I study how we generalise what we know so that we can perform in new environments. I'm interested in how attention -- selecting and prioritising parts of sensory information -- relates to that ability to generalise. I try to understand this through people's actions in very basic computer games, through questions about how they think, and through non-invasive recordings of brain activity; but I'm curious about how other perspectives, from rodent work and computational modelling, can help us develop more precise explanations.