Recent progress in artificial intelligence has renewed interest in building systems that learn and think like people. Many advances have come from using deep neural networks trained end-to-end in tasks such as object recognition, video games, and board games, achieving performance that equals or even beats that of humans in some respects. Despite their biological inspiration and performance achievements, these systems differ from human intelligence in crucial ways. We review progress in cognitive science suggesting that truly human-like learning and thinking (...) machines will have to reach beyond current engineering trends in both what they learn and how they learn it. Specifically, we argue that these machines should build causal models of the world that support explanation and understanding, rather than merely solving pattern recognition problems; ground learning in intuitive theories of physics and psychology to support and enrich the knowledge that is learned; and harness compositionality and learning-to-learn to rapidly acquire and generalize knowledge to new tasks and situations. We suggest concrete challenges and promising routes toward these goals that can combine the strengths of recent neural network advances with more structured cognitive models. (shrink)

The book presents the case that cognitive science should turn its attention to developing theories of human cognition that cover the full range of human perceptual, cognitive, and action phenomena. Cognitive science has now produced a massive number of high-quality regularities with many microtheories that reveal important mechanisms. The need for integration is pressing and will continue to increase. Equally important, cognitive science now has the theoretical concepts and tools to support serious attempts at unified theories. The argument is made (...) entirely by presenting an exemplar unified theory of cognition both to show what a real unified theory would be like and to provide convincing evidence that such theories are feasible. The exemplar is SOAR, a cognitive architecture, which is realized as a software system. After a detailed discussion of the architecture and its properties, with its relation to the constraints on cognition in the real world and to existing ideas in cognitive science, SOAR is used as theory for a wide range of cognitive phenomena: immediate responses ; discrete motor skills ; memory and learning ; problem solving ; language ; and development. The treatments vary in depth and adequacy, but they clearly reveal a single, highly specific, operational theory that works over the entire range of human cognition, SOAR is presented as an exemplar unified theory, not as the sole candidate. Cognitive science is not ready yet for a single theory – there must be multiple attempts. But cognitive science must begin to work toward such unified theories. (shrink)

This paper presents a “frame” representation for common sense knowledge and uses it to formalize our knowledge of “mundane” painting (walls; not portraits). These frames. while designed to aid a computer program to understand stories about the painting process, should be of use to programs which attempt to actually carry out the activity. The paper stresses a “deep” understanding of the activity so that the representation indicates not only what steps to carry out, but also how to do them, and (...) why they should be done. To accomplish this, while at the same time preserving modularity and nonredundancy, a system of interframe pointers is introduced (the COMES‐FROM and LEADS‐TO pointers) which explain how or why something is done in tens of knowledge given in other frames. The paper proceeds by steadily deepening an initial English‐like description of the activity, and a context free grammar for the representation is included. (shrink)

We advance a model of human probability judgment and apply it to the design of an extrapolation algorithm. Such an algorithm examines a person's judgment about the likelihood of various statements and is then able to predict the same person's judgments about new statements. The algorithm is tested against judgments produced by thirty undergraduates asked to assign probabilities to statements about mammals.

We show how an epistemology informed by cognitive science promises to shed light on an ancient problem in the philosophy of mathematics: the problem of exactness. The problem of exactness arises because geometrical knowledge is thought to concern perfect geometrical forms, whereas the embodiment of such forms in the natural world may be imperfect. There thus arises an apparent mismatch between mathematical concepts and physical reality. We propose that the problem can be solved by emphasizing the ways in which the (...) brain can transform and organize its perceptual intake. It is not necessary for a geometrical form to be perfectly instantiated in order for perception of such a form to be the basis of a geometrical concept. (shrink)

Preparing words in speech production is normally a fast and accurate process. We generate them two or three per second in fluent conversation; and overtly naming a clear picture of an object can easily be initiated within 600 msec after picture onset. The underlying process, however, is exceedingly complex. The theory reviewed in this target article analyzes this process as staged and feedforward. After a first stage of conceptual preparation, word generation proceeds through lexical selection, morphological and phonological encoding, phonetic (...) encoding, and articulation itself. In addition, the speaker exerts some degree of output control, by monitoring of self-produced internal and overt speech. The core of the theory, ranging from lexical selection to the initiation of phonetic encoding, is captured in a computational model, called WEAVER++. Both the theory and the computational model have been developed in interaction with reaction time experiments, particularly in picture naming or related word production paradigms, with the aim of accounting for the real-time processing in normal word production. A comprehensive review of theory, model, and experiments is presented. The model can handle some of the main observations in the domain of speech errors (the major empirical domain for most other theories of lexical access), and the theory opens new ways of approaching the cerebral organization of speech production by way of high-temporal-resolution imaging. (shrink)

Many people have noted the similarities between case theories in linguistics and frame representations in artificial intelligence. In particular, the cases of a verb seem to correspond to the slots of a frame. This has led many people, including Fillmore [1977] and Winston [1977] to assert that cases and slots are one and the same. This hypothesis has not attracted much attention, probably because the notion of “slot” in frame representations is so underconstrained that the theory would seem to be (...) content free. In this note, we wish to show that this is not true. For one thing, the case‐slot identity theory can explain the difficulty linguists have had in deciding how many cases there are. Furthermore, its adoption will place certain interesting, and plausible constraints on the frame and case theories compatible with it. (shrink)

Establishing reasonable, prior distributions remains a significant obstacle for the construction of probabilistic expert systems. Human assessment of chance is often relied upon for this purpose, but this has the drawback of being inconsistent with axioms of probability. This article advances a method for extracting a coherent distribution of probability from human judgment. The method is based on a psychological model of probabilistic reasoning, followed by a correction phase using linear programming.

Expert reasoning combines voluminous domain‐specific knowledge with more general factual and strategic knowledge. Whereas expert system builders have recognized the need for specificity and problem‐solving researchers the need for generality, few attempts have been made to develop expert reasoning engines combining different kinds of knowledge at different levels of generality. This paper reports on the FERMI project, a computer‐implemented expert reasoner in the natural sciences that encodes factual and strategic knowledge in separate semantic hierarchies. The principled decomposition of knowledge according (...) to type and level of specificity yields both power and cross‐doman generality, as demonstrated in FERMI's ability to apply the same principles of invariance and decomposition to solve problems in fluid statics, DC‐circuits, and centroid location. Hierarchical knowledge representation and problem‐solving principles are discussed, and illustrative problem‐solving traces are presented. (shrink)

Although the connectionist approach has lead to elegant solutions to a number of problems in cognitive science and artificial intelligence, its suitability for dealing with problems in knowledge representation and inference has often been questioned. This paper partly answers this criticism by demonstrating that effective solutions to certain problems in knowledge representation and limited inference can be found by adopting a connectionist approach. The paper presents a connectionist realization of semantic networks, that is, it describes how knowledge about concepts, their (...) properties, and the hierarchical relationship between them may be encoded as an interpreter‐free massively parallel network of simple processing elements that can solve an interesting class of inheritance and recognition problems extremely fast—in time proportional to the depth of the conceptual hierarchy. The connectionist realization is based on an evidential formulation that leads to principled solutions to the problems of exceptions and conflicting multiple inheritance situations during inheritance, and the best‐match or partial‐match computation during recognition. The paper also identifies constraints that must be satisfied by the conceptual structure in order to arrive at an efficient parallel realization. (shrink)