LIP Lab Methusalem Overview Meeting 2026

This OSF page contains the slides presented at the LIP Lab Methusalem Overview meeting held at de Krook on February 4th, 2026. The meeting aimed to present the Lab's research funded by the Methusalem grant (). Program Overview: 12:00-12:45: Welcome & Sandwiches 12:45-13:10: Introduction, Jan De Houwer 13:10-13:50: General behavioral science: A framework for studying behavior in all natural and artificial systems, Jan De Houwer & Alejandro Macías Based on the idea that science aims to understand phenomena by relating them in certain ways to other phenomena using the scientific method, scientific activities can be organized in terms of the nature of the to-be-understood (i.e., target) phenomenon, the nature of the other phenomenon that the target phenomenon is related to, and the way in which these phenomena are related. Scientists can differ also with regard to the criteria they use to evaluate understanding but must adopt the scientific method for achieving understanding. We put forward a general framework for behavioral science that is situated within this general framework for science. The framework is delineated in two ways: (1) with regard to the criteria that are used to evaluate understanding; (2) with regard to the nature of the target phenomena that are examined. More specifically, we (1) opted for a pragmatic perspective that highlights the aim of influencing behavior and (2) put forward a functional definition of behavior that allows researchers to describe the nature of a wide variety of phenomena that behavioral science aims to understand. These ideas qualify as a framework for behavioral science in that they provide guidance for conducting and organizing research within behavioral science. The framework is general in that the scientific perspective and definitions on which it is built can be applied to many areas of behavioral science, thereby providing common ground for scientists who study behavior in different scientific disciplines. We illustrate the framework by discussing how it can be applied to the study of group behavior. 13:50-14:30: An integrative Cognitive Theory for Behavior (Therapy): What Are We Aiming For?, Yannick Boddez The need to improve treatments for psychological suffering remains considerable. Current approaches span highly heterogeneous analytic frameworks (e.g., functional analysis grounded in classical and operant conditioning; identification of maladaptive thought patterns) and a broad yet fragmented set of interventions (e.g., exposure, cognitive restructuring, defusion). Their justification and further development typically rest on theoretical models that themselves form a scattered landscape. We therefore propose a general yet precise model of human behavior that integrates key insights from established theories. We show (1) how the model helps explain a wide range of overt and covert behaviors linked to psychological suffering (e.g., worry, substance use, avoidance) and (2) how the model unifies effective elements of existing therapies and offers a structured, step-by-step guide for intervention design. We will discuss how ideas and research findings gathered in the context of the Methusalem project led to these ideas and which further steps we propose. 14:30-15:00: Coffee break 15:00-15:40: Developing new methods for testing and training relational behavior and its contextual control, Martin Finn Stimuli can be related in a variety of ways (e.g., same as, more than, opposite to etc.) and according to any property (e.g., size, color, value, speed, number etc). It is therefore critical that relational behavior is determined by two kinds of relational contextual cues; Crels that specify the relation between stimuli (e.g., sameness, opposition, comparison etc) and Cfuncs that specify the functional dimension according to which stimuli are related (e.g., color, monetary value). In recent years our team has explored the development and modification of Crels and Cfuncs. This work has leveraged existing tasks to ask questions about the functional properties of Crels (Finn et al., 2025), has studied the establishment of new Crels (Delabie et al., 2022), and has developed new tasks to establish Cfuncs (Finn & De Houwer, 2021; Finn et al., 2023). Building on this work we have developed measures of the abilities for relating stimuli and for changing behavior based on relations between stimuli (Finn et al. 2026) and explored the relationship between these two facets of relational behavior (Cummins et al., 2026). Future work will further refine these measures, develop interventions for relational behavior, and test the added benefit of these measures for improving relational thinking and behavior. 15:40-16:00: Towards a Computational Model of Arbitrarily Applicable Relational Responding, Matthias Raemaekers Relational Frame Theory (RFT) is one of many theoretical perspectives that consider relational abilities a cornerstone of human cognition. RFT posits that arbitrarily applicable relational responding (AARR), the ability to respond to relations between stimuli that are not defined by their physical properties but by the context, is the foundational unit. Despite spectacular advances, state-of-the-art artificial intelligence (AI) is still challenged by the human capacity for AARR. Furthermore, the data- and model-size-driven approach behind recent advances is incompatible with RFT’s description of the development of AARR. I describe an alternative approach towards building a computational model of AARR in humans, heavily inspired by RFT. We take a step-by-step approach. We start by modelling AARR as observed in adults who have already acquired the ability to AARR and can apply it to learn about novel arbitrarily applied relations (step 3) and to establish novel contextual cues (step 2). To do so, we combine traditional computational reinforcement learning algorithms with hardcoded knowledge of relational derivation and contextual cues. We demonstrate how this approach allows us to model human behavior in matching-to-sample procedures like those employed in seminal demonstrations of the arbitrary, flexible and generative nature of AARR (e.g., Steele & Hayes, 1991). I discuss implications of our findings, and directions for future research on developing a more flexible and scalable model of AARR, and on modelling the long-term development of AARR (step 1). 16:00-16:20: Training relational behavior to improve cognitive skills: Evaluating and adapting the SMART program, Zita Meijer Cognitive development is a strong predictor of later educational and life outcomes, highlighting the importance of identifying structured learning processes that enhance cognitive development. In line with Relational Frame Theory (RFT), recent research emphasizes the role of relational responding skills in the development of human cognition and language. Among interventions targeting relational responding, the SMART program (Strengthening Mental Abilities through Relational Training) stands out. SMART applies progressive learning phases in which participants are presented with relational premises involving arbitrary stimuli, followed by yes/no questions that assess derived relational responding. Although initial findings suggest positive effects, methodological limitations in the existing literature, such as small sample sizes and absence of control conditions, underscore the need for further evaluation of the program. Additionally, studies often lack assessments of real-world outcomes, focusing predominantly on IQ. Our work also indicates that the practical use of the program is not always straightforward, as we encountered challenges with children’s engagement and motivation, which prompted us to implement gamification elements in the program. Based on these considerations, this project aims to further study SMART’s training effects with more methodological rigor and focusing on specific cognitive abilities and determining far-transfer effects. We will present results from a randomized controlled trial in older adults, and multiple single-case experimental designs and a mixed method study in children aged 10-12, demonstrating that the training effects are not as robust as suggested by previous literature. We will also outline upcoming research plans to further evaluate and adapt the SMART program. 16:20-16:50: Coffee break 16:50-17:30: How relational knowledge shapes impressions of others, Marine Rougier Person perception research shows that people readily attribute traits from minimal relational information. One key source is pre-existing beliefs about how traits relate. The attractiveness halo effect illustrates this: attractive people are judged as more sociable or vain, but not necessarily more intelligent. These patterns of selectivity reflect observers’ prior learning about the relations between traits. In recent work, we showed that changing a target’s attractiveness systematically shifts trait attributions in line with these patterns of selective relations (Rougier & De Houwer, 2024). Relational information guiding person perception can also be learned during an experiment. According to the related features principle (Hughes et al., 2020), when a target shares a feature with a source (e.g., a frame color or spatio-temporal pairing), observers assume they share other features. For example, pairing faces with positive or negative stimuli biases visual memory for facial positivity/negativity (Rougier & De Houwer, 2023). We also found effects of shared features on selective patterns of attractiveness- and height-based attributions (Hughes et al., 2024; Rougier et al., 2023): faces sharing a feature with attractive or tall individuals were later judged as similar not only in attractiveness or height but also in traits that are stereotypically related to attractiveness (e.g., vanity) or height (e.g., leadership). In future research, we will try to map patte