This review is intended primarily to provide cognitive benchmarks and perhaps a new mindset for behavioral neuroscientists who study memory. Forgetting, defined here broadly as all types of decreases in acquired responding to stimulus-specific eliciting cues, is commonly attributed to one or more of the following families of mechanisms: (1) (4) associative interference by information similar to, but different from the target information, (2) spontaneous decay of memory with increasing retention intervals, (3) displacement from short-term memory by irrelevant information, and (4) inadequate retrieval cues at test. I briefly review each of these families and discuss data suggesting that many apparent instances of spontaneous forgetting and displacement from short-term memory can be viewed as variants of inadequate retrieval cues and associative interference. The potential for recovery of target information from each of these families of forgetting without further relevant training is then reviewed, with a conclusion that most forgetting is due to retrieval failure as opposed to irreversible erasure of memory. The more general point is made that there are logical problems with ever talking about attenuating or erasing a memory as a consequence of conventional forgetting or disrupted consolidation/reconsolidation. Consideration is then given to the frequently overlooked but highly beneficial consequences of most forgetting. Lastly, the major variables that moderate forgetting are summarized, including (a) the similarities of the target information including training context to the explicit retrieval cues and context present at test, (b) the similarities of potentially interfering acquired information to the retrieval cues and context present at test, and (c) the retention interval for the target information relative to that for the potentially interfering information. Appropriate manipulation of these variables can reduce forgetting, and increase forgetting when desired.

这篇综述的主要目的是为研究记忆的行为神经科学家提供认知基准，或许还有一种新的思维方式。遗忘，在这里广义地定义为对刺激特异性诱发线索的获得性反应的所有类型的减少，通常归因于以下一个或多个机制家族：（1）（4）由类似于但不同的信息引起的联想干扰目标信息，（2）随着保留间隔的增加，记忆自发衰减，（3）不相关信息从短期记忆中移位，以及（4）测试时检索线索不足。我简要回顾了这些家庭中的每一个，并讨论了表明许多明显的自发遗忘和短期记忆移位的例子可以被视为检索线索不足和联想干扰的变体。然后回顾了在没有进一步相关培训的情况下从这些遗忘家族中的每一个中恢复目标信息的潜力，得出的结论是，大多数遗忘是由于检索失败而不是不可逆的记忆擦除造成的。更普遍的观点是，由于传统的遗忘或破坏的巩固/重新巩固，曾经谈论衰减或擦除记忆存在逻辑问题。然后考虑大多数遗忘的经常被忽视但非常有益的后果。最后，总结了缓解遗忘的主要变量，包括（a）目标信息的相似性，包括训练上下文与测试中存在的显式检索线索和上下文，（b）潜在干扰获取信息与检索线索和上下文的相似性在测试时，以及 (c) 目标信息相对于潜在干扰信息的保留间隔。适当操纵这些变量可以减少遗忘，并在需要时增加遗忘。(c) 目标信息相对于潜在干扰信息的保留间隔。适当操纵这些变量可以减少遗忘，并在需要时增加遗忘。(c) 目标信息相对于潜在干扰信息的保留间隔。适当操纵这些变量可以减少遗忘，并在需要时增加遗忘。