类别学习多重系统理论的研究与展望

介绍了国外学者为验证类别学习的多重系统理论所做的行为实验研究及主要结果,包括反馈训练、延迟反馈、反应位置、间断的类别刺激分布等因素对信息整合的类别学习的影响,和数字Stroop任务、序列记忆扫描任务、类别数量等因素对基于规则的类别学习的影响;同时,作者指出了实验中没有很好地解决两种类别结构之间的难度差异等问题,并提出了今后需要进一步研究的课题     

Conditioned Inhibition Produced by Extinction of a Conditioned Stimulus

Four experiments used a conditioned taste aversion procedure to examine the potential for CS-alone extinction treatment to produce a conditioned stimulus that possesses inhibitory properties. In Experiment 1, saccharin was paired with LiCl, and then saccharin was presented alone for several trials to produce extensive behavioral extinction. Animals receiving this treatment were retarded in reacquiring conditioned responding to saccharin relative to control subjects receiving conditioning to the flavor for the first time. In Experiment 2, the extinguished saccharin stimulus was shown to decrease conditioned responding to a known excitor when the two stimuli were presented in compound as a summation test. Experiments 3A and 3B replicated the findings of Experiments 1 and 2 while providing evidence that the effects were not due to the differential effects of neophobia during testing. These three experiments revealed that an extinguished conditioned excitor passes retardation and summation tests for conditioned inhibition. Experiment 4 found that extinction of a known excitor was slowed when the excitor was extinguished in compound with a previously extinguished conditioned stimulus. That is, an extinguished CS provided protection from extinction to another CS, a finding also consistent with the view that extinction produces conditioned inhibition.     

学习方式和抑制能力对概率类别学习的影响

采用经典的色词Stroop任务和天气预报任务范式,探讨反馈学习和观察学习两种方式以及不同抑制能力对概率类别学习的影响。实验采用2(学习方式)x2(抑制能力)被试间设计,随机选取204名大学生被试。结果显示:反馈学习的成绩显著高于观察学习;高抑制能力者的成绩显著高于低抑制能力者;反馈学习的成绩与意识知识出现分离。这表明,学习方式影响概率类别学习,反馈学习的效果优于观察学习;抑制能力影响概率类别学习,高抑制能力者的学习效果更佳;概率类别学习中同时存在内隐和外显两种学习系统。     

认知神经和神经心理角度的多重类别学习系统

归类是一项非常重要的认知技能,近十多年来归类的研究尤其关注类别学习的多重系统问题.本文回顾了近年来国外对类别学习的认知神经和神经心理研究,结果表明不同的类别学习任务下的学习机制和所依赖的神经生理基础是不同的,支持类别学习的多系统理论.     

Birth of an Abstraction: A Dynamical Systems Account of the Discovery of an Elsewhere Principle in a Category Learning Task

Human participants and recurrent (“connectionist”) neural networks were both trained on a categorization system abstractly similar to natural language systems involving irregular (“strong”) classes and a default class. Both the humans and the networks exhibited staged learning and a generalization pattern reminiscent of the Elsewhere Condition (Kiparsky, 1973). Previous connectionist accounts of related phenomena have often been vague about the nature of the networks’ encoding systems. We analyzed our network using dynamical systems theory, revealing topological and geometric properties that can be directly compared with the mechanisms of non‐connectionist, rule‐based accounts. The results reveal that the networks “contain” structures related to mechanisms posited by rule‐based models, partly vindicating the insights of these models. On the other hand, they support the one mechanism (OM), as opposed to the more than one mechanism (MOM), view of symbolic abstraction by showing how the appearance of MOM behavior can arise emergently from one underlying set of principles. The key new contribution of this study is to show that dynamical systems theory can allow us to explicitly characterize the relationship between the two perspectives in implemented models.     

Conditioned Inhibition of Cyclophosphamide-Induced Taste Aversion

Rats in an experimental group received trials during which 1 flavor (saccharin) was always followed by cyclophosphamide, an immunosuppressive drug, but another (vanilla) was not. An unconditioned stimulus-only group served as a control. Flavor-preference tests revealed that conditioned excitation and conditioned inhibition occurred in the conditioned group subjects but not in the control group subjects. This demonstration suggests that a conditioned inhibitor might be used to modify conditioned and unconditioned immune system functions, for example, natural killer-cell activity.     

Implicit Learning in a Simple Cued Reaction-Time Task

Implicit learning tasks usually involve the learning of complex rules. While this does reduce the likelihood of subjects becoming aware of the relationship to be learned, it also raises the possibility of explaining improved performance in terms of explicit processes. The current experiments are an attempt to develop a task which shows evidence of implicit learning, but which involves the learning of a very simple rule and so avoids these alternative explanations. In two experiments, we exposed subjects to learning trials in which a target letter (or shape) was immediately preceded by a cue letter (or shape) in otherwise random nine-letter (or 15-shape) sequences. In a test phase, subjects responded more quickly to cued than uncued targets if the learning phase had involved reliable cue–target pairings, but not following random control pairings. This was true of subjects who were classified as aware and those classified as unaware of the cue–target relationship.     

类别学习中的分类和推理

该文介绍了类别学习中的分类和推理两种任务,并从学习的条件、过程、结果和发展等方面的归纳了当前研究的最新进展。表明了类别的分类学习和推理学习有相同的形式,但学习的信息处理过程和学习的结果不同。分类学习关注类别间的区分性信息,更可能是样例学习结果;推理学习更为关注单个类别内部的共同性信息,更可能是原型学习结果。这方面的结论强化了基于解释的观点。     

Flight Feedback and Learning a Projectile Task

Over KR acquisition trials of projecting an object a criterion distance, the hypotheses were tested that feedback from the flight of a projected object (flight feedback) would facilitate response recognition but not recall. Visual and auditory flight feedback variables each had two levels, flight feedback present or absent, forming a 2 × 2 × 61 trials design with 14 Ss per cell. Results supported the hypotheses, with flight flight feedback not facilitating response recall, though both flight feedback variables were strong determinants of response recognition. This latter finding was qualified by a discussion of the distinction between response recognition and recognition of response outcome.     

6岁儿童的类别学习能力:类别表征、注意和分类策略

探讨了6岁儿童的类别学习能力、类别表征和分类策略。62名儿童参加了实验,实验1采用了"5/4模型"类别结构,实验2采用了"3/3类别结构"。结果发现:6岁儿童已经具备了一定的类别学习能力;相对于原型表征,6岁儿童更倾向于进行样例表征;6岁儿童在注意上具有定位在高典型性特征维度上的能力,但不具备定位在区分性特征维度上的能力;在类别学习策略上主要采用单维度分类策略和规则加例外的分类策略。     

Retention of Conditioned Inhibition Produced by Extinction

Rats in a flavor aversion procedure were used to examine the extent to which an extinguished CS can pass a retardation and summation test for conditioned inhibition after either a short or a long retention interval. Rats were given a single pairing of a flavored solution with LiCl, and then received a large number of CS-alone extinction trials. Following extinction, some subjects received a 3-week retention interval following extinction while others did not. Extinction produced evidence of conditioned inhibition on both tests when a short retention interval occurred following extinction. However, conditioned inhibition was attenuated by the long retention interval when assessed with a retardation test, but no effect of the long retention interval was observed when inhibition was evaluated with a summation test.     

类别学习的SUSTAIN模型

SUSTAIN模型是关于类别学习的类群或群集规则的选择模型,它强调多层次的类别子结构和模型的结构搜索功能.模型的运行以类别相似性为基础,从最简单的规则开始,有灵活的参数及其运算过程,与其他主要的类别学习模型相比有更多的优势.因此,SUSTAIN模型是迄今解释人类类别学习的最优模型.     

关系类别及其学习机制

关系类别是指类别的成员身份由一个共同的关系结构决定。其意义来自于其与其他实体的关系或其内部的一系列成分间的关系,关系类别强调关系结构的系统性。文章重点介绍和评述了相似性种类,关系类别和实体类别的联系和差异,关系类别的学习过程及机制———结构—映射理论。未来的研究方向主要是探索新的研究范式,以深入理解关系类别的学习;深入探讨校准和映射过程与其他基本认知过程的相互作用,为关系类别的学习提供更一致的认知观点。     

基于规则的类别学习和信息整合的类别学习

该文介绍了类别学习的双重系统理论:基于规则的类别学习指的是学习一种可以用言语表述的规则,是一种外显学习系统;信息整合的类别学习指的是学习一种难以用言语表述的规则,是一种内隐学习系统。同时介绍了类别学习的主要理论模型,以及支持双系统理论的实证研究,包括行为、临床和认知神经实验的研究结果。     

类别学习中分类学习与推理学习的对比研究

本研究对类别学习中分类学习与推理学习进行了对比.实验中要求被试从成对呈现的项目中选择一个更典型的A,并且要求被试将典型的A和典型的B画出来.测试中所呈现项目诊断性程度或典型性程度不同.研究结果表明,分类学习者主要受诊断性信息的影响,而推理学习者主要受典型性信息的影响.即,分类学习者关注类别间信息,而推理学习者关注类别内信息.     

非监控类别学习的分类策略与表征

类别学习是人类对不同类别加以归类的过程。类别信息的表征、分类策略运用的特点一直是类别学习研究的重点。非监控类别学习可分为直接的非监控类别学习和间接的非监控类别学习。直接的非监控类别学习(非限制任务, 限制任务)中被试的分类策略具有分类“单维度倾向”策略特点,类别变异程度会影响类别表征; 间接的非监控类别学习更倾向形成相似性表征, 直接的非监控类别学习则为基于规则表征。现有的非监控类别学习的理论对分类策略和表征的解释仍显薄弱, 不同学习任务下类别迁移和知识效应的研究还存在不足, 未来研究还需要进一步验证知识效应对非监控类别学习的认知加工过程的影响、探索影响类别表征形成的因素等问题。     

Computer-Based Performance Monitoring and Productivity in a Multiple Task Environment

Computer-based performance monitoring (CPM) provides managers with the ability to continuously, and unobtrusively, monitor the work performed by their employees. This paper examines the impact that CPM has on productivity when people are monitored on only a portion of the work that they perform. In a simulated work setting, subjects worked on two computerized tasks, and were led to believe that their work on one, both, or none of the tasks would be electronically monitored. People who were monitored only on a relatively simple task tended to work at a faster rate on both their monitored and non-monitored tasks, in comparison to people who were not monitored at all. People who were monitored only on a task that was more moderate in level of difficulty did not work any faster or more accurately than people who were not monitored. The results of the current study suggests that under certain conditions, employees may not discriminate between monitored and non-monitored work. Recommendations are offered to managers who are considering implementing CPM in their workplace.     

Comparison Versus Contrast: Task Specifics Affect Category Acquisition

A large literature has documented that comparison and contrast lead to better performance in a variety of tasks. However, studies of comparison and contrast present contradictory conclusions as to when and how these processes benefit learners. Across four studies, we examined how the specifics of the comparison and contrast task affect performance by systematically manipulating the feature variation and category structure in a category extension task performed by 3‐year‐old and 4‐year‐old children. Studies 1 (n = 48, M = 42.6 months) and 2 (n = 48, M = 42. 4 months) investigated comparison and contrast with high‐density categories. Studies 3A (n = 60, M = 43.47 months), 3B (n = 48, M = 53. 2 months) and 4 (n = 48, M = 53.7 months) investigated comparison and contrast with low‐density categories. Results indicated both category structure and feature variation affect the efficacy of comparison and contrast. Copyright © 2012 John Wiley & Sons, Ltd.     

类别学习的阻碍效应与双机制理论

通过对类别学习中的阻碍效应进行系统研究,尝试性的提出了类别学习的双机制理论。三个实验分别考察：当样例特征随机呈现时;定义特征固定呈现在样例首位时;以及刺激材料为知觉图形材料时,类别学习中的阻碍效应。三个实验的研究结果都发现：在类别学习中存在一定程度的阻碍效应,支持类别学习同时存在联结学习机制和认知学习机制的双机制观点。     

Sensory Feedback in the Learning of a Novel Motor Task

The role of different forms of feedback is examined in learning a novel motor task. Five groups of ten subjects had to learn the voluntary control of the abduction of the big toe, each under a different feedback condition (proprioceptive feedback, visual feedback, EMG feedback, tactile feedback, force feedback). The task was selected for two reasons. First, in most motor learning studies subjects have to perform simple movements which present hardly any learning problem. Second, studying the learning of a new movement can provide useful information for neuromuscular reeducation, where patients often also have to learn movements for which no control strategy exists. The results show that artificial sensory feedback (EMG feedback, force feedback) is more powerful than “natural” (proprioceptive, visual, and tactile) feedback. The implications of these results for neuromuscular reeducation are discussed.