Online research in philosophy

Representations have been viewed as the essential concern of cognitive science, yet few studies have examined how people create, perceive, and attribute meaning to new representational forms. How does the learner relate instructions he doesn't yet understand to features on the computer screen he can't yet parse into objects and relations? Linguistic schema models assume that the world comes pre-represented, already parameterized into objective features; reasoning operates on a stream of "perceptually obvious" symbols. In such an exclusively linguistic cognitive model, inference and comprehension rests on nothing but more words--definitions, causal relationships, classifications. Although it is well-known that such models are "ungrounded," that the symbols have no meaning to the program itself, little attempt has been made to find out how people create symbolic forms. What are the non-linguistic processes that control attention, affect reconceptualization, and correlate disparate ways of seeing? I present an example of human learning that falls outside linguistic schema theories, illustrating representation creation as perceptual interaction at both interpersonal and gestural-material levels. I focus on sequences of activity in which students' interpretation of what constitutes a representational language (and what it means) changes as they construct models of what they are seeing and doing. This social-perceptual analysis complements linguistic schema theories of novice-expert differences with more detailed learning mechanisms, emphasizing especially the nature of perception. This perspective leads to new experimentation and ways of observing and understanding student interaction with today's instructional programs.

A computational theory of induction must be able to identify the projectible predicates, that is to distinguish between which predicates can be used in inductive inferences and which cannot. The problems of projectibility are introduced by reviewing some of the stumbling blocks for the theory of induction that was developed by the logical empiricists. My diagnosis of these problems is that the traditional theory of induction, which started from a given (observational) language in relation to which all inductive rules are formulated, does not go deep enough in representing the kind of information used in inductive inferences. As an interlude, I argue that the problem of induction, like so many other problems within AI, is a problem of knowledge representation. To the extent that AI-systems are based on linguistic representations of knowledge, these systems will face basically the same problems as did the logical empiricists over induction. In a more constructive mode, I then outline a non-linguistic knowledge representation based on conceptual spaces. The fundamental units of these spaces are "quality dimensions". In relation to such a representation it is possible to define "natural" properties which can be used for inductive projections. I argue that this approach evades most of the traditional problems.

Many different enterprises go under the title of semantics or semantic theory. For each of these, there must be a correspondingly different conception of pragmatics, at least in those cases where such a distinction is admitted. On the relevance-theoretic view, which is the primary focus of this paper, the distinction between semantics and pragmatics is a distinction between two types of cognitive process employed in understanding utterances: decoding and inference. The decoding process is performed by an autonomous linguistic system, the parser or language perception module. Having identified a particular acoustic stimulus as linguistic, this system executes a series of deterministic grammatical computations, or mappings, resulting in an output representation, which is the semantic representation, or logical form, of the sentence or phrase employed in the utterance. It is a structured string of concepts, which has both logical and causal properties. The second type of cognitive process, the pragmatic inferential process, integrates the linguistic contribution with other readily accessible information in order to reach a confirmed interpretive hypothesis concerning the speaker's informative intention. This inferential phase of interpretation is constrained and guided by the communicative principle of relevance, which licences a hearer to look for an interpretation which interacts fruitfully with his cognitive system and does not put him to any unjustifiable processing effort.

In this paper I introduce a formalism for natural language understandingbased on a computational implementation of Discourse RepresentationTheory. The formalism covers a wide variety of semantic phenomena(including scope and lexical ambiguities, anaphora and presupposition),is computationally attractive, and has a genuine inference component. Itcombines a well-established linguistic formalism (DRT) with advancedtechniques to deal with ambiguity (underspecification), and isinnovative in the use of first-order theorem proving techniques.The architecture of the formalism for natural language understandingthat I advocate consists of three levels of processing:underspecification, resolution, andinference. Each of these levels has a distinct function andtherefore employs a different kind of semantic representation. Themappings between these different representations define the interfacesbetween the levels.

Many researchers have suggested that premise interpretation errors can account, at least in part, for errors on categorical syllogisms. However, although it is possible to show that people make such errors in simple inference tasks, the evidence for them is far less clear when actual syllogisms are administered. Part of the problem is due to the lack of clear predictions for the solutions that would be expected when using modified quantifiers, assuming that correct inferences are made from them. This paper presents the expected solutions for Gricean, reversible, and reversible Gricean interpretations, and evaluates these using three datasets (two currently available, and one new). The evidence supported the adoption of reversible and reversible Gricean interpretations, but not Gricean interpretations on their own. These results suggest that the categorical syllogism task tends to induce different quantifier interpretations from those identified in simple inference tasks.

This chapter investigates the computational consequences of a broadly Gricean view of language use as intentional activity. In this view, dialogue rests on coordinated reasoning about communicative intentions. The speaker produces each utterance by formulating a suitable communicative intention. The hearer understands it by recognizing the communicative intention behind it. When this coordination is successful, interlocutors succeed in considering the same intentions— that is, the same representations of utterance meaning—as the dialogue proceeds. In this paper, I emphasize that these intentions can be formalized; we can provide abstract but systematic representations that spell out what a speaker is trying to do with an utterance. Such representations describe utterances simultaneously as the product of our knowledge of grammar and as actions chosen for a reason. In particular, they must characterize the speaker’s utterance in grammatical terms, provide the links to the context that the grammar requires, and so arrive at a contribution that the speaker aims to achieve. Because I have implemented this formalism, we can regard it as a possible analysis of conversational processes at the level of computational theory. Nevertheless, this analysis leaves open what the nature of the biological computation involved in inference to intentions is, and what regularities in language use support this computation.

How can computers distinguish the coherent from the unintelligible, recognize new information in a sentence, or draw inferences from a natural language passage? Computational semantics is an exciting new field that seeks answers to these questions, and this volume is the first textbook wholly devoted to this growing subdiscipline. The book explains the underlying theoretical issues and fundamental techniques for computing semantic representations for fragments of natural language. This volume will be an essential text for computer scientists, linguists, and anyone interested in the development of computational semantics.

I argue that, pace Chomsky (2000, 2003), standard theories of linguistic competence are committed to taking talk of representations seriously, in particular, to recognizing that the “of x” clause that invariably follows “representation” is a way of specifying that representation’s intentional content. One reason to insist upon intentional content in such cases is that the “x” in “of x” may not exist (as in "of Zeus"). This issue is especially relevant to linguistics since, recapitulating considerations raised by many linguists, I go on to argue that most of the SLEs themselves seldom, if ever exist: it is doubtful there are many, if any, tokens of them in space and time; indeed, their existence is by and large inessential to the needs of either communication or serious linguistic theory. All that linguistic theory requires to be real in this regard are the representations, presumably entokened in people’s brains, understood, however, in terms of their intentional contents.

cal practice: the enterprise of specifying information about the world for use in computer systems. Knowledge representation as a field also encompasses conceptual results that call practitioners’ attention to important truths about the world, mathematical results that allow practitioners to make these truths precise, and computational results that put these truths to work. This chapter surveys this practice and its results, as it applies to the interpretation of natural language utterances in implemented natural language processing systems. For a broader perspective on such technical practice, in all its strengths and weaknesses, see (Agre 1997). Knowledge representation offers a powerful general tool for the science of language. Computational logic, a prototypical formalism for representing knowledge about the world, is also the model for the level of logical form that linguists use to characterize the grammar of meaning (Larson and Segal 1995). And researchers from (Schank and Abelson 1977) to (Shieber 1993) and (Bos to appear) have relied crucially on such representations, and the inference methods associated with them, in articulating accounts of semantic relations in language, such as synonymy, entailment, informativeness and contradiction. The new textbooks (Blackburn and Bos 2002a, Blackburn and Bos 2002b) provide an excellent grounding in this research, and demonstrate how deeply computational ideas from knowledge representation can inform pure linguistic study. In this short chapter, I must leave much of..

Logic has its roots in the study of valid argument, but while traditional logicians worked with natural language directly, modern approaches first translate natural arguments into an artificial language. The reason for this step is that some artificial languages now have very well developed inferential systems. There is no doubt that this is a great advantage in general, but for the study of natural reasoning it is a drawback that the original linguistic forms get lost in translation. An alternative approach would be to develop a general theory of the natural logic behind human reasoning and human information processing by studying formal logics that operate directly on linguistic representations. That this is possible we will try to make plausible in this paper. It will turn out that one level of representation, that of Logical Form, can meaningfully be identified with the language of an existing and well-understood logic, a restricted form of the theory of types. It is not difficult to devise inference systems for this language, and it is thus possible to study reasoning systems that are based directly on language.