Figure 1.

Outline of experimental procedures, example behavior, and schematic of hypotheses regarding how mPFC cells may alter responses and mediate behaviors as a result of flexible learning. A, Rabbits (n = 8) received standard training for trace eyeblink conditioning until asymptotic. Inset shows the standard training protocol and example eyelid response after learning (upward deflection indicates closure). Waterfall plots show traces of eyelid position for each trial of a session for two representative rabbits (trial 1 is at top, x-axis is time: gray bar, 500 ms tone; arrowhead, US onset). Eyelid closures prior to US onset are CRs. After acquisition, rabbits experienced four types of training sessions each week: Day 1, standard training; Day 2, pharmacological blockade of CRs (M, muscimol infusion into the cerebellum; note absence of CRs after infusion for both example sessions; markers show histologically verified infusion sites around Bregma −19.5; scale bar, 1 mm); Day 3, tone-only extinction training (E) followed by reacquisition (R; note the decrease in CRs during E training and the fast reinstatement of behavior during R training); and Day 4, different tone (9.6 kHz) and interval training (750 ms DTI; note the temporal shift in CR onsets during DTI). After Day 4, recording tetrodes were advanced to isolate a different population of mPFC cells, with daily manipulations repeated the following week. B, Schematic of persistent and “trace” (CR feedback responses) single-unit spike responses observed in the mPFC during trace conditioning (left) and the hypothesized response changes if persistent cells mediate flexible learning.