Individual curiosity modulates exploration in sequential book selection.

Information seeking is widely understood as a curiosity-driven exploration behavior similar to resource foraging. However, it remains unclear whether specific exploration decision mechanisms, such as reward generalization and directed exploration, extend beyond physical consumption domains to the semantic spaces of knowledge and information. How do people choose what to read when navigating vast and unfamiliar content landscapes? We study sequential book selection to investigate this question. At times, readers exploit familiar books they expect to enjoy; at other times, they explore novel options to discover potential rewards. Using a large-scale real-world dataset and a controlled behavioral experiment, we show that book selection is guided by structured generalization across a semantic embedding space and by directed exploration toward options with high uncertainty. Moreover, individual differences in curiosity modulate this exploration-exploitation tradeoff, promoting exploratory reading and increasing enjoyment. These findings demonstrate that core computational mechanisms underlying foraging generalize to epistemic domains and shape real-world media selection.