Sequential information maximization can explain eye movements in an object learning task