Nina Poth criticizes Tenenbaum Griffiths’ Bayesian model of perceptual generalization for its “semantic opaqueness” – roughly, it is unclear what the model’s probabilistic states are about – which suggests a connection with AI opacity. The first half of this paper analyzes the connection and suggests “semantically deficient models of cognition” (SEDMOC) as improved terminology. Crucially, for some purposes and in some contexts, models of cognition should reveal to cognitive scientists how semantic content produces behavior. While this is not at issue in the standard opacity debate, semantic opacity is one potential source of deficiency. The second half of this paper investigates hybrid models that combine Bayesian reasoning with geometric similarity (more specifically with Conceptual Spaces), thereby avoiding deficiency, and proposes to let the probability function shape spaces in a hierarchical model, outlining a PCA-based abstraction mechanism. It is argued that the proposed mechanism may introduce opaque dimensions which, however, are not semantically deficient. Both the terminological shift and the PCA-based abstraction mechanism contribute to better model-based explanations of cognitive behavior.
This editorial introduces a special issue collecting seven papers from the seventh Philosophy of Language and Mind (PLM) conference, held at Charles University, Prague, in August 2024. The papers span topics from singular reference and presupposition to speech-act theory, response-dependent theories of meaning, autistic masking and self-understanding, and Lewisian self-ascription, converging on a shared theme: how meaning and understanding are shaped and renegotiated across individuals, contexts, and social norms.
This paper clarifies and then indicates a conflict between, on the one hand, influential and longstanding ideas in the epistemology of imagination and, on the other, an emerging view of the accuracy conditions of remembering and the content of episodic memory—namely, alethism. Although we explore several ways to resolve the conflict, we take the most plausible to be to reject alethism. Drawing on assumptions made on both sides of debates about the epistemic limits of imagination, the paper constitutes an extended argument in favour of (‘strong’) authenticism about the accuracy conditions of remembering. Authenticists can maintain that experience is necessary for certain forms of imagining and also that episodic memory is what supports and explains this fact by virtue of providing a privileged store of experiential knowledge.
Shame is often portrayed as an evolutionarily ancient social emotion that manages reputation through appeasement and withdrawal. Yet it also seems capable of moral autonomy. It can arise in private, persist without witnesses, and motivate self-criticism and repair even when social approval is absent or explicitly rejected. This dual profile generates the tension between heteronomous shame, which is responsive to social evaluation, and autonomous shame, which is responsive to self-endorsed moral standards. Building on Kurth’s identity-based account, I develop a constructionist predictive processing reconstruction that makes the underlying architecture explicit. On my view, shame is constructed when temporally deep self-models and audience models jointly generate an identity-level disturbance. Whether an episode takes a heteronomous or autonomous shape depends on how a predictive control system allocates precision across competing expectations, such as audience-facing cues and moral identity priors that structure self-conception over time. This framework models graded, mixed, and shifting shame profiles within a single control architecture, rather than requiring distinct kinds of shame. Finally, I argue that the evolutionary evidence often cited in support of a dedicated shame module can be accommodated by selection for domain-general predictive and social-regulatory capacities, with shame emerging as one way those capacities are recruited in norm-governed social niches.
Many traditional metaethicists think that laypeople believe morality is objective. This has led many experimental philosophers and psychologists to investigate laypeople’s beliefs about moral objectivity. In this literature on folk moral objectivism, it has been proposed that the construct validity of one of its survey-based methods, the disagreement task, could be improved by including dedicated options for error theory and non-cognitivism, the two main non-objectivist theories in traditional metaethics. In this paper, I argue that, if the aim of the disagreement task is understood as capturing laypeople’s implicit metaethical commitments, this proposal involves misunderstandings of traditional metaethics. Both error theory and non-cognitivism predict that laypeople would not select the dedicated options for these theories. In particular, error theory predicts that, if true, laypeople would select the option indicating a belief in moral objectivity, whereas non-cognitivism, by itself, either yields the same prediction or makes no prediction at all.
Despite our proclivity to make either tacit or overt passage of time judgments in daily life, insufficient work has been done to uncover the neurocognitive mechanisms underpinning such determinations. A prima facie promising account of passage of time judgments derives from the predictive processing framework. More precisely, Hohwy et al. (2016) propose that the reported speed of the experienced passage of time is a function of how rapidly the perceptual system "distrusts the present" current perceptual hypothesis in favour of a new one. We will propose that this account suffers from an incapacity to explain the relevant empirical data related to passage of time judgments and also possesses its own internal shortcomings. In its place, we offer a predictive processing theory of passage of time judgments based on "error dynamics", according to which the sense of time moving quickly is grounded in the cognitive system’s subdoxastic detection that it is minimising prediction error at a faster rate than it expected. If, on the other hand, prediction error is being minimised at a slower rate than expected, time will be sensed to be moving slowly, and a more protracted PoTJ will be reported. We will demonstrate the internal consistency of this proposition, as well as its ability to account for the key empirical findings any model of passage of time judgments ought to explain.
Current AI-assisted debiasing interventions assume a defeatist premise: biased minds cannot be changed, only contained. These “outcome-focused” tools mask demographic information or correct decisions post hoc, leaving underlying psychological mechanisms untouched. We argue that this pessimism is premature. The disappointing record of “change-based” interventions partly stems from a mismatch between simple intervention designs and the pluralistic, dynamically interacting memory systems that sustain bias. Drawing on 4E cognitive science, Intergroup Contact Theory, multiple memory systems research, and the Heterarchical Control Network model of implicit bias, we articulate six design principles for next-generation interventions and demonstrate how Generative AI and immersive Virtual Reality make it possible to scale optimal debiasing conditions. Using recruitment as our example, we sketch a training module where recruiters engage in intergroup interaction under controlled yet ecologically valid conditions. Finally, we address the ethical risks that arise when AI is used not merely to sanitize outputs, but to transform minds.
A pressing question for the philosophy of personal identity in the digital age is the extent to which people can be identical to their various digital self-representations, from virtual avatars and social media profiles to AI digital duplicates and future mind uploads. This article addresses this question in the case of virtual avatars: user-controlled, visual representations of self in online environments like video games and virtual reality worlds. I interpret the metaphysical relationship between avatar and user implied by the ‘avatar identity question’ through the lens of the extended self thesis. Drawing on the narrative theory of identity, I argue for the avatar extended self (AES): the view that virtual avatars can become constitutive components of a user’s personhood. After establishing criteria for avatar-user narrative integration and responding to objections to the view, I examine its ethical implications, arguing that AES raises the ethical significance of virtual actions involving one’s avatar and, in so doing, sheds light on the question of what grounds the moral weight of virtual phenomena.
Most research in experimental metaethics suggests high levels of interpersonal and intrapersonal variation in support for moral realism and antirealism among nonphilosophers (i.e., people without significant training in philosophy). These findings challenge the assumption that nonphilosophers share a uniform commitment to moral realism. Recent evidence challenges the validity of these studies by demonstrating that many people do not interpret questions about metaethics as intended. Pölzler and Wright offer a potential solution: train participants to understand the relevant metaethical distinctions. They devised a set of paradigms that feature training exercises, comprehensive instructions, and detailed response options designed to mitigate unintended interpretations. I argue that their efforts were inadequate, and that the training required to ensure participants interpret stimuli as intended would need to be so extensive that it would effectively transform participants into novice philosophers. This leads to a dilemma: if training is unable to ensure adequate rates of intended interpretation, then it can’t resolve the threat to validity it was designed to address, while if it succeeds, participants will no longer represent nonphilosophers, which undermines the study’s external validity. Since research on other philosophical topics may likewise suffer from low rates of intended interpretations, e.g., free will and consciousness, this dilemma may have broader implications for experimental philosophy and psychology.
Autistic people often hide or alter their behavior in order to meet societal expectations and fit in—a phenomenon known as masking. Masking has important benefits for neurodivergent individuals, from increasing social acceptance to enhancing their career opportunities and ability to navigate social hierarchies. But it also comes with significant costs. It often leads to mental and emotional exhaustion, increased anxiety, delayed or missed diagnosis, or even suicidal ideation. In this paper, we focus on a relatively underexplored cost of masking, namely, its negative impact on subjects’ self-understanding. We appeal to the framework of mindshaping, which suggests that our thoughts, behaviors, and mental states are shaped by the social norms and expectations to which we are exposed, and argue that masking can be conceptualized as a natural reaction of neurodivergent individuals to neurotypical mindshaping. As a result, we suggest, masking commonly leads to an experience that can be characterized as “self-mask ambiguity”—a failure to demarcate and differentiate oneself from one’s mask. In light of this, we consider ways in which masking and self-mask ambiguity can diminish individuals’ self-understanding. We finish by discussing how diagnosis and unmasking can lead to a deepening of one’s self-understanding.
We do not only use language to avow our emotions but also to deny having them. Negative avowals – utterances such as “I’m not happy,” “I’m not scared,” or “I’m not angry” – raise an important but neglected question: how can the absence of a mental state be expressed in speech? I argue that negative avowals can, like positive ones, function as explicit expressive acts. This phenomenon, which I call expressive denegation, shows that speakers can make mental state absence itself linguistically manifest. To address the resulting puzzle, I develop propositional expressivism, on which what is expressed is not a mental state but a proposition about the speaker’s mental condition – namely, that she is (not) in the state named. This framework preserves the core insight of avowal expressivism while extending it to negative avowals.
A new explanatory challenge of time perception presented by Gerardo Viera (2020) lies in reconciling three different ways time appears perceptually unified with the empirical data of how the mechanisms underlying time perception are fragmented and dissociable. Viera (2020) argues that existing models of time perception that locate this unity at the level of mechanisms fail to meet this explanatory challenge. I propose that this challenge can be met if we adopt Pedersen’s (2024) hybrid theory of temporal binding. This theory accounts for the temporal organisation of perception without taking it to be a function of synchronous processing or some centralised temporal integration mechanism. Instead, this organisation depends on the attentional modulation of hybrid retrodictive and predictive sensory processing. Effectively, the accounts suggest that temporal organisation of perception is achieved through selective attentional modulation that biases perception towards representing some subset sensory features that make up coherent and behaviourally useful perceptions. I explain how we can use this theory to account for the different kinds of perceived unity of time, proposed by Viera, while also accounting for the disunified organisation of the underlying mechanisms. This view entails that we should deflate the phenomenological richness of this perceived unity relative to the limits and selectivity of attention.
Actions expressive of emotion are widely thought to pose a problem for the view that actions admit of rational explanation. Supposing the standard, instrumentalist view of practical rationality, such actions seem not to admit of practical reasons. In this paper, I assess a distinctive line of response to this problem. As Helm (2001, 2016) and Bennett (2016, 2021) have proposed, we can make sense of expressive actions as rational since they are responses to a specific type of evaluative reason, which differs from paradigm instrumental reasons. Whilst sympathetic to this line of response, I argue that Helm and Bennett go awry in developing it since they rely on an inadequate view of the expressive character of expressive actions. As they suppose, emotions are at least in part a specific form of evaluation and expressive actions express this evaluation. This view, I propose, fails to make expressive actions intelligible as rational responses to evaluative reasons. I indicate an alternative way of developing the view, on which it lives up to its promise.
Examining some recent projects in the philosophy of conspiracy theory theory (the academic study of conspiracy theories) that portray belief in conspiracy theories as in some sense generally bad, I argue that such projects are prone to failure. They either rely on overly restrictive definitions of what counts as a ‘conspiracy theory’ (i.e. they turn out to be about a subset of such theories), or overly permissive definitions of what falls under the label ‘conspiracy theory’ (i.e. the account ends up including a lot of beliefs we typically—and for good reason—tend not to consider problematic). I then argue that a further problem is that many of these projects tend to change the subject or topic of debate rather than work with the emerging philosophical consensus (AKA particularism) of a non-pejorative definition of ‘conspiracy theory’.
Bilinguals often choose more utilitarian options when reasoning in a second language. This pattern is commonly read as improved rationality through emotional distance. I argue the better explanation is processing cost and reduced expressive capacity: these nudge people toward simpler, outcome-counting heuristics and thinner deontic justification. The critique has prior defenders, who drew on the somatic marker hypothesis to show that native-language reasoning tends to produce more ethical choices. The present paper extends that line of argument in a specific direction. Model-based decomposition studies show that the foreign-language effect reduces both deontological and utilitarian inclinations simultaneously; the apparent utilitarian shift arises because deontological sensitivity drops further in high-conflict dilemmas, not because utilitarian motivation increases. This dissociation is the empirical backbone my justificatory thinning account requires. Emotions here are not mere noise but epistemic inputs that supply moral concepts and appraisals; their attenuation under expressive constraint matters for how decisions are justified, not only chosen. I propose a processing-cost mechanism with testable predictions by proficiency and dilemma type, a five-item coding scheme for justificatory texture, and practical safeguards for multilingual deliberation – including native-language stages at key decision points, slower pacing with interpreters, and records that separate outcomes from justifications.
In this paper I defend the normative power account of forgiveness against the charge that it cannot accommodate the intimate connection between forgiveness and emotion. I look at the grounds for thinking that emotion is central to forgiveness. I then show how philosophers have developed this idea in accounts which make emotional changes definitive of forgiveness. Against such views, I show that serious problems face any attempt to make emotional change definitive of forgiveness. Next, I explain the normative power account and some of its advantages; I show that the normative power account escapes the problems of the emotional change view. I then consider the objection to the normative power account mentioned above: that it cannot adequately explain the role of emotion in forgiveness. I develop this objection by reference to a recent discussion by Luke Russell. I consider two initial responses to this objection that can be made by a proponent of the normative power account: that forgiveness does not always need to be emotional; and that forgiveness can involve undertaking a commitment to emotional change. I then develop the normative power account, explaining two ways in which forgiveness can be ‘expressive.’ The normative power of forgiveness may require emotion for its exercise. And forgiveness may be a powerful expressive action through which the forgiver marks their commitment to an ongoing relationship.
Language utterances are meaningful because they mean something to someone. This truism motivates regarding linguistic meaning as a response-dependent property of language utterances. But the properties called ‘response-dependent’ in the literature differ substantially. This paper discusses three kinds with respect to their metaphysics, calling them by their more precise bynames ‘judgment-dependent’, ‘response-dispositional’, and (genuinely) ‘response-dependent’. Looking at what it takes for something to have such a property, we can devise a test to determine which of the three kinds of property we are dealing with. This test is then used to corroborate Goldberg’s (2012) finding that Davidson’s two accounts of linguistic meaning, the Radical Interpretation (RI) account and the account from Language Learning (LL) by triangulation, while both response-dependent (in the broad sense), differ in kind. The test shows that (RI) is an account of meaning as a judgment-dependent property, (LL) as a response-dependent property. But whereas Goldberg argued for giving up (LL) and retaining (RI) in order to have a single account of linguistic meaning, this paper argues in favour of (LL) over (RI). It shows that (i) (RI) cannot exist without (LL), and (ii) (LL) covers more ground than (RI) because it can account for both speaker and listener meaning. So, while (RI) may evolve out of (LL) and has its place in some situations, (LL) is the basis we need in order to account for linguistic meaning.
This paper examines the structure and philosophical implications of the types of perceptions described by the constructions "perceiving in" and "perceiving through": the perception of emotions thanks to emotional expression, picture-based perception and perception thanks to physical traces. Despite their apparent differences, it is argued that these cases share four fundamental characteristics: they are specifically mediated perceptions (mediated through physical changes, pictures, or imprints), direct and non-inferential in nature, involve objects which are not in the perceiver's line of sight, and are dependent on their objects. A unified account of these three cases is provided which analyzes their underlying relations of mediation. The investigation begins with trace-based perception as the basic case, arguing that it allows us to understand both emotional and pictorial perception. This approach pays particular attention to the neglected ontological structure of media, traces and informational transmission.
The representation-hungry challenge is a popular challenge leveled against anti-representational approaches to the mind sciences. It contends that any general anti-representationalist conclusion is presently premature, as anti-representationalists have not yet offered a non-representational account of representation-hungry cognition; that is, cognition aimed at absent or abstract targets. Anti-representationalist tried to face this challenge in a number of different ways, but these attempts failed to convince representationalists. Hence, the need for an alternative way to handle the challenge. Here, I provide such an alternative: rather than facing the representation-hungry challenge, I will try to defuse it, showing that it does arise in any meaningful form. I’ll do so by arguing that the challenge hinges on a presupposition that is factually false, and so that the challenge fails to arise. Additionally, I will argue that even if the challenge were to arise, it won’t be a serious challenge to anti-representationalism. For, given certain, widespread representationalists assumptions, must concede that anti-representationalists have already met the challenge, and that no additional explanatory work is required on their part.
This paper is an attempt to give shape to the idea of an ‘expressive emotion.’ An emotion is expressive, I argue, if it is best explained by reference to an expressive action. An expressive action is an action that has expressive or symbolic properties that relate non-arbitrarily to the significance of a situation, and which non-instrumentally marks the situation as in some way significant. I argue for six main conclusions about expressive emotions and their relation to expressive action. First, their psychological role is the same as that of expressive actions, namely, to lift situations out of the mundane and mark them as in some way significant. Second, their phenomenology is best explained by reference to the symbolism of an expressive action. Third, they centrally involve an attitude to an expressive action as called for in the situation. Fourth, the conditions of fittingness specified by the emotion’s formal object are shared with the conditions of appropriateness of an expressive action. Fifth, they are given expression in the relevant expressive action; this is the action towards which the emotion has a motivational tendency. Sixth, they are emotions for which there can be non-instrumental reasons; and the nature of such reasons is best explained by reference to an expressive action. I argue that reference to expressive action helps to explain how emotions are to be individuated; in what their unity consists; and how there can be non-instrumental reasons for emotions. I do not claim that the expressive theory applies to everything that is called ‘emotion.’ Emotions that I am interested in include pride, shame, guilt, joy, disappointment, awe. In this paper I focus on the emotion of guilt, arguing that it is best understood in relation to the expressive action of dissociation (from wrongdoing).
