Most theories of concepts take concepts to be structured bodies of information used in categorization and inference. This paper argues for a version of atomism, on which concepts are unstructured symbols. However, traditional Fodorian atomism is falsified by polysemy and fails to provide an account of how concepts figure in cognition. This paper argues that concepts are generative pointers, that is, unstructured symbols that point to memory locations where cognitively useful bodies of information are stored and can be deployed to (...) resolve polysemy. The notion of generative pointers allows for unresolved ambiguity in thought and provides a basis for conceptual engineering. (shrink)

How feasible are conceptual engineering projects of social concepts that aim for the engineered concept to be deployed in people's ordinary conceptual practices? Predominant frameworks on the psychology of concepts that shape work on stereotyping, bias, and machine learning have grim implications for the prospects of conceptual engineers: conceptual engineering efforts are ineffective in promoting certain social‐conceptual changes. Since conceptual components that give rise to problematic social stereotypes are sensitive to statistical structures of the environment, purely conceptual change won't be (...) possible without corresponding world change. This tradition, however, tends to ignore that concepts don't only encode statistical, but also causal information. Paying attention to this feature of concepts, I argue, shows that conceptual engineering is not only possible. There is an imperative to conceptually‐engineer. (shrink)

In this paper, I develop an essentialist model of the semantics of slurs. I defend the view that slurs are a species of kind terms: Slur concepts encode mini-theories which represent an essence-like element that is causally connected to a set of negatively-valenced stereotypical features of a social group. The truth-conditional contribution of slur nouns can then be captured by the following schema: For a given slur S of a social group G and a person P, S is true of (...) P iff P bears the “essence” of G—whatever this essence is—which is causally responsible for stereotypical negative features associated with G and predicted of P. Since there is no essence that is causally responsible for stereotypical negative features of a social group, slurs have null-extension, and consequently, many sentences containing them are either meaningless or false. After giving a detailed outline of my theory, I show that it receives strong linguistic support. In particular, it can account for a wide range of linguistic cases that are regarded as challenging, central data for any theory of slurs. Finally, I show that my theory also receives convergent support from cognitive psychology and psycholinguistics. (shrink)

I discuss Edouard Machery’s claim that philosophers and psychologists when using the term ‘concept’ are really theorizing about different things. This view is not new, but it has never been developed or defended in detail. Once spelled out, we can see that Machery is right that the psychological literature uses a different notion of concept. However, Machery fails to acknowledge that the two notions are not only compatible but complementary. This fits more with the traditional view according to which philosophers (...) and psychologists are merely interested in different aspects of the same kind. The main aim of this paper is then to show how precisely the two notions of ‘concept’ relate. Distinguishing them resolves the long-standing debate on whether concepts can be prototypes and allows me to formulate success conditions of a theory of categorization that are independent of the success conditions of a theory of concepts. (shrink)

The meaning that expressions take on particular occasions often depends on the context in ways which seem to transcend its direct effect on context-sensitive parameters. ‘Truth-conditional pragmatics’ is the project of trying to model such semantic flexibility within a compositional truth-conditional framework. Most proposals proceed by radically ‘freeing up’ the compositional operations of language. I argue, however, that the resulting theories are too unconstrained, and predict flexibility in cases where it is not observed. These accounts fall into this position because (...) they rarely, if ever, take advantage of the rich information made available by lexical items. I hold, instead, that lexical items encode both extension and non-extension determining information. Under certain conditions, the non-extension determining information of an expression e can enter into the compositional processes that determine the meaning of more complex expressions which contain e. This paper presents and motivates a set of type-driven compositional operations that can access non-extension determining information and introduce bits of it into the meaning of complex expressions. The resulting multidimensional semantics has the tools to deal with key cases of semantic flexibility in appropriately constrained ways, making it a promising framework to pursue the project of truth-conditional pragmatics. (shrink)

Mind &Language, Volume 35, Issue 1, Page 90-114, February 2020.

How are biases encoded in our representations of social categories? Philosophical and empirical discussions of implicit bias overwhelmingly focus on salient or statistical associations between target features and representations of social categories. These are the sorts of associations probed by the Implicit Association Test and various priming tasks. In this paper, we argue that these discussions systematically overlook an alternative way in which biases are encoded, that is, in the dependency networks that are part of our representations of social categories. (...) Dependency networks encode information about how features in a conceptual representation depend on each other. This information determines the degree of centrality of a feature for a conceptual representation. Importantly, centrally encoded biases systematically disassociate from those encoded in salient-statistical associations. Furthermore, the degree of centrality of a feature determines its cross-contextual stability: in general, the more central a feature is for a concept, the more likely it is to survive into a wide array of cognitive tasks involving that concept. Accordingly, implicit biases that are encoded in the central features of concepts are predicted to be more resilient across different tasks and contexts. As a result, the distinction between centrally encoded and salient-statistical biases has important theoretical and practical implications. (shrink)

This paper presents and evaluates the early Husserl’s account of typicality. In the Logical Investigations, Husserl holds that the meaning of ordinary language (common) names is sensitive to typicality: this meaning depends on typical examples which vary in different contexts and are more or less similar to one another. This seems to entail that meanings, which according to Husserl are concepts, are “fluctuating” (schwankend) and vague. Prima facie, such a claim contravenes his theory of ideal meanings, or concepts, which are (...) “fixed” (fest) and sharp. However, Husserl wants to save this theory. He claims that the fluctuation and vagueness in question are not to be found in the meaning itself, or the concept, but rather derive from the act of meaning. Thus, he apparently manages to make room for typicality in ordinary language while accepting only fixed and sharp meanings. After presenting Husserl’s theory, I evaluate it and ask whether he will still be committed, despite his own claims, to accepting prototype concepts to account for typicality in ordinary language. (shrink)

Large language models (LLMs) such as OpenAI’s ChatGPT reflect, and can potentially perpetuate, social biases in language use. Conceptual engineering aims to revise our concepts to eliminate such bias. We show how machine learning and conceptual engineering can be fruitfully brought together to offer new insights to both conceptual engineers and LLM designers. Specifically, we suggest that LLMs can be used to detect and expose bias in the prototypes associated with concepts, and that LLM de-biasing can serve conceptual engineering projects (...) that aim to revise such conceptual prototypes. At present, these de-biasing techniques primarily involve approaches requiring bespoke interventions based on choices of the algorithm’s designers. Thus, conceptual engineering through de-biasing will include making choices about what kind of normative training an LLM should receive, especially with respect to different notions of bias. This offers a new perspective on what conceptual engineering involves and how it can be implemented. And our conceptual engineering approach also offers insight, to those engaged in LLM de-biasing, into the normative distinctions that are needed for that work. (shrink)

It is widely presumed that intuitions about thought experiments can help overturn philosophical theories. It is also widely presumed, albeit implicitly, that if thought experiments play any epistemic role in overturning philosophical theories, it is via intuition. In this paper, I argue for a different, neglected epistemic role of philosophical thought experiments, that of improving some reasoner’s appreciation both of what a theory’s predictions consist in and of how those predictions tie to elements of the theory. I call this role (...) theory clarification. I show that theory clarification does not proceed via intuition, and I argue that it is only in conjunction with theory clarification that intuitions about thought experiments can help overturn philosophical theories. I close by sketching how a more radical view might be true, on which thought experiments help justify the rejection of philosophical theories exclusively by clarifying theories, not by any intuitions those thought experiments might generate. (shrink)

This paper aims to reassess a notion in the works of the later Husserl that is both historically important and philosophically insightful, but remains understudied, namely, that of type. In opposition to a standard reading which treats Husserl’s type presentations as pre-conceptual habits, this paper argues that these representations are a specific kind of concept. More precisely, it shows that Husserl’s account of type presentations is akin to the contemporary prototype theory of concepts. This is historically important, since the predecessor (...) of the prototype theory is usually said to be Wittgenstein. From a philosophical standpoint, the paper shows that Husserl has an innovative account of the connection between type concepts and their extension. Contrary to the standard view of extensions as sets and thus sharp entities, Husserl develops a correlationalist theory of concepts, according to which, for the specific characteristics in the structure of a concept, there are corresponding characteristics in the arrangement of its members, and vice versa. According to this theory, while sharp concepts lead to sharp extensions, vague concepts such as (proto)type concepts lead to vague extensions. The paper presents this understanding of Husserl in detail and explains its philosophical significance. (shrink)