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     

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.     

学习和记忆的无意识研究

内隐学习和内隐记忆的研究代表了人类学习和记忆的无意识过程。在过去的40年里,内隐学习和内隐记忆的研究经历了：研究对象从人工材料走向真实生活,理论观点从分离走向协同,研究方法从单一走向多样化,以及人工神经网络模型中学习和记忆过程的模拟等。它不仅对学习和记忆本身的心理机制得到了更多的理解,而且还为整个心理学特别是认知心理学的研究开辟了广阔的前景。具体表现为多重记忆的划分、无意识研究的异军突起、研究方法的突破扩展和交叉学科的融会贯通     

材料呈现方式和复习与否对中文电子笔记学习学习效果的影响

笔记策略的“转换假说”指出：将听觉材料写下来比将视觉材料写下来的学习效果更好。本研究以此为基础,以104名大学生为被试,采用多元方差分析的方法,探讨了材料呈现方式和复习与否对拼音键入的中文电子笔记的即时和延时学习效果的影响。结果发现,电子笔记在视觉材料上比在听觉材料上的即时学习效果更优;在间隔一周后的延时重测中,复习后的延时学习效果显著优于不复习,但复习在视觉材料组和听觉材料组中的促进作用无显著差异。     

A tutorial on kernel methods for categorization

The abilities to learn and to categorize are fundamental for cognitive systems, be it animals or machines, and therefore have attracted attention from engineers and psychologists alike. Modern machine learning methods and psychological models of categorization are remarkably similar, partly because these two fields share a common history in artificial neural networks and reinforcement learning. However, machine learning is now an independent and mature field that has moved beyond psychologically or neurally inspired algorithms towards providing foundations for a theory of learning that is rooted in statistics and functional analysis. Much of this research is potentially interesting for psychological theories of learning and categorization but also hardly accessible for psychologists. Here, we provide a tutorial introduction to a popular class of machine learning tools, called kernel methods. These methods are closely related to perceptrons, radial-basis-function neural networks and exemplar theories of categorization. Recent theoretical advances in machine learning are closely tied to the idea that the similarity of patterns can be encapsulated in a positive definite kernel. Such a positive definite kernel can define a reproducing kernel Hilbert space which allows one to use powerful tools from functional analysis for the analysis of learning algorithms. We give basic explanations of some key concepts—the so-called kernel trick, the representer theorem and regularization—which may open up the possibility that insights from machine learning can feed back into psychology.     

Age differences in associative and strategic processes in human conditional learning

Research indicates associative and strategic deficits mediate age related deficits in memory, whereas simple associative processes are independent of strategic processing and strategic processes mediate resistance to interference. The present study showed age-related deficits in a contingency learning task, although older participants' resistance to interference was not disproportionately affected. Recognition memory predicted discrimination, whereas general cognitive ability predicted resistance to interference, suggesting differentiation between associative and strategic processes in learning and memory, and age declines in associative processes. Older participants' generalisation of associative strength from existing to novel stimulus–response associations was consistent with elemental learning theories, whereas configural models predicted younger participants' responses. This is consistent with associative deficits and reliance on item-level representations in memory during later life.     

Dissociable behavioural outcomes of visual statistical learning

ABSTRACT

Statistical learning refers to the extraction of probabilistic relationships between stimuli and is increasingly used as a method to understand learning processes. However, numerous cognitive processes are sensitive to the statistical relationships between stimuli and any one measure of learning may conflate these processes; to date little research has focused on differentiating these processes. To understand how multiple processes underlie statistical learning, here we compared, within the same study, operational measures of learning from different tasks that may be differentially sensitive to these processes. In Experiment 1, participants were visually exposed to temporal regularities embedded in a stream of shapes. Their task was to periodically detect whether a shape, whose contrast was staircased to a threshold level, was present or absent. Afterwards, they completed a search task, where statistically predictable shapes were found more quickly. We used the search task to label shape pairs as “learned” or “non-learned”, and then used these labels to analyse the detection task. We found a dissociation between learning on the search task and the detection task where only non-learned pairs showed learning effects in the detection task. This finding was replicated in further experiments with recognition memory (Experiment 2) and associative learning tasks (Experiment 3). Taken together, these findings are consistent with the view that statistical learning may comprise a family of processes that can produce dissociable effects on different aspects of behaviour.     

Research on intelligent cognitive function enhancement of intelligent robot based on ant colony algorithm

In this paper, ant colony algorithm is studied to improve the visual cognitive function of intelligent robots. Based on the detailed understanding of the research status in this field at home and abroad, and learning from cognitive science and neurobiology research results, a solution is proposed from the perspective of ant colony algorithm based on human brain structure and function. By simulating the process of autonomous learning controlled by human long-term memory and its working memory, a visual strangeness-driven growth long-term memory autonomous learning algorithm is proposed. This method takes incremental self-organizing network as long-term memory structure, and combines with visual strangeness internal motivation Q learning method in working memory. The visual knowledge acquired by self-learning is accumulated into long-term memory continuously, thus realizing the ability of self-learning, memory and intelligence development similar to human beings. The experimental results show that the robot can learn visual knowledge independently, store and update knowledge incrementally, and improve its intelligence development, classification and recognition ability compared with the method without long-term memory. At the same time, the generalization ability and knowledge expansion ability are also improved.     

Name that percussive tune: Associative memory and amplitude envelope†

A series of experiments demonstrated novel effects of amplitude envelope on associative memory, with tones exhibiting naturally decaying amplitude envelopes (e.g., those made by two wine glasses clinking) better associated with target objects than amplitude-invariant tones. In Experiment 1 participants learned associations between household objects and 4-note tone sequences constructed of spectrally matched pure tones with either “flat” or “percussive” amplitude envelopes. Those hearing percussive tones correctly recalled significantly more sequence–object associations. Experiment 2 demonstrated that participants hearing percussive tones learned the associations more quickly. Experiment 3 used “reverse percussive” tones (percussive tones played backwards) to test whether differences in overall energy might account for this effect, finding they did not lead to the same level of performance as percussive tones. Experiment 4 varied the envelope at encoding and retrieval to determine which stage of the task was most affected by the envelope manipulation. Participants hearing percussive tones at both encoding and retrieval performed significantly better than the other three groups (i.e., flat at encoding/percussive at retrieval, etc.). We conclude that amplitude envelope plays an important role in learning and memory, a finding with relevance to psychological research on audition and associative memory, as well as practical relevance for improving human–computer interface design.     

Learning and encoding higher order rules in neural networks

Some researchers state that whereas neural networks are fine for pattern recognition and categorization, complex rule formation requires a separate “symbolic” level. However, the human brain is a connectionist system and, however imperfectly, does complex reasoning and inference. Familiar modeling principles (e.g., Hebbian or associative learning, lateral inhibition, opponent processing, neuromodulation) could recur, in different combinations, in architectures that can learn diverse rules. These rules include, for example, “go to the most novel object,” “alternate between two given objects,” and “touch three given objects, without repeats, in any order.” Frontal lobe damage interferes with learning all three of those rules. Hence, network models of rule learning and encoding should include a module analogous to the prefrontal cortex. They should also include modules analogous to the hippocampus for episode setting and analogous to the amygdala for emotional evaluation.     

Aging neuromodulation impairs associative binding: a neurocomputational account

Relative to young adults, older adults are particularly impaired in episodic memory tasks requiring associative binding of separate components into compound episodes, such as tasks requiring item-context and item-item binding. This associative-binding deficit has been attributed to senescent changes in frontal-hippocampal circuitry but has not been formally linked to impaired neuromodulation involving this circuitry. Previous neurocomputational work showed that impaired neuromodulation could result in less distinct neurocognitive representations. Here we extend this computational principle to simulate aging-related deficits in associative binding. As expected, networks with simulated deficiency in neuromodulation resulted in less distinct internal representations than did networks simulating the processing and performance of young adults, and were also more impaired under task conditions that required associative binding. The findings suggest that senescent changes in neuromodulatory mechanisms may play a basic role in aging-related impairment in associative binding by reducing the efficacy of distributed conjunctive coding.     

Multiple realization and methodological pluralism

Multiple realization was once taken to be a challenge to reductionist visions, especially within cognitive science, and a foundation of the “antireductionist consensus.” More recently, multiple realization has come to be challenged on naturalistic grounds, as well as on more “metaphysical” grounds. Within cognitive science, one focal issue concerns the role of neural plasticity for addressing these issues. If reorganization maintains the same cognitive functions, that supports claims for multiple realization. I take up the reorganization involved in language dysfunctions to deal with questions concerned with multiple realization and neural plasticity. Beginning with Broca’s case for localization and the nineteenth century discussion of “reorganization,” and returning to more recent evidence for neural plasticity, I argue that, in the end, there is substantial support for multiple realization in cognitive systems; I further argue that this is wholly consistent with a recognition of methodological pluralism in cognitive science.     

Mind,Consciousness, and Free Will

This commentary discusses how philosophy and science can collaborate to understand the human mind, considering dialogues involving three philosophers and three cognitive scientists. Their topics include the relation of philosophy and science, the nature of mind, the problem of consciousness, and the existence of free will. I argue that philosophy is more general and normative than science, but they are interdependent. Philosophy can build on the cognitive sciences to develop a theory of mind I call “multilevel materialism,” which integrates molecular, neural, mental, and social mechanisms. Consciousness is increasingly being understood as resulting from neural mechanisms. Scientific advances make the traditional concept of free will implausible, but “freeish” will is consistent with new theories of decision making and action resulting from brain processes. Philosophers should work closely with scientists to address profound problems about knowledge, reality, and values.     

Electrodermal pavlovian conditioning with prepared and unprepared stimuli

One-trial learning referred to by Guthrie has been suggested to occur in autonomic conditioning, if the conditional stimuli (CSs) are so-called prepared ones. To test this idea, half of 28 subjects were given spider or snake slides as “prepared” CSs, while the remainder were given neutral slides as “unprepared” CSs. A shock was employed as the unconditional stimulus (UCS), with a CS-UCS interval of 8 seconds. Electrodermal activity and probe reaction times were the dependent measures of conditioning, conceived in cognitive, information-processing terms as the learning of the CS/UCS contingency. Evidence for the usual CS/UCS contingency learning emerged in both indicators, and during both acquisition and extinction, but none for one-trial learning, perhaps because the UCS was insufficiently aversive.     

Impaired Pavlovian conditioned inhibition in offenders with personality disorders

Certain types of violent offending are often accompanied by evidence of personality disorders (PDs), a range of heterogeneous conditions characterized by disinhibited behaviours that are generally described as impulsive. The tasks previously used to show impulsivity deficits experimentally (in borderline personality disorder, BPD) have required participants to inhibit previously rewarded responses. To date, no research has examined the inhibition of responding based on Pavlovian stimulus–stimulus contingencies, formally “conditioned inhibition” (CI), in PDs. The present study used a computer-based task to measure excitatory and inhibitory learning within the same CI procedure in offenders recruited from the “personality disorder” and the “dangerous and severe personality disorder” units of a high-security psychiatric hospital. These offenders showed a striking and statistically significant change in the expression of inhibitory learning in a highly controlled procedure: The contextual information provided by conditioned inhibitors had virtually no effect on their prepotent associations. Moreover, this difference was not obviously attributable to nonspecific cognitive or motivational factors. Impaired CI would reduce the ability to learn to control associative triggers and so could provide an explanation of some types of offending behaviour.     

Steps to a “Properly Embodied” cognitive science

Cognitive systems research has predominantly been guided by the historical distinction between emotion and cognition, and has focused its efforts on modelling the “cognitive” aspects of behaviour. While this initially meant modelling only the control system of cognitive creatures, with the advent of “embodied” cognitive science this expanded to also modelling the interactions between the control system and the external environment. What did not seem to change with this embodiment revolution, however, was the attitude towards affect and emotion in cognitive science. This paper argues that cognitive systems research is now beginning to integrate these aspects of natural cognitive systems into cognitive science proper, not in virtue of traditional “embodied cognitive science”, which focuses predominantly on the body’s gross morphology, but rather in virtue of research into the interoceptive, organismic basis of natural cognitive systems.     

Computer-based instruction and the practical aspects of memory

This article examines the relationship between basic memory research and computer-based instruction (CBI). The results of basic memory research have helped determine the current directions of computer-assisted instruction (CAI) research and development. These directions have included examining the utility of such learning systems for facilitating memory outcomes and developing CBI systems based on cognitive instructional principles. Examples of such systems include the emerging generation of hypermedia and artificial tutoring systems. Inconclusive results, however, have been found regarding CBI's effects on memroy phenomena. The CBI literature has thus not established the utility of any memory theory for naturally occurring phenomena. Several plausible reasons for this “null finding” are discussed.     

Towards learning agents with personality traits: Modeling Openness to Experience

Recent advances in neurosciences and cognitive sciences show us that the human neocortex is not a slave to the experiences from our perception and that the memories stored in hippocampus are goal weighted during the replay of the experiences for the purpose of re-learning from them. Temporal difference reinforcement learning systems that use neural networks as function approximators rely on an experience replay memory structure similar to the hippocampus. We bring forward this similarity and present a novel way of using a goal weighted prioritization of the memory that is biologically inspired. Furthermore, we introduce a novel prioritization criteria called Variety of Experience Index, or VEI, for weighting the selection of the experiences that are stored in the replay memory. Weighting the experiences based on two different extremes of VEI can behaviourally modify the agent’s learning process, generating different types of learning agents that exhibit different personality traits along the dimension of Openness to Experience.     

情景记忆成功年老化的神经机制

随着年龄的增长, 大部分老年人的情景记忆会出现衰退, 但也会有一部分老年人的情景记忆表现出成功的年老化, 即记忆成绩较好或随增龄的衰退程度较小。脑保持理论、神经去分化理论、认知储备理论以及神经补偿理论分别从不同角度解释了情景记忆成功年老化的神经机制。基于选择性优化与补偿模型对现有理论进行整合, 发现情景记忆成功年老化可能与个体的认知储备水平直接相关：高认知储备的老年人能够对情景记忆相关的脑区和脑网络进行优化且具备更强的神经补偿能力, 因而其脑功能(比如, 神经表征和神经加工通路的特异性)可能会保持地更好。未来研究需要更多地采用纵向设计来考察各理论之间的关系及其影响因素, 从而更好地解释记忆成功年老化的神经机制并为提升老年人的脑与认知健康提供支持。     

“认知科学”与“第二代认知科学”刍议

以计算隐喻为核心假设的传统认知心理学以及联结主义心理学均不能克服离身心智（disembodied mind）的根本缺陷,当代认知心理学正面临着新的范型转换。以具身性和情境性为重要特征的第二代认知科学将日受重视,并促使认知神经科学进入新的发展阶段。作者认为在身心关系上应该坚持生理只是心理的必要条件,而非充分条件的立场,克服生理还原论的危险;应该重新审视基于二元论的生理机制这种说法;心理学传统中的科学主义和人文主义有可能在第二代认知科学强调认知情境性的基础上达成某种融合;第一代认知科学对意识的研究是不成功的,因为对知觉、注意、记忆、思维等心理过程的研究不能代替意识的研究,同时还应避免以意识内容的研究取代心理学研究的倾向。第二代认知科学中的动力系统理论关于变量（因素）之间的偶合（coupling）关系完全不同于变差分析中的变量之间的交互作用关系”,其动力系统模式可能更有助于破解意识的产生（涌现）之谜,并引发心理学研究的方法论的变革新潮。第二代认知科学的兴起将启发人们对身心关系、生理还原论、意识研究在心理学中的地位、人工智能对心智完全模拟的可能性等重大问题重新思考