Goals as reward-producing programs

Abstract

People are remarkably capable of generating their own goals, beginning with child’s play and continuing into adulthood. Despite considerable empirical and computational work on goals and goal-oriented behaviour, models are still far from capturing the richness of everyday human goals. Here we bridge this gap by collecting a dataset of human-generated playful goals (in the form of scorable, single-player games), modelling them as reward-producing programs and generating novel human-like goals through program synthesis. Reward-producing programs capture the rich semantics of goals through symbolic operations that compose, add temporal constraints and allow program execution on behavioural traces to evaluate progress. To build a generative model of goals, we learn a fitness function over the infinite set of possible goal programs and sample novel goals with a quality-diversity algorithm. Human evaluators found that model-generated goals, when sampled from partitions of program space occupied by human examples, were indistinguishable from human-created games. We also discovered that our model’s internal fitness scores predict games that are evaluated as more fun to play and more human-like. To enable artificial agents to generate human-like goals, a model must capture the complexity and diversity of human goals. Davidson et al. model playful goals from a naturalistic experiment as reward-producing programs, mapping an agent’s behaviour to goal success. They then develop a computational model to generate diverse human-like goals.