Explore the vast range of titles available.

In a study of quantifier-scope priming, Chemla and Bott (2015) found evidence suggesting that, while representations of quantifiers’ relative scope can be primed, a scope inversion operation cannot. We identify a confound in their materials. In Experiment 1, we replicate their finding with this confound intact. In Experiment 2, we remove the confound and find that all priming disappears. This confound demonstrates how structural priming paradigms can be sensitive to many dimensions of similarity, pointing to a need for task-specific controls. We conclude that the prior study does not provide evidence concerning the priming of either relative scope representations or operations. While priming of scope representations has been independently found in other paradigms, the jury is still out on Chemla and Bott’s more novel finding – the absence of priming of a scope inversion operation.

The relationship between Carnap’s Logical Syntax of Language (hereafter LSL) ([1934] 1937) and Wittgenstein’s Tractatus Logico-Philosophicus (TLP) ([1921] 1922) has been interpreted in several ways during past decades. One of the interpretations has gained keen advocates among Carnap scholars. It was originally provoked by what Caranp said in LSL, and it consists of two parts. First, it indicates that in TLP the possibility of speaking about the logical form of a language within the same language (which happens to be the only language that there is) had been foresworn by Wittgenstein, but Carnap proved him wrong by producing a book (LSL) written exactly in the manner which had been proscribed by Wittgenstein. This is the debate about the possibility of speaking about logical form. Second, Wittgenstein’s dogmatism with regard to the existence of a unique correct grammar at the foundation of the language has been contrasted with Carnap’s open-mindedness in conceiving a boundless ocean of possibilities for constructing logical systems. Interestingly enough, Wittgenstein rambled with rage in reaction to Carnap’s view about the LSL-TLP relationship. But unlike Carnap’s view, which led to a dominant interpretation of the relationship, Wittgenstein’s testimony about the case has been strangely ignored in the history of analytic philosophy. In this paper, I try to make an inquiry about the grounds for Wittgenstein’s dissatisfaction with the Carnapian reading of the LSL-TLP relationship. I will show that Wittgenstein was not totally unfair in his judgment, and that some salient aspects of LSL (recognized as the anti-Tractarian aspects of the work) could be best understood in the light, or rather the gloom, of TLP, and bear a significant resemblance to it. This, however, does not need to diminish the logical and historical significance of LSL.

This paper is intended to address the work “Against logical form”, authored by Johnson-Laird in 2010. In it, based on the theory of mental models, Johnson-Laird claims that the way people interpret sentences in natural language has nothing to do with logic. This is because that action is not related to logical forms. According to him, the mental activity is mainly linked to semantics and pragmatics. However, here, following arguments provided by López-Astorga, I try to show that the theory of mental models can be linked to syntactic structures. My main point is made by dealing with an argument given in the mentioned Johnson-Laird’s paper. In principle, that argument seems to undermine proposals such as that of López-Astorga. However, the present work proposes otherwise.

Kolodny argues that none of them work. The best way to resolve the paradox is to give a semantics for deontic modals and indicative conditionals that lets all see how the argument can be invalid even with its obvious logical form. This requires rejecting the general validity of at least one classical deduction rule.

The dual strategy model proposes that people use one of two potential ways of processing information when making inferences. The statistical strategy generates a rapid probabilistic estimate based on associative access to a wide array of information, while the counterexample strategy uses a more focused representation, allowing for a search for potential counterexamples. In the following studies, we explore the hypothesis that individual differences in strategy use are related to the ability to make rapid intuitive logical judgments. In Study 1, we show that this is the case for rapid judgments requiring a distinction between simple logical form and for a novel form of judgment, the ability to identify inferences that are not linked to their premises (non sequiturs). In Study 2, we show that strategy use is related to the ability to make the kinds of rapid logical judgments previously examined over and above contributions of working memory capacity. Study 3 shows that strategy use explains individual variability in rapid logical responding with belief-biased inferences over and above the contribution of IQ. The results of Studies 2 and 3 indicate that under severe time constraint cognitive capacity is a very poor predictor of reasoning, while strategy use becomes a stronger predictor. These results extend the notion that people can make rapid intuitive “logical” judgments while highlighting the importance of strategy use as a key individual difference variable.

This paper outlines a truth-conditional view of logical form, that is, a view according to which logical form is essentially a matter of truth-conditions. Section 1 provides some preliminary clarifications. Section 2 shows that the main motivation for the view is the fact that fundamental logical relations such as entailment or contradiction can formally be explained only if truth-conditions are formally represented. Sections 3 and 4 articulate the view and dwell on its affinity with a conception of logical form that has been defended in the past. Sections 5-7 draw attention to its impact on three major issues that concern, respectively, the extension of the domain of formal explanation, the semantics of tensed discourse, and the analysis of quantification. Este artículo esboza una concepción veritativo-condicional de la forma lógica, es decir, una concepción de acuerdo con la cual la forma lógica es esencialmente una cuestión de condiciones de verdad. La sección 1 proporciona algunas clarificaciones preliminares. La sección 2 muestra que la principal motivación para esta concepción es el hecho de que hay relaciones lógicas fundamentales, como la implicación o la contradicción, que sólo pueden explicarse formalmente si las condiciones de verdad se representan formalmente. Las secciones 3 y 4 articulan dicha concepción y profundizan en su afinidad con una concepción de la forma lógica que ha sido defendida en el pasado. Las secciones 5 a 7 destacan su impacto sobre tres asuntos principales que conciernen, respectivamente, a la extensión del dominio de las explicaciones formales, la semántica del discurso temporalizado, y el análisis de la cuantificación.

In this paper I provide five separate responses, one for each of the contributed papers, in order to clarify some crucial aspects of the view defended in my book.

Compositional generalization benchmarks for semantic parsing seek to assess whether models can accurately compute for novel sentences, but operationalize this in terms of (LF) prediction. This raises the concern that semantically irrelevant details of the chosen LFs could shape model performance. We argue that this concern is realized for the COGS benchmark (Kim and Linzen, ). COGS poses generalization splits that appear impossible for present-day models, which could be taken as an indictment of those models. However, we show that the negative results trace to incidental features of COGS LFs. Converting these LFs to semantically equivalent ones and factoring out capabilities unrelated to semantic interpretation, we find that even baseline models get traction. A recent variable-free translation of COGS LFs suggests similar conclusions, but we observe this format is not semantically equivalent; it is incapable of accurately representing some COGS meanings. These findings inform our proposal for ReCOGS, a modified version of COGS that comes closer to assessing the target semantic capabilities while remaining very challenging. Overall, our results reaffirm the importance of compositional generalization and careful benchmark task design.

This paper outlines an account of concessive conditionals that rests on two main ideas. One is that the logical form of a sentence as used in a given context is determined by the content expressed by the sentence in that context. The other is that a coherent distinction can be drawn between a reading of ‘if’ according to which a conditional is true when its consequent holds on the supposition that its antecedent holds, and a stronger reading according to which a conditional is true when its antecedent supports its consequent. As we will suggest, the logical form of concessive conditionals can be elucidated by relying on this distinction.

Landgrebe and Smith (Synthese 198(March):2061–2081, 2021) present an unflattering diagnosis of recent advances in what they call language-centric artificial intelligence—perhaps more widely known as natural language processing: The models that are currently employed do not have sufficient expressivity, will not generalize, and are fundamentally unable to induce linguistic semantics, they say. The diagnosis is mainly derived from an analysis of the widely used Transformer architecture. Here I address a number of misunderstandings in their analysis, and present what I take to be a more adequate analysis of the ability of Transformer models to learn natural language semantics. To avoid confusion, I distinguish between inferential and referential semantics. Landgrebe and Smith (2021)’s analysis of the Transformer architecture’s expressivity and generalization concerns inferential semantics. This part of their diagnosis is shown to rely on misunderstandings of technical properties of Transformers. Landgrebe and Smith (2021) also claim that referential semantics is unobtainable for Transformer models. In response, I present a non-technical discussion of techniques for grounding Transformer models, giving them referential semantics, even in the absence of supervision. I also present a simple thought experiment to highlight the mechanisms that would lead to referential semantics, and discuss in what sense models that are grounded in this way, can be said to understand language. Finally, I discuss the approach Landgrebe and Smith (2021) advocate for, namely manual specification of formal grammars that associate linguistic expressions with logical form.