The document discusses three experiments that examine how visual short-term memory (VSTM) representations of bound object features influence long-term learning. [Experiment 1] found that long-term memory traces were generated even though bindings in VSTM are fragile. [Experiment 2] maximized exposure to repeated arrays but found no increase in memory performance. [Experiment 3] used a probed recall procedure and found that learning only occurred for fully bound objects presented in the "repeat-all" condition.