Controlling the probability of a series of seemingly random events is the key to mimicking the human brain to optimize neuromorphic learning.