“…Subjects trained on repeated reversals may become able to generalize that reward contingencies can be reversed, which is sometimes referred to as the principle of reversal (Shettleworth, 1998). In deterministic two-choice serial reversal-learning tasks, subjects may optimize performance by acquiring the win-stay, lose-shift (WSLS) strategy (Warren, 1966; for discussion, see Bessemer & Stollnitz, 1971), in which the subject repeats the response from the previous trial if that choice was rewarded (i.e., if win , then stay ) but switches to the alternative response if that choice was not rewarded (i.e., if lose , then shift ).…”