In 3 experiments, participants were trained in an associative learning paradigm in which they learned the relation between consumption of certain foodstuffs and the type of allergic reaction shown by a fictional patient. Experiment 1 demonstrated the learned predictiveness effect, showing that cues that had served as good predictors of outcomes in an initial phase of training were especially effective in a test given after a second phase of training in which learning about the same cues, but with different outcomes, had been required. Experiment 2 showed that this effect could be obtained when the two phases of training occurred in reverse order, so that the critical cues were established as good or bad predictors only after the associations tested in the final test had been acquired. This learned predictiveness effect cannot be explained by an enhancement of the associability of the predictive cues that facilitated learning about them in phase two. This encouraged us to consider 2 alternatives to associability for explaining learned predictiveness: (a) that training a cue as a good predictor increases its effective salience, thus enhancing its power to evoke responding on test and (b) that learned predictiveness is the result of a nonattentional process in which subjects integrate information acquired in the separate phases of training. Support for the latter came from Experiment 3, which showed that a modified test procedure, designed to reduce the tendency to integrate across phases, eliminated the learned predictiveness effect. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

中文翻译：

---

解释学习的预测性：注意力和联想结构整合的作用。

在3个实验中，参与者接受了相关学习范例的训练，他们在其中学习了某些食物的消耗与虚构患者表现出的过敏反应类型之间的关系。实验1证明了学习到的预测效果，表明在训练的初始阶段充当良好结果预测指标的线索在训练的第二阶段之后进行的测试中特别有效，在该测试中，学习相同的线索但结果不同，是必需的。实验2表明，当训练的两个阶段以相反的顺序进行时，可以获得这种效果，因此只有在获得了最终测试中测试的关联之后，才能将关键线索确定为好预测或坏预测。无法通过增强有助于在第二阶段中对其进行学习的预测线索的关联性来解释所学到的预测效果。这鼓励我们考虑两种关联性的替代方法来解释学习的预测能力：（a）将线索训练为良好的预测因子可以提高其有效显着性，从而增强其激发测试反应的能力；（b）学习的预测能力是以下结果的结果一个无注意的过程，在该过程中，受试者将在单独的培训阶段中获得的信息进行整合。对后者的支持来自实验3，该实验表明修改后的测试程序旨在减少跨阶段集成的趋势，消除了学习到的预测效果。（PsycINFO数据库记录（c）2019 APA，保留所有权利）。