This essay continues my investigation of `syntactic semantics': the theory that, pace Searle's Chinese-Room Argument, syntax does suffice for semantics (in particular, for the semantics needed for a computational cognitive theory of natural-language understanding). Here, I argue that syntactic semantics (which is internal and first-person) is what has been called a conceptual-role semantics: The meaning of any expression is the role that it plays in the complete system of expressions. Such a `narrow', conceptual-role semantics is the appropriate sort of semantics to account (from an `internal', or first-person perspective) for how a cognitive agent understands language. Some have argued for the primacy of external, or `wide', semantics, while others have argued for a two-factor analysis. But, although two factors can be specified---one internal and first-person, the other only specifiable in an external, third-person way---only the internal, first-person one is needed for understanding how someone understands. A truth-conditional semantics can still be provided, but only from a third-person perspective.

This document is a draft of an article for the Encyclopedia of Language and Linguistics, 2nd Edition (Elsevier, forthcoming). This article describes the Turing Test for determining whether a computer can think. It begins with a description of an “imitation game ” for discriminating between a man and a woman, discusses variations of the Test, standards for passing the Test, and experiments with real Turing-like tests (including Eliza and the Loebner competition). It then considers what a 1 computer must be able to do in order to pass a Turing Test, including whether written linguistic behavior is a reasonable replacement for “cognition”, what counts as understanding natural language, the role of world knowledge in understanding natural language, and the philosophical implications of passing a Turing Test, including whether passing is a sufficient demonstration of cognition, briefly discussing two counterexamples: a table-lookup program and the Chinese Room Argument.

We define a knowledge representation and inference formalism that is well suited to natural language processing. In this formalism every subformula of a formula is closed. We motivate this by observing that any formal language with (potentially) open sentences is an inappropriate medium for the representation of natural language sentences. Open sentences in such languages are a consequence of the separation of variables from their quantifier and type constraints, typically in the antecedents of rules. This is inconsistent with the use of descriptions and noun phrases corresponding to variables in language. Variables in natural language are constructions that are typed and quantified as they are used. A consequence of this is that variables in natural language may be freely reused in dialog. This leads to the use of pronouns and discourse phenomena such as ellipsis involving reuse of entire subformulas. We present an augmentation to the representation of variables so that variables are ...

Computationalism, the notion that cognition is computation, is a working hypothesis of many AI researchers and Cognitive Scientists. Although it has not been proved, neither has it been disproved. In this paper, I give some refutations to some well-known alleged refutations of computationalism. My arguments have two themes: people are more limited than is often recognized in these debates; computer systems are more complicated than is often recognized in these debates. To underline the latter point, I sketch the design and abilities of a possible embodied computer system. 1 Artificial Intelligence and Computationalism There are several disparate goals pursued by Artificial Intelligence (AI) researchers: computational psychology, computational philosophy, and advanced Computer Science [Shapiro, 1992]. In this paper, I will concentrate on computational philosophy, which could also be called "the computational study of cognition." AI is often thought of in the popular press as a technolo...

I advocate a theory of "syntactic semantics" as a way of understanding how computers can think (and how the Chinese-Roam-Argument objection to the Turing Test can be overcome): (1) Semantics, considered as the study of relations between symbols and meanings, can be turned into syntax- a study of relations among symbols (including meanings)- and hence syntax (i.e., symbol manipulation) can suffice for the semantical enterprise (contra Searle). (2) Semantics, considered as the process of understanding one domain (by modeling it) in terms of another, can be viewed recursively: The base case of semantic understanding- understanding a domain in terms of itself- is "syntactic understanding." (3) An internal (or "narrow"), first-person point of view makes an external (or "wide"), third-person point of view otiose for purposes of understanding cognition.

It is sometimes necessary to reason non-monotonically, to withdraw previously held beliefs. Default rules and defeasible reasoning have been frequently used to handle such situations. Some years ago, J. Terry Nutter proposed a form of defeasible reasoning involving additional truthvalues that would avoid non-monotonicity in many situations. We adopt and implement much of Nutter's underlying concept, but use a structure of case frames in our knowledge representation to avoid the need for more than the standard truth values. Our implementation captures most of the advantages of the original proposal and in some cases allows more flexibility.

What is the computational notion of "implementation"? It is not individuation, instantiation, reduction, or supervenience. It is, I suggest, semantic interpretation. This document is Technical Report 97-15 (Buffalo: SUNY Buffalo Department of Computer Science) and Technical Report 97-5 (Buffalo: SUNY Buffalo Center for Cognitive Science). 1 INTRODUCTION Consider the relationships among algorithms, computer programs, and the computers that execute them. An algorithm is (roughly) a procedure for computing a function (for more details, see Soare 1996; Rapaport, forthcoming). A program is a more specific and detailed textual expression of an algorithm, expressed in a programming language. Often, it is said that the program "implements" the algorithm. A computer process is an algorithm being executed (see Rapaport 1988, 1995; Smith 1997). It is a physical device (a computer) behaving in a certain way ; the way is described (or specified) by the program, and the physical device running the ...

This essay explores the implications of the thesis that implementation is semantic interpretation. Implementation is (at least) a ternary relation: I is an implementation of an ‘Abstraction ’ A in some medium M. Examples are presented from the arts, from language, from computer science and from cognitive science, where both brains and computers can be understood as implementing a ‘mind Abstraction’. Implementations have side effects due to the implementing medium; these can account for several puzzles surrounding qualia. Finally, an argument for benign panpsychism is developed.

0). The best reply to the actual Chinese Room scenario is the Korean Professor Argument (Rapaport, 1988) which identifies a weakness in the Chinese Room scenario (the person in the room understands the instructions to be followed) before re-describing the scenario in a form acceptable to CCS and AI. 3 More generally, though, AI addresses this problem through a form of `Systems reply': intelligence, awareness and consciousness arise from computational processes and interactions between these processes. Whether these processes are mental, physical or behavioural is irrelevant, in that the system as a whole moves through various states, where the next state of the system is determined by the current state of the system and any input it receives. The stress here is on functionality and cognitive architecture (Putnam, 1967; Fodor and Pylyshyn, 1988): The states a system goes through are representational states, and a cognitive architecture is an architectur

s. Clinical notes are usually free-text, often with little structure and irregular grammar. They contain many terms from a specialized vocabulary and grammar (e.g., the use of `present' as a non-transitive verb, as in "The child presented to the hospital with the heart murmur"), often with multiple ways of specifying the same concept. There are often misspellings, errors in wording, and mistranscribed words in dictated notes (e.g., the name `Kay Seal' for the chemical `KCl' (potassium chloride)). This makes the process of encoding clinical notes difficult. Although the free-text data could simply be captured in an unprocessed form, this fails to provide any of the benefits that CPRs have over paper records and may even be less efficient. In such a capture, no attempt is made to understand the text or to allow anything but simple word searches on the information it contains. Another method of encoding is to control the vocabulary that clinicians can use. Instead of allowing free-text e