Many chemistry educators have adopted the process-oriented guided instructional learning (POGIL) pedagogy. However, it is not clear which aspects of POGIL are the most important in terms of actual learning. We compared 354 first-year undergraduate psychology students' learning in physiological psychology using four teaching methods: control, POGIL, POGIL without reporting [no report out (NRO)], and POGIL run by untrained graduate students [new facilitator (NF)]. Student activities were identical across POGIL variations and highly similar for control. Participants' knowledge was evaluated before (pretest), immediately after (posttest), and 2 wk later (followup). Control and POGIL groups showed no improvement at posttest, whereas NRO and NF groups both recalled more material than at pretest ("P" = 0.002 and "P" less than 0.0005, respectively). In a surprise test 2 wk later, control ("P" less than 0.0005), NRO ("P" = 0.03), and NF ("P" less than 0.0005) groups recalled less than at posttest. The POGIL group showed the smallest drop in knowledge ("P" = 0.05). Importantly, the control group's knowledge was below pretest levels ("P" less than 0.0005), whereas the POGIL, NRO, and NF groups' knowledge was not. Self assessment of knowledge was consistent across groups at pretest, but POGIL participants had the lowest confidence at posttest and 2 wk later. At followup, the control, NRO, and NF groups showed greater confidence in their knowledge than the POGIL group ("P" = 0.03, "P" = 0.002, and "P" = 0.004, respectively). POGIL and its variations appear to consolidate existing knowledge against memory decay even when student confidence does not match performance.