This paper proposes a novel learning to learn method, called learning to learn iterative search algorithm(LISA), for signal detection in a multi-input multi-output (MIMO) system. The idea is to regard the signal detection problem as a decision making problem over tree. The goal is to learn the optimal decision policy. In LISA, deep neural networks are used as parameterized policy function. Through training, optimal parameters of the neural networks are learned and thus optimal policy can be approximated. Different neural network-based architectures are used for fixed and varying channel models. LISA provides soft decisions and does not require any information about the additive white Gaussian noise. Simulation results show that LISA achieves state-of-the-art detection performance. Particularly, LISA obtains near maximum likelihood detection performance in both fixed and varying channel models under QPSK modulation, and performs significantly better than classical detectors and recently proposed deep/machine learning based detectors under various modulations and signal to noise levels.