In recognition memory, participants often fail to change their criterion for making a “studied” response from one trial to the next based on learning strength, even when they are given obvious cues to identify each test item as studied often (“strong”) or studied a single time (“weak”) (e.g., Stretch & Wixted, 1998). In three experiments we tested the hypothesis that participants produce robust item-by-item shifts only when responding did not involve significant response interference (Simon, Acosta, Mewaldt, & Speidel, 1976). In our three experiments, participants studied lists of words studied once (weak) or five times (strong). In Experiment 1, both strong and weak words appeared at test under the questions “Was this studied at all?” or “Was this studied five times?” Participants were randomly assigned to conditions using two keys to respond “yes” or “no”, or using four keys with one set of “yes” and “no” per question. Four-key participants were expected to shift their criteria while 2-key participants could not due to response interference, though results showed that both conditions were capable of criterion-shifting. In Experiment 2 test items appeared on either the left or the right side of the screen; only strong words appeared on the right and only weak words on the left. Participants went through one study-test cycle with four response keys, and one with two. Regardless of the testing conditions, participants did not shift their criteria in the 2-key condition while participants in the 4-key condition did shift their criteria. Finally, Experiment 3 fully crossed 2 or 4 key conditions with blocked or unblocked presentation of test items. Previous experiments have found both number of response keys and blocking of presentations to have an effect on ability to criterion-shift (Hicks & Starns, 2014; Starns & Olchowski, submitted; Verde & Rotello, 2007). Experiment 3 confirmed that number of response keys has a significant effect on criterion-shifting and that it is separate from any effect of blocking. All three experiments suggest that response interference is not the driving force behind criterion-shifting. A new explanation is proposed.