Response effects, also called action effects, are events produced as a consequence of a response. Morin and Grant (1955) conducted an 8-choice task in which, when one of the response keys was pressed, a feedback light (i.e., response effect) was lit in a row below the stimulus lights. Across participants, stimulus-response and response-effect mappings varied in compatibility, ranging from perfectly corresponding, to random, to perfectly mirror-reversed mapping. After several practice sessions, the feedback lights were removed in a transfer session, and response times increased greatly, particularly for the mappings without much structure, indicating reliance on the feedback during practice. We revisited this paradigm by means of 2 experiments that examined the influence on acquisition and transfer performance of task instruction, reliability of the visual feedback, and task difficulty. Lower task difficulty and unreliable visual feedback resulted in superior performance in the transfer session, indicating better learning of the stimulus-response mapping and less reliance on the feedback during learning. Task instructions that emphasized learning of the mapping or indicated that there would be a later test without the feedback lights did not influence participants' performance in the transfer session. These findings are discussed in relation to several of Healy and colleagues' principles of skill acquisition and transfer and in the context of contemporary research on response-effect associations.