The theoretical framework presented in this article explains expert performance as the end result of individuals ' prolonged efforts to improve performance while negotiating motivational and external constraints. In most domains of expertise, individuals begin in their childhood a regimen of effortful activities (deliberate practice) designed to optimize improvement. Individual differences, even among elite performers, are closely related to assessed amounts of deliberate practice. Many characteristics once believed to reflect innate talent are actually the result of intense practice extended for a minimum of 10 years. Analysis of expert performance provides unique evidence on the potential and limits of extreme environmental adaptation and learning. Our civilization has always recognized exceptional individuals, whose performance in sports, the arts, and science is vastly superior to that of the rest of the population. Speculations on the causes of these individuals ' extraordinary abilities and performance are as old as the first records of their achievements. Early accounts commonly attribute these individuals' outstanding performance to divine intervention, such as the

Goal-directed cognition is often discussed in terms of specialized memory structures like the "goal stack." The goal-activation model presented here analyzes goal-directed cognition in terms of the general memory constructs of activation and associative priming. The model embodies three predictive constraints: (1) the interference level, which arises from residual memory for old goals; (1) the strengthening constraint, which makes predictions about time to encode a new goal; and (3) the priming constraint, which makes predictions about the role of cues in retrieving pending goals. These constraints are formulated algebraically and tested through simulation of latency and error data from the Tower of Hanoi, a means-ends puzzle that depends heavily on suspension and resumption of goals. Implications of the model for understanding intention superiority, postcompletion error, and effects of task interruption are discussed.

This paper presents a skill learning model CLARION. Different from existing models of mostly high-level skill learning that use a top-down approach (that is, turning declarative knowledge into procedural knowledge through practice), we adopt a bottom-up approach toward low-level skill learning, where procedural knowledge develops first and declarative knowledge develops later. Our model is formed by integrating connectionist, reinforcement, and symbolic learning methods to perform on-line reactive learning. It adopts a two-level dual-representation framework (Sun, 1995), with a combination of localist and distributed representation. We compare the model with human data in a minefield navigation task, demonstrating some match between the model and human data in several respects.

It is argued that working memory limitations are best defined in terms of the complexity of relations that can be processed in parallel. Relational complexity is related to processing loads in problem solving, and discriminates between higher animal species, as well as between children of different ages. Complexity is defined by the number of dimensions, or sources of variation, that are related. A unary relation has one argument and one source of variation, because its argument can be instantiated in only one way at a time. A binary relation has two arguments, and two sources of variation, because two argument instantiations are possible at once. Similarly, a ternary relation is three dimensional, a quaternary relation is four dimensional, and so on. Dimensionality is related to number of chunks, because both attributes on dimensions and chunks are independent units of information of arbitrary size. Empirical studies of working memory limitations indicate a soft limit which corresponds to processing one quaternary relation in parallel. More complex concepts are processed by segmentation or conceptual chunking. Segmentation entails breaking tasks into components which do not exceed processing capacity, and which are processed serially. Conceptual chunking entails "collapsing" representations to reduce their dimensionality and consequently their processing load, but at the cost of making some relational information inaccessible. Parallel distributed processing implementations of relational representations show that relations with more arguments entail a higher computational cost, which corresponds to empirical observations of higher processing loads in humans. Empirical evidence is presented that relational complexity discriminates between higher species...

Officer praised dog for barking at object." Enables Detect Drugs out FK Initiates Retrieval 5 6 Missing Figure 10. Forgetting to fill the tank with gas A=actual intention; E=expectation; Q=question; C=context; I=index; G=goal Tank Out of Gas Tank Full Tank Low Fill Tank Should have filled up with gas when tank low Expectation What Action to Do? KEY: G = goal; I = index; C = context; Q = question; E = expectation; A = actual intention Results At Store connections with related concepts. Other learning goals take multiple arguments. For instance, a knowledge differentiation goal (Cox & Ram, 1995) is a goal to determine a change in a body of knowledge such that two items are separated conceptually. In contrast, a knowledge reconciliation goal (Cox & Ram, 1995) is one that seeks to merge two items that were mistakenly considered separate entities. Both expansion goals and reconciliation goals may include or spawn a knowledge organization goal (Ram, 1993) that seeks to reorganize the existing knowledge so that it is made available to the reasoner at the appropriate time, as well as modify the structure or content of a concept itself. Such reorganization of knowledge affects the conditions under which a particular piece of knowledge is retrieved or the kinds of indexes associated with an item in memory.

Coached problem solving is known to be effective for teaching cognitive skills. Simple forms of coached problem solving are used in many ITS. This paper first considers how university physics can be taught via coached problem solving. It then discusses how coached problem solving can be extended to support two other forms of learning: conceptual learning and meta learning.

Cognitive skills acquisition is acquiring the ability to solve problems in intellectual tasks, where success is determined more by the subjects' knowledge than their physical prowess. This chapter reviews reseach conducted in the last ten years on cognitive skill acquisition. It covers the initial stages of acquiring a single principle or rule, the initial stages of acquiring a collection of interacting pieces of knowledge, and the final stages of acquiring a skill, wherein practice causes increases speed and accuracy.

It is now possible to build machine learning systems whose behavior is consistent with data from human students. How can education use such simulated students? Applications that help three user groups are discussed. Teachers can practice the art of tutoring byhaving them teach a simulated student. Using a simulation instead of a real student allows teachers to see how their actions affect that student's knowledge, to undo their actions, and to try their skills on students with varying prior knowledge and learning strategies. Students can learn in collaboration with a simulated student. Because the simulated student can be simultaneously an expert and a colearner, it can scaffold and guide the human's learning in subtle ways. Instructional developers can test their instruction on simulated students. Unlike formativeevaluations with real students, a simulation-based evaluation can indicate exactly what piece of the instruction caused which pieces of knowledge, and thus help developers troubleshoot their instructional designs early in the design process. For each of these three areas of application, inherent technical limitations, existing systems and prospective systems are discussed.

Developers of intelligent tutoring systems would like to know what human tutors do and which activities are responsible for their success in tutoring. We address these questions by comparing episodes where tutoring does and does not cause learning. Approximately 125 hr of tutorial dialog between expert human tutors and physics students are analyzed to see what features of the dialog are associated with learning. Successful learning appears to require that the student reach an impasse. When students were not at an impasse, learning was uncommon regardless of the tutorial explanations employed. On the other hand, once students were at an impasse, tutorial explanations were sometimes associated with learning. Moreover, for different types of knowledge, different types of tutorial explanations were associated with learning different types of knowledge. In principle, advances in Artificial Intelligence (AI) should make it easy to build tutoring systems that emulate human tutors. However, it is not yet clear what human tutors do. Although an initial picture exists due to existing studies, one goal of this research is to add more detail to this picture. We argue that it makes sense Requests for reprints should be sent to Kurt VanLehn, Learning Research and Development Center,

this paper, we describe a framework for goal-driven learning and its relationship to prior and current theories from each of these perspectives.