Results are presented that demonstrate the learning and fine-tuning of search strategies using connectionist mechanisms. Previous studies of strategy learning within the symbolic, production-rule formalism have not addressed fine-tuning behavior. Here a two-layer connectionist system is presented that develops its search from a weak to a task-specific strategy and fine-tunes its performance. The system is applied to a simulated, realtime, balance-control task. We compare the performance of one-layer and two-layer networks, showing that the ability of the two-layer network to discover new features and thus enhance the original representation is critical to solving the balancing task.

We formalize a model for supervised learning of action strategies in dynamic stochastic domains and show that PAC-learning results on Occam algorithms hold in this model as well. We then identify a class of rule-based action strategies for which polynomial time learning is possible. The representation of strategies is a generalization of decision lists; strategies include rules with existentially quantified conditions, simple recursive predicates, and small internal state, but are syntactically restricted. We also study the learnability of hierarchically composed strategies where a subroutine already acquired can be used as a basic action in a higher level strategy. We prove some positive results in this setting, but also show that in some cases the hierarchical learning problem is computationally hard. 1 Introduction We formalize a model for supervised learning of action strategies in dynamic stochastic domains, and study the learnability of strategies represented by rule-based syste...

Abstract. An architecture is described for designing systems that acquire and manipulate large amounts of unsystematized, or so-called commonsense, knowledge. Its aim is to exploit to the full those aspects of computational learning that are known to offer powerful solutions in the acquisition and maintenance of robust knowledge bases. The architecture makes explicit the requirements on the basic computational tasks that are to be performed and is designed to make these computationally tractable even for very large databases. The main claims are that (i) the basic learning and deduction tasks are provably tractable and (ii) tractable learning offers viable approaches to a range of issues that have been previously identified as problematic for artificial intelligence systems that are programmed. Among the issues that learning offers to resolve are robustness to inconsistencies, robustness to incomplete information and resolving among alternatives. Attribute-efficient learning algorithms, which allow learning from few examples in large dimensional systems, are fundamental to the approach. Underpinning the overall architecture is a new principled approach to manipulating relations in learning systems. This approach, of independently quantified arguments, allows propositional learning algorithms to be applied systematically to learning relational concepts in polynomial time and in a modular fashion.

Suppose that we wish to learn from examples and counter-examples a criterion for recognizing whether an assembly of wooden blocks constitutes an arch. Suppose also that we have preprogrammed recognizers for various relationships e.g. on-top-of(x; y), above(x; y), etc. and believe that some possibly complex expression in terms of these base relationships should suffice to approximate the desired notion of an arch. How can we formulate such a relational learning problem so as to exploit the benefits that are demonstrably available in propositional learning, such as attribute-efficient learning by linear separators, and error-resilient learning? We believe that learning in a general setting that allows for multiple objects and relations in this way is a fundamental key to resolving the following dilemma that arises in the design of intelligent systems: Mathematical logic is an attractive language of description because it has clear semantics and sound proof procedures. However, as a basis for large programmed systems it leads to brittleness because, in practice, consistent usage of the various predicate names throughout a system cannot be guaranteed, except in application areas such as mathematics where the viability of the axiomatic method has been demonstrated independently. In this paper we develop the following approach to circumventing this dilemma. We suggest that brittleness can be overcome by using a new kind of logic in which each statement is learnable. By allowing the system to learn rules empirically from the environment, relative to any particular programs it may have for recognizing some base predicates, we enable the system to acquire a set of statements approximately consistent with each other and with the world, without the need for a globally knowledgeable and consistent programmer. We illustrate

The authors propose a reinforcement-learning mechanism as a model for recurrent choice and extend it to account for skill learning. The model was inspired by recent research in neurophysiological studies of the basal ganglia and provides an integrated explanation of recurrent choice behavior and skill learning. The behavior includes effects of differential probabilities, magnitudes, variabilities, and delay of reinforcement. The model can also produce the violation of independence, preference reversals, and the goal gradient of reinforcement in maze learning. An experiment was conducted to study learning of action sequences in a multistep task. The fit of the model to the data demonstrated its ability to account for complex skill learning. The advantages of incorporating the mechanism into a larger cognitive architecture are discussed.

This paper argues for an explanation of the mechanistic (computational) basis of consciousness that is based on the distinction between localist (symbolic) representation and distributed representation, the ideas of which have been put forth in the connectionist literature. A model is developed to substantiate and test this approach. The paper also explores the issue of the functional roles of consciousness, in relation to the proposed mechanistic explanation of consciousness. The model, embodying the representational difference, is able to account for the functional role of consciousness, in the form of the synergy between the conscious and the unconscious. The fit between the model and various cognitive phenomena and data (documented in the psychological literatures) is discussed to accentuate the plausibility of the model and its explanation of consciousness. Comparisons with existing models of consciousness are made in the end.

This article describes the path-mapping theory of how humans integrate analogical mapping and general problem solving. The theory posits that humans represent analogs with declarative roles, map analogs by lower-level retrieval of analogous role paths, and coordinate mappings with higher-level organizational knowledge. Implemented in the ACT-R cognitive architecture, the path-mapping theory enables models of analogical mapping behavior to incorporate and interface with other problem-solving knowledge. Path-mapping models thus can include task-specific skills such as encoding analogs or generating responses, and can make behavioral predictions at the level of real-world metrics such as latency or correctness. We show that the path-mapping theory can successfully account for the major phenomena addressed by previous theories of analogy. We also describe a path-mapping model that can account for subjects’ incremental eye-movement and typing behavior in a story-mapping task. We discuss extensions and implications of this work to other areas of analogy and problem-solving research.

The interactivist approach to development generates a framework of types of constraints on what can be constructed. The four constraint types are based on: (1) what the constructed systems are about; (2) the representational relationship itself; (3) the nature of the systems being constructed; and (4) the process of construction itself. We give illustrations of each constraint type. Any developmental theory needs to acknowledge all four types of constraint; however, some current theories conflate different types of constraint, or rely on a single constraint type to explicate development. Such theories will be inherently unable to explain important aspects of development.

Research in computational cognitive modeling investigates the nature of cognition through developing process-based understanding by specifying computational models of mechanisms (including representations) and processes. In this enterprise, a cognitive architecture is a domaingeneric computational cognitive model that may be used for a broad, multiple-level, multipledomain analysis of behavior. It embodies generic descriptions of cognition in computer algorithms and programs. Developing cognitive architectures is a difficult but important task. In this article, discussions of issues and challenges in developing cognitive architectures will be undertaken, and an example cognitive architecture (CLARION) will be described. 1

In an ideal design process, designers envision a configuration of components prior to determining dimensions or sizes of these components. Given the breadth of suppliers and production methods that exist today, most engineered artifacts are a mix of both custom-made parts and OEM (original equipment manufacturer) parts. The design of any future artifact must be carefully planned to take advantage of the diverse set of possibilities. We conjecture that computational design tools could be developed to help designers navigate the design space in creating configurations from detailed specifications of function. In this research, a methodology is developed that extracts design knowledge from an expanding online library of components in the form of grammar rules. From an initial implementation of forty-five rules compiled from 15 components extracted from three products, we demonstrate a computational process that builds a new design configuration by borrowing concepts from how common functions are solved in related designs.