Sidman (1994, 2000) hypothesized that equivalence relations are a direct outcome of reinforcement contingencies. This theory is problematic because contingencies do not always result in equivalence. Sidman proposed that equivalence relations may conflict with analytic units, the other outcome of contingencies (e.g., in conditional discriminations with common responses/reinforcers). This conflict may result in a generalized class breakdown and a failure to pass equivalence tests. This is more likely in nonhumans, very young humans, etc. The conflict can also result in a selective class breakdown and success in equivalence tests. This occurs after experience shows the organism the necessity and utility of this process. The nature of that experience and the class breakdown processes were not described by Sidman. I explored the implications of the following hypotheses for Sidman’s theory. First, conditional discriminations with a common response/reinforcer result in a generalized class breakdown when participants fail to discriminate emergent relations incompatible with contingencies from those compatible. Second, learning to discriminate between the two requires a history of multiple exemplar training (MET). This implies that equivalence class breakdown is a common response to exemplars that have nothing in common except their relations. This, however, contradicts Sidman’s views about the impossibility of such process in the absence of a complex verbal repertoire. If that type of learning from MET is possible, then the possibility that MET results in the selective formation of equivalence classes must be admitted, and the utility of hypothesizing that equivalence is a direct outcome of reinforcement contingencies can be questioned.

Sidman (1994, 2000) 假设等价关系是强化偶然事件的直接结果。这个理论是有问题的，因为偶然事件并不总是导致等价。西德曼提出，等价关系可能与分析单位、意外事件的其他结果（例如，在具有共同反应/强化物的条件歧视中）发生冲突。这种冲突可能会导致广义的类崩溃和无法通过等效性测试。这种情况在非人类、非常年轻的人类等中更有可能发生。冲突还可能导致选择性的类别崩溃和等价测试的成功。当经验向有机体表明该过程的必要性和实用性之后，就会发生这种情况。西德曼没有描述这种经历的性质和阶级分解过程。我探讨了以下假设对西德曼理论的影响。首先，当参与者无法区分与意外事件不相容的紧急关系和相容的关系时，具有共同反应/强化剂的条件歧视会导致普遍的阶级崩溃。其次，学习区分两者需要多样本训练（MET）的历史。这意味着等价类分解是对除了关系之外没有任何共同点的范例的常见反应。然而，这与西德曼的观点相矛盾，西德曼认为在缺乏复杂的语言能力的情况下这一过程是不可能的。如果从 MET 进行这种类型的学习是可能的，那么必须承认 MET 导致选择性形成等价类的可能性，并且假设等价是强化意外事件的直接结果的效用可能会受到质疑。