It is still unknown of how do people understand isolated negative sentences with uncertain discrete result, such as the Cartoon isn’t Mickey Mouse. Both schema-plus-tag model and two-step simulation hypothesis, being the most accepted hypothesis, cannot answer this problem. So anchor–based activation and satisfaction constrained model, ABASM, was proposed to solve this problem. According to ABASM, when processing isolated negative sentences with uncertain discrete result, participants would firstly represent the negated state of affairs, Mickey Mouse, and then search the final results. If there were alternatives of the affairs, participants would accept the alternatives as the final results of the affairs, else they would accept the representation of the negated state of affairs with negative marker (×+ N) and its alternatives as final representation. But ABASM did not explain how did the alternatives be searched and whether all of them would be accepted. Therefore, we would fill the void in the current research. In the three experiments, a sentence–verification paradigm was adopted to explore the processing mechanism of negated sentence with uncertain discrete results. In these experiments, participants were required to hear the sentences, such as the Cartoon is/isn’t Mickey Mouse. Then, 250,750 or 1500ms later after listening the sentences, pictures were presented instantly in the center of the screen. Participants were demanded to press the key to verify whether the objects in the pictures appeared in the former sentences. Three kinds of probing pictures were designed in these experiments. The first kind was anchor picture, short for A, which depicting the negated state of affairs of negated sentences, such as Mickey Mouse for the sentence the Cartoon isn’t Mickey Mouse; the second kind was high-related picture, short for HR, which depicting something with high relationship with anchor picture, such as Donald Duck; the last kind was LR, which depicting something with low relationship with anchor picture, such as Smurf. Results of Experiment 1 shown that both of the data pattern of the affirmative and negated sentences were “HR = LR, A < HR and A < LR”, which suggested that, in the early stage, participants simulate the actual state of affairs directly in affirmative sentences condition, and simulate the negated state of affairs in negated sentences condition. Results of Experiment 2 show that, the data pattern of the affirmative also was “HR = LR, A < HR and A < LR”, and that of negative sentence was “A=HR=LR”, which suggested that the searching stage have finished, furthermore, the search domain covered the LR. Results of Experiment 3 showed that the data pattern of the affirmative also was “HR = LR, A < HR and A < LR” and that of negated sentence was “HR < A, HR < LR and HR = LR”. This finding suggested that, in the 1500ms delay condition, participants only accept HR being the final results and stop the processing. The results of the three experiments indicated that, when processing the negative sentences with uncertain discrete results, participants would first simulate an anchor, then search all the alternates quickly. And if there were the alternates supplied, participants would accept the high-related alternatives being the final result. The results support anchor-based activation and satisfaction constrained model as well as All-search hypothesis and Limited-accept hypothesis.