obtaining a plurality of demonstration sequences, each of the demonstration sequences being a sequence of images of an environment while a respective instance of the reinforcement learning task is being performed;

for each demonstration sequence, processing each image in the demonstration sequence through an image processing neural network comprising a plurality of hidden layers to determine feature values for a respective set of features for the image from activations generated by one or more of the hidden layers;

determining, from the demonstration sequences, a partitioning of the reinforcement learning task into a plurality of subtasks, wherein each image in each demonstration sequence is assigned to a respective subtask of the plurality of subtasks;

determining, from the feature values for the images in the demonstration sequences, a respective set of discriminative features for each of the plurality of subtasks; and

training the agent to perform the reinforcement learning task using one or more perception-based rewards computed from image feature values for the discriminative features for the subtasks.