Theory learning as stochastic search in the language of thought

We present an algorithmic model for the development of children's intuitive theories within a hierarchical Bayesian framework, where theories are described as sets of logical laws generated by a probabilistic context-free grammar. We contrast our approach with connectionist and other emergentist approaches to modeling cognitive development. While their subsymbolic representations provide a smooth error surface that supports efficient gradient-based learning, our symbolic representations are better suited to capturing children's intuitive theories but give rise to a harder learning problem, which can only be solved by exploratory search. Our algorithm attempts to discover the theory that best explains a set of observed data by performing stochastic search at two levels of abstraction: an outer loop in the space of theories and an inner loop in the space of explanations or models generated by each theory given a particular dataset. We show that this stochastic search is capable of learning appropriate theories in several everyday domains and discuss its dynamics in the context of empirical studies of children's learning.     

Understanding the language of thought

Conclusions To conclude then, our question was, What determines the content of a thought? This question gets its punch by being set against the background assumption that thoughts are internal physical occurrences. The answer I have proposed is that thoughts can be classified in four importantly different ways. Introspection yields syntactic categories. These are important for cognitive processing, but they do not correspond to contents. A second way of categorizing thoughts is in terms of narrow content. This is determined by the functional role of the thought in rational architecture together with the way in which that rational architecture is tied to the world through input correlations. Narrow contents are indexical, so to get truth bearers we must augment the narrow contents with the values of the indexical parameters. Propositional content can be taken to consist of pairs of narrow contents and values for indexical parameters. Finally, thoughts can be classified in terms of that clauses. This kind of classification does not uniquely determine propositional content, but describes it in a more general way.     

Conceptual clusters in figurative language production

Although most prior research on figurative language examines comprehension, several recent studies on the production of such language have proved to be informative. One of the most noticeable traits of figurative language production is that it is produced at a somewhat random rate with occasional bursts of highly figurative speech (e.g., Corts & Pollio, 1999). The present article seeks to extend these findings by observing production during speech that involves a very high base rate of figurative language, making statistically defined bursts difficult to detect. In an analysis of three Baptist sermons, burst-like clusters of figurative language were identified. Further study indicated that these clusters largely involve a central root metaphor that represents the topic under consideration. An interaction of the coherence, along with a conceptual understanding of a topic and the relative importance of the topic to the purpose of the speech, is offered as the most likely explanation for the clustering of figurative language in natural speech.     

Conceptual knowledge attenuates viewpoint dependency in visual object recognition

Learning verbal semantic knowledge for objects has been shown to attenuate recognition costs incurred by changes in view from a learned viewpoint. Such findings were attributed to the semantic or meaningful nature of the learned verbal associations. However, recent findings demonstrate surprising benefits to visual perception after learning even noninformative verbal labels for stimuli. Here we test whether learning verbal information for novel objects, independent of its semantic nature, can facilitate a reduction in viewpoint-dependent recognition. To dissociate more general effects of verbal associations from those stemming from the semantic nature of the associations, participants learned to associate semantically meaningful (adjectives) or nonmeaningful (number codes) verbal information with novel objects. Consistent with a role of semantic representations in attenuating the viewpoint-dependent nature of object recognition, the costs incurred by a change in viewpoint were attenuated for stimuli with learned semantic associations relative to those associated with nonmeaningful verbal information. This finding is discussed in terms of its implications for understanding basic mechanisms of object perception as well as the classic viewpoint-dependent nature of object recognition.     

An overview of the controversial issues in the psychology of language and thought

This article is an overview of the most important controversial issues in the field of psycholinguistics. Seven questions were posed in a dialogue interview to each of the following distinguished scholars in the field: Noam Chomsky, Charles Osgood, Jean Piaget, Ulrich Neisser, and Marcel Kinsbourne. The authors have provided a critical discussion of the issues as well as a summary of the different points of view of each of the above mentioned individuals. The complete dialogues will be published together in a forthcoming book entitledDialogues in the Psychology of Language and Thought.This paper was originally written in collaboration with G. Voyat for and will appear as the introduction of a forthcoming book entitledDialogues in the Psychology of Language and Thought, R. W. Rieber, ed. (Plenum, New York, 1981). The reader should consult the above-mentioned book for a complete picture of the dialogues discussed in this paper.     

Operational Noology as a new methodology for the study of thought and language: theoretical aspects and possible practical applications

In this article, the author presents a new methodology for the study of two fundamental components of consciousness, that is thought and language. The fundamental presupposition that forms the basis of this methodology is that thought is not simply a passive "reflection" of an external "reality", but also (and especially) something active, i.e. that the fundamental components of thought are sequences of operations, amongst which the ones of attention play a key role. These sequences of elementary mental operations are called mental categories, and are the meanings of all the words that do not seem to indicate something physical (first of all, all the "grammatical" words, that is conjunctions, prepositions, articles, pronouns, fundamental verbs like "to be", "to have" etc., the main adverbs, and, in the large number of languages that have a more or less rich morphology, all morphemes (the ones which indicate cases, in languages that have cases, the number and the gender of nouns and adjectives, moods and tenses of the verb etc.). The author proposes a list of these elementary mental operations and he shows how it is possible, basing ourselves on them, to identify the meanings of these words, which are indispensable for any linguistic expression. The author also mentions a possible short-term practical application of these theories, i.e. a device in order to improve the quality of machine translation. He also formulates the hypothesis that these theories could allow us to look in a new way at the problem of the (partial) artificial reproduction of the human activity of thought and language.     

Conceptual models and task representation in using a command language

In most cases, there is no direct correspondence between the objectives that users of an interactive device may be led to formulate and the commands that enable them to attain those objectives. In order to set up what Young (1981) calls a “mapping” between the “task and action arenas”, users must (1) reorganize their objectives in order to adapt them to the capabilities of the device and (2) if necessary, combine several commands to reach a given objective. In doing so, they are assumed to rely on a “conceptual model” of the device, a notion whose cognitive status remains to be clarified. In this article, the editing keys of the MS/DOS operating system are taken as an example and described according to two different conceptual models. The links between the task and action arenas, in these two conceptual models and in a third one called the “mixed model”, are analysed in detail. Finally, the performance of three groups of subjects, while editing a line of text, are compared. All three of the dependent variables analysed are shown to vary according to how the objective is assumed to be formulated in the three conceptual models. Some hypotheses are presented concerning the cognitive mechanisms involved in such a situation.     

The problem of language and thought in schizophrenia: A review

A slightly different version of this article will also appear as a chapter in H. J. Vetter & R. W. Rieber, (Eds.) (in press).The psychopathology of language and thought. New York: Plenum Press.     

Conceptual engineering as concept preservation

In the burgeoning philosophical literature on conceptual engineering improving our concepts is typically portrayed as the hallmark activity of the field. However, Herman Cappelen has challenged the idea that we can know how and why conceptual changes occur well enough to actively intervene in revising our concepts; the mechanisms of conceptual change are typically inscrutable to us. If the ‘inscrutability challenge’ is correct, the practical aspect of conceptual engineering may seem to be undermined, but I argue that endorsing such pessimism would be a mistake. Even if the inscrutability challenge is correct, conceptual engineers often have good reasons to try to preserve existing concepts. I examine several cases where concept preservation is important and draw lessons about this activity for conceptual engineers.     

Odors are expressible in language,as long as you speak the right language

From Plato to Pinker there has been the common belief that the experience of a smell is impossible to put into words. Decades of studies have confirmed this observation. But the studies to date have focused on participants from urbanized Western societies. Cross-cultural research suggests that there may be other cultures where odors play a larger role. The Jahai of the Malay Peninsula are one such group. We tested whether Jahai speakers could name smells as easily as colors in comparison to a matched English group. Using a free naming task we show on three different measures that Jahai speakers find it as easy to name odors as colors, whereas English speakers struggle with odor naming. Our findings show that the long-held assumption that people are bad at naming smells is not universally true. Odors are expressible in language, as long as you speak the right language.     

Some observations on the relationship between language and verbal thought in a psychotic adolescent

Abstract

Janine Chasseguet-Smirgel: The Ego Ideal. Free Association Books, 1985. Hardback £18.50; paperback £8.95.

Creativity and Perversion. Free Association Books, 1985. Hardback £11.95; paperback £6.95.

Jerome Kagan: The Nature of the Child. Published by Basic Books, 1984. Hardback, £15.85. Also by Harper and Row. Paperback, £4.95.

Julia Segal: Phantasy in Everyday Life. Published by Pelican, 1985. Paperback £2.95.