To take an appropriate action, animals are often required to shift previously unrewarded choice to alternatives. Although the lateral habenula (LHb) and anterior cingulate cortex (ACC) have been implicated in monitoring negative outcomes, how their neuron signals contribute to behavioral shift remains poorly understood. To address this issue, we recorded single-unit activity in the LHb and ACC in monkeys performing a reversal learning task. While the monkey was gazing a fixation point, two saccadic targets were presented on both the left and the right sides of the point. The monkey was required to choose one of the targets. Choosing one target was followed by reward with 50% probability, while choosing the other was followed by no-reward. The reward-position contingency was fixed within a block of 20 to 40 trials, and then reversed without any instruction. The monkey learned to choose the rewarded direction by trial-and-error, and shifted the choice to the alternative if the choice was repeatedly followed by no-reward. We recorded the activity of 62 LHb neurons (41 in monkey A and 21 in monkey E) and 359 ACC neurons (256 in monkey A and 103 in monkey E). Ninety-two percent (57/62) of the LHb neurons and 35% (125/359) of the ACC neurons were more strongly activated by no-reward outcome than by reward outcome. We found that the no-reward response of ACC neurons reflected several past outcomes, and that this response signaled whether the monkey would shift the current choice to the alternative in the next trial. On the other hand, the response of LHb neurons mainly monitored no-reward outcome in the current trial. The response latency was significantly shorter in LHb than in ACC neurons (LHb, 207.3ms ± 18.1ms; ACC, 580.2ms ± 38.1ms; Wilcoxon rank-sum test, P < 0.01). Our findings suggest that LHb neurons quickly detect the current negative outcome, whereas the signal in ACC neurons is suitable for regulating the subsequent behavioral shift.