Giovanni Petri

Northeastern University London offers a unique PhD opportunity in the interdisciplinary field of network science, focusing on higher-order coordination and emergent communication across human, AI, and animal systems. As both a UK university and the European campus of Northeastern University (Boston), NU London provides students with access to international research networks and the chance to engage with overseas campuses during their doctoral studies. This project, part of the ERC Consolidator Grant “RUNES” (Reconstruction and Unification of Neural and Ecological Systems), investigates how group interaction structures shape coordination, norm formation, and the emergence of communication systems. The research spans three cognitive system classes: human populations, multi-agent AI systems, and animal societies. The approach integrates online behavioural experiments, multiagent simulations, and analysis of animal communication data. Key research areas include: Social coordination on group interaction structures: Extending the naming-game framework to higher-order topologies, the project designs and runs large-scale online coordination experiments, varying group size, overlap, and clustering. The aim is to test how group-level interaction structure shifts tipping-point dynamics and convention formation, using higher-order contagion models and information-theoretic decompositions. Emergent communication in multi-agent AI systems: The project explores how interaction topology affects the complexity and compositionality of communication protocols in multiagent systems. It leverages recent findings on LLM agent populations and uses reinforcement-learning agents for mechanistic validation, measuring communication complexity with information-theoretic tools. Animal communication and collective coordination: Using data from Project CETI (sperm whale vocalisations and behaviour) and collaborating labs (zebrafish collective behaviour with neural activity markers), the project applies higher-order statistical analysis to predict collective outcomes beyond pairwise models. The project culminates in a cross-system synthesis, examining whether topological and information-theoretic features predicting coordination in one system transfer to another—a foundational question for complex systems science. The successful candidate will join the NP Lab within the Network Science Institute, led by Professor Giovanni Petri, and interact with a vibrant, interdisciplinary research environment. The supervisory team includes Professor Giovanni Petri (NU London) and Dr. Zhongtian Sun (University of Kent). The PhD is aligned with the Network Science programme and is full-time. Funding: The position is fully funded, covering tuition fees, an annual stipend, and a London allowance (UKRI rates) for 3.5 years. Eligibility: Applicants must have a Bachelor’s degree in a relevant subject (2:1 or 1st class, essential) and preferably a Master’s degree in Physics, Mathematics, Computer Science, or a related quantitative field. Strong modelling, computational, and code development skills are required, with experience in network science, multi-agent systems, statistical mechanics, or computational social science. Good programming skills in at least one of Python, Julia, Rust, or C++ are essential. IELTS 7 overall (with at least 6.5 in each component) or equivalent is required for non-native English speakers. Applications are open to UK and international students, but visa costs cannot be supported. Application deadline: April 30, 2026. How to apply: Send a CV, a 2-page research proposal related to the project topics, and a covering letter stating how you meet the requirements and your interest in the research. Apply via the provided link and reference your application ‘R139425’. Informal enquiries can be directed to Professor Giovanni Petri ([email protected]). This PhD offers a stimulating environment for candidates excited by interdisciplinary research at the intersection of network science, collective dynamics, and communication systems.

Northeastern University London (NU London) is offering multiple fully funded PhD scholarships as part of a major investment to accelerate interdisciplinary research in the humanities, social sciences, and digital sciences. Each scholarship is fully funded for three and a half years at UKRI rates, covering full course fees, an annual stipend (with an additional London allowance), and associated training costs. NU London is a UK university governed by UK higher education regulations and serves as the European campus of Northeastern University, a top-tier research-intensive institution based in Boston, USA. The university is renowned for its high-impact research, interdisciplinarity, experiential learning, and global connections, with campuses across the United States and Canada. This PhD project, part of the ERC Consolidator Grant “RUNES” (Reconstruction and Unification of Neural and Ecological Systems), investigates the geometric and topological structures in neural representations that enable or constrain compositional cognition. The research aims to determine whether similar signatures appear across biological brains and artificial neural networks. The successful candidate will work at the intersection of mechanistic interpretability, computational neuroscience, and cognitive AI, exploring how the internal geometry of neural representations shapes the capabilities and limitations of both brains and deep networks. Recent studies have shown that Vision-Language Models (VLMs) exhibit compositional failures paralleling known human cognitive constraints, such as the binding problem. The geometric overlap of concept vectors in latent representation space correlates with specific error patterns, suggesting that the geometry of representations constrains compositional cognition across both biological and artificial systems. The project will employ mechanistic interpretability tools—concept extraction, causal tracing, activation patching—alongside geometric and information-theoretic methods to bridge AI interpretability with neuroscience. The research programme spans four interconnected directions: (i) using mechanistic interpretability methods to characterize how the geometry of representations in open-weight VLMs and LLMs shapes compositional success and failure; (ii) contributing to the RUNES neuroimaging programme by applying computational neuroscience tools to fMRI data from multitasking and binding experiments; (iii) building a cross-system bridge to test whether representational signatures distinguishing compositional success from failure in AI systems also appear in neural population codes during analogous cognitive tasks; and (iv) developing new interpretability methods that capture group-level structure in neural network circuits beyond current pairwise approaches. The successful candidate will join the NP Lab within the Network Science Institute, led by Professor Giovanni Petri, and is expected to interact with group members, postdocs, graduate and PhD students, and work across a range of applications in network and topological data analysis. The research environment is interdisciplinary and vibrant, with international collaboration and networking opportunities, dedicated research space, and access to the wider Northeastern University network in North America. Eligibility requirements include a Bachelor’s degree in a relevant subject (2:1 or 1st class), a Master’s degree in Physics, Mathematics, Computer Science, or related fields (strongly recommended), strong computational and modelling skills (including proficiency in Python and experience with deep learning frameworks), and experience or strong interest in mechanistic interpretability, representational geometry, or computational neuroscience. Candidates should be comfortable working across AI, neuroscience, and mathematics, and possess excellent communication and collaboration skills. English language proficiency at IELTS 6.5 overall (with at least 6.5 in each component) or equivalent is required for non-native speakers. Applications are open to UK and international students, but visa costs cannot be supported. To apply, send a CV, a 2-page research proposal related to the project topics, and a covering letter explaining how you meet the requirements and your interest in the research. Apply via the provided link and reference your application ‘R139422’. Informal enquiries can be directed to Professor Giovanni Petri before the application deadline. The panel will shortlist candidates on a rolling basis and reserves the right to close the post before the deadline once a suitable applicant is found.