Does processing level at retrieval moderate the testing effect? Evidence of an asymmetry between study-based encoding and retrieval-based encoding

Retrieval often enhances memory more than restudy (i.e., the testing effect), reflecting the encoding (or re-encoding) effects of retrieval. The present study assessed how similar these retrieval-based encoding processes are to more typical study-based encoding processes. Deep (semantic) processing at encoding benefits memory more than shallow (non-semantic) processing (i.e., levels of processing). If retrieval-based encoding operates similarly, an analogous levels of retrieval effect should occur, with semantic retrieval increasing the benefits of retrieval more than non-semantic retrieval. Participants studied short lists of words, each followed by restudy or retrieval trials, in preparation for a recognition test taken immediately (Experiment 1) or two days later (Experiments 1 and 2). Participants restudied words alongside phonemic or semantic cues, or retrieved words given those cues. In Experiment 2, participants completed two additional rounds of restudy or retrieval. To isolate the effects of processing level on retrieval, testing effects were calculated for each level by subtracting performance in each restudy condition from the corresponding retrieval condition. There was no evidence of a levels of retrieval effect - the benefit from phonemic retrieval over phonemic restudy was either similar (Experiment 1) or greater (Experiment 2) than the benefit from semantic retrieval over semantic restudy.