Projects
A research platform for studying how locally adapting neuron-like units act together in a body and world. Compare Falandays and SORN models on tracking, Pong and CartPole tasks, and examine how different local mechanisms shape collective behaviour. Recorded evaluations preserve settings, seeds, results and provenance for further investigation. Developed by Ben Gaskin, Polyphony Bruna, Ian Jackson and William O’Hearn through the Diverse Intelligences Summer Institute (DISI) 2026, with support from DISI and the John Templeton Foundation. Presented as a late-breaking abstract at ALIFE 2026 in Waterloo, Canada.
A browser-based tool for evolving visual worlds by breeding and mutating GLSL shaders. Choose worlds that interest you, breed variations and follow the results through successive generations. GLSL shader programs generate their appearance and motion, while visual descriptors guide the search towards varied outcomes. Your selections give the process an aesthetic direction. You explore procedural images by responding to what appears, with each generation supplying material for the next choice. Worlds can be saved locally or shared by link.
A special session at ALIFE 2026 in Waterloo, co-organised by Ben Gaskin and Simon McGregor. It brings together experimental, theoretical and position papers that use synthetic systems to examine philosophical questions about life, agency, emergence and simulation. Building a system gives an idea a concrete form: its assumptions become explicit, and their consequences can be explored. The session develops this approach to artificial life as a practice of experimental philosophy.
A workshop at ALIFE 2026 on the origins and evolution of learning, co-organised by Jason Yoder, Anselmo Pontes, Austin Ferguson and Ben Gaskin. It explores how evolution moves from specific, hard-wired responses towards more flexible forms of adaptation. Three questions guide the discussion: how to identify learning, which mechanisms enable it and which ecological conditions favour its emergence. Together, they connect the study of learning to the evolutionary circumstances that make it useful and possible.
An ALIFE 2026 paper exploring how sensing and movement arise together in soft bodies made of identical contractile cells. On the plateau terrains used in evolution, selected specimens approach an edge, distinguish a cliff from a goal and reverse at cliffs. The study examines these behaviours through the interaction of cells, body and terrain, making the body itself a subject of inquiry into sensing. The companion page presents figures and animations, with public code and specimen records for closer investigation.
A JAX-native reimplementation of Evolution Gym, the soft-body robot benchmark introduced by Bhatia and colleagues at NeurIPS 2021. It brings experiments with evolving robot bodies into JAX’s functional programming model, where compilation, vectorisation and iteration compose within an experiment. Robot and terrain are represented separately, supporting batched evaluation across many morphologies. The library provides a basis for exploring how body shape and control work together, and for building evolutionary experiments around those questions.
An interactive reconstruction of Kolmogorov and Uspensky’s 1958 model of computation. Information lives in a finite typed complex, and each computational step replaces a small part of that structure around an active node. A worked tree program makes the rules visible as they operate, so you can follow both the calculation and the changing organisation of memory. The project offers a concrete way into the formalism and its account of computation as local transformation.
Interactive simulations of selected Braitenberg vehicles, each built by coupling light sensors directly to motors. The sign of each connection and the choice of which sensor drives which motor shape how a vehicle moves around a light source. These small wiring differences produce behaviour that observers may describe as fear, aggression, love or preference. Watching the mechanisms and movements together offers a way to explore how we interpret behaviour and explain it in terms of simpler parts.
Use Up and Down to browse. Right Arrow moves to links. Escape releases a kept-open preview.