The CDS Summer School offers a range of talks, covering a variety of topics across multiple disciplines.
Decisions and the Mind
Decisions and AI
Decisions and Societies
Decisions and Mental Health
Decisions and Evolution
Meet our Chairs
Our chairs come from a diverse range of academic disciplines and will each be leading their own days during the summer school, focused on their respective field of interest.
Gabriele Bellucci
Director of the Center for Decision Sciences
Anand Subramoney
Lecturer in Computer Science
Andreu Casas
Lecturer in Political Communication
Nura Sidarus
Lecturer in Cognitive Neuroscience
Monday
Gabriele Bellucci, Computational Cognitive Modeling (Bellucci_computational_cognitive_modeling.mp4; talk1.pdf)
Decision-making is a fundamental cognitive process that underpins behavior across domains—from neuroscience and psychology to artificial intelligence and economics. Here, I will introduce the range of computational formalism and methods that are currently used in the psychology, neuroscience, and cognitive sciences. These will include reinforcement learning, Bayesian methods, machine learning techniques, and neural networks. I will show in particular how these models have been applied to explain specific cognitive phenomena and understand their neural correlates.
Gabriele Bellucci, Cognitive Models of Social Decision Making (Bellucci_cognitive_models_of_social_decision_making.mp4; talk2.pdf)
Social decision-making involves navigating complex environments where choices affect—and are affected by—others. Understanding the cognitive mechanisms behind such decisions is critical for fields ranging from psychology and neuroscience to economics, sociology, and artificial intelligence. Here, I will focus on those cognitive models that have been applied to capture social behaviors in humans, such fairness, trust, and prosociality. We will explore models that propose a formalization of social cognitive abilities (e.g., theory of mind), social preferences, and social learning. You will gain insight into the latest tools and theories used to model social cognition and decision-making, and develop a deeper understanding of how humans—and increasingly, machines—make decisions in social environments.
Carolina Feher da Silva, Reinforcement Learning: Model-Free or Muddled Models? (talk4.pdf)
An influential theory in decision neuroscience posits a competition between a low-effort, “model-free” habit system and a complex, “model-based” goal-directed system. This talk presents behavioural and neuroimaging evidence that challenges this long-standing dual-system view, arguing that humans primarily use model-based inference when a task is properly understood.
We demonstrate that the widely-reported influence of model-free learning is likely an artefact of task misunderstanding. By providing clearer, story-based instructions for the classic two-stage task, we observe a dramatic shift to almost purely model-based choices. Crucially, this increase in correct model-based behaviour is accompanied by reduced mental effort and physiological markers of cognitive load, suggesting that standard abstract instructions promote a more cognitively costly, incorrect model-based approach rather than a less effortful model-free one. Furthermore, our neuroimaging analyses reveal that previous reports of combined model-free and model-based prediction errors in the ventral striatum are likely spurious, with more appropriate models showing no reliable evidence for a model-free signal.
These findings compel a re-evaluation of influential learning theories. We argue that what has been interpreted as low-effort, model-free behaviour may in fact reflect the higher cognitive cost of grappling with a misunderstood environment. This suggests the simple dichotomy between model-free and model-based systems is inadequate for explaining the complexities of human choice.
Matteo Lisi, Disentangling Bias and Noise in Human Confidence (Lisi_disentangling_bias_and_noise_in_human_confidence.mp4; talk5.pdf)
Every choice we make in life is accompanied by a sense of confidence – a subjective feeling about how likely it is that we are choosing the best possible course of action. Importantly, confidence is not only used to critically re-evaluate past decisions but also plays a pivotal role in guiding future behaviour when outcomes are not immediately available. In this talk, I will present a psychophysical ‘dual-decision’ method that allows us to assess the calibration of confidence, revealing systematic biases of over- and under-confidence. I will then show how this approach can be used to separate confidence errors into bias and noise, uncovering domain-specific patterns: under-confidence in perceptual tasks, but increased confidence noise in knowledge-based tasks. These results shed light on the computational constraints shaping human confidence.
Keiji Ota, Modelling Probability Distortion (Ota_models_of_decision_making.mp4; talk3.pdf)
People distort probability in decision under risk. In this hands-on tutorial, students will learn how to model probability distortion using the R package RStan. The session will begin with a brief theoretical overview of probability distortion. We will then walk through a model-fitting pipeline: simulating hypothetical data, testing parameter recovery, fitting the model to real data and interpreting the result.
Matteo Lisi, Signal Detection Theory (Lisi_signal_detection_theory.mp4)
Tuesday
Chris Watkins, Large Language Models (LLM)
Robin Schiewer, Reinforcement Learning and Model-Based Reinforcement Learning (MBRL_Intro.pdf; Schiewer_RL_model_based_RL.mp4)
Robin Schiewer, Reinforcement Learning
David Kappel, Bayesian Models of Uncertainty (david-bayes-by-backprop-expanded.pdf; Kappel_bayesian_model_of_uncertainty.mp4)
Agnieszka Mensfelt, Agentic Artificial Intelligence (agentic_artificial_intelligence.pdf; Mensfelt_AAI.mp4)
“Agentic AI” has recently become a popular term in discussions about the future of artificial intelligence, attracting interest from researchers, developers, and the public. But the idea of agents — systems that can sense, plan, and act in the world — is not new; it has been a central concept in AI, robotics, and cognitive science for decades. This talk introduces the fundamental ideas behind agents, explains how modern “agentic AI” systems fit into this framework, and clarifies common misconceptions about what these systems can — and cannot — currently do. We will examine what it means for an AI system to be agentic, review examples of current agent architectures, and discuss major challenges such as aligning AI with human goals, making decisions under uncertainty, and ensuring long-term reliability. The goal is to equip students with a solid foundation for understanding and critically engaging with the rapidly evolving field of agentic AI.
Agnieszka Mensfelt, Agentic Artificial Intelligence (Mensfelt_AAI_practical.mp4)
Wednesday
Sílvia Majó-Vázquez, Leveraging digital tracking data to study polarisation across countries
Andreu Casas, Text-as-data in political science (Casas_text_as_data_political_science.mp4)
Andreu Casas, Text-as-data in political science (Casas_text_as_data_political_science_practical.mp4)
Chris Hanretty, Classifying sentences in parliamentary corpora using fine-tuned large language models
Parliaments record large quantities of spontaneously generated speech.
Because this speech is spontaneously generated, it can reveal aspects of
the speaker’s psychology. In this talk, I describe how I fine tuned a
large language model to classify sentences spoken in different
parliaments for their temporal focus — broadly, whether they were about
the past, present or future. I describe the initial sample of hand-coded
sentences, the training process, and some of the practical difficulties
encountered. I use this to show how parliamentarians’ future focus
declines as they enter old age.
Andreu Casas, Images-as-data in political science
Andreu Casas, Multimodal Analysis in Political Science
Thursday
Jennifer Dykxhoorn, Social determinants of mental health
Over the last 30 years, a growing body of evidence has highlighted how people’s mental health is shaped by broad societal factors. The conditions where people are born, grow, live, work, and age shape mental health and structural factors generate and perpetuate intergenerational cycles of disadvantage and poor mental health. In this talk, Dr Jen Dykxhoorn will focus on several of the most pervasive social determinants of mental health across the life course, with a particular focus on marginalized groups exposed to multiple intersecting social risk factors. You can read more about Dr. Dykxhoorn’s research at www.mentalhealthepi.com.
Francesco Scaramozzino, How do we decide what is real? Computational accounts of psychosis and psychotic-like experiences (Scaramazzino_decision_making_in_psychosis.mp4)
Psychiatry faces the critical challenge of mapping neurobiological mechanisms onto first-person experiences, namely, psychiatric symptoms. Psychosis is typically characterised by beliefs and/or perceptions that deviate from socially shared interpretations of reality. In recent years, information-theoretic approaches have been used to link the emergence of such experiences to neurobiological mechanisms, providing a promising theoretical foundation for the naturalistic study of psychotic phenomenology.
In this context, we outline how predictive processing and attractor network models contribute to understanding psychosis along its continuum, encompassing clinical symptomatology and psychotic-like experiences. These frameworks assign specific informational roles and contents to neurobiological mechanisms and, crucially, link these neurocomputational processes to the phenomenological alterations observed in psychosis. We argue that integrating perspectives from predictive processing, the free-energy principle, attractor network theory, and the entropic brain theory provides a coherent theoretical foundation for studying psychotic phenomenology across clinical and non-clinical populations.
Our focus will be on two key mechanisms: the modulation of precision in neural signalling and the stability of attractor states as reflected in brain dynamics. These processes may play a critical role in the brain’s efforts to minimise uncertainty (i.e., free energy), in shaping the structure and content of perceptions and beliefs, and to sustain stable interpretations of reality.
Kelly Donegan, Multimodal Analysis in Political Science (CDS_TakingResearchOnline_Workshop_KRD.pptx; Donegan_mobile_and_app_data.mp4)
Alex Pike, Using computational models to understand mental health: promise and problems (CDS_Pike.pptx; Pike_computational_psychiatry_and_mental_health.mp4)
Computational psychiatry is a maturing field that focuses on applying computational models of learning and decision-making to mental health. In this talk, I will highlight some findings from my group that underlie the diverse ways that modelling can be used to gain a mechanistic understanding of mental health concerns, including a meta-analysis of reinforcement learning approaches to depression and anxiety. However, this approach is not without limitations – so I will go on to present data that provides a specific test of some of the implicit assumptions we make as a field, which has somewhat alarming implications – but also some promising ones.
Alex Lloyd, Co-production and involvement of public and lived experience experts (CDS_CoProduction_AL.pptx; Lloyd_co_production_and_involvement_of_public_and_lived_experience_experts.mp4)
Involving individuals with lived experience is increasingly being recognised as important in the study of psychopathology. In recognition of the value of collaborating with experts by experience, international funders are increasingly mandating that mental health research is developed by teams that include individuals from the population of study. Yet, there is limited understanding of how to implement Patient and Public Involvement (PPI), including co-production and co-design, in fields such as computational psychiatry. Further, there is heterogeneity in how such methods are implemented and reported, limiting reproducibility. In this session, I will outline finings of a scoping review examining how PPI methods are implemented in mental health research, providing a practical introduction into how these methods can be adopted in computational psychiatry. I will also briefly outline emerging findings from a Delphi study into facilitators and barriers to diversity in PPI, which can help to ensure experts by experience who collaborate on research reflect the diversity of populations that experience mental health problems.
Friday
Fabio Manfredini, The neurogenomic approach to understand social behaviour in animals (manfredini_gene_expression_and_social_behaviors.mp4)
Animals display a wide range of behaviours that are often defined as “complex” because they are transient, extremely plastic and often involve the participation of multiple individuals. This is surely the case for social behaviours that have evolved under the pressure of intricate networks of interactions among participants. It has always fascinated scientists to understand what the molecular basis for these behaviours is, i.e. what are the genes involved and how are they regulated to produce the behaviour that we observe at the phenotypic level? This is the power of neurogenomics, which consists in the analysis of global patterns of gene expression and regulation at the brain level. In my lecture I will provide a framework for this discipline, present key case studies in different organism and illustrate what tools can be used to achieve the ambitious goal of linking genes and social behaviour.
Gabriele Bellucci, Probing the Evolution of Culture (talk2.pdf; Bellucci_probing_the_evolution_of_culture.mp4)
Will Hoppitt, Studying Social Transmission using Networks
Network Based Diffusion Analysis (NBDA) is a statistical technique for studying the social transmission of behaviour, or other information. It was developed to help inform and resolve debates about whether some non-human animals also have culture, but its scope has become much broader. I will explain the background and rationale behind the technique and show how it has been extended in various ways.
Michael Chimento, Statistical signatures of social transmission using STbayes (Chimento_bayesian_models_of_social_transmission.mp4)
This workshop will be a hands-on tutorial where students will learn how to use the R package STbayes to create and fit Bayesian models of social transmission. I will briefly introduce the Bayesian approach to modelling. We will then walk through simulating transmission data, running an analysis pipeline, and interpreting model output. This is intended as a high-level tutorial, and a statistical background is not required for participation. Basic familiarity with R will be helpful.
Equity, Diversity, and Inclusion (EDI) in Research Panel
The goal of this panel is to provoke reflection and promote action on how we can all work to make our research more inclusive and diverse. We will consider how EDI issues impact our research workplaces and cultures, namely through inequalities in representation, or the impact that has on the experiences of researchers. Moreover, we will discuss how EDI issues interact with the research itself, for example in terms of who gets to participate; whose perspectives, voices or issues can be heard; or who gets to shape what or how research is done. Inevitably, these two dimensions – work places and the work itself – interact, yet highlighting the different issues involved can reveal more opportunities for action, at individual, institutional, or international levels.
Chaired by Nura Sidarus, Lecturer in Cognitive Neuroscience, RHUL; Chair of Psychology Department’s EDI Team, RHUL; Chair of the Gender Equality Group, RHUL; Chair of Women in Cognitive Science Europe (WiCS-E).
Vanessa Pinfold – Research Director and Co-Founder of the McPin Foundation
Vanessa has worked in mental health research for over 25 years, and has published studies on stigma and discrimination, families and carers, experiences of the mental health system, wellbeing networks and co-production in mental health research. She is now prioritising developing peer research methods through collaborative or co-production approaches. Vanessa currently chairs the Alliance of Mental Health Research Funders and is co-founder and research director at McPin, responsible for overseeing the work of the charity.
She previously worked at Rethink Mental Illness and the KCL IOPPN and has a PhD from University of Nottingham.
Tiarna Lee – PhD in the School of Biomedical Engineering and Imaging Sciences at King’s College London
Tiarna Lee completed a PhD in fairness for AI used for cardiac imaging at King’s College London. Her work focuses on racial bias in cardiac magnetic resonance image segmentation.
Romana Saleh – Youth advisor on the ReSET project.
Romana is a final year undergraduate student studying psychology with neuroscience at the University of Reading. She has previous experience as a youth advisor on the ReSET project at UCL and was a co-author of ‘No decision about me, without me: collaborating with children and young people in mental health research’, a systematic review into how co-production, co-design, and Patient and Public Involvement (PPI) is implemented in youth mental health research.
Attendee Talks
Caroline Pioger, Experiential value neglect remains robust across changes in options’ and outcomes’ representation (Presentation_CDS_080925.pdf)
In decision-making, values attached to options can stem from past experiences with rewards and punishments (experiential) or explicit descriptions of outcomes and probabilities (descriptive). According to the common currency hypothesis, we encode these subjective values on the same scale. Most studies examine decision-making within either experiential or descriptive options separately, but what about when individuals choose between the two? Garcia et al. (2023) examined such hybrid choices by asking participants to choose between learned experiential options and symbolically described ones, and found systematic neglect of experiential options. This study explores, through five experiments, whether experiential neglect arises from differences in options’ and outcomes’ representations. Our findings show that experiential neglect persisted across all conditions, further challenging the dominant theory and suggesting that the neglect is primarily driven by memory retrieval cost. Furthermore, we found patterns of reduced sensitivity to losses, especially in contexts involving comparative decision-making.
Bonan Zhao, Discovering hidden laws in innovation by recombination
People create new things from combining existing things – whether mixing water and flour or feeding large datasets into neural networks. My research investigates how people hypothesize rules and theories for recombination, test those hypotheses, and communicate our findings to each other. Crucially, as people are constrained by cognitive capacity, I draw ideas from resource-rationality and information theory to formalize how people solve the problem of innovation by recombination in an open-ended space under channel capacity constraints. In a grid-world crafting game, we find empirical evidence that people rely on their inductive biases to explore recombinations, and reveal an inverse-U shaped relationship between the compressibility of recombination rules and people’s communication effort.
Tiia Ladvelin, A daily study on emotions and smartphone use at work
Smartphones are essential in working life, but also a source of distraction. The study examines the impact of emotions on smartphone use during work and how this predicts job performance. Grounded in Pekrun’s (2006) cognitive-motivational model of emotion effects, we investigate how discrete emotions, such as excitement or anxiety, may support or impair staying on task. Over eight workdays, 150 participants complete a daily emotion induction task at their workplace, describing a work-related emotional memory. They then return to their tasks for 30 minutes. Surveys, administered before and after this period, assess emotions, smartphone use metrics, and perceived performance. We employ Multilevel Structural Equation Modelling to capture how within-person fluctuations in emotions predict smartphone use in naturalistic settings. The research contributes to our understanding of the within-person processes behind job performance and the role of emotion in everyday decision-making.
Miriam Vignando, Normative receptor density differentially correlates of mismatch negativity alterations in Parkinson’s disease with visual hallucinations (summerschool-pres_vignando2.key)
Visual hallucinations (VH) in Parkinson’s disease (PD) predict poor outcomes, yet underlying mechanisms remain unclear. Reduced mismatch negativity (MMN) in PD-VH suggests a predictive coding deficit. We applied dynamic causal modelling (DCM) to MMN EEG data and examined neurochemical correlates using normative receptor density maps.
Group differences in a ventral visual stream microcircuit model were estimated with parametric empirical Bayes. Source-reconstructed MMN signals were regressed on receptor densities across regions, correcting for multiple comparisons and spatial autocorrelation. Individual-level effects were tested with linear mixed modelling.
PD-VH showed decreased top-down and increased bottom-up connectivity. Serotonergic density positively correlated with MMN, whereas dopaminergic and cholinergic densities correlated negatively. The 5-HT2A association aligns with increased bottom-up visual connectivity in PD-VH and supports a role for receptor-enriched models in linking neurochemistry to neurophysiological phenotypes in neurodegeneration.
Niharika Madala, Exploring the Role of Arousal Dysregulation and Interoception in Risk-taking Behaviour among Individuals with ADHD traits and Comorbid Anxiety
Risky decision-making in Attention Deficit Hyperactivity Disorder (ADHD) has traditionally been linked to executive functioning deficits, but recent research highlights the role of physiological dysregulation, particularly in arousal and interoception. A mismatch between cortical and peripheral arousal can produce brain hypo-arousal alongside heightened emotional arousal during decision-making and in response to feedback, impairing sustained attention and goal-directed behaviour. The present study investigates how individual differences in ADHD, interoception, and anxiety influence arousal and risk-taking behaviour, and examines how physiological arousal, feedback sensitivity, and impulsivity interact to drive risky decisions. Participants completed questionnaires on impulsivity, anxiety, ADHD traits, and interoception, and performed the modified Columbia Card Task (CCT) while physiological arousal was recorded using Galvanic Skin Response (GSR). Data analysis is ongoing to clarify how arousal regulation and individual differences shapes risk preferences. Findings aim to inform underlying mechanisms driving risk preferences and strategies for managing risk-taking behaviours in individuals with elevated ADHD traits.
center-decision-sciences.com 