Machine learning and the prediction of suicide in psychiatric populations: a systematic review.

Machine learning (ML) has emerged as a promising tool to enhance suicidal prediction. However, as many large-sample studies mixed psychiatric and non-psychiatric populations, a formal psychiatric diagnosis emerged as a strong predictor of suicidal risk, overshadowing more subtle risk factors specific to distinct populations. To overcome this limitation, we conducted a systematic review of ML studies evaluating suicidal behaviors exclusively in psychiatric clinical populations. A systematic literature search was performed from inception through November 17, 2022 on PubMed, EMBASE, and Scopus following the PRISMA guidelines. Original research using ML techniques to assess the risk of suicide or predict suicide attempts in the psychiatric population were included. An assessment for bias risk was performed using the transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD) guidelines. About 1032 studies were retrieved, and 81 satisfied the inclusion criteria and were included for qualitative synthesis. Clinical and demographic features were the most frequently employed and random forest, support vector machine, and convolutional neural network performed better in terms of accuracy than other algorithms when directly compared. Despite heterogeneity in procedures, most studies reported an accuracy of 70% or greater based on features such as previous attempts, severity of the disorder, and pharmacological treatments. Although the evidence reported is promising, ML algorithms for suicidal prediction still present limitations, including the lack of neurobiological and imaging data and the lack of external validation samples. Overcoming these issues may lead to the development of models to adopt in clinical practice. Further research is warranted to boost a field that holds the potential to critically impact suicide mortality.

Dissecting abstract, modality-specific and experience-dependent coding of affect in the human brain.

Emotion and perception are tightly intertwined, as affective experiences often arise from the appraisal of sensory information. Nonetheless, whether the brain encodes emotional instances using a sensory-specific code or in a more abstract manner is unclear. Here, we answer this question by measuring the association between emotion ratings collected during a unisensory or multisensory presentation of a full-length movie and brain activity recorded in typically developed, congenitally blind and congenitally deaf participants. Emotional instances are encoded in a vast network encompassing sensory, prefrontal, and temporal cortices. Within this network, the ventromedial prefrontal cortex stores a categorical representation of emotion independent of modality and previous sensory experience, and the posterior superior temporal cortex maps the valence dimension using an abstract code. Sensory experience more than modality affects how the brain organizes emotional information outside supramodal regions, suggesting the existence of a scaffold for the representation of emotional states where sensory inputs during development shape its functioning.

Verbal lie detection using Large Language Models.

Human accuracy in detecting deception with intuitive judgments has been proven to not go above the chance level. Therefore, several automatized verbal lie detection techniques employing Machine Learning and Transformer models have been developed to reach higher levels of accuracy. This study is the first to explore the performance of a Large Language Model, FLAN-T5 (small and base sizes), in a lie-detection classification task in three English-language datasets encompassing personal opinions, autobiographical memories, and future intentions. After performing stylometric analysis to describe linguistic differences in the three datasets, we tested the small- and base-sized FLAN-T5 in three Scenarios using 10-fold cross-validation: one with train and test set coming from the same single dataset, one with train set coming from two datasets and the test set coming from the third remaining dataset, one with train and test set coming from all the three datasets. We reached state-of-the-art results in Scenarios 1 and 3, outperforming previous benchmarks. The results revealed also that model performance depended on model size, with larger models exhibiting higher performance. Furthermore, stylometric analysis was performed to carry out explainability analysis, finding that linguistic features associated with the Cognitive Load framework may influence the model's predictions.

How Male and Female Literary Authors Write About Affect Across Cultures and Over Historical Periods.

A wealth of literature suggests the existence of sex differences in how emotions are experienced, recognized, expressed, and regulated. However, to what extent these differences result from the put in place of stereotypes and social rules is still a matter of debate. Literature is an essential cultural institution, a transposition of the social life of people but also of their intimate affective experiences, which can serve to address questions of psychological relevance. Here, we created a large corpus of literary fiction enriched by authors' metadata to measure the extent to which culture influences how men and women write about emotion. Our results show that even though before the twenty-first century and across 116 countries women more than men have written about affect, starting from 2000, this difference has diminished substantially. Also, in the past, women's narratives were more positively laden and less arousing. While the difference in arousal is ubiquitous and still present nowadays, sex differences in valence vary as a function of culture and have dissolved in recent years. Altogether, these findings suggest that historic evolution is associated with men and women writing similarly about emotions and reveal a sizable impact of culture on the affective characteristics of the lexicon. Supplementary Information: The online version contains supplementary material available at 10.1007/s42761-023-00219-9.

Distinguishing Fine Structure and Summary Representation of Sound Textures from Neural Activity.

The auditory system relies on both local and summary representations; acoustic local features exceeding system constraints are compacted into a set of summary statistics. Such compression is pivotal for sound-object recognition. Here, we assessed whether computations subtending local and statistical representations of sounds could be distinguished at the neural level. A computational auditory model was employed to extract auditory statistics from natural sound textures (i.e., fire, rain) and to generate synthetic exemplars where local and statistical properties were controlled. Twenty-four human participants were passively exposed to auditory streams while the electroencephalography (EEG) was recorded. Each stream could consist of short, medium, or long sounds to vary the amount of acoustic information. Short and long sounds were expected to engage local or summary statistics representations, respectively. Data revealed a clear dissociation. Compared with summary-based ones, auditory-evoked responses based on local information were selectively greater in magnitude in short sounds. Opposite patterns emerged for longer sounds. Neural oscillations revealed that local features and summary statistics rely on neural activity occurring at different temporal scales, faster (beta) or slower (theta-alpha). These dissociations emerged automatically without explicit engagement in a discrimination task. Overall, this study demonstrates that the auditory system developed distinct coding mechanisms to discriminate changes in the acoustic environment based on fine structure and summary representations.

Reconstructing individual responses to direct questions: a new method for reconstructing malingered responses.

Introduction: The false consensus effect consists of an overestimation of how common a subject opinion is among other people. This research demonstrates that individual endorsement of questions may be predicted by estimating peers' responses to the same question. Moreover, we aim to demonstrate how this prediction can be used to reconstruct the individual's response to a single item as well as the overall response to all of the items, making the technique suitable and effective for malingering detection. Method: We have validated the procedure of reconstructing individual responses from peers' estimation in two separate studies, one addressing anxiety-related questions and the other to the Dark Triad. The questionnaires, adapted to our scopes, were submitted to the groups of participants for a total of 187 subjects across both studies. Machine learning models were used to estimate the results. Results: According to the results, individual responses to a single question requiring a "yes" or "no" response are predicted with 70-80% accuracy. The overall participant-predicted score on all questions (total test score) is predicted with a correlation of 0.7-0.77 with actual results. Discussion: The application of the false consensus effect format is a promising procedure for reconstructing truthful responses in forensic settings when the respondent is highly likely to alter his true (genuine) response and true responses to the tests are missing.

Information load dynamically modulates functional brain connectivity during narrative listening.

Narratives are paradigmatic examples of natural language, where nouns represent a proxy of information. Functional magnetic resonance imaging (fMRI) studies revealed the recruitment of temporal cortices during noun processing and the existence of a noun-specific network at rest. Yet, it is unclear whether, in narratives, changes in noun density influence the brain functional connectivity, so that the coupling between regions correlates with information load. We acquired fMRI activity in healthy individuals listening to a narrative with noun density changing over time and measured whole-network and node-specific degree and betweenness centrality. Network measures were correlated with information magnitude with a time-varying approach. Noun density correlated positively with the across-regions average number of connections and negatively with the average betweenness centrality, suggesting the pruning of peripheral connections as information decreased. Locally, the degree of the bilateral anterior superior temporal sulcus (aSTS) was positively associated with nouns. Importantly, aSTS connectivity cannot be explained by changes in other parts of speech (e.g., verbs) or syllable density. Our results indicate that the brain recalibrates its global connectivity as a function of the information conveyed by nouns in natural language. Also, using naturalistic stimulation and network metrics, we corroborate the role of aSTS in noun processing.

A modality-independent proto-organization of human multisensory areas.

The processing of multisensory information is based upon the capacity of brain regions, such as the superior temporal cortex, to combine information across modalities. However, it is still unclear whether the representation of coherent auditory and visual events requires any prior audiovisual experience to develop and function. Here we measured brain synchronization during the presentation of an audiovisual, audio-only or video-only version of the same narrative in distinct groups of sensory-deprived (congenitally blind and deaf) and typically developed individuals. Intersubject correlation analysis revealed that the superior temporal cortex was synchronized across auditory and visual conditions, even in sensory-deprived individuals who lack any audiovisual experience. This synchronization was primarily mediated by low-level perceptual features, and relied on a similar modality-independent topographical organization of slow temporal dynamics. The human superior temporal cortex is naturally endowed with a functional scaffolding to yield a common representation across multisensory events.

Acquired Pedophilia: international Delphi-method-based consensus guidelines.

Idiopathic and acquired pedophilia are two different disorders with two different etiologies. However, the differential diagnosis is still very difficult, as the behavioral indicators used to discriminate the two forms of pedophilia are underexplored, and clinicians are still devoid of clear guidelines describing the clinical and neuroscientific investigations suggested to help them with this difficult task. Furthermore, the consequences of misdiagnosis are not known, and a consensus regarding the legal consequences for the two kinds of offenders is still lacking. The present study used the Delphi method to reach a global consensus on the following six topics: behavioral indicators/red flags helpful for differential diagnosis; neurological conditions potentially leading to acquired pedophilia; neuroscientific investigations important for a correct understanding of the case; consequences of misdiagnosis; legal consequences; and issues and future perspectives. An international and multidisciplinary board of scientists and clinicians took part in the consensus statements as Delphi members. The Delphi panel comprised 52 raters with interdisciplinary competencies, including neurologists, psychiatrists, neuropsychologists, forensic psychologists, expert in ethics, etc. The final recommendations consisted of 63 statements covering the six different topics. The current study is the first expert consensus on a delicate topic such as pedophilia. Important exploitable consensual recommendations that can ultimately be of immediate use by clinicians to help with differential diagnosis and plan and guide therapeutic interventions are described, as well as future perspectives for researchers.

ANKK1 and TH gene variants in combination with paternal maltreatment increase susceptibility to both cognitive and attentive impulsivity.

Recent scientific findings suggest that dopamine exerts a central role on impulsivity, as well as that aversive life experiences may promote the high levels of impulsivity that often underlie violent behavior. To deepen our understanding of the complex gene by environment interplay on impulsive behavior, we genotyped six dopaminergic allelic variants (ANKK1-rs1800497, TH-rs6356, DRD4-rs1800955, DRD4-exonIII-VNTR, SLC6A3-VNTR and COMT-rs4680) in 655 US White male inmates convicted for violent crimes, whose impulsivity was assessed by BIS-11 (Barratt Impulsiveness Scale). Furthermore, in a subsample of 216 inmates from the whole group, we also explored the potential interplay between the genotyped dopaminergic variants and parental maltreatment measured by MOPS (Measure of Parental Style) in promoting impulsivity. We found a significant interaction among paternal MOPS scores, ANKK1-rs1800497-T allele and TH-rs6356-A allele, which increased the variance of BIS-11 cognitive/attentive scores explained by paternal maltreatment from 1.8 up to 20.5%. No direct association between any of the individual genetic variants and impulsivity was observed. Our data suggest that paternal maltreatment increases the risk of attentive/cognitive impulsivity and that this risk is higher in carriers of specific dopaminergic alleles that potentiate the dopaminergic neurotransmission. These findings add further evidence to the mutual role that genetics and early environmental factors exert in modulating human behavior and highlight the importance of childhood care interventions.

HTR1B genotype and psychopathy: Main effect and interaction with paternal maltreatment.

Psychopathy is a condition characterized by atypical emotions and socially maladaptive behavioral patterns. Among incarcerated people, psychopathy has been associated with higher rates of crimes, recidivism, and resistance to treatment. Many studies have indicated significant heritability of psychopathic traits, but little is known about the specific contribution of genes and their interaction with adverse experiences in life. Considering the primary role that serotonin plays in cognition and emotion, we investigated TPH2-rs4570625, 5-HTTLPR, MAOA-uVNTR, HTR1B-rs13212041 and HTR2A-rs6314 as risk factors for psychopathy in the largest sample of institutionalized individuals studied so far, consisting of 793 US White male incarcerated adults, and in a replication sample of 168 US White male incarcerated adolescents. In a subgroup of the adult sample, the interaction between genetics and parenting style, assessed by the Measure of Parental Style (MOPS) questionnaire, was also evaluated. The HTR1B-rs13212041-T/T genotype, as compared to HTR1B-rs13212041-C allele, predicted higher psychopathy scores in both the adult and the adolescent samples. The interaction between HTR1B-rs13212041-T/T genotype and paternal MOPS scores, investigated in a subgroup of the adult sample, was an even stronger predictor of higher levels of psychopathy than either the genetics or the environment taken individually. Overall, these data, obtained in two independent samples, shed new light on neurobiological correlates of psychopathy with promising implications both at a clinical and forensic level.

COVID-19 and Biomedical Experts: When Epistemic Authority is (Probably) Not Enough.

This critical essay evaluates the potential integration of distinct kinds of expertise in policymaking, especially during situations of critical emergencies, such as the COVID-19 pandemic. This article relies on two case studies: (i) herd immunity (UK) and (ii) restricted access to ventilators for disabled people (USA). These case studies are discussed as examples of experts' recommendations that have not been widely accepted, though they were made within the boundaries of expert epistemic authority. While the fundamental contribution of biomedical experts in devising public health policies during the COVID-19 pandemic is fully recognized, this paper intends to discuss potential issues and limitations that may arise when adopting a strict expert-based approach. By drawing attention to the interests of minorities (disenfranchized and underrepresented groups), the paper also claims a broader notion of "relevant expertise." This critical essay thus calls for the necessity of wider inclusiveness and representativeness in the process underlying public health policymaking.

Default and control network connectivity dynamics track the stream of affect at multiple timescales.

In everyday life, the stream of affect results from the interaction between past experiences, expectations and the unfolding of events. How the brain represents the relationship between time and affect has been hardly explored, as it requires modeling the complexity of everyday life in the laboratory setting. Movies condense into hours a multitude of emotional responses, synchronized across subjects and characterized by temporal dynamics alike real-world experiences. Here, we use time-varying intersubject brain synchronization and real-time behavioral reports to test whether connectivity dynamics track changes in affect during movie watching. The results show that polarity and intensity of experiences relate to the connectivity of the default mode and control networks and converge in the right temporoparietal cortex. We validate these results in two experiments including four independent samples, two movies and alternative analysis workflows. Finally, we reveal chronotopic connectivity maps within the temporoparietal and prefrontal cortex, where adjacent areas preferentially encode affect at specific timescales.

Interactions between auditory statistics processing and visual experience emerge only in late development.

The auditory system relies on local and global representations to discriminate sounds. This study investigated whether vision influences the development and functioning of these fundamental sound computations. We employed a computational approach to control statistical properties embedded in sounds and tested samples of sighted controls (SC) and congenitally (CB) and late-onset (LB) blind individuals in two experiments. In experiment 1, performance relied on local features analysis; in experiment 2, performance benefited from computing global representations. In both experiments, SC and CB performance remarkably overlapped. Conversely, LB performed systematically worse than the other groups when relying on local features, with no alterations on global representations. Results suggest that auditory computations tested here develop independently from vision. The efficiency of local auditory processing can be hampered in case sight becomes unavailable later in life, supporting the existence of an audiovisual interplay for the processing of auditory details, which emerges only in late development.

Emotion Regulation Failures Are Preceded by Local Increases in Sleep-like Activity.

Emotion self-regulation relies both on cognitive and behavioral strategies implemented to modulate the subjective experience and/or the behavioral expression of a given emotion. Although it is known that a network encompassing fronto-cingulate and parietal brain areas is engaged during successful emotion regulation, the functional mechanisms underlying failures in emotion suppression (ES) are still unclear. In order to investigate this issue, we analyzed video and high-density EEG recordings of 20 healthy adult participants during an ES and a free expression task performed on two consecutive days. Changes in facial expression during ES, but not free expression, were preceded by local increases in sleep-like activity (1-4 Hz) in brain areas responsible for emotional suppression, including bilateral anterior insula and anterior cingulate cortex, and in right middle/inferior frontal gyrus (p < .05, corrected). Moreover, shorter sleep duration the night before the ES experiment correlated with the number of behavioral errors (p = .03) and tended to be associated with higher frontal sleep-like activity during ES failures (p = .09). These results indicate that local sleep-like activity may represent the cause of ES failures in humans and may offer a functional explanation for previous observations linking lack of sleep, changes in frontal activity, and emotional dysregulation.

Complementing canonical fMRI with functional Quantitative Susceptibility Mapping (fQSM) in modern neuroimaging research.

Functional Quantitative Susceptibility Mapping (fQSM) allows for the quantitative measurement of time-varying magnetic susceptibility across cortical and subcortical brain structures with a potentially higher spatial specificity than conventional fMRI. While the usefulness of fQSM with General Linear Model and "On/Off" paradigms has been assessed, little is known about the potential applications and limitations of this technique in more sophisticated experimental paradigms and analyses, such as those currently used in modern neuroimaging. To thoroughly characterize fQSM activations, here we used 7T MRI, tonotopic mapping, as well as univariate (i.e., GLM and population Receptive Field) and multivariate (Representational Similarity Analysis; RSA) analyses. Although fQSM detected less tone-responsive voxels than fMRI, they were more consistently localized in gray matter. Also, the majority of active gray matter voxels exhibited negative fQSM response, signaling the expected oxyhemoglobin increase, whereas positive fQSM activations were mainly in white matter. Though fMRI- and fQSM-based tonotopic maps were overall comparable, the representation of frequency tunings in tone-sensitive regions was significantly more balanced for fQSM. Lastly, RSA revealed that frequency information from the auditory cortex could be successfully retrieved by using either methods. Overall, fQSM produces complementary results to conventional fMRI, as it captures small-scale variations in the activation pattern which inform multivariate measures. Although positive fQSM responses deserve further investigation, they do not impair the interpretation of contrasts of interest. The quantitative nature of fQSM, its spatial specificity and the possibility to simultaneously acquire canonical fMRI support the use of this technique for longitudinal and multicentric studies and pre-surgical mapping.

Oscillatory signatures of Repetition Suppression and Novelty Detection reveal altered induced visual responses in early deafness.

The ability to differentiate between repeated and novel events represents a fundamental property of the visual system. Neural responses are typically reduced upon stimulus repetition, a phenomenon called Repetition Suppression (RS). On the contrary, following a novel visual stimulus, the neural response is generally enhanced, a phenomenon referred to as Novelty Detection (ND). Here, we aimed to investigate the impact of early deafness on the oscillatory signatures of RS and ND brain responses. To this aim, electrophysiological data were acquired in early deaf and hearing control individuals during processing of repeated and novel visual events unattended by participants. By studying evoked and induced oscillatory brain activities, as well as inter-trial phase coherence, we linked response modulations to feedback and/or feedforward processes. Results revealed selective experience-dependent changes on both RS and ND mechanisms. Compared to hearing controls, early deaf individuals displayed: (i) greater attenuation of the response following stimulus repetition, selectively in the induced theta-band (4-7 Hz); (ii) reduced desynchronization following the onset of novel visual stimuli, in the induced alpha and beta bands (8-12 and 13-25 Hz); (iii) comparable modulation of evoked responses and inter-trial phase coherence. The selectivity of the effects in the induced responses parallels findings observed in the auditory cortex of deaf animal models following intracochlear electric stimulation. The present results support the idea that early deafness alters induced oscillatory activity and the functional tuning of basic visual processing.

Translational Application of a Neuro-Scientific Multi-Modal Approach Into Forensic Psychiatric Evaluation: Why and How?

A prominent body of literature indicates that insanity evaluations, which are intended to provide influential expert reports for judges to reach a decision "beyond any reasonable doubt," suffer from a low inter-rater reliability. This paper reviews the limitations of the classical approach to insanity evaluation and the criticisms to the introduction of neuro-scientific approach in court. Here, we explain why in our opinion these criticisms, that seriously hamper the translational implementation of neuroscience into the forensic setting, do not survive scientific scrutiny. Moreover, we discuss how the neuro-scientific multimodal approach may improve the inter-rater reliability in insanity evaluation. Critically, neuroscience does not aim to introduce a brain-based concept of insanity. Indeed, criteria for responsibility and insanity are and should remain clinical. Rather, following the falsificationist approach and the convergence of evidence principle, the neuro-scientific multimodal approach is being proposed as a way to improve reliability of insanity evaluation and to mitigate the influence of cognitive biases on the formulation of insanity opinions, with the final aim to reduce errors and controversies.

Idiopathic and acquired pedophilia as two distinct disorders: an insight from neuroimaging.

Pedophilia is a disorder of public concern because of its association with child sexual offense and recidivism. Previous neuroimaging studies of potential brain abnormalities underlying pedophilic behavior, either in idiopathic or acquired (i.e., emerging following brain damages) pedophilia, led to inconsistent results. This study sought to explore the neural underpinnings of pedophilic behavior and to determine the extent to which brain alterations may be related to distinct psychopathological features in pedophilia. To this aim, we run a coordinate based meta-analysis on previously published papers reporting whole brain analysis and a lesion network analysis, using brain lesions as seeds in a resting state connectivity analysis. The behavioral profiling approach was applied to link identified regions with the corresponding psychological processes. While no consistent neuroanatomical alterations were identified in idiopathic pedophilia, the current results support that all the lesions causing acquired pedophilia are localized within a shared resting state network that included posterior midlines structures, right inferior temporal gyrus and bilateral orbitofrontal cortex. These regions are associated with action inhibition and social cognition, abilities that are consistently and severely impaired in acquired pedophiles. This study suggests that idiopathic and acquired pedophilia may be two distinct disorders, in line with their distinctive clinical features, including age of onset, reversibility and modus operandi. Understanding the neurobiological underpinnings of pedophilic behavior may contribute to a more comprehensive characterization of these individuals on a clinical ground, a pivotal step forward for the development of more efficient therapeutic rehabilitation strategies.

The detection of faked identity using unexpected questions and choice reaction times.

The identification of faked identities, especially within the Internet environment, still remains a challenging issue both for companies and researchers. Recently, however, latency-based lie detection techniques have been developed to evaluate whether the respondent is the real owner of a certain identity. Among the paradigms applied to this purpose, the technique of asking unexpected questions has proved to be useful to differentiate liars from truth-tellers. The aim of the present study was to assess whether a choice reaction times (RT) paradigm, combined with the unexpected question technique, could efficiently detect identity liars. Results demonstrate that the most informative feature in distinguishing liars from truth-tellers is the Inverse Efficiency Score (IES, an index that combines speed and accuracy) to unexpected questions. Moreover, to focus on the predictive power of the technique, machine-learning models were trained and tested, obtaining an out-of-sample classification accuracy of 90%. Overall, these findings indicate that it is possible to detect liars declaring faked identities by asking unexpected questions and measuring RTs and errors, with an accuracy comparable to that of well-established latency-based techniques, such as mouse and keystroke dynamics recording.