 The hippocampus is known to play a key role in memory formation and retrieval. Recent research suggests that it may also be involved in predicting future events based on past experiences. This predictive ability has been modeled using a predictive map called the successor representation, SR, which captures patterns in hippocampal activity. A new study shows that these patterns can emerge from a recurrent neural network if its synaptic weights match the transition probabilities of the network. Additionally, the network's gain can control the length of the prediction horizon. These findings suggest that the SR is more accessible in neural circuits than previously thought, and could potentially support a wide variety of cognitive functions. This article was authored by Qing Fang, Dimitri Aronov, LF Abbott, and others.