A hybrid model of learning to solve physics problems

Abstract

The current state of the debate between proponents of connectionism and defenders of traditional, symbolic modelling is used as the basis of an argument for the potential usefulness of hybrid models, in which principles of both connectionist and symbolic models are combined. The possible merits of hybrid models are illustrated by showing how a hybrid model of aspects of expertise development in physics problem solving might be superior to existing models. In the hybrid model, which is capable of solving simple mechanics problems and of learning from its own solutions, a distributed memory module is linked to a classical production system. In the distributed memory module, associations between problem statements and useful subgoals are stored. These subgoals serve to guide the inference process carried out by the production system, making the system's solutions more efficient as learning proceeds. The results of a simulation experiment indicate that the trained system's solutions closely correspond to those generated by human experts, and that the hybrid model probably offers a better account of the development and nature of expert reasoning than existing symbolic models.     

Motivated learning for the development of autonomous systems

A new machine learning approach known as motivated learning (ML) is presented in this work. Motivated learning drives a machine to develop abstract motivations and choose its own goals. ML also provides a self-organizing system that controls a machine’s behavior based on competition between dynamically-changing pain signals. This provides an interplay of externally driven and internally generated control signals. It is demonstrated that ML not only yields a more sophisticated learning mechanism and system of values than reinforcement learning (RL), but is also more efficient in learning complex relations and delivers better performance than RL in dynamically-changing environments. In addition, this paper shows the basic neural network structures used to create abstract motivations, higher level goals, and subgoals. Finally, simulation results show comparisons between ML and RL in environments of gradually increasing sophistication and levels of difficulty.     

自主安全动机与安全绩效的多水平模型

自主安全动机对安全绩效的积极作用已引起研究者的关注, 但自主动机与安全绩效多水平的作用机制还未得到深入研究。结合自我决定理论与多水平动机过程, 个体与团队自主安全动机通过安全目标, 进而影响安全绩效; 在安全工作情境中通过对自主支持、工作重塑的干预促进自主安全动机。研究拟结合认知测量和问卷测量, 通过纵向研究、实验、准实验研究等方法探讨上述问题。研究结果一方面可拓展自我决定理论与安全绩效的研究; 另一方面为安全管理实践, 尤其是安全培训提供理论依据, 对高风险企业安全绩效的改善有积极的现实意义。     

Kano model of autonomous driving user acceptance according to driver characteristics: A survey study

The development of advanced technology has revolutionized human life. In this regard, autonomous driving, a core technology currently being developed, is changing rapidly. In addition to improving technology, the acceptance of technology users must be secured. Most relevant studies conducted hitherto have involved evaluation using acceptance elements defined based on the technology acceptance model and the unified theory of acceptance and use of technology. In this study, 21 elements associated with the acceptance of autonomous driving are defined. The Kano model is used to classify the acceptance elements into five attributes and to propose guidelines for improving acceptance. Driver characteristics are classified based on four human factors, which are used to investigate differences in acceptance between groups. A Google survey and fieldwork were completed by 187 participants. Contrary to previous studies, no significant gender differences are observed in the current study. In terms of age, many obstacles are encountered in securing autonomous driving acceptance from the elderly driver group. Additionally, a more conservative tendency is indicated by people with more driving experience. The results of this study reveal important points for identifying elements that hinder future sustainability and commercialization of autonomous driving, thereby facilitating its further technological development.     

A computer model of inflection learning

This paper presents a computer program that creates and modifies inflectional paradigms. A paradigm is a linguistic device used to index lexical entries that share a common inflection along a given dimension such as tense or case. Paradigms are useful because they offer a convenient method for predicting the form of a novel word and for explaining some aspects of language development. The model presented here is based on Pinker’s (1984) inflection learning theory. The model covers a subset of the rules for learning inflectional paradigms that are outlined in the theory. The primary concern of the model is to understand how elements that are crucial to the theory, such as lexical entries and feature equations, may interact to form and expand word-specific paradigms. The model also represents an attempt to assess implementation details not specified by the theory. A brief overview of the model is given, followed by a sample run. Finally, some practical and theoretical considerations, such as the structure of paradigms, paradigm assignment to novel words, and suppletion, are discussed.     

A process model of the understanding of uncertain conditionals

To build a process model of the understanding of conditionals we extract a common core of three semantics of if-then sentences: (a) the conditional event interpretation in the coherencebased probability logic, (b) the discourse processingtheory of Hans Kamp, and (c) the game-theoretical approach of Jaakko Hintikka. The empirical part reports three experiments in which each participant assessed the probability of 52 if-then sentencesin a truth table task. Each experiment included a second task: An n-back task relating the interpretation of conditionals to working memory, a Bayesian bookbag and poker chip task relating the interpretation of conditionals to probability updating, and a probabilistic modus ponens task relating the interpretation of conditionals to a classical inference task. Data analysis shows that the way in which the conditionals are interpreted correlates with each of the supplementary tasks. The results are discussed within the process model proposed in the introduction.     

Variations on active learning

Although current theories of memory emphasize the fact that learning is an active process, there is little consensus on the role of active learning. This paper attempts to provide a more precise characterization of active learning in the acquisition of computer skills. The analysis of verbal protocols suggests that active learning varies on at least two dimensions: schema orientation and learning strategy. Schema orientation refers to the type of information the learner is acquiring and varies from relational information to procedural information. Learning strategy describes how the information is acquired. Some people prefer to be internally driven and form their own plans for learning. Others are externally driven and allow the instructional materials to guide their learning. These differences need to be taken into account in describing learning and in designing instructional materials.     

A cognitive probability model for learning

A quantitative model for the behavior of albino rats in choice-making situations is presented. The model, which is based upon a cognitive conceptualization of the learning process, is shown to yield predictions which are equivalent to those produced by the linear operator stochastic models at the asymptotic limit but which differ from these during early trials in the learning situation.     

A unified model for perceptual learning

Perceptual learning in adult humans and animals refers to improvements in sensory abilities after training. These improvements had been thought to occur only when attention is focused on the stimuli to be learned (task-relevant learning) but recent studies demonstrate performance improvements outside the focus of attention (task-irrelevant learning). Here, we propose a unified model that explains both task-relevant and task-irrelevant learning. The model suggests that long-term sensitivity enhancements to task-relevant or irrelevant stimuli occur as a result of timely interactions between diffused signals triggered by task performance and signals produced by stimulus presentation. The proposed mechanism uses multiple attentional and reinforcement systems that rely on different underlying neuromodulators. Our model provides insights into how neural modulators, attentional and reinforcement learning systems are related.     

A model for serial verbal learning

A model for analyzing the learning process with a special emphasis on serial-position effect is proposed. This model consists of two analyses, one being an analysis of the learning process of each item in a list by a stochastic method, and the other being an analysis of serial-position effect in terms of pro- and retroactive inhibitions, and of forgetting. The model is experimentally verified, and moreover, it is found that the model permits prediction of the results of many experiments with lists of various lengths and varying difficulty.The author wishes to acknowledge help received during discussion with Prof. T. Indow.     

A vigilance model for latent learning

The author proposes a heuristic model for latent learning. It is concluded that to regard academic learning as qualitatively different from other forms of learning is to deny evolutionary continuity. Academic learning is not a unitary process governed by a single set of parameters. In addition, it is observed that the problem of student motivation may very well turn out to be purely academic. The instructional technique for a captive audience of a class may be so structured as to make the direction of attention irresistible, the performance of a response, when needed, compelling, and the acquisition of knowledge inevitable. Vigilance is an instance of innate foundation. Its most striking characteristics are its universality in the animal world, its ready evocation by a wide range of stimuli, and its apparent behavioral and physiological manifestations. The last two are the natural resources for objective investigation, and the first may well be the basis of broad and valid generalizations.     

A markov model for discrimination learning

A theory for discrimination learning which incorporates the concept of an observing response is presented. The theory is developed in detail for experimental procedures in which two stimuli are employed and two responses are available to the subject. Applications of the model to cases involving probabilistic and nonprobabilistic schedules of reinforcement are considered; some predictions are derived and compared with experimental results.This research was supported by a grant from the National Science Foundation.     

A model of probabilistic category learning.

A new connectionist model (named RASHNL) accounts for many "irrational" phenomena found in nonmetric multiple-cue probability learning, wherein people learn to utilize a number of discrete-valued cues that are partially valid indicators of categorical outcomes. Phenomena accounted for include cue competition, effects of cue salience, utilization of configural information, decreased learning when information is introduced after a delay, and effects of base rates. Experiments 1 and 2 replicate previous experiments on cue competition and cue salience, and fits of the model provide parameter values for making qualitatively correct predictions for many other situations. The model also makes 2 new predictions, confirmed in Experiments 3 and 4. The model formalizes 3 explanatory principles: rapidly shifting attention with learned shifts, decreasing learning rates, and graded similarity in exemplar representation.     

A model of memory in concept learning

A concept learning model was developed and tested in two conjunctive attribute identification tasks. The model includes assumptions about the focus of attention, decision making, and memory for stimulus information and prior decisions. Predictions are made about how S changes his hypothesis following an error. Procedures in both tasks allowed inference of the subject's current hypothesis. The hypothesis selections and error statistics were in the majority of cases accurately predicted by the model. Deviations from predictions on hypothesis sampling occurred for naive Ss but not for trained Ss who were required to state a hypothesis on each trial.     

Application of a model to paired-associate learning

The proposal is made to consider a paired-associate item as becoming conditioned to its correct response in all-or-none fashion, and that prior to this conditioning event the subject guesses responses at random to an unlearned item. These simple assumptions enable the derivation of an extensive number of predictions about paired-associate learning. The predictions compare very favorably with the results of an experiment discussed below.This research was supported by a grant, M-3849, from the National Institutes of Mental Health, United States Public Health Service.     

Perspective taking promotes action understanding and learning

People often learn actions by watching others. The authors propose and test the hypothesis that perspective taking promotes encoding a hierarchical representation of an actor's goals and subgoals-a key process for observational learning. Observers segmented videos of an object assembly task into coarse and fine action units. They described what happened in each unit from either the actor's, their own, or another observer's perspective and later performed the assembly task themselves. Participants who described the task from the actor's perspective encoded actions more hierarchically during observation and learned the task better.