (57,919 female and 46,327 male meioses) (Table 1). The vast majority of meioses are derived from individuals of European ancestry, representing 93.7% of all meioses, and 6.3% are from other origins including African American (1.6%), East Asian (1.8%) and Latino American (1.5%) (Supplementary Table 1). The combined data set consists of 2,338,628 female and 999,007 male recombination events (Table 1, Supplementary Tables 2 and 3). The boundaries of these recombination events define a total of 833,754 and 18,039 single nucleotide polymorphism (SNPs) intervals on the autosomal and X chromosomes, respectively.
|
study
| 100.0 |
We constructed high-resolution female and male maps, by modelling the recombination rate in each inter-SNP interval using a Bayesian Markov Chain Monte Carlo (MCMC) procedure based on the assumption that the fine-scale recombination rate is similar across individuals of a given sex (Methods and Supplementary Methods). By combining information across multiple families, this method allows the refinement and localization of recombination events, and therefore allows recombination rates to be estimated at a finer scale. Applying this method to female and male separately, we obtained female and male mean posterior rates for each inter-SNP interval as well as credible intervals (CI). In addition to the resulting ‘refined' sex-specific genetic maps, we generated a sex-averaged refined map by averaging female and male rates in each interval. After refinement, the median resolution of autosomal recombination events is 26.9 kb, with 28.6% of events resolved to within 10 kb, compared to 34.7 kb resolution and 19.5% of events within 10 kb in the unrefined map (Fig. 1a). The female and male refined maps have similar median resolutions of autosomal recombination events of 27.3 kb and 26.1 kb, respectively.
|
study
| 100.0 |
To assess the accuracy of our posterior estimation of recombination rates, we compared the refined maps to maps generated using standard methods. The female and male genetic maps constructed with our method are in good agreement with previous pedigree-based maps24 (Supplementary Fig. 1), although it should be noted that we have reused the data use to generate these maps. A cleaner comparison can be made with the HapMap map15 generated from patterns of linkage disequilibrium (LD), where our sex-averaged refined map shows a Pearson r2>0.93 at the megabase scale (Fig. 1b). At the fine scale, where LD-based maps are expected to have higher resolution, the sex-averaged refined map shows an increased correlation to the HapMap map compared to the unrefined map and to the earlier pedigree-based maps. For example, at the 5 kb scale, the Pearson correlation between our map and the HapMap map is r2=0.62, whereas for the unrefined map it is 0.49, and respectively 0.46 and 0.44 for the Campbell et al.2 map and the deCODE 2010 map4. Similarly, the refined map shows improved correlation at scales below 200 kb to the LD-based CEU map generated from 1,000 Genomes Project Illumina OMNI array data16 (Supplementary Fig. 2b). Refinement also improves correlation for a refined Icelandic map constructed using data from Kong et al.7 only (Supplementary Fig. 3).
|
study
| 100.0 |
To evaluate the potential impact of the different ancestries in the refined map, we build a refined European map applying our method to the recombination events inferred in parents of European ancestry, representing 94.3% of female and 93% of male meioses included in the combined data set (Supplementary Table 1). The refined female, male and sex-averaged maps are highly correlated to the European maps at all scales (Supplementary Fig. 4). At the 20 kb scale, Pearson's r2 between the refined and European maps is above 0.96. At the 1 Mb scale and beyond, they have near perfect correlation. The two maps also have the same Pearson correlation to the HapMap and OMNI CEU maps suggesting that genome wide the refined estimates of recombination rates are mostly representative of Europeans.
|
study
| 100.0 |
We next tested the ability of our method to narrow down the location of recombination events to known recombination hotspots inferred from the HapMap data17. In the refined map, the mean recombination rate around hotspot centres is much higher than for the unrefined map, and is much more comparable to the rates estimated by HapMap (Fig. 1c). Likewise, the refined map shows marked improvement in the localization of events compared to previous pedigree maps (Supplementary Fig. 2c). Increased recombination rates are observed at LD-inferred hotspots in both males and females (Fig. 1d). While the elevation in recombination rate observed at hotspots is higher in females, the fold increase over the background rate is similar between the sexes (females: 19.5 cM per Mb/2.0 cM per Mb=9.75. Males: 12.3 cM per Mb/1.3 cM per Mb=9.46).
|
study
| 100.0 |
Compared to the unrefined map, the refined map shows a higher proportion of recombination concentrated in a smaller fraction of total sequence (Fig. 1e). In the sex-averaged refined map, 80% of recombination is localized in 10.6% of the genome, which is comparable to 13.5% in the HapMap, and much more concentrated than the 38.5% observed for the unrefined map. This is consistent with a better localization of events to hotspots of recombination in the refined map. Male recombination occurs in a narrower portion of the genome than female, with 80% of male recombination occurring in 6.1% of sequence compared to 8.8% for female. This is consistent with male recombination being slightly more concentrated in hotspots2.
|
study
| 100.0 |
Our sex-specific genetic maps allowed us to investigate sexual dimorphism in recombination rate at high resolution. We first sought to quantify the magnitude of differences in rate across the human genome (Fig 2a). We find 13.7% of the autosomal genome has a recombination rate difference of at least 2 cM per Mb between the sexes, with 9.3% of the genome being at least 2 cM per Mb higher in females, and 4.4% being at least 2 cM per Mb higher in males. At the more extreme end, regions showing at least 5 cM per Mb difference account for 4.3% and 2.1% of the genome in females and males, respectively.
|
study
| 100.0 |
We identified localized regions of the autosomal genome with significant sex differences in recombination rate while taking into account local uncertainties in rate estimates (Fig. 2b). Specifically, we computed CI around the mean posterior estimates of recombination rates within each SNP interval. We defined dimorphic regions as 10 kb windows around SNP intervals with significant sex difference in rate at the 99% level (Fig. 2b, Methods). Doing so, we identified 6,377 putative dimorphic regions, including 5,262 regions hotter in females and 1,115 hotter in males. By definition, these regions show significant sex differences in crossover rate, with a minimum difference of 2 cM per Mb, and a median of 9.7 cM per Mb (median rate difference for regions hotter in females: 9.2 cM per Mb and males: 12.4 cM per Mb) (Supplementary Fig. 5). We observe a higher number of regions recombining in females, which is consistent with higher genome-wide female recombination rates and male recombination being concentrated in a smaller number of hotspots. We next defined sex-specific hotspots as dimorphic regions with high recombination rate in one sex (mean rate>10 cM per Mb), but very low in the other (maximum rate across all intervals within the region<1 cM per Mb) (Fig. 2b). We find 304 female specific hotspots and 147 male specific hotspots. Applying different criteria for hotspot definition consistently yielded a small number of sex-specific hotspots relative to the tens of thousands of recombination hotspots estimated in LD maps1517 (Supplementary Table 4). Thus, in the autosomes, only a fairly low number of genomic regions have recombination limited to one sex and suppressed in the other. Dimorphic regions, including sex-specific hotspots, appear to cluster into regions that are more recombinogenic in one sex or the other (Supplementary Fig. 6). Most obviously, male recombinogenic regions are concentrated in subtelomeric regions, with 50% located within 8 Mb of chromosome ends.
|
study
| 100.0 |
We use wavelet analysis to characterize the variation in female and male recombination rate along chromosomes at a range of scales, and assess the scale-specific differences between the sexes. Wavelet analysis allows a sequence of observations to be transformed into a series of coefficients that describe changes in the signal at each location and successively broader scales. In Fig. 3a, we illustrate wavelet transformation applied to the female recombination rate, the male rate and their difference along a 2 Mb region of chromosome 10. Sex differences in recombination rate are revealed at multiple scales, ranging from differences at specific hotspots to variation over much larger scales. This highlights an advantage of wavelet analysis, which avoids choosing an arbitrary window size at which to compare female and male recombination rates.
|
study
| 100.0 |
We performed genome-wide wavelet analysis using female and male refined maps interpolated in non-overlapping bins of 1 kb for each chromosome. More specifically, rate estimates were log10 transformed and the discrete wavelet transformation (DWT) was applied using the simplest wavelet Haar function for scales from 2 kb to 16 Mb (see Methods). We used the detail coefficients of the DWT to quantify the proportion of the variance in the original signal that is captured at a particular scale, and cannot be attributed to any other orthogonal scales. This measure, known as the power spectrum, is shown for female and male recombination rates, as well as their difference in Fig. 3b (Supplementary Fig. 7a). The three power spectra are very similar to each other (as are the per-chromosome power spectra—Supplementary Fig. 7b–d) and show that, genome wide, over 93% of the heterogeneity in the log recombination rate signals is captured at scales below 1 Mb (94% and 96% for male and female, respectively). Furthermore, while the fine scales (2, 4 and 8 kb) explains 28% and 26% of the variance in the male and female rate signals, the greatest contribution comes from intermediate scales (16, 32 and 64 kb), with 44% and 49% of variance explained for male and female rate, respectively.
|
study
| 100.0 |
The wavelet decomposition allows investigation of the correlation between the localized changes in the sex-specific recombination rates at each scale. The variation in female recombination rate, as captured by the detail coefficients, is significantly correlated to the variation in male recombination rate from the finest scale studied, 2 kb, to the broadest, 16 Mb (Fig. 3c, Supplementary Fig. 8). The degree of correlation increases from r2=13.3% at 2 kb scale to a maximum of r2=36.8% at 512 kb scale.
|
study
| 100.0 |
As such, local variations in the recombination rate that can be attributed to broad scales are more strongly correlated between the sexes than those at the fine scale. Conversely, removing variation that can be attributed to the fine scale increases the correlation in the residual rate estimates between the sexes. Specifically, by comparing the smoothing coefficients at each scale, we see that the correlation between male and female recombination rates increases as a function of scale (Supplementary Fig. 9). As such, we conclude that the majority of the differences in recombination rate between males and females can be attributed to the fine scale.
|
study
| 100.0 |
Given the availability of both of female and male recombination rate estimates, we aimed to assess the scale-specific relationship between the sex-specific rates and known recombination correlates. To do this, we performed a multiple linear regression on the wavelet coefficients at each scale using a selection of known recombination correlates as predictor variables (see Methods). In our initial analysis, we included GC content, exon density, density of THE1B repeat and density of the expected 13-base PRDM9 binding motif (CCNCCNTNNCCNC) as predictors within the model, all of which are known covariates with the sex-averaged recombination rate51518192021. All of these previously associated predictors of recombination show correlations with both male and female recombination rates at some scales (Fig. 4). However, some differences in the strength of correlation emerge. For example, genome-wide GC content appears to be more strongly associated with female rate than male recombination rate over a wide range of scales (Fig. 4c, Supplementary Table 5). Conversely, THE1B elements appear to be more recombinogenic in males between scales 4 and 64 kb (Fig. 4c, Supplementary Fig. 10); a pattern that is not observed for other repeat classes (Supplementary Fig. 11). While PRDM9 motif density shows a high-positive association to both male and female recombination rate, it shows little influence on the difference between female and male rates, with the only significant (but fairly weak) correlation found at the 4 kb scale (P=1.8 × 10−5, two-sided t-test).
|
study
| 100.0 |
Previous reports have shown that recombination tend to be suppressed within genes4151719. Our linear model analysis of the wavelet transformations reveals a more complex relationship between exon density and the sex-specific rates. At intermediate to large scales, exon density has a negative effect on both female and male recombination rates, whereas it has a positive effect on recombination rates at fine scales (Fig. 4a,b, Supplementary Fig. 12). To further investigate this pattern, we plotted average recombination rates around the transcription start site (TSS) of GENCODE genes. At the fine scale, we observe an elevation in female recombination rate at the TSS that is absent in males (Fig. 5a). This peak of recombination at promoter regions is consistent with previous report based on human sex-averaged rates15, but appears to be driven by female recombination alone. This fine-scale pattern of sexual dimorphism holds with normalized recombination rates, and both for genes located in subtelomeric and centromeric regions (Supplementary Fig. 13). The elevation of female recombination rate at the TSS appears to be associated with the co-localization of the PRDM9 motif, with a 30% rate increase from background rate being observed for genes with a 13-mer degenerate motif within 5 kb of the TSS, and no such elevation for genes without a motif within 5 kb (Fig. 5b). Furthermore, the female peak appears associated more strongly to the motif location then the TSS location itself as the recombination peak is shifted away from the TSS with increasing distance to nearest motif (Supplementary Fig. 14), which suggests a role for PRDM9 in the observed dimorphism.
|
study
| 100.0 |
Sexually dimorphic recombination in functional regions may arise as a consequence of sex differences in DNA double-strand breaks (DSBs) that initiate recombination and/or their downstream resolution. To investigate further, we used published male recombination initiation maps22, and find that male DSBs are significantly depleted around the TSS, with 4.4% of them overlapping a male DSB peak within 1 kb compared to an expected 5.1% (CI: 4.8–5.5%) overlap at a random position (Supplementary Fig. 15), providing independent evidence for suppressed male recombination at the promoter region of genes. Finally, we investigated recombination at CpG islands, which are known sites of transcription initiation. In females, we observe an elevation of recombination rate at these sites that is again absent in males (Supplementary Fig. 16). These different lines of evidence suggest that female meiosis is permissive of crossover activity at the site of transcription initiation whereas male meiosis suppresses recombination at these sites.
|
study
| 100.0 |
In females, oogenesis is initiated during early embryonic development but remains in a state of arrest until ovulation. Conversely in males, spermatogenesis occurs continuously following the first wave at puberty. As such, differences in recombination may reflect structural or epigenetic differences in the genome at the time of crossover formation. Given PRDM9's role as an H3 K4 trimethyltransferase, we investigated the association of H3K4me3 peaks with male and female recombination. Since data are currently lacking for oocytes, we focused on a testis sample22.
|
study
| 100.0 |
We found male and female recombination rate to be strongly associated with density of H3K4me3 marks (Fig. 6a, Supplementary Fig. 17), and that H3K4me3 marks containing at least one degenerate PRDM9 motif are more recombinogenic in both sexes than those lacking a motif (Fig. 6b). H3K4me3 marks in testis are not all interchangeable to the marks associated with transcription in other tissue. The H3K4me3 marks specific to testis show increased rate in both sexes compared to marks seen in other ENCODE tissues alone (Fig. 6c), while H3K4me3 marks found in other ENCODE cell lines but not in testis show no effect on the recombination rate at all. As such, we predict at least some overlap in H3K4me3 marks between spermatocytes and oocytes during the initialization of recombination, likely due to PRDM9 action.
|
study
| 100.0 |
Recent years have seen great strides in our understanding of the molecular basis of recombination. This has been notably enabled by the availability of fine-scale and genome-wide assays that allow measurement of recombination rates and related factors in humans, mice and other species1152122232425. However, less progress has been made in understanding the underpinnings of sexual dimorphism in recombination. On one hand, this is because fine scale estimates of recombination are most easily obtained from patterns of LD, which are intrinsically sex averaged. On the other hand, experimental approaches for characterizing meiotic recombination rates tend to be biased towards males, owing to the ability to characterize recombination events or DSBs in sperm. Although small-scale studies have already been conducted2526, the difficulty of obtaining sufficient oocytes to characterize large numbers of completed crossover events in females means that such approaches are unlikely to be applied at scale, at least in human, for the foreseeable future.
|
review
| 99.7 |
The huge quantity of human genotyping data in pedigrees now provides an alternative means to study sex-specific patterns of recombination. In this study, we have exploited the large quantity of available data to construct sex-specific genetic maps that begin to approach the resolution of LD-based sex-averaged maps. Our fine-scale map of recombination highlights that a substantial fraction of the human genome recombines at different rates in females and males, and these differences are largely attributable to the fine scale. At this scale, our sex-specific maps reveal two consistent dimorphic patterns, first that THE1B elements are more recombinogenic in males and second, that female recombination drives an elevation of recombination in the promoter region of genes. As our maps are mainly representative of Europeans, it remains to be determined if these dimorphic patterns are found in other populations. Given the fundamental role of recombination in evolution, the observed dimorphism may have implications for understanding the role of recombination in shaping patterns of human genetic diversity. As more data is collected, it will become possible to not only build maps partitioned by sex, but also in terms of other factors known to influence recombination, including both genetic and environmental factors (such as parental age).
|
study
| 99.94 |
We assembled a collection of recombination events inferred from genome-wide SNPs data in human pedigrees comprising a total of 104,246 informative meioses (57,919 female and 46,327 male meioses). This collection is derived from six sets of recombination data, including one that we have previously analysed2, as well as five publicly available data sets from previous studies67111213 (Supplementary Methods).
|
study
| 100.0 |
The 22 autosomes and X chromosome were split into 833,754 and 18,039 SNP intervals respectively, as defined by the boundaries of all recombination events found in the combined data set. We estimated female and male recombination rates within each of these SNPs intervals. First, we build female and male unrefined maps based on the maximum likelihood estimates of rates (Supplementary Methods). Second, we implemented a MCMC procedure to build the refined maps. We assumed that recombination rate in each interval was independent from every other SNP interval. We modelled the underlying recombination rate within a SNP interval as a random variable drawn from a gamma distribution, with the prior parameters selected via inspection of the unrefined map. We used a Gibbs sampler to iteratively sample the SNP interval location of each recombination event, and then resampled recombination rates consistent with the event assignment. For each SNP interval, we obtained a mean posterior estimate of the recombination rate and the associated CI by running 1.3 million iterations, and discarding the first 300,000 as burn-in. We sampled from the chain every 100 iterations to obtain 10,000 samples. We applied this method separately to female and male recombination events, obtaining separate female and male refined maps. Our method to refine the estimates of recombination rates is available here: github.com/auton1/rMCMC. The recombination maps generated can be downloaded here: https://github.com/cbherer/Bherer_etal_SexualDimorphismRecombination..
|
study
| 100.0 |
To identify regions of the genome with significant rate difference between the sexes, we used the CI of posterior recombination rates obtained with our mapping procedure. At the 99% level, we found 6,936 intervals with significantly higher rate in females than males and conversely, 1,427 intervals with significantly higher rates in males than females. To retain only well-localized dimorphic intervals, those smaller than 100 bp (n=250) and larger than 10 kb (n=475) were excluded. We defined dimorphic regions as 10 kb windows centred around this set of high-confidence and well-localized dimorphic SNPs intervals. Adjacent dimorphic intervals were merged and considered as one larger dimorphic interval. We verified that female recombinogenic regions contained only dimorphic SNP interval with higher rate in females, and conversely for males recombinogenic regions. We excluded 10 kb regions with recombination rate difference between the sexes below 2 cM per Mb. We also excluded regions overlapping gaps, as defined in the UCSC table. In the cases of overlap between regions, we kept the region with highest difference in rate between the sexes.
|
study
| 100.0 |
We used the estimation of female and male recombination rates from the refined map and interpolated the map in non-overlapping bins of 1 kb across each of the 22 autosomes. Bins were set to `̀NA'' if they overlapped a centromere, gaps in the assembly, or were located in low-resolution regions of our map (where inter-SNPs intervals size>50 kb). For each bin, we also computed a number of genomic and epigenomic annotations, as listed in Supplementary Table 6. GC content and CpG content were computed using the hg19 build of human reference genome. The percentage of bins in CpG islands was computed using the CpG island track from the UCSC Genome Browser. SNP density in Europeans was computed using the 1000 Genomes Phase 3 variants list16. We computed the exon percentage as the proportion of each 1 kb bin overlapping transcripts listed in the Gencode gene list (v19). Per cent overlap with a given repeat family or repeat elements were computed using the UCSC repeat masker table. We obtained DNA motif locations using our program (available here: https://github.com/auton1/motiflocation) applied to the reference genome. Per cent overlap with histone peaks was computed using the ENCODE data, as well as H3k4me3 peaks from testis sample kindly provided K. Brick and published in ref. 22.
|
study
| 100.0 |
For the wavelet analysis, each signal was padded with NAs to obtain a number of bins equal to a power of two. As such, any wavelet coefficient influenced by edge effects would also be NA, and hence excluded from analysis. Using each annotation as the signal, we applied a DWT, as implemented in the WMTSA R package, and using the Haar wavelet function to yield detail and smoothing coefficients for scales from 2 to 3,278 kb27.
|
study
| 99.94 |
The recombination maps generated in this study are available for download at: https://github.com/cbherer/Bherer_etal_SexualDimorphismRecombination. The refined genetic maps include: the refined maps, the refined European maps, and the refined Icelandic map. For each of them, we share the female, male and sex-averaged estimates of recombination rates.
|
study
| 98.3 |
Childhood onset of delinquent behaviour and severe family problems, including child maltreatment and neglect, are associated with a variety of adverse outcomes in young adulthood [1–6]. These childhood problems are important risk factors for later delinquent behaviour and hamper psychological functioning [1, 3, 4, 7–17]. So far, childhood risk factors of adulthood problems have been studied either within delinquent populations [1–3, 9, 13, 18–21] or in populations of young adults who experienced maltreatment and out-of-home placements in their childhood [3, 22]. These studies focused predominantly on the severity, age of onset and persistence of delinquent behaviour and on maltreatment and family interferences by, for example, the Child Protection Services (CPS; Dutch: Raad voor de Kinderbescherming). However, such childhood problems are closely interrelated and the presence of multiple problems in childhood drastically increases the probability of adverse adult outcomes [19, 23, 24]. Therefore, studies should focus on combinations of risk factors in young children [13, 25, 26], instead of focusing on single risk factors, and assess to what extent these combinations can predict outcomes later in life. In this way, it may be possible to distinguish among youth risk profiles which may help tailor primary, secondary and tertiary prevention strategies. The present study tackled these issues by retrospectively studying combined risk factors and long-term outcomes of both childhood judicial and civil CPS interferences in multi-problem young adults.
|
study
| 99.8 |
Young adulthood is considered a distinct developmental stage comprising major psychological [27–29], social and neurobiological changes that are critical for a healthy transition towards adulthood [31–33]. In most cases, young adults (aged 18–27) who experienced severe psychological, family and judicial problems since childhood encounter difficulties during this transition in becoming self-sufficient adults [32–35]. Previous studies have provided evidence that these vulnerable young adults are at high risk of an accumulation of several problems such as unemployment, psychological problems, early parenthood, and court involvement [34, 36–38]. Furthermore, a majority of these young adults suffer from substance use disorder [39, 40], and lack social support [33, 34]. This group with multiple and intertwined problems has been called multi-problem young adults, and is increasingly recognized as warranting specific scientific attention in order to inform and help improve professional support [33, 41]. An important aspect in this respect is to understand the development of the childhood problems that culminate in these multi-problem young adults.
|
review
| 99.7 |
In general, childhood problems as risk factors of later delinquent behaviour and mental health problems are widely studied. These risk factors are often distinguished on the individual and family level [2, 9, 12, 13]. Individual risk factors as intellectual disability, disruptive behaviour, psychological problems and an early onset of substance use are related to the development of antisocial behaviour [2, 42–44] later in life, and to mental health problems in adulthood as well . Other risk factors in this respect are low school achievement and truancy [46, 47]. Important risk factors on the family level are inadequate parenting, low social economic status, maltreatment and neglect, mental health problems and substance abuse of parents . All these factors may have contributed in their own unique way to the various problems of young adults.
|
review
| 99.9 |
Many multi-problem young adults have demonstrated delinquent behaviour and severe family problems during childhood [1, 22, 48–50] and, therefore, are likely to have underwent CPS interference during their youth. In The Netherlands, there are two main reasons for a child to receive a CPS investigation: to request a civil or a penal measure. It is not uncommon for children to receive multiple CPS interferences during their lives . Therefore, the characteristics of CPS interference differ among children [21, 51–53]. Multi-problem young adults are likely to have experienced several judicial, school and family problems simultaneously [19, 23, 24], for which the timing, the number and the intensity of CPS investigations may vary . CPS characteristics can be seen as static risk factors for deviant development since children who underwent CPS interference have an elevated risk of developing delinquent behaviour and mental health problems in young adulthood [1, 3, 8, 21, 48, 55, 56]. The annual arrest rate for young adults who as children had been referred to CPS is more than four times higher than the national rate for 18- to 24-year olds and 50% of this young adult population have experienced mental health problems .
|
study
| 85.7 |
Whereas all children who were exposed to severe family problems and/or who were involved in juvenile delinquency have an elevated risk of adult problem behaviour [1, 6, 15, 50, 58–61], the extent to which these problems persist and evolve into young adulthood differs substantially [7, 61, 62]. This might indicate heterogeneous profiles of the concurrent childhood problems. Several studies investigated and aimed to reduce the heterogeneity of problems within comparable populations of high-risk youths by exploring profiles [9, 13]. A study by Haapasalo found two groups of young adult offenders with CPS interventions: an early onset multiple intervention group and a late onset group who had fewer interventions . A study by Dembo et al. in high-risk youths reported two classes based on self-report data; one with a low prevalence and the other with a high prevalence of problems in family and peer relations, psychological functioning and education . Furthermore, Geluk et al. distinguished three profiles in childhood arrestees, differing in the extent of problems in peer relations, psychological functioning and authority conflicts. So, exploring profiles proved useful in ordering these childhood problems into several homogenous classes concerning the onset, the prevalence and the extent of the problems. However, these studies did not explore specifically if and how these childhood classes may contribute to a deviant development into (young) adulthood.
|
review
| 99.7 |
Although CPS does not provide treatment, CPS interference is directly related to extensive contact with judicial, mental health and social services [48, 63] and CPS may refer their clients to appropriate care, if necessary. However, many (young) adults with a childhood history of CPS interference still experience serious problems, even after repeated intervention [3, 48, 49, 64, 65]. As such, it seems that the effectiveness of current secondary prevention and intervention practices during childhood is limited in this population. Therefore, retrospectively identifying classes of interrelated static risk factors of CPS interference within a relatively unstudied population of multi-problem young adults may prove useful for more effective problem recognition and screening purposes in childhood [26, 54]. Finally, relating these childhood classes to delinquency and mental health problems in young adulthood may give useful indications for the prevention of the escalation of these problems to clinical practice [48, 49].
|
review
| 99.7 |
The present study aims to explore whether groups of CPS characteristics in childhood can be identified within a sample of multi-problem young adults. Furthermore, the associations between class membership and both self-reported delinquency and psychological functioning in young adulthood are investigated. Based on the literature, we expect multi-problem young adults to have a significant prevalence of CPS interference. Within this group we expect to find distinct latent classes differing in the onset, number and intensity of judicial and civil interferences and in the extent of family problems [7, 9]. Lastly, it is hypothesized that classes of CPS interference in youths relate differently to current psychological dysfunctioning and current severity of delinquent behaviour in multi-problem young adults [1, 65, 66].
|
study
| 100.0 |
In 2014–2016 a total of 596 multi-problem young adults were recruited in Rotterdam, The Netherlands. All participants were male, between 18 and 27 years old (mean age 21.7), and had sufficient knowledge of the Dutch language to understand the study procedure and the questionnaires. This study was part of a larger study in which participants were recruited from two sites. The first site was a municipal agency (Dutch: Jongerenloket) where young adults between the ages of 18 and 27 can apply for social welfare. Every year over 4000 intakes are carried out by so-called youth coaches . During this intake, the level of self-sufficiency of the young adult is assessed on eleven life domains with the validated Self-Sufficiency Matrix—Dutch version (SSM-D) [68–70], based on the American version of the SSM , on a five-point scale with scores ranging from 1 (acute problems) to 5 (completely self-sufficient). Participants were eligible when they adhered to the following definition: (a) a score of 1 or 2 on the domains Income and Daytime Activities, (b) a maximum score of 3 on at least one of the following domains: Addiction, Mental health, Social network, Justice and (c) a minimum score of 3 on the domain Physical health . Eligible young adults were asked to cooperate voluntarily. As a part of a larger study, N = 436 participants were recruited in this way . The second site was multimodal day treatment program New Opportunities (Dutch: De Nieuwe Kans; DNK). Multi-problem young adults also signed up to DNK themselves or were referred to DNK directly by youth care, probation services, mental health services or social organizations. Therefore, additional participants were recruited directly from DNK (N = 160). From the total study sample (N = 596), 99.3% (N = 592) gave informed consent to conduct the register and record research. Of the N = 592, 65.9% (N = 390) was matched to a record in the CPS system.
|
study
| 100.0 |
The study was performed by the VU University Medical Center Department of Child and Adolescent Psychiatry and approved by the Medical Ethics Review Committee of VU University Medical Center.1 Participants gave informed consent before voluntary participation after a member of the research team had provided oral information accompanied by written information. After informed consent, trained (junior) researchers administered questionnaires.
|
other
| 99.94 |
Interference with CPS was checked in the CPS system Kinderbescherming Bedrijfs Processen Systeem (KBPS) using first names, surname and date of birth of the participants. This resulted in a match of 65.9% (N = 390) of the total sample (N = 592); 34.1% (N = 202) did not match to a record in the system. For a part of the latter group it is uncertain whether they truly never had CPS contact or whether their record has been destroyed, since CPS is legally required to destroy records of clients that reach age 24. This applies to N = 98 of the N = 202 that did not match to a record in the system. For the other N = 104 (51.5% of N = 202), it was certain that they did not have CPS interference, since they were younger than 24 years old. The CPS files consist of all documents received and sent by the CPS concerning the child and a selection of judicial and police report data . Data were extracted from April 2015 to August 2016 by trained (junior) researchers. To test the inter-rater reliability, 19 randomly selected files were scored by two independent raters, showing a substantial inter-rater reliability (κ = 0.72) [74, 75].
|
study
| 100.0 |
The register and record research was conducted at CPS and the data were extracted between April 2015 and August 2016. CPS monitors children between 0 and 18 years old when there are serious concerns regarding their home situation and upbringing. In families with severe parenting problems a child welfare investigator can perform a civil protection investigation of the home environment of the child, at the request of CPS. At the request of the court, CPS mediates when parents break up and disagree about arrangements concerning their children. Moreover, CPS can initiate a judicial or truancy investigation for youth suspected of an offence or truancy. The investigation report with recommendations on (mandatory) service use or a suitable penalization is delivered to the court .
|
other
| 99.9 |
Socio-demographic characteristics were assessed with a structured self-report questionnaire. Ethnicity was based on the country of birth of the respondent and at least one of his parents. A respondent was classified as non-Dutch if he or one of his parents was not born in The Netherlands . Ethnicity was recoded into a dichotomous variable (Dutch ethnicity vs. other ethnicity). Educational level was classified into three levels: maximum primary education, achievement of junior secondary education and senior secondary education attainment. Family problems in youth were assessed with the single item ‘Did you suffer from problems that existed in the family you grew up with? (Yes/No)’. Police contact of family members in youth was assessed with the single item ‘Did family members you grew up with have police contact? (Yes/No)’. Prior service use was assessed with the single item ‘Did you previously use services? (Yes/No)’. Frequency of service use was assessed with the single item ‘Which services did you have contact with?’ (e.g., youth care, probation services, child protection services). This was recoded into a frequency score defined as the number of self-reported services.
|
study
| 100.0 |
Several variables were obtained from the CPS records. All variables were divided into categories to perform the latent class analysis (LCA), as it is a condition for this analysis to use categorical variables. The variables Age of first CPS report, Type of investigation, Number of investigations, Child maltreatment, Age of onset of delinquent behaviour and Family supervision order were used as indicators to execute the LCA. Age of first CPS report in which date of the first CPS investigation was recoded into four categories: no report, below age 13, 13 or 14 years old, age 15 up to 18. The CPS records provided information on three types of investigations: offence investigation, protection investigation and truancy investigation. Type of investigation was recoded into a variable that contained five categories: no investigation, protection investigation, offence investigation, truancy investigation, several types of investigations. Number of CPS investigations was recoded into three categories: no investigation, one or two investigations, at least three investigations. Child maltreatment was extracted from the record when a professional ascertained child maltreatment (Yes/No). Domestic violence was observed and registered by a professional (Yes/No). The verdict of the court to impose a family supervision order was included in the record (Yes/No). Out-of-home placement was also included in the record in the verdict of the court (Yes/No). Age of onset of delinquent behaviour: the date of the first offence was registered based on the police report. Using this date combined with the date of birth, the age of first offence was computed. This variable was recoded into four categories: no offence, first offence below age thirteen, first offence between 13 and 14 years of age, and first offence at age 15 or older.
|
study
| 100.0 |
The Dutch version of the Adult Self Report (ASR) was assessed orally and filled out by the researcher to obtain current psychological functioning. ASR part VIII consists of 123 items on internalizing and externalizing problems during the previous 6 months. The reliability of the questionnaire is good, with a Cronbach’s α of 0.83. In this study the ASR total problem score and the scores of nine subscales were used as outcome measures. The subscales are: anxious/depressed, withdrawn, somatic complaints (internalizing problems); intrusive, rule-breaking and aggressive behaviour (externalizing problems); thought problems, attention problems and substance use. The prevalence of serious dysfunctioning on all subscales is presented in Table 1. The mean scale scores per class as outcome measure are based on percentile scores (Table 5).Table 1Descriptive characteristics in percentages (N = 390) Socio-demographic characteristics Mean age21.7 years old Born in The Netherlands Yes76.6 Dutch ethnicity Yes12.6 Educational level Primary36.5 Junior secondary44.7 Senior secondary17.5 Other1.3Family characteristics Family problems in youth Yes63.2 Police contact of family members in youth Yes19.0Service use Service use Yes83.3 Frequency of service use None16.2 Once28.0 2 or 336.5 4 or more19.3Prevalence serious dysfunctioning (%)a Psychological functioning previous 6 months (ASR) Total problems29.8 Anxious/depressed30.8 Withdrawn51.2 Somatic complaints29.3 Intrusive7.7 Rule-breaking behaviour44.7 Aggressive behaviour28.0 Attention problems30.6 Thought problems34.2 Substance use53.0Delinquent behaviour from onset till young adulthood (SRD) Committed at least one offence Yes93.3 Destruction/public order offence Yes62.6 Property offence Yes85.9 Aggression/violent offence Yes73.1 Drug offence Yes59.2Delinquent behaviour previous 6 months (SRD) (N = 179)b Committed at least one offence Yes63.0 Destruction/public order offence Yes10.8 Property offence Yes27.1 Aggression/violent offence Yes21.6 Drug offence Yes21.0 aPrevalence of serious dysfunctioning is based on percentile scores in the borderline (between the 84th and 90th percentiles) and clinical range (above the 90th percentile) bSelf-reported delinquency in the previous 6 months has been added during the study and measured in 179 participants
|
study
| 100.0 |
The frequency and seriousness of delinquent behaviour were investigated orally and filled out by a researcher using the Dutch version of the Self-report Delinquency Scale (SRD) . This questionnaire has 29 items (including two items of violation: fare dodging and lighting fireworks when prohibited) and the internal consistency of the total score is excellent with Cronbach’s α = 0.85 [79, 81]. The questionnaire explored the frequency of offences committed both during the respondent’s lifetime and in the previous 6 months. In addition, the items were also divided into four different offence categories: destruction/public order offences (5 items, Cronbach’s α = 0.64), property offences (11 items, Cronbach’s α = 0.79), aggression/violent offences (8 items, Cronbach’s α = 0.7) and drug offences (3 items, Cronbach’s α = 0.72) . The frequencies per offence category were recoded into dichotomous variables (Yes/No), due to the skewed distribution of the data. Lifetime and previous 6 months’ prevalence are presented in Table 1. Mean scores based on the frequencies of offences in the previous 6 months were used as outcome measure (see Table 5). The 27 items (excluding two items of violation) add up to one total delinquency score reflecting the multiplication of the seriousness of the offences and their frequency. The seriousness is divided into minor and serious offences based on applicable legal penalties; minor offences have a maximum custodial sentence of 48 months (score 1) and serious offences have a minimum custodial sentence of 48 months (score 2) [79, 80].
|
study
| 99.94 |
In order to detect classes of childhood correlates Latent Class Analysis (LCA) was performed. LCA is a useful method for analysing the relationships among observed variables, when each observed variable is categorical, in a heterogeneous population assumed to be comprised of a set of latent classes . LCA was performed with the program Statistical Analysis System (SAS) version 9.3. The six CPS childhood indicators mentioned above were entered into the LCA. Analyses were conducted using PROC LCA 1.2.6 for SAS 9.3 . Good qualification quality was established taking into account the Bayesian information criterion (BIC), entropy and Akaike information criterion (AIC) . The entropy value ranges between 0 and 1; a value approaching 1 indicates a clear description of the classes . Subsequently, item response probability scores on all indicators were used to interpret the classes. Lastly, to explore differences among classes derived from the LCA on current psychological functioning and delinquent behaviour, One-Way Analyses of Variance and Post Hoc t-tests with Bonferroni correction were performed with Statistical Packages for the Social Sciences, version 22 for Windows .
|
study
| 100.0 |
Table 1 shows the self-reported socio-demographic and family characteristics, service use, current psychological functioning and delinquent behaviour of multi-problem young adults with CPS interference in youth. It shows that many young adults had problems in youth; 63.2% had problems in their family, 83.3% reported prior service use and 93.3% committed an offence. During the previous 6 months, 53.0% had serious substance use problems and 63.0% committed an offence.
|
study
| 99.94 |
Table 2 shows the descriptive results of the childhood CPS correlates in percentages. After referral to CPS, 84.9% of participants were investigated. In 21.0% of the participants the first CPS investigation was below the age of thirteen and 39.0% had their first investigation at age fifteen or older. Almost half of the group (43.9%) had one or two CPS investigations and 41.5% had at least three CPS investigations. Judicial investigations were conducted in 75.0% of the group and protection investigations in 40.0% of participants. Multiple types of investigations were conducted in 32.6% of participants of which 50.0% first had a protection investigation and 40.0% first had a judicial investigation. Truancy investigations rarely occurred separately (1.8%). Child maltreatment was registered in 29.5% of the CPS reports and the CPS records reported domestic violence in 16.4% of the cases. Protection measures taken by the juvenile court were investigated as well; 33.6% of participants underwent a family supervision order and 22.1% an out-of-home placement. In 88.5% of the CPS records childhood delinquency was registered and 23.3% committed their first offence below age 13.Table 2Frequencies of childhood correlates CPS records (N = 390)%Age of the first CPS report No report15.1 First report below age 1321.0 First report age 13 or 1424.9 First report age 15 or older39.0Number of CPS investigations None14.6 1 or 243.9 3 or more41.5Type of CPS investigation No investigation14.9 Protection investigation8.0 Judicial investigation42.7 Truancy investigation1.8 Multiple types of investigations32.6Registered child maltreatment Yes29.5Domestic violence Yes16.4Family supervision order Yes33.6Out-of-home placement Yes22.1Age at onset of delinquent behaviour No offence10.5 Below age 1323.3 Age 13 or 1433.6 Age 15 or older32.6
|
study
| 100.0 |
The first step conducted for the LCA involved identifying the number of latent classes that best fit the data on six childhood indicators. Table 3 presents the fit indices after carrying out several class models. Based on the entropy (0.95) and the BIC value (692.03), the four-class models fitted best. The five-class model, however, had the lowest value of the AIC (417.74). Models distinguishing six or more classes all performed worse on all indicators. Based on these findings and the interpretability of the resulting latent class model, we decided that the four-class model had the best fit for these data.Table 3Model fit sizes of latent class analysis of childhood correlates (N = 390)ModelEntropyAICBICDf21.001009.571124.5893030.93597.93772.4491540.95458.02692.0390050.91417.74711.24885 AIC Akaike information criteria, BIC Bayesian information criteria; Df degrees of freedom
|
study
| 100.0 |
In order to interpret the latent classes, item response probabilities of the indicators were examined for each latent class. Table 4 presents the item-response probabilities and the proportions of the classes.Table 4Item response probabilities LCA (N = 390)Class1 (N = 175)2 (N = 120)3 (N = 57)4 (N = 38)Class size proportions44.9%30.8%14.6%9.7%Family supervision order Yes0.02 0.84 0.02 0.70 No 0.98 0.16 0.98 0.30Registered child maltreatment Yes0.14 0.57 0.02 0.59 No 0.86 0.43 0.98 0.41Age at onset of delinquent behaviour No offence0.000.000.31 0.62 Below age 130.200.420.050.10 Age 13 or 140.410.370.180.11 Age 15 or older0.390.210.460.18Age of the first CPS report No report0.010.01 0.997 0.00 First report below age 130.040.440.00 0.60 First report age 13 or 140.290.340.000.15 First report age 15 or older 0.67 0.210.000.25Number of CPS investigations None0.000.00 0.997 0.00 1 or 2 0.68 0.130.00 0.94 3 or more0.32 0.87 0.000.06Type of CPS investigation No investigation0.000.00 0.997 0.03 Protection investigation0.000.000.00 0.85 Judicial investigation 0.89 0.040.000.12 Truancy investigation0.040.000.000.00 Multiple types of investigations0.07 0.95 0.000.00Current psychological functioning and delinquent behaviour per group
|
study
| 99.94 |
The first class, labelled as the late CPS/penal investigation group (44.9%) (Fig. 1), did not experience maltreatment or a family supervision order in childhood. They all committed at least one offence2 and their first offence was at age 13 or 14. Their first judicial CPS report was executed at age fifteen or older (late CPS interference) and they had a maximum of two, solely judicial, reports.Fig. 11-Late CPS/penal investigation group
|
other
| 97.9 |
A majority of the second class, labelled as the early CPS/multiple investigation group (30.8%) (Fig. 2), experienced maltreatment in childhood which often resulted in at least one family supervision order pronounced by the court. They had their first report at a young age, below age 13 (early CPS interference) and had three or more CPS investigations, due to various causes (judicial and/or family and/or truancy investigations), since they often committed their first offence below age thirteen.Fig. 22-Early CPS/multiple investigation group
|
other
| 99.8 |
The third class, labelled as the late CPS interference without investigation group (14.6%) (Fig. 3), did not experience any severe family problems such as maltreatment or family supervision orders. If they committed an offence, it was at age 15 or older (late CPS interference). CPS decided mostly not to investigate the child and they often did not have any reports in their record.Fig. 33-Late CPS interference without investigation group
|
study
| 99.25 |
The fourth class, labelled as the early CPS/family investigation group (9.7%) (Fig. 4), had early CPS interference below age thirteen (early CPS interference), due to severe family problems such as maltreatment which resulted mostly in at least one family supervision order. CPS decided to investigate their situations once or twice, which were specifically protection investigations. Participants in this group were not likely to commit any offence. Fig. 44-Early CPS/family investigation group
|
study
| 99.9 |
Table 5 presents results of the ANOVA and post hoc comparisons between LCA class membership on current psychological functioning. There was a significant difference among classes on anxious/depressive problems (p = 0.035), a borderline significant difference on intrusive problems (p = 0.056) and a significant difference on substance use (p = 0.029). The post hoc test showed that participants of the early CPS/family investigation group reported significantly more anxious/depressive problems than participants of the early CPS/multiple investigation group (p = 0.022). Moreover, the early CPS/family investigation group reported more substance abuse than the late CPS interference without investigation group (borderline significant; p = 0.056).Table 5Results of ANOVA comparisons among classes on current self-reported psychological functioning and delinquent behaviour (N = 390)Class1 (N = 175)2 (N = 120)3 (N = 57)4 (N = 38)FpM (SD)M (SD)M (SD)M (SD)Psychological functioninga Total psychological problems61.4 (26.0)61.5 (25.8)59.8 (28.2)71.1 (22.8)1.710.164 Anxious/depressed69.2 (18)66.3 (16)69.4 (18)75.8 (18)2.88b 0.035** Withdrawn79.0 (17.2)78.1 (16.8)73.2 (18.7)80.8 (16.6)1.970.118 Somatic complaints68.1 (16.4)67.8 (16.1)69.2 (17.4)72.6 (16.7)0.900.439 Intrusive55.7 (1)59.3 (1)55.7 (1)57.8 (2)2.55c 0.056* Rule-breaking behaviour78.6 (16.8)79.9 (16.8)78.4 (16.2)82.6 (17.6)0.710.549 Aggressive behaviour67.7 (16.1)67.2 (15.5)68.5 (17)74.2 (16.9)1.970.118 Attention problems73.4 (14.3)74 (14.5)72.3 (14.5)77.7 (14.7)1.180.317 Thought problems74.3 (17.5)73.2 (16.3)72.1 (17.3)79.2 (16.5)1.520.208 Substance used 78.0 (18)81 (19)73.9 (19)83.9 (18)3.04e 0.029**Class 1 (N = 74) 2 (N = 59) 3 (N = 25) 4 (N = 21)FpM (SD)M (SD)M (SD)M (SD)Delinquency previous 6 months Total delinquency3.5 (8.1)7.1 (11.5)6.0 (13.2)2.2 (5.3)2.10.101 Destruction/public order offence0.09 (0.3)0.14 (0.4)0.00 (0)0.19 (0.4)1.890.133 Property offence0.22 (0.4)0.37 (0.5)0.27 (0.5)0.18 (0.4)1.720.165 Aggression/violent offence0.20 (0.4)0.27 (0.4)0.15 (0.4)0.23 (0.4)0.610.609 Drug offence0.57 (0.5)0.65 (0.5)0.54 (0.5)0.61 (0.5)1.350.261 aNormal functioning (score < 84), borderline range (score 84-90), clinical range (above 90) bSignificant difference between early CPS/family investigation group and early CPS/multiple investigation group cSignificant difference between early CPS/multiple investigation group and late CPS/penal investigation group dClass 1; N = 174 eSignificant difference between early CPS/family investigation group and late CPS interference without investigation group* p < 0.10, ** p < 0.05, *** p < 0.01
|
study
| 100.0 |
The purpose of this study was twofold. The first aim was to retrospectively identify distinct classes in multi-problem young adults based on childhood CPS characteristics. This resulted in four latent classes: a late CPS/penal investigation group (44.9%), an early CPS/multiple investigation group (30.8%), a late CPS interference without investigation group (14.6%) and an early CPS/family investigation group (9.7%). The second aim was to explore whether these classes differed on current young adult psychological functioning and delinquent behaviour. The early CPS/family investigation group reported significantly more problematic anxiousness/depression problems than the other groups. Substance use differed significantly among groups, although post hoc tests only revealed borderline significant differences. No differences in current delinquent behaviour were reported among the classes.
|
study
| 100.0 |
In our sample of multi-problem young adults, 65.9% had one or more CPS interference(s) during their childhood versus 1% of the total population of Dutch children in 2016 . Furthermore, 29.5% in the current sample underwent maltreatment versus 3% of Dutch youth that was in danger of any type of maltreatment in 2010 . Thus, the prevalence of CPS interferences and severe family problems is, as expected, clearly higher in this population of multi-problem young adults than in the general population. One should note, however, that these percentages are not completely comparable, since the prevalence in the current study was not limited to 1 year. The high prevalence of CPS interference in multi-problem young adults matches their self-reported problems in childhood quite adequately: 83.3% reported service use in their youth and 63.2% reported family problems. As expected, multi-problem young adults also experience heterogeneous problems in their current functioning. This extends findings in other studies [88–90] that argue that different forms of problem behaviour (such as mental health problems, delinquency and substance use) with an onset in childhood are interrelated and may be seen as symptoms of a general disposition toward deviant behaviour through life, by some referred to as problem behaviour syndrome (PBS) . How PBS is expressed may vary over time and across contexts. For children with PBS, the transition to adulthood typically occurs in the context of severe family problems and interference by multiple justice/care/and child welfare systems [41, 66]. Therefore, they may experience a differential pathway into adulthood in which more tailor-made specialized care is needed to support their adopting adult responsibilities such as independent living . This way, they may be prevented from growing into multi-problem young adults. Our first findings underline the importance of gaining more insight into the childhood onset of the problem heterogeneity of multi-problem young adults in order to enhance effective tailor-made intervention.
|
study
| 99.94 |
The present study confirmed several distinct classes of risk factors for adult problem behaviour in addition to earlier studies [3, 9, 13]. Dembo et al. 9 and Geluk et al. 13 identified two and three classes, respectively, differing in the extent of problem behaviour; Haapasalo reported two classes differing in age of onset and number of CPS interventions. A first distinction in the identified classes in the current study indeed occurred between early (below age 13) and late (from age 15) CPS involvement. The early CPS/multiple investigation group had the earliest onset of delinquent behaviour (below age 13). Several studies show that early onset delinquents are more at risk for problems in young adulthood, such as mental health problems, substance abuse, drug related and violent delinquent behaviour, than later onset delinquents [20, 61]. Furthermore, the early CPS/multiple investigation group underwent the most CPS investigations and is, therefore, also comparable to the early onset group in the Haapasalo study , in which the offenders demonstrated more problems during their youth and were in greater need of CPS interventions such as placement in foster care.
|
study
| 99.94 |
Regarding the long term outcomes of childhood CPS interference specifically, the early CPS/family investigation group reported the most anxious/depression problems and the most substance abuse in young adulthood. Maltreatment, family supervision and other severe family problems in childhood have repeatedly been shown to be robust risk factors for mental health problems in (young) adulthood [7, 16]. For example, according to Thornberry et al. , childhood maltreatment is indeed strongly related to later substance abuse and internalizing problems. Although the early CPS/family investigation was the smallest identified group (9.7%), they seem to have followed the most adverse developmental pathway into young adulthood. It is possible that CPS failed to provide appropriate interventions for this group, since the CPS involvement was not as intensive as for the early onset/multiple investigation group. Moreover, the early CPS/family group was the only group that did not engage in delinquent behaviour in childhood/adolescence. This may have caused them to stay unnoticed for a longer period of time. However, traumatic events in the child’s family environment may have already occurred long before the first CPS interference and are associated with an increased likelihood of adverse adult outcomes [7, 16]. Besides a broader focus on the problems of the child itself, children with solely civil CPS interference may benefit from more attention to treatment of the problems of the parents. Interventions could be aimed at strengthening their parenting capabilities and resources. Adopting such a ‘two-generation approach’ has shown promising results in preventing family and childhood problems from growing worse .
|
review
| 99.75 |
No significant differences among classes in current delinquent behaviour were found among groups. The late CPS/penal group was the largest group in our sample (44.9%); their first CPS investigation was at age 15 or older and the age of onset of their delinquent behaviour varied between ages 13 and 15. All multi-problem young adults showed a strong tendency for persisting in and/or developing criminal behaviour into adulthood, notwithstanding their distinct childhood histories. Moreover, since the group without CPS investigations also reported delinquent behaviour in adulthood, all forms of CPS interference (even marginal contact) should be considered risk factors for later antisocial behaviour. In addition, the late CPS/penal children proved to be a group without severe family problems, at least according to the CPS data. Steinberg noted that adolescent onset offenders often manifest less severe patterns of family pathology and mental health problems than life course persistent offenders . In our sample, both late onset CPS groups indeed reported fewer mental health problems in young adulthood than the early onset groups. A follow-up study should be conducted to explore whether these differences in problem behaviour among groups still persist into (middle) adulthood. Finally, since all groups persisted in their delinquent behaviour, children with CPS interference should be targeted as a high-risk population in need of specialized interventions aimed at reducing the criminogenic risk factors associated with recidivism.
|
study
| 100.0 |
Like any other study, this study has some limitations. First, the CPS record investigation in the current study was not performed using a validated instrument, because an applicable instrument was not available. However, CPS investigations are standardized and in order to optimize and monitor the quality of the data, inter-rater reliability was analysed and found to be substantial. Second, registered offence data, and in particular data on the first offence, is likely to be under reported, as a minority of juvenile delinquents is actually convicted . Still, in this sample officially recorded and self-reported delinquency data are, while not exactly similar, quite comparable, both showing a high prevalence of delinquent behaviour. Third, in this study, self-report questionnaires were also used to investigate socio-demographic characteristics and psychological functioning. To achieve good reliability, a validated self-report psychological functioning questionnaire is used and anonymity and privacy of participants was emphasized before and during the assessment of questionnaires. Fourth, a majority of 87.4% of participants in this study have a non-Dutch ethnicity. In our case, non-Dutch ethnicity refers to an amalgam of cultural backgrounds, for example Surinamese, Antillean, Moroccan and Turkish. However, due to small sample sizes per ethnic subgroup, it was not possible to perform separate analyses. Fifth, generalizability of study results to an international context is not straightforward, because of different service system organizations. In Great-Britain and the United States of America, for example, child protection service and the judicial youth system are more separate systems than in The Netherlands [93, 94]. Scandinavian countries have more comparable systems to the Dutch system, although those systems are more based on prevention. For instance, in Sweden voluntary and involuntary services are not divided as in The Netherlands . And lastly, LCA is an exploratory data-driven method and the findings per class represent probabilities on latent indicators.
|
study
| 100.0 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.