Stagewise cognitive development: an application of catastrophe theory.

In this article an overview is given of traditional methodological approaches to stagewise cognitive developmental research. These approaches are evaluated and integrated on the basis of catastrophe theory. In particular, catastrophe theory specifies a set of common criteria for testing the discontinuity hypothesis proposed by Piaget. Separate criteria correspond to distinct methods used in cognitive developmental research. Such criteria are, for instance, the detection of spurts in development, bimodality of test scores, and increased variability of responses during transitional periods. When a genuine stage transition is present, these criteria are expected to be satisfied. A revised catastrophe model accommodating these criteria is proposed for the stage transition in cognitive development from the preoperational to the concrete operational stage.     

U-shaped learning and frequency effects in a multi-layered perceptron: implications for child language acquisition

A three-layer back-propagation network is used to implement a pattern association task in which four types of mapping are learned. These mappings, which are considered analogous to those which characterize the relationship between the stem and past tense forms of English verbs, include arbitrary mappings, identity mappings, vowel changes, and additions of a suffix. The degree of correspondence between parallel distributed processing (PDP) models which learn mappings of this sort (e.g., Rumelhart & McClelland, 1986, 1987) and children's acquisition of inflectional morphology has recently been at issue in discussions of the applicability of PDP models to the study of human cognition and language (Pinker & Mehler, 1989; Bever, in press). In this paper, we explore the capacity of a network to learn these types of mappings, focusing on three major issues. First, we compare the performance of a single-layered perceptron similar to the one used by Rumelhart and McClelland with a multi-layered perceptron. The results suggest that it is unlikely that a single-layered perceptron is capable of finding an adequate solution to the problem of mapping stems and past tense forms in input configurations that are sufficiently analogous to English. Second, we explore the input conditions which determine learning in these networks. Several factors that characterize linguistic input are investigated: (a) the nature of the mapping performed by the network (arbitrary, suffixation, identity, and vowel change); (b) the competition effects that arise when the task demands simultaneous learning of distinct mapping types; (c) the role of the type and token frequency of verb stems; and (d) the influence of phonological subregularities in the irregular verbs. Each of these factors is shown to have selective consequences on both successful and erroneous performance in the network. Third, we outline several types of systems which could result in U-shaped acquisition, and discuss the ways in which learning in multi-layered networks can be seen to capture several characteristics of U-shaped learning in children. In general, these models provide information about the role of input in determining the kinds of errors that a network will produce, including the conditions under which rule-like behavior and U-shaped learning will and will not emerge. The results from all simulations are discussed in light of behavioral data on children's acquisition of the past tense and the validity of drawing conclusions about the acquisition of language from models of this sort.     

Repetition priming of words, pseudowords, and nonwords

In 5 experiments, the authors assessed repetition priming for words, pseudowords, and nonwords using a task that combines an implicit perceptual fluency measure and a recognition memory assessment for each list item. Words and pseudowords generated a consistently strong repetition effect even when there was a failure to recognize the stimulus. In 2 of the experiments, the repetition effect for nonwords was reliably above chance even when there was a failure to recognize the stimulus. The authors propose a parallel distributed processing (PDP) model based on the work of J. McClelland and D. Rumelhart (1985) as a way to understand the mechanisms potentially responsible for the pattern of findings. Although the error-driven nature of learning in the model results in a poor fit to the nonword priming data, this is not endemic to all PDP models. Using a model based on Hebbian learning, the authors instantiate a property that they believe is characteristic of implicit memory--that learning is primarily based on the strengthening of connections between units that become active during the processing of a stimulus. This model provides a far more satisfactory account of the data than does the error-driven model.     

Phonology: A Review and Proposals from a Connectionist Perspective

A parallel distributed processing (PDP) model of phonological processing is developed, including components to support repetition, auditory processing, comprehension, and language production. From the performance of the PDP reading model of Plaut, McClelland, Seidenberg, and Patterson (1996), it is inferred that the acoustic-articulatory motor pattern associator that supports repetition provides the basis for phonological sequence knowledge. From the observation that many patients make phonemic paraphasic errors in language production, as in repetition, it is argued that there must be a direct link between distributed concept representations (lexical semantic knowledge) and this network representation of sequence knowledge. In this way, both lexical semantic and phonotactic constraints are brought to bear on language production. The literature on phonological function in normal subjects (slip-of-the-tongue corpora) and in patients with aphasia is critically reviewed from this perspective. The relationship between acoustic and articulatory motor representations in the process of phonetic perception is considered. Repetition and reproduction conduction aphasia are reviewed in detail and extended consideration is given to the representation of auditory verbal short-term memory in the model. Finally, the PDP model is reconciled with information processing models of phonological processing, including that of Lichtheim, and with current knowledge of the anatomic localization of phonological processing. Although no simulations of the model were run, a number of simulation studies are proposed.     

On simulating one-trial learning using morphological neural networks

“Learning once, remembering forever”, this wonderful cognitive phenomenon sometimes occurs in the learning process of human beings. Psychologists call this psychological phenomenon “one-trial learning”. The traditional artificial neural networks can simulate the psychological phenomenon of “implicit learning”, but can’t simulate the cognitive phenomenon of “one-trial learning”. Therefore, cognitive psychology gives a challenge to the traditional artificial neural networks. From two aspects of theory and practice in this paper, the possibility of simulating this kind of psychological phenomenon was explored by using morphological neural networks. This paper takes advantage of morphological associative memory networks to realize the simulation of “one-trial learning” for the first time, and gives 5 simulating practical examples. Theoretical analysis and simulation experiments show that the morphological associative memory networks are a higher effective machine learning method, and can better simulate the cognitive phenomenon of “one-trial learning”, therefore provide a theoretical basis and technological support for the study of intelligent science and cognitive science.     

The Newell Test for a theory of cognition

Newell (1980; 1990) proposed that cognitive theories be developed in an effort to satisfy multiple criteria and to avoid theoretical myopia. He provided two overlapping lists of 13 criteria that the human cognitive architecture would have to satisfy in order to be functional. We have distilled these into 12 criteria: flexible behavior, real-time performance, adaptive behavior, vast knowledge base, dynamic behavior, knowledge integration, natural language, learning, development, evolution, and brain realization. There would be greater theoretical progress if we evaluated theories by a broad set of criteria such as these and attended to the weaknesses such evaluations revealed. To illustrate how theories can be evaluated we apply these criteria to both classical connectionism (McClelland & Rumelhart 1986; Rumelhart & McClelland 1986b) and the ACT-R theory (Anderson & Lebiere 1998). The strengths of classical connectionism on this test derive from its intense effort in addressing empirical phenomena in such domains as language and cognitive development. Its weaknesses derive from its failure to acknowledge a symbolic level to thought. In contrast, ACT-R includes both symbolic and sub-symbolic components. The strengths of the ACT-R theory derive from its tight integration of the symbolic component with the sub-symbolic component. Its weaknesses largely derive from its failure, as yet, to adequately engage in intensive analyses of issues related to certain criteria on Newell's list.     

A connectionist model of a continuous developmental transition in the balance scale task

A connectionist model of the balance scale task is presented which exhibits developmental transitions between ‘Rule I’ and ‘Rule II’ behavior [Siegler, R. S. (1976). Three aspects of cognitive development. Cognitive Psychology,8, 481–520.] as well as the ‘catastrophe flags’ seen in data from Jansen and van der Maas [Jansen, B. R. J., & van der Maas, H. L. J. (2001). Evidence for the phase transition from Rule I to Rule II on the balance scale task. Developmental Review, 21, 450–494]. The model extends a connectionist model of this task [McClelland, J. L. (1989). Parallel distributed processing: Implications for cognition and development. In R. G. M. Morris (Ed.), Parallel distributed processing: Implications for psychology and neurobiology (pp. 8–45). Oxford: Clarendon Press] by introducing intrinsic variability into processing and by allowing the network to adapt during testing in response to its own outputs. The simulations direct attention to several aspects of the experimental data indicating that children generally show gradual change in sensitivity to the distance dimension on the balance scale. While a few children show larger changes than are characteristic of the model, its ability to account for nearly all of the data using continuous processes is consistent with the view that the transition from Rule I to Rule II behavior is typically continuous rather than discrete in nature.     

Attaining landmark status: Rumelhart and McClelland's PDP Volumes and the Connectionist Paradigm

In 1986, David Rumelhart and James McClelland published their two‐volume work, Parallel distributed processing: Explorations in microcognition, Volume 1: Foundations and Volume 2: Psychological and biological models. These volumes soon become classic texts in both connectionism, specifically, and in the cognitive science field more generally. Drawing on oral histories, book reviews, translations, citation records, and close textual analysis, this paper analyzes how and why they attained landmark status. It argues that McClelland and Rumelhart's volumes became classics largely as a result of a confluence of rhetorical factors. Specifically, the PDP Volumes appeared at a kairotic moment in the history of connectionism, publishing dynamics that facilitated their circulation played an important role, and the volumes were ambiguous about the relationship between model and brain in a manner that enabled them to address an expansive audience. In so doing, this paper offers insight into both the history of cognitive science and rhetoric's role in establishing classic texts.     

Parallel distributed processing and neuropsychology: A neural network model of Wisconsin card sorting and verbal fluency

Neural networks can be used as a tool in the explanation of neuropsychological data. Using the Hebbian Learning Rule and other such principles as competition and modifiable interlevel feedback, researchers have successfully modeled a widely used neuropsychological test, the Wisconsin Card Sorting Test. One of these models is reviewed here and extended to a qualitative analysis of how verbal fluency might be modeled, which demonstrates the importance of accounting for the attentional components of both tests. Difficulties remain in programming sequential cognitive processes within a parallel distributed processing (PDP) framework and integrating exceedingly complex neuropsychological tests such as Proverbs. PDP neural network methodology offers neuropsychologists co-validation procedures within narrowly defined areas of reliability and validity.     

内隐学习的人工神经网络模型

近年来,人工神经网络模型常被用来模拟各种心理活动,从而为心理学的一些相关理论提供丰富的证据,内隐学习也不例外。基于权重调整来学习正确反应的人工神经网络模型和内隐学习的两大本质特征间有着极为相应的匹配,因此,人工神经网络模型特别适用于内隐学习研究。到目前为止,针对两种较为普遍的内隐学习任务,已经相应地出现了两种使用较为广泛的神经网络模型——自动联系者和简单循环网络     

Active learning: effects of core training design elements on self-regulatory processes, learning, and adaptability

This article describes a comprehensive examination of the cognitive, motivational, and emotional processes underlying active learning approaches; their effects on learning and transfer; and the core training design elements (exploration, training frame, emotion control) and individual differences (cognitive ability, trait goal orientation, trait anxiety) that shape these processes. Participants (N = 350) were trained to operate a complex, computer-based simulation. Exploratory learning and error-encouragement framing had a positive effect on adaptive transfer performance and interacted with cognitive ability and dispositional goal orientation to influence trainees' metacognition and state goal orientation. Trainees who received the emotion-control strategy had lower levels of state anxiety. Implications for development of an integrated theory of active learning, learner-centered design, and research extensions are discussed.     

Competitive Learning: From Interactive Activation to Adaptive Resonance

Functional and mechanistic comparisons are made between several network models of cognitive processing: competitive learning, interactive activation, adaptive resonance, and back propagation. The starting point of this comparison is the article of Rumelhart and Zipser (1985) on feature discovery through competitive learning. All the models which Rumelhart and Zipser (1985) have described were shown in Grossberg (1976b) to exhibit a type of learning which is temporally unstable. Competitive learning mechanisms can be stabilized in response to an arbitrary input environment by being supplemented with mechanisms for learning top-down expectancies, or templates; for matching bottom-up input patterns with the top-down expectancies; and for releasing orienting reactions in a mismatch situation, thereby updating short-term memory and searching for another internal representation. Network architectures which embody all of these mechanisms were called adaptive resonance models by Grossberg (1976c). Self-stabilizing learning models are candidates for use in real-world applications where unpredictable changes can occur in complex input environments. Competitive learning postulates are inconsistent with the postulates of the interactive activation model of McClelland and Rumelhart (1981), and suggest different levels of processing and interaction rules for the analysis of word recognition. Adaptive resonance models use these alternative levels and interaction rules. The self-organizing learning of an adaptive resonance model is compared and contrasted with the teacher-directed learning of a back propagation model. A number of criteria for evaluating real-time network models of cognitive processing are described and applied.     

汉语句子格角色分配的一种神经网络方法

提出了一个基于分布式表征的计算模型,通过并行分布加工方式完成六类汉语句子的格角色分配任务。模型是一个四层的前传网络,包括输入层（词的分布式表征层）,两个隐层,输出层（格角色层）;其中第一隐层的一部分反馈到输入层。模型采用误差反传算法,通过提供学习样本和目标输出,不断调整三个权值矩阵,使得网络稳定时能得到正确的结果。经过训练后的网络具有一定的稳定性和鲁棒性。还对这种方法与传统的符号处理方法作了比较和分析。     

A Dynamic Model of Cognitive Growth in a Population: Spatial Tasks and Conservation

A dynamic model of cognitive growth is developed that is applicable to cross-sectional studies of growth in a population, and that predicts the distribution of scores as a function of age. This model modifies the theory of P. van Geert (1991, Psychological Review, 98, 3-53) for the cognitive growth of an individual under limited resources, by taking into account the effect of schooling or training. The modified theory provides an explanation for the puzzling effect reported by C. J. Brainerd (1977, Psychological Bulletin, 84, 919-939), that the gain in scores obtained in concept training experiments is independent of the initial state of the learner. In developing the modified theory, it was found that the assumption of delayed feedback is unnecessary. To apply the model to populations, it was assumed that the ability parameters that appear in the growth function are distributed normally in the population. A distribution function G(s; t) was derived that predicts the distributions of scores s at age t in terms of a small number of parameters. Although the abilities are assumed to be distributed normally, the score distributions are often bimodal. The model was applied to the studies of H. Thomas and A. Lohaus (1993, Monographs of the Society for Research in Child Development, 58, 1-169) on the growth of the concepts of verticality and horizontality, and to the study of P. M. Bentler (1970, Perceptual and Motor Skills, 31, 855-859) on the growth of the concept of conservation. In these applications the score distributions are clearly bimodal, but there is no evidence of abrupt transitions or catastrophes. This shows that the presence of a "catastrophe flag" (bimodal distributions) does not necessarily imply the presence of a catastrophe. Copyright 1999 Academic Press.     

Social network schemas and the learning of incomplete networks

Social networks that are missing relations among some of their members--termed incomplete networks--have been of critical theoretical and empirical interest in sociological research on weak ties and structural holes but typically have been overlooked in social psychological studies of network learning. Five studies tested for schematic processing differences in the encoding and recalling of incomplete networks. In Studies 1 and 2, prior knowledge of missing relations facilitated learning an unfamiliar, incomplete network. Study 3 ruled out differences in general pattern recognition ability as an explanation. Study 4 manipulated the degree of familiarity with missing relations, which produced predicted differences in learning rates. Finally, Study 5 examined how improved learning of an incomplete network affected a strategic organizational choice. The findings suggest that people can become schematic for complex, incomplete social networks.     

A Symbolic Model of the Nonconscious Acquisition of Information

This article presents counter evidence against Smolensky's theory that human intuitive/nonconscious congnitive processes can only be accurately explained in terms of subsymbolic computations carried out in artificial neural networks. We present symbolic learning models of two well-studied, complicated cognitive tasks involving nonconscious acquisition of information: learning production rules and artificial finite state grammars. Our results demonstrate that intuitive learning does not imply subsymbolic computation, and that the already well-established, perceived correlation between “conscious” and “symbolic” on the one hand, and between “nonconscious” and “subsymbolic” on the other, does not exist.     

Challenging the widespread assumption that connectionism and distributed representations go hand-in-hand

One of the central claims associated with the parallel distributed processing approach popularized by D.E. Rumelhart, J.L. McClelland and the PDP Research Group is that knowledge is coded in a distributed fashion. Localist representations within this perspective are widely rejected. It is important to note, however, that connectionist networks can learn localist representations and many connectionist models depend on localist coding for their functioning. Accordingly, a commitment to distributed representations should be considered a specific theoretical claim regarding the structure of knowledge rather than a core principle, as often assumed. In this paper, it is argued that there are fundamental computational and empirical challenges that have not yet been addressed by distributed connectionist theories that are readily accommodated within localist approaches. This is highlighted in the context of modeling word and nonword naming, the domain in which some of the strongest claims have been made. It is shown that current PDP models provide a poor account of naming monosyllable items, and that distributed representations make it difficult for these models to scale up to more complex language phenomena. At the same time, models that learn localist representations are shown to hold promise in supporting many of the core reading and language functions on which PDP models fail. It is concluded that the common rejection of localist coding schemes within connectionist architectures is premature.     

融合视角下创造力的认知神经机制

创造力的认知神经机制是近年来心理学研究领域的前沿和热点问题。通过融合创造力整体宏观视角和创造性产生过程的微观视角,对创造力的认知神经机制进行了综述。宏观视角下,创造力主要涉及α波和大脑前额叶、内外侧颞叶以及外侧顶叶; 微观视角下,在创造力产生过程中主要涉及α波序列位置效应以及默认网络和执行控制网络的功能耦合。未来研究方向应该结合多模态脑成像数据库,利用机器算法来探究创造力的本质; 关注青少年群体创造力的纵向发展趋势; 结合分子遗传学研究,探究与创造力有关的基因问题。     

远距离规则的内隐学习使用了何种记忆存储器：来自神经网络模拟的证据

关于远距离规则的知识是如何被内隐学习的,研究尚未得出结论。该研究通过采用和人类被试相同的实验材料和程序,考察了简单循环网络模型（SRN）对两种汉语声调远距离规则——倒映和逆行规则的内隐学习。结果发现：1.在广泛的参数范围上,SRN能够学会倒映和逆行规则,表明模型的记忆缓冲器可以模拟人类远距离规则的内隐学习;2.SRN对倒映规则的学习比对逆行规则的学习更好,表明在功能上远距离规则的内隐学习可能优先使用了先进先出的记忆存储器及信息加工模式。该研究为探究远距离规则内隐学习的机制提供了新的证据和视角。