We introduce Incentive Design: a new class of problems for equilibrium verification in multi-agent systems. In our model, agents attempt to maximize their utility functions, which are expressed as formulae in LTL[F], a quantitative extension of Linear Temporal Logic with functions computable in polynomial time. We assume agents are rational, in the sense that they adopt strategies consistent with game theoretic solution concepts such as Nash equilibrium. For each solution concept we consider, we analyze the problems of verifying whether an incentive scheme achieves a societal objective and finding one that does so, whether it be social welfare or any other aggregate measure of collective well-being. We study both static and dynamic incentive schemes, showing that the latter are more powerful than the former. Finally, we solve the incentive verification and synthesis problems for all the solution concepts we consider, and analyze their complexity.

Training reinforcement learning (RL) agents using scalar reward signals is often infeasible when an environment has sparse and non-Markovian rewards. Moreover, handcrafting these reward functions before training is prone to misspecification. We learn non-Markovian finite task specifications as finite-state ‘task automata’ from episodes of agent experience within environments with unknown dynamics. First, we learn a product MDP, a model composed of the specification’s automaton and the environment’s MDP (both initially unknown), by treating it as a partially observable MDP and employing a hidden Markov model learning algorithm. Second, we efficiently distil the task automaton (assumed to be a deterministic finite automaton) from the learnt product MDP. Our automaton enables a task to be decomposed into sub-tasks, so an RL agent can later synthesise an optimal policy more efficiently. It is also an interpretable encoding of high-level task features, so a human can verify that the agent’s learnt tasks have no misspecifications. Finally, we also take steps towards ensuring that the automaton is environment-agnostic, making it well-suited for use in transfer learning.

We introduce and study a computational version of the principal-agent problem -- a classic problem in Economics that arises when a principal desires to contract an agent to carry out some task, but has incomplete information about the agent or their subsequent actions. The key challenge in this setting is for the principal to design a contract for the agent such that the agent's preferences are then aligned with those of the principal. We study this problem using a variation of Boolean games, where multiple players each choose valuations for Boolean variables under their control, seeking the satisfaction of a personal goal formula. In our setting, the principal can only observe some subset of these variables, and the principal chooses a contract which rewards players on the basis of the assignments they make for the variables that are observable to the principal. The principal's challenge is to design a contract so that, firstly, the principal's goal is achieved in some or all Nash equilibrium choices, and secondly, that the principal is able to verify that their goal is satisfied. In this paper, we formally define this problem and completely characterise the computational complexity of the most relevant decision problems associated with it.