Sarah A Fisher

In Concepts at the Interface, Shea characterizes concepts as “plug and play devices” that seamlessly integrate information from across our psychology. They are thus positioned to orchestrate heterogeneous thought processes, ranging from general reasoning that depends on the language-like compositionality of conceptual representations, to the construction of multimodal simulations based on highly specific knowledge and experience. I will argue here that Shea’s account can help reinvigorate an influential but beleaguered paradigm in cognitive science: dual process theory. Drawing on Shea’s analysis, I explore three ways of distinguishing between type 1 “intuitive” processing and type 2 “reflective” processing. In the pluralistic spirit of his book, I show how each of these distinctions (and potentially others) could explain different portions of the empirical data.

Since Austin’s introduction of the locutionary-illocutionary-perlocutionary distinction, it has been a matter of debate within speech act theory whether illocutionary acts like promising, warning, refusing and telling require audience ‘uptake’ in order to be performed. Philosophers on different sides of this debate have tried to support their positions by appealing to hypothetical scenarios, designed to elicit intuitive judgements about the role of uptake. However, philosophers’ intuitions appeared to remain deadlocked, while laypeople’s intuitions have not yet been probed. To begin rectifying that, we ran two experiments probing lay intuitions about the implications of uptake failure. Overall, we found that participants’ responses were skewed towards agreement that speech acts were performed, despite the lack of uptake. There were, however, significant differences across the four different speech act types we investigated (with the highest levels of agreement found for refusing, followed by warning, then telling, and finally promising). We also obtained evidence of complex effects relating to the (high or low) stakes involved in the scenarios. While this study only represents an initial exploration of intuitions about uptake, our results form a basis for further research into their nature and significance, across a range of speech acts, scenarios, and experimental designs.

The debate between Semantic Minimalism and Radical Contextualism is standardly characterized as concerning truth-evaluability – specifically, whether or not sentences require rich contextualization in order to express complete, truth-evaluable contents. In this paper, I examine the notion of truth-evaluability, considering which kinds of mappings it might require from worldly states of affairs to truth-values. At one end of the spectrum, an exhaustive notion would require truth-evaluable contents to map all possible states of affairs to truth-values. At the other end, a liberal notion would require only that truth-evaluable contents map at least one possible state of affairs to at least one truth-value. I show that both Minimalists and Radical Contextualists rely on some intermediate, moderately strict notion of truth-evaluability, falling between these two poles. I consider four ways in which such a notion could be defined. However, I argue that each of these is ultimately implausible, giving us no reason to favour a moderately strict notion of truth-evaluability over the liberal alternative. This suggests that the debate must shift to more moderate ground; rather than concerning the in principle possibility of truth-evaluable contents, it fundamentally hinges on their explanatory value. More generally, paying close attention to the notion of truth-evaluability allows us to tease apart distinct strands in the Minimalism-Contextualism debate, and gain a better appreciation of what is at stake.

Artificially generated content threatens to seriously disrupt the public sphere. Generative AI massively facilitates the production of convincing portrayals of fabricated events. We have already begun to witness the spread of synthetic misinformation, political propaganda, and non-consensual intimate deepfakes. Malicious uses of the new technologies can only be expected to proliferate over time. In the face of this threat, social media platforms must surely act. But how? While it is tempting to think they need new sui generis policies targeting synthetic content, we argue that the challenge posed by generative AI should be met through the enforcement of general platform rules. We demonstrate that the threat posed to individuals and society by AI-generated content is no different in kind from that of ordinary harmful content—a threat which is already well recognised. Generative AI massively increases the problem but, ultimately, it requires the same approach. Therefore, platforms do best to double down on improving and enforcing their existing rules, regardless of whether the content they are dealing with was produced by humans or machines.

Newly powerful large language models have burst onto the scene, with applications across a wide range of functions. We can now expect to encounter their outputs at rapidly increasing volumes and frequencies. Some commentators claim that large language models are bullshitting, generating convincing output without regard for the truth. If correct, that would make large language models distinctively dangerous discourse participants. Bullshitters not only undermine the norm of truthfulness (by saying false things) but the normative status of truth itself (by treating it as entirely irrelevant). So, do large language models really bullshit? I argue that they can, in the sense of issuing propositional content in response to fact-seeking prompts, without having first assessed that content for truth or falsity. However, I further argue that they need not bullshit, given appropriate guardrails. So, just as with human speakers, the propensity for a large language model to bullshit depends on its own particular make-up.

In a recent case, a Facebook user in Iran posted “death to Khamenei,” which the platform removed as a violation of its policy against threats and incitement. Facebook ultimately overturned the decision on the grounds that the speech, while contravening its rules, was newsworthy. Yet the company’s Oversight Board offered a distinct rationale for allowing the post: “death to Khamenei” was not a threat or an incitement at all, but rather a rhetorical expression of criticism, disdain, or disgust. Who was right? Drawing on longstanding distinctions in the philosophy of language, we argue that the answer depends on whether platform rules should primarily target an utterance’s illocutionary force or instead its perlocutionary effects. Using this case study to trace broader lessons for the proper governance of online speech, we argue that platforms’ rules should primarily target illocutionary force, only considering perlocutionary effects in cases where illocutionary force is ambiguous.

Synthetic content, which has been produced by generative artificial intelligence, is beginning to spread through the public sphere. Increasingly, we find ourselves exposed to convincing ‘deepfakes’ and powerful chatbots in our online environments. How should we mitigate the emerging risks to individuals and society? This article argues that labelling synthetic content in public forums is an essential first step. While calls for labelling have already been growing in volume, no principled argument has yet been offered to justify this measure (which inevitably comes with some additional costs). Rectifying that deficit, I conduct a close examination of our epistemic and expressive interests in identifying synthetic content as such. In so doing, I develop a cumulative case for social media platforms to enforce a labelling duty. I argue that this represents an important element of good platform governance, helping to shore up the integrity of our contemporary public discourse, which takes place increasingly online.

This chapter explores an injustice that arises as people switch between languages (‘interlinguistic injustice’). It is common that speakers moving between languages encounter mismatches between the distinct sets of hermeneutical resources available to them. We argue that, for speakers from less powerful linguistic communities, or those with marginalised social identities, these mismatches become systematic barriers to communicating significant areas of their social experience in a hegemonic tongue. By expanding on the literature on epistemic injustice, we argue that these barriers amount to an injustice with a distinctive epistemic dimension that has yet to be recognised. We start by showing that linguistic mismatches can take the form of ‘voids’ or ‘surpluses’, can reflect greater or lesser (mis)alignment in surrounding networks of meanings, and can affect speakers’ ability to contribute to a linguistic community’s collective understanding. Crucially, given the inequalities built into the structures of languages and societies, the epistemic harms are unlikely to fall equally: marginalised speakers are most likely to lose testimonial clout. We end by discussing some mitigation strategies, examining their applicability to different forms of linguistic mismatch, and their distribution of cognitive burdens between speakers and audiences.

According to some philosophers, a sentence's semantics can fail to constitute a complete propositional content, imposing mere constraints on such a content. Recently, Daniel Harris has begun developing a formal constraint semantics. He claims that the semantic values of sentences constrain what speakers can literally say with them—and what hearers can know about what was said. However, that claim is undermined by his conception of semantics as the study of a psychological module. I argue instead that semantic constraints should be understood as properties of public languages.

Since Austin’s introduction of the locutionary-illocutionary-perlocutionary distinction, it has been a matter of debate within speech act theory whether illocutionary acts like promising, warning, refusing and telling require audience ‘uptake’ in order to be performed. Philosophers on different sides of this debate have tried to support their positions by appealing to hypothetical scenarios, designed to elicit intuitive judgements about the role of uptake. However, philosophers’ intuitions appeared to remain deadlocked, while laypeople’s intuitions have not yet been probed. To begin rectifying that, we ran two experiments probing lay intuitions about the implications of uptake failure. Overall, we found that participants’ responses were skewed towards agreement that speech acts were performed, despite the lack of uptake. There were, however, significant differences across the four different speech act types we investigated (with the highest levels of agreement found for refusing, followed by warning, then telling, and finally promising). We also obtained evidence of complex effects relating to the (high or low) stakes involved in the scenarios. While this study only represents an initial exploration of intuitions about uptake, our results form a basis for further research into their nature and significance, across a range of speech acts, scenarios, and experimental designs.

According to standard philosophical and clinical understandings, pain is an essentially mental phenomenon. In a challenge to this standard conception, a recent burst of empirical work in experimental philosophy, such as that by Justin Sytsma and Kevin Reuter, purports to show that people ordinarily conceive of pain as an essentially bodily phenomenon—specifically, a quality of bodily disturbance. In response to this bodily view, other recent experimental studies have provided evidence that the ordinary concept of pain is more complex than previously assumed: rather than tracking only bodily or only mental aspects of pain, the ordinary concept of pain can actually track either of these aspects. The polyeidic analysis of the folk concept of pain, as proposed by Emma Borg et al., captures this complexity. Whereas previous empirical support for the polyeidic view has focused on the context-sensitivity of the folk concept of pain, here we discuss individual differences in people’s ‘pain priors’—namely, their standing tendencies to think of pain in relatively mind-centric or body-centric ways. We describe a preliminary empirical study and present a small number of findings, which will be explored further in future work. The results we discuss are part of a larger programme of work which seeks to integrate philosophical pain research into clinical practice. For example, we hypothesise that variations in how patients with chronic pain are thinking about pain could help predict their responses to treatment.

This article refines a foundational tenet of rational choice theory known as the principle of description invariance. Attempts to apply this principle to human agents with imperfect knowledge have paid insufficient attention to two aspects: first, agents’ epistemic situations, i.e. whether and when they recognize alternative descriptions of an object to be equivalent; and second, the individuation of objects of description, i.e. whether and when objects count as the same or different. An important consequence is that many apparent ‘framing effects’ may not violate the principle of description invariance, and the subjects of these effects may not be irrational.

It is common in philosophy of language to recognise two different kinds of linguistic meaning: literal or conventional meaning, on the one hand, versus communicated or conveyed meaning, on the other. However, once we recognise these two types of meaning, crucial questions immediately emerge; for instance, exactly which meanings should we treat as the literal (semantic) ones, and exactly which appeals to a context of utterance yield communicated (pragmatic), as opposed to semantic, content? It is these questions and, specifically, how we should model the relationship between semantic content and utterance context, that is the topic of this chapter. We explore five contemporary answers to this modelling question, considering the benefits and challenges of each, before closing by examining some potential new directions for debates in this area.

This article surveys the latest research on risky-choice framing effects, focusing on the implications for rational decision-making. An influential program of psychological research suggests that people's judgements and decisions depend on the way in which information is presented, or ‘framed’. In a central choice paradigm, decision-makers seem to adopt different preferences, and different attitudes to risk, depending on whether the options specify the number of people who will be saved or the corresponding number who will die. It is standardly assumed that such responses violate a foundational tenet of rational decision-making, known as the principle of description invariance. We discuss recent theoretical and empirical research that challenges the dominant ‘irrationalist’ narrative. These approaches typically pay close attention to how decision-makers represent decision problems (including their interpretation of numerical quantifiers or predicate choice) and they highlight the need for a more robust characterization of the description invariance principle. We conclude by indicating avenues for future research that could bring us closer to a complete—and potentially rationalizing—explanation of framing effects.

Human judgements are affected by the words in which information is presented —or ‘framed’. According to the standard gloss, ‘framing effects’ reveal counter-normative reasoning, unduly affected by positive/negative language. One challenge to this view suggests that number expressions in alternative framing conditions are interpreted as denoting lower-bounded (minimum) quantities. However, it is unclear whether the resulting explanation is a rationalising one. I argue that a number expression should only be interpreted lower-boundedly if this is what it actually means. I survey how number expressions might be assigned lower-bounded meanings, due to their conventional semantics or pragmatic enrichment in context. I argue that deciding between these possibilities requires foundational philosophical input.

Framing effects occur when people respond differently to the same information, just because it is conveyed in different words. For example, in the classic ‘Disease Problem’ introduced by Amos Tversky and Daniel Kahneman, people’s choices between alternative interventions depend on whether these are described positively, in terms of the number of people who will be saved, or negatively in terms of the corresponding number who will die. In this paper, I discuss an account of framing effects based on ‘fuzzy-trace theory’. The central claim of this account is that people represent the numbers in framing problems in a ‘gist-like’ way, as ‘some’; and that this creates a categorical contrast between ‘some’ people being saved (or dying) and ‘no’ people being saved (or dying). I argue that fuzzy-trace theory’s gist-like representation, ‘some’, must have the semantics of ‘some and possibly all’, not ‘some but not all’. I show how this commits fuzzy-trace theory to a modest version of a rival ‘lower bounding hypothesis’, according to which lower-bounded interpretations of quantities contribute to framing effects by rendering the alternative descriptions extensionally inequivalent. As a result, fuzzy-trace theory is incoherent as it stands. Making sense of it requires dropping, or refining, the claim that decision-makers perceive alternatively framed options as extensionally equivalent; and the related claim that framing effects are irrational. I end by suggesting that, whereas the modest lower bounding hypothesis is well supported, there is currently less evidence for the core element of the fuzzy trace account.

According to standard philosophical and clinical understandings, pain is an essentially mental phenomenon (typically, a kind of conscious experience). In a challenge to this standard conception, a recent burst of empirical work in experimental philosophy, such as that by Justin Sytsma and Kevin Reuter, purports to show that people ordinarily conceive of pain as an essentially bodily phenomenon—specifically, a quality of bodily disturbance. In response to this bodily view, other recent experimental studies have provided evidence that the ordinary (‘folk’) concept of pain is more complex than previously assumed: rather than tracking only bodily or only mental aspects of pain, the ordinary concept of pain can actually track either of these aspects. The polyeidic (or ‘many ideas’) analysis of the folk concept of pain, as proposed by Emma Borg et al., captures this complexity. Whereas previous empirical support for the polyeidic view has focused on the context-sensitivity of the folk concept of pain, here we discuss individual differences in people’s ‘pain priors’—namely, their standing tendencies to think of pain in relatively mind-centric or body-centric ways. We describe a preliminary empirical study and present a small number of findings, which will be explored further in future work. The results we discuss are part of a larger programme of work which seeks to integrate philosophical pain research into clinical practice. For example, we hypothesise that variations in how patients with chronic pain are thinking about pain could help predict their responses to treatment.

I use the psychological phenomenon of ‘attribute framing’ as a case study for exploring philosophical conceptions of semantics and the semantics-pragmatics divide. Attribute frames are pairs of sentences that use contradictory expressions to predicate the same property of an individual or object. Despite their equivalence, pairs of attribute frames have been observed to induce systematic variability in hearers’ responses. One explanation of such framing effects appeals to the distinct ‘reference point information’ conveyed by alternative frames. Although this information is taken to be pragmatic, a live debate in the philosophy of language concerns the extent to which pragmatic information can really be held apart from semantics. Therefore, in this article I examine whether – and how – reference point information might be thought to intrude on the semantics of an attribute frame. I describe two ways in which the information might be considered broadly semantic, due to its being directly communicated or conventional. However, I argue that the framing data in fact support a narrower conception of semantics, and a semantics-pragmatics distinction that separates standing sentence meaning from enriched communicated meaning.

The debate between Semantic Minimalism and Radical Contextualism is standardly characterized as concerning truth-evaluability—specifically, whether or not sentences require rich contextualization in order to express complete, truth-evaluable contents. In this paper, I examine the notion of truth-evaluability, considering which kinds of mappings it might require from worldly states of affairs to truth-values. At one end of the spectrum, an exhaustive notion would require truth-evaluable contents to map all possible states of affairs to truth-values. At the other end, a liberal notion would require only that truth-evaluable contents map at least one possible state of affairs to at least one truth-value. I show that both Minimalists and Radical Contextualists rely on some intermediate, moderately strict notion of truth-evaluability, falling between these two poles. I consider four ways in which such a notion could be defined. However, I argue that each of these is ultimately implausible, giving us no reason to favour a moderately strict notion of truth-evaluability over the liberal alternative. This suggests that the debate must shift to more moderate ground; rather than concerning the in principle possibility of truth-evaluable contents, it fundamentally hinges on their explanatory value. More generally, paying close attention to the notion of truth-evaluability allows us to tease apart distinct strands in the Minimalism-Contextualism debate, and gain a better appreciation of what is at stake.

This paper considers one aspect of the relationship between language and thought, focusing on a theory proposed by José Luis Bermúdez. Bermúdez argues that language is required for any kinds of thinking that involve thinking about thoughts, or what he calls 'intentional ascent'. I argue, to the contrary, that Bermúdez gives us no reason to suppose language is necessary for all instances of thinking about thoughts. Whilst I am broadly sympathetic to supra-communicative approaches to language, I seek to deny Bermúdez’s particular thesis; I do so by showing that alternative representational formats, specifically maps, could play qualitatively the same role that language is supposed to play in his account of thinking about thoughts.