Search

Autism spectrum disorder (ASD) is widely heterogeneous with respect to its etiology, behavioral and neural phenotypes. This large heterogeneity represents a barrier to accurate diagnosis and treatment allocation for many autistic individuals. The diagnosis is given based on two core symptoms: deficits in social cognition and restrictive and repetitive patterns of behavior. The latter manifesting in highly circumscribed interests (Baraskewich et al., 2021). While some previous research has treated social cognitive deficits and rigidity as separable dimensions of autism (Happé & Ronald, 2008), a high level of rigidity could impact social cognition, specifically learning about the preferences of people with more variable interests. This would ultimately result in less accurate representations of social knowledge, such as peer preferences, which have been shown to guide learning about others (Rosenblau et al., 2021). Leveraging computational modeling, this study aims to investigate the relationship between autistic individuals’ restricted self-preferences/interests and their social learning strategies (i.e., learning about others’ preferences). In this study we focus on early adolescence, a period of rapid social development, which increases the sophistication of social decision-making abilities (Mills et al., 2014).The main aim of the study is to stratify early autistic adolescents (age range 8-12 years) into subgroups based on their self-preference and social learning profiles. To this end, we will compare subgroups of autistic adolescents to non-autistic peers.

Personality traits describe how a person behaves across situations. Previous studies have shown that patients with borderline personality disorder (BPD) have difficulties in recognizing positive social signals and show hypersensitivity to negative social signals (Gunderson et al., 2018). Crucially, patients with BPD evaluate others and themselves more negatively and more inconsistently in comparison to healthy controls (Korn et al., 2016). It is unclear how this bias affects learning about others’ personality traits. Here we use a learning task (which builds on our previous studies: (Frolichs et al., 2021; Korn et al., 2012, 2016; Rosenblau et al., 2018, 2021) to investigate potential group differences between patients with BPD and healthy controls. Specifically, we investigate how women with BPD and those from a healthy control group learn about the personality traits of BPD patients and healthy controls (HC). We also investigate how healthy controls learn about individuals with BPD versus individuals without BPD since healthy controls evaluate the personality traits of patients with BPD more negatively than they evaluate the traits of other healthy participants (Korn et al., 2016).

Slowing the spread of COVID-19 requires people to actively change their lives and follow best practices for social distancing and hygiene. Optimistic beliefs about the personal impact of COVID-19 could increase or decrease compliance with precautionary actions such as social distancing. Insights on adaptive or maladaptive consequences of optimistic beliefs might lead to better forecasts and policies. Additionally, research on optimism about COVID-19 can inform theories on optimism in general. In the case of COVID-19, Optimistic beliefs about COVID-19 might change in unprecedented manners due to the rapid, severe, and wide-ranging impacts of the COVID-19 pandemic. In a pilot study, we tested individuals’ beliefs about infection probabilities and anticipated social distancing practices in the UK, the USA and in Germany on 16.03.2020, i.e., before strict curfews had been implemented in these countries (see a preprint for an initial summary of these pilot data https://psyarxiv.com/epcyb/). Individuals estimated the probability of getting infected with COVID-19, and of infecting others if infected as lower for themselves than for a similar other. This optimism could be linked to the estimated frequency of direct social contacts as well as to the necessity and ease of reducing these.

Using a longituginal design, we study how optimistic beliefs about the personal consequences of COVID-19 change during the pandemic, in the US, the UK and in Germany. In this preregistration, we specify how we will analyse independent datasets we collected to replicate our own study.

Autism Spectrum Disorder (ASD) is associated with a heterogenous presentation across individuals which hinders the development of mechanistic evaluation procedures and treatments. However, emerging theories and empirical research have connected altered predictive abilities to social and non-social deficits of autistic individuals (Friston, Lawson, & Frith, 2013; Pellicano & Burr, 2012; Sinha et al., 2014; Van de Cruys et al., 2014). The reinforcement learning (RL) framework provides a robust mathematical framework to study how people adapt their predictions to a given environment. The simplest form, the Rescorla Wagner model, posits that individuals optimize predictions by updating them based on the prediction error (PE), the mismatch between initial expectations and environmental outcomes (Behrens, Hunt, & Rushworth, n.d.; Dunne & O’Doherty, 2013; Niv & Langdon, 2016). Our previous work corroborates aberrant prediction error encoding during social learning in autistic adolescents (Rosenblau, Korn, Dutton, Lee, & Pelphrey, 2021). When making judgements about a peer’s preferences, specifically, typically developing (TD) adolescents relied on knowledge of mean preferences of their peers as well as RL. Autistic individuals, however, relied solely on their own preferences to make these preference judgements. It remains unknown whether differences in PE encoding and predictive abilities of autistic individuals are domain-general or socially-specific. Adolescence is an important period for forging social relationships with peers and consequently for social learning. While TD adolescents actively seek out peer relationships, the brain regions supporting social information processing undergo dramatic development, which influences social learning and decision making in this age group (Blakemore & Robbins, 2012). This study investigates whether neural and behavioral mechanisms differ when TD adolescents learn about non-social information, which would point to socially specific changes in adolescence. This proposal further examines differences between TD and autistic adolescents in learning in the social and non-social domain. Specifically, this study will test whether groups differ in learning about a more and less predictable individual, someone who resembles the average peer more and less, and how participants switch between blocks in which they infer more or less predictable information. The goal of this manipulation is to test cognitive flexibility in specifically social versus nonsocial settings. Finally, this study will elucidate cerebellar contributions to PE encoding and learning, in specifically social settings. It has been posited that the cerebellum posterior lobe (CPL) may play a role in encoding PEs. Moreover, this region is structurally and functionally connected with the medial prefrontal cortex, a key region for representing and updating information about other people (Castelli, Frith, Happe, & Frith, n.d.; Y-J Yang et al., n.d.). As prior research has identified atypical recruitment of mPFC and structural abnormalities of CPL in ASD, it is critical to identify how functional connectivity differences between CPL-mPFC scale with variability in learning strategies in autism (Kelly & Strick, 2003; Salmi et al., 2010; Stoodley & Schmahmann, 2010).

Demographics for all samples, a list of all questions, additional details for statistical tests that were only summarised in the main text (including visualisations), and author contributions

Slowing the spread of COVID-19 requires people to actively change their lives and follow best practices for social distancing and hygiene. On 16.03.2020, we tested individuals’ beliefs about infection probabilities and abilities to practice social distancing in UK, USA and Germany. Given the rapidly evolving situation, we report here initial analyses in a preliminary brief manuscript. We found that individuals show an optimism bias: they estimate the probability of getting infected with the virus, and of infecting others if infected themselves as lower for themselves than for someone similar to them. We link this optimism to the estimated frequency of direct social contacts as well as to the necessity and ease of reducing these. Our ongoing studies will assess how these biases change over the course of the COVID-19 outbreak, and how adaptive it is.

Real life social settings provide us with a multitude of dynamic social information (e.g. facial expressions, tone of voice, body language). Our ability to process and integrate this information to a whole is crucial for understanding others’ mental states (i.e. mentalizing) and subsequently formulating an adequate social response. Individuals with Autism Spectrum Disorders (ASD) are characterized by severe social impairments, which have been linked to aberrant neural and behavioral processing of social stimuli. However, given that typical laboratory settings differ greatly from real life social interactions, it remains unclear how the experimental results obtained with mostly abstract and static stimuli can be generalized to dynamic social information processing in naturalistic settings. Study 1 introduced two naturalistic video-based behavioral tasks for a performance based assessment of implicit (i.e. spontaneous) and explicit (i.e. controlled, more cognitively demanding) mental state inferences. The aim of this study was to characterize implicit and explicit social cognitive processes – how they are to be differentiated in typically developed individuals and individuals with ASD. Study 2 investigated the neural mechanisms of spontaneous, naturalistic mentalizing in individuals with ASD and typically developed controls. Furthermore, the study aimed at relating neural processing of mental states in naturalistic settings to accurate implicit mental state inferences as measured with the behavioral task introduced in study 1. Study 3 aimed at extending insights on dynamic social information processing to the auditory modality. Here, neural processing of emotional prosody (i.e. tone of voice) and behavioral prosody recognition performance were investigated with newly developed naturalistic tasks in individuals with ASD and typically developed controls. Similar to study 2, in a next step, the relationship between neural and behavioral prosody processing was investigated. In typically developed individuals, implicit and explicit social cognitive processes were distinguishable to a certain degree on the behavioral and neural level. Neural activity in the amygdala correlated with accurate implicitsocial behavior, and activity in a network including fronto-temporal regions such as the inferior frontal gyrus (IFG) predicted explicit social behavior. Individuals with ASD performed lower on tasks addressing implicit and explicit social cognitive processes, whereby these different sets of processes were closely linked. On the neural level, the social impairments of individuals with ASD were associated with reduced activity of the amygdala and the superior temporal sulcus (STS) and a reduced relationship between neural processing of social information and accurate social behavior compared to controls. On the basis of the current literature and the three empirical studies of this dissertation, I propose a neurocognitive model of typical and impaired social information processing in naturalistic settings. The proposed model aims at advancing the current literature by integrating behavioral and neuroscientific research on dynamic social information processing in typically developed individuals and individuals characterized by severe social impairments., Alltagssituationen beinhalten eine Vielzahl dynamischer, sozialer Informationen (z.B., Gesichtsausdrücke, Stimmklang (Prosodie), Körpersprache). Die Fähigkeit diese Informationen zu verarbeiten und zu einem Ganzen zu integrieren, ist eine wichtige Voraussetzung für die Erkennung mentaler Zustände beim Gegenüber und ermöglicht somit dass wir in sozialen Situationen angemessen reagieren. Menschen mit Autismus- Spektrum-Störungen (ASD) haben schwerwiegende soziale Beeinträchtigungen, die mit einer gestörten Verarbeitung sozialer Reize auf der Verhaltens- und der Hirnebene im Zusammenhang gebracht wurden. Da die Operationalisierung sozialer Kognition in experimentellen Untersuchungen stark von sozialer Kognition im Alltag abweicht, bleibt es unklar ob und in wie fern Ergebnisse experimenteller Studien, mit abstraktem und statischem Stimulusmaterial, auf dynamische soziale Informationsverarbeitung im Alltag generalisierbar sind. In Studie 1 wurden zwei naturalistische video-basierte Verhaltensmaße entwickelt, um implizite (spontane) und explizite (kognitiv anspruchsvolle) mentale Attribution objektiv und direkt messbar zu machen. Ziel der Studie war es herauszufinden, ob und wie weit sich implizite und explizite soziale Kognition unterscheiden lassen in typisch entwickelten Personen und Menschen mit ASD. In Studie 2 wurden die neuronalen Mechanismen spontaner mentaler Attribution mit einer naturalistischen video-basierten Aufgabe untersucht. Des Weiteren wurde der Zusammenhang zwischen den neuronalen Mechanismen spontaner mentaler Attribution und der Performanz impliziter mentaler Attribution untersucht. Die Untersuchung der Performanz sozial kognitiver Prozesse erfolgte mit Hilfe der impliziten Verhaltensaufgabe, die in Studie 1 eingeführt wurde. In Studie 3 sollten Erkenntnisse bezüglich der sozialen Informationsverarbeitung im Hinblick auf auditorische Verarbeitung sozialer Reize generalisiert werden. Entsprechend wurde die Verarbeitung emotionaler Prosodie auf der Verhaltens- und Hirnebene mit neu entwickelten naturalistischen Aufgaben bei neurotypischen Personen und Personen mit ASD untersucht. Des Weiteren wurde auch hier derZusammenhang zwischen neuronaler Verarbeitung von Prosodie und Performanz untersucht. Bei typisch entwickelten Personen konnten implizite und explizite sozialkognitive Prozesse voneinander unterschieden werden, sowohl auf der Verhaltens- als auch auf neuronaler Ebene. Neuronale Aktivität in der Amygdala korrelierte mit impliziter sozialer Kognition auf der Verhaltensebene. Des Weiteren sagte die Aktivität eines fronto-temporalen Netzwerks, welches den inferioren frontalen Gyrus (IFG) umfasste, akkurates explizites Verhalten voraus. Menschen mit ASD wiesen niedrigere Performanzwerte in impliziter und expliziter sozialer Kognition auf und implizite und explizite sozial kognitive Performanz waren miteinander signifikant korreliert. Auf der neuronalen Ebene waren die sozialen Beeinträchtigungen von Menschen mit ASD mit verringerter Aktivität im superioren temporalen Sulcus (STS) korreliert und gingen außerdem mit einem geringerem Zusammenhang zwischen neuronaler Aktivität und sozial kognitiver Performanz einher im Vergleich zu typisch entwickelten Personen. Auf der Grundlage der aktuellen Forschungslage und der Befunde dieser Dissertation, stelle ich ein neurokognitives Modell typischer und beeinträchtigter sozialer Informationsverarbeitung dynamisch naturalistischer Information auf. Das vorgeschlagene Modell soll die aktuelle Forschungslage erweitern, indem es Verhaltens- und neuronale Befunde zur sozialen Informationsverarbeitung dynamischer Reize bei typisch entwickelten Personen und Personen mit schwerer sozialer Beeinträchtigung integriert.