How does responding to an object affect explicit memory for visual information? The close theoretical relationship between action and perception suggests that items that require a response should be better remembered than items that require no response. However, conclusive evidence for this claim is lacking, as semantic coherence, category size, and trial frequency often differ between stimuli that require a response and those that do not. Here we showed that memory is affected by response requirements, even when confounding factors were eliminated. Participants viewed a stream of images and encoded them into memory. During encoding, some images required a response, whereas others did not. Although all images were task relevant, images that were overtly responded to (e.g., with a button press) were better remembered than those that were not. However, the action itself was not critical to the memory advantage. Covertly counted images were better remembered than those that were not. Moreover, when participants pressed a button for most images, images that required withholding a button press were remembered better than the others. We conclude that the need to modify an ongoing activity results in improved memory.