This article addresses an open problem in the area of cognitive systems and architectures: namely the problem of handling (in terms of processing and reasoning capabilities) complex knowledge structures that can be at least plausibly comparable, both in terms of size and of typology of the encoded information, to the knowledge that humans process daily for executing everyday activities. Handling a huge amount of knowledge, and selectively retrieve it according to the needs emerging in different situational scenarios, is an important (...) aspect of human intelligence. For this task, in fact, humans adopt a wide range of heuristics (Gigerenzer & Todd) due to their “bounded rationality” (Simon, 1957). In this perspective, one of the requirements that should be considered for the design, the realization and the evaluation of intelligent cognitively-inspired systems should be rep- resented by their ability of heuristically identify and retrieve, from the general knowledge stored in their artificial Long Term Memory (LTM), that one which is synthetically and contextually relevant. This requirement, however, is often neglected. Currently, artificial cognitive systems and architectures are not able, de facto, to deal with complex knowledge structures that can be even slightly comparable to the knowledge heuris- tically managed by humans. In this paper I will argue that this is not only a technological problem but also an epistemological one and I will briefly sketch a proposal for a possible solution. (shrink)

Todd and Rabern (2021) mount an argument that – contra both Thomason’s (1970) supervaluationism and MacFarlane’s (2014) relativism – an “open future” view is incompatible with the principle they call “Retro-closure”, according to which today’s rain implies that yesterday it was true that it would rain a day later. In a recent piece, MacFarlane replies. This paper has two aims. First, I argue that MacFarlane’s response to Todd and Rabern is unsuccessful on its own terms. Second, I attempt (...) to clarify Todd and Rabern’s overall argument, and explain how MacFarlane’s replies should be construed within the overall dialectic. The intended result: if you want an "open future", then one's best option is a modified Peirceanism (Todd 2021); if one wants Retro-closure, one's best option is one on which there is a determinate "Thin Red Line" (a view sometimes called "Ockhamism"). However, one cannot have what supervaluationism and relativism both promise, viz., a view that preserves both. (shrink)

The principle of Conditional Excluded Middle has been a matter of longstanding controversy in both semantics and metaphysics. The principle suggests (among other things) that for any coin that isn't flipped, there is a fact of the matter about how it would have landed if it had been flipped: either it would have landed heads, or it would have landed tails. This view has gained support from linguistic evidence indicating that ‘would’ commutes with negation (e.g., ‘not: if A, would C’ (...) is equivalent to ‘if A, would not C’). There is, however, a long list of operators that similarly appear to commute with negation, even though the corresponding excluded middle principles are indefensible. We suggest that the data supporting Conditional Excluded Middle is best explained as a pragmatic effect. (shrink)

Various philosophers have wanted to move from a (P.F.) “Strawsonian” understanding of the “practices of moral responsibility” to a non-skeptical result. I focus on a strategy moving from a “response-dependent” theory of responsibility. I aim to show that a key analogy associated with this strategy fails to support a compatibilist result. It seems clear that nothing could show that nothing we have been laughing at has really been funny. If “the funny” is similar to “the blameworthy”, then perhaps it would (...) follow that nothing could show that no one we have ever blamed has really been blameworthy. The comparison is interesting, but inconclusive: even if we grant that these properties are normatively similar, the latter has substantive empirical presuppositions, whereas the former does not. One important upshot: even if the standards operative in our practices cannot be mistaken, it could nevertheless be that no one has ever met these standards Ruling out this possibility seemingly requires re-engaging in all of the so-called "metaphysical" debates Strawsonians are keen to avoid. (shrink)

This book launches a sustained defense of a radical interpretation of the doctrine of the open future. Patrick Todd argues that all claims about undetermined aspects of the future are simply false.

Recently, philosophers have turned their attention to the question, not when a given agent is blameworthy for what she does, but when a further agent has the moral standing to blame her for what she does. Philosophers have proposed at least four conditions on having “moral standing”: -/- 1. One’s blame would not be “hypocritical”. 2. One is not oneself “involved in” the target agent’s wrongdoing. 3. One must be warranted in believing that the target is indeed blameworthy for the (...) wrongdoing. 4. The target’s wrongdoing must some of “one’s business”. -/- These conditions are often proposed as both conditions on one and the same thing, and as marking fundamentally different ways of “losing standing.” Here I call these claims into question. First, I claim that conditions (3) and (4) are simply conditions on different things than are conditions (1) and (2). Second, I argue that condition (2) reduces to condition (1): when “involvement” removes someone’s standing to blame, it does so only by indicating something further about that agent, viz., that he or she lacks commitment to the values that condemn the wrongdoer’s action. The result: after we clarify the nature of the non-hypocrisy condition, we will have a unified account of moral standing to blame. Issues also discussed: whether standing can ever be regained, the relationship between standing and our "moral fragility", the difference between mere inconsistency and hypocrisy, and whether a condition of standing might be derived from deeper facts about the "equality of persons". (shrink)

It is widely accepted that there is what has been called a non-hypocrisy norm on the appropriateness of moral blame; roughly, one has standing to blame only if one is not guilty of the very offence one seeks to criticize. Our acceptance of this norm is embodied in the common retort to criticism, “Who are you to blame me?”. But there is a paradox lurking behind this commonplace norm. If it is always inappropriate for x to blame y for a (...) wrong that x has committed, then all cases in which x blames x (i.e. cases of self-blame) are rendered inappropriate. But it seems to be ethical common-sense that we are often, sadly, in position (indeed, excellent, privileged position) to blame ourselves for our own moral failings. And thus we have a paradox: a conflict between the inappropriateness of hypocritical blame, and the appropriateness of self-blame. We consider several ways of resolving the paradox, and contend none is as defensible as a position that simply accepts it: we should never blame ourselves. In defending this startling position, we defend a crucial distinction between self-blame and guilt. (shrink)

At least since Aristotle's famous discussion of the sea-battle tomorrow in On Interpretation 9, philosophers have been fascinated by a rich set of interconnecte.

There is a familiar debate between Russell and Strawson concerning bivalence and ‘the present King of France’. According to the Strawsonian view, ‘The present King of France is bald’ is neither true nor false, whereas, on the Russellian view, that proposition is simply false. In this paper, I develop what I take to be a crucial connection between this debate and a different domain where bivalence has been at stake: future contingents. On the familiar ‘Aristotelian’ view, future contingent propositions are (...) neither true nor false. However, I argue that, just as there is a Russellian alternative to the Strawsonian view concerning ‘the present King of France’, according to which the relevant class of propositions all turn out false, so there is a Russellian alternative to the Aristotelian view, according to which future contingents all turn out false, not neither true nor false. The result: contrary to millennia of philosophical tradition, we can be open futurists without denying bivalence. (shrink)

I provide a manipulation-style argument against classical compatibilism—the claim that freedom to do otherwise is consistent with determinism. My question is simple: if Diana really gave Ernie free will, why isn't she worried that he won't use it precisely as she would like? Diana's non-nervousness, I argue, indicates Ernie's non-freedom. Arguably, the intuition that Ernie lacks freedom to do otherwise is stronger than the direct intuition that he is simply not responsible; this result highlights the importance of the denial of (...) the principle of alternative possibilities for compatibilist theories of responsibility. Along the way, I clarify the dialectical role and structure of “manipulation arguments”, and compare the manipulation argument I develop with the more familiar Consequence Argument. I contend that the two arguments are importantly mutually supporting and reinforcing. The result: classical compatibilists should be nervous—and if PAP is true, all compatibilists should be nervous. (shrink)

In this paper, I introduce a problem to the philosophy of religion – the problem of divine moral standing – and explain how this problem is distinct from (albeit related to) the more familiar problem of evil (with which it is often conflated). In short, the problem is this: in virtue of how God would be (or, on some given conception, is) “involved in” our actions, how is it that God has the moral standing to blame us for performing those (...) very actions? In light of the recent literature on “moral standing”, I consider God’s moral standing to blame on two models of “divine providence”: open theism, and theological determinism. I contend that God may have standing on open theism, and – perhaps surprisingly – may also have standing, even on theological determinism, given the truth of compatibilism. Thus, if you think that God could not justly both determine and blame, then you will have to abandon compatibilism. The topic of this paper thus sheds considerable light on the traditional philosophical debate about the conditions of moral responsibility. (shrink)

At least since Aristotle’s famous 'sea-battle' passages in On Interpretation 9, some substantial minority of philosophers has been attracted to the doctrine of the open future--the doctrine that future contingent statements are not true. But, prima facie, such views seem inconsistent with the following intuition: if something has happened, then (looking back) it was the case that it would happen. How can it be that, looking forwards, it isn’t true that there will be a sea battle, while also being true (...) that, looking backwards, it was the case that there would be a sea battle? This tension forms, in large part, what might be called the problem of future contingents. A dominant trend in temporal logic and semantic theorizing about future contingents seeks to validate both intuitions. Theorists in this tradition--including some interpretations of Aristotle, but paradigmatically, Thomason (1970), as well as more recent developments in Belnap, et. al (2001) and MacFarlane (2003, 2014)--have argued that the apparent tension between the intuitions is in fact merely apparent. In short, such theorists seek to maintain both of the following two theses: (i) the open future: Future contingents are not true, and (ii) retro-closure: From the fact that something is true, it follows that it was the case that it would be true. It is well-known that reflection on the problem of future contingents has in many ways been inspired by importantly parallel issues regarding divine foreknowledge and indeterminism. In this paper, we take up this perspective, and ask what accepting both the open future and retro-closure predicts about omniscience. When we theorize about a perfect knower, we are theorizing about what an ideal agent ought to believe. Our contention is that there isn’t an acceptable view of ideally rational belief given the assumptions of the open future and retro-closure, and thus this casts doubt on the conjunction of those assumptions. (shrink)

In response to criticism, we often say – in these or similar words – “Let’s see you do better!” Prima facie, it looks like this response is a challenge of a certain kind – a challenge to prove that one has what has recently been called standing. More generally, the data here seems to point a certain kind of norm of criticism: be better. Slightly more carefully: One must: criticize x with respect to standard s only if one is better (...) than x with respect to standard s. In this paper, I defend precisely this norm of criticism – an underexplored norm that is nevertheless ubiquitous in our lives, once we begin looking for it. The be better norm is, I hope to show, continuously invoked in a wide range of ordinary settings, can undergird and explain the widely endorsed non-hypocrisy condition on the standing to blame, and apparent counterexamples to the norm are no such counterexamples at all. I further contend that, given some plausible principles, my previous “moral commitment” account of the moral standing to blame will be extensionally equivalent to the be better norm. (shrink)

These are my materials (a short precis, and replies to Alan Rhoda and Daniel Rubio) for an invited symposium on my book _The Open Future: Why Future Contingents are All False_ (OUP, 2021) in IJPR. [The commentaries from Rhoda/Rubio are available on their respective PhilPapers profiles.].

Various philosophers have long since been attracted to the doctrine that future contingent propositions systematically fail to be true—what is sometimes called the doctrine of the open future. However, open futurists have always struggled to articulate how their view interacts with standard principles of classical logic—most notably, with the Law of Excluded Middle. For consider the following two claims: Trump will be impeached tomorrow; Trump will not be impeached tomorrow. According to the kind of open futurist at issue, both of (...) these claims may well fail to be true. According to many, however, the disjunction of these claims can be represented as p ∨ ~p—that is, as an instance of LEM. In this essay, however, I wish to defend the view that the disjunction these claims cannot be represented as an instance of p ∨ ~p. And this is for the following reason: the latter claim is not, in fact, the strict negation of the former. More particularly, there is an important semantic distinction between the strict negation of the first claim [~] and the latter claim. However, the viability of this approach has been denied by Thomason, and more recently by MacFarlane and Cariani and Santorio, the latter of whom call the denial of the given semantic distinction “scopelessness”. According to these authors, that is, will is “scopeless” with respect to negation; whereas there is perhaps a syntactic distinction between ‘~Will p’ and ‘Will ~p’, there is no corresponding semantic distinction. And if this is so, the approach in question fails. In this paper, then, I criticize the claim that will is “scopeless” with respect to negation. I argue that will is a so-called “neg-raising” predicate—and that, in this light, we can see that the requisite scope distinctions aren’t missing, but are simply being masked. The result: a under-appreciated solution to the problem of future contingents that sees and as contraries, not contradictories. (shrink)

P.F. Strawson’s (1962) “Freedom and Resentment” has provoked a wide range of responses, both positive and negative, and an equally wide range of interpretations. In particular, beginning with Gary Watson, some have seen Strawson as suggesting a point about the “order of explanation” concerning moral responsibility: it is not that it is appropriate to hold agents responsible because they are morally responsible, rather, it is ... well, something else. Such claims are often developed in different ways, but one thing remains (...) constant: they meant to be incompatible with libertarian theories of moral responsibility. The overarching theme of this paper is that extant developments of “the reversal” face a dilemma: in order to make the proposals plausibly anti-libertarian, they must be made to be implausible on other grounds. I canvas different attempts to articulate a “Strawsonian reversal”, and argue that none is fit for the purposes for which it is intended. I conclude by suggesting a way of clarifying the intended thesis: an analogy with the concept of funniness. The result: proponents of the “reversal” need to accept the difficult result that if we blamed small children, they would be blameworthy, or instead explain how their view escapes this result, while still being a view on which our blaming practices “fix the facts” of moral responsibility. (shrink)

Перевод статьи: Davies T., Chandler R. Online deliberation design: Choices, criteria, and evidence // Democracy in motion: Evaluating the practice and impact of deliberative civic engagement / Nabatchi T., Weiksner M., Gastil J., Leighninger M. (eds.). -- Oxford: Oxford univ. press, 2013. -- P. 103-131. А. Кулик. -/- Вниманию читателей предлагается обзор эмпирических исследований в области дизайна онлайн-форумов, предназначенных для вовлечения граждан в делиберацию. Размерности дизайна определены для различных характеристик делиберации: назначения, целевой аудитории, разобщенности участников в пространстве и во времени, (...) среды коммуникации и организации делиберативного процесса. После краткого обзора критериев оценки вариантов дизайна рассматриваются эмпирические данные, соотносящиеся с каждым из вариантов. Эффективность онлайн-делиберации зависит от того, насколько условия коммуникации соотносятся с заданиями делиберации. Компромиссы, как, например, между анонимным или идентифицируемым участием, предполагают различные дизайны в зависимости от цели делиберации и состава участников. Выводы исследования получены на материале существующих технологий и могут измениться по мере коэволюции технологий и пользователей. (shrink)

There is a longstanding argument that purports to show that divine foreknowledge is inconsistent with human freedom to do otherwise. Proponents of this argument, however, have for some time been met with the following reply: the argument posits what would have to be a mysterious non-causal constraint on freedom. In this paper, I argue that this objection is misguided – not because after all there can indeed be non-causal constraints on freedom (as in Pike, Fischer, and Hunt), but because the (...) success of the incompatibilist's argument does not require the real possibility of non-causal constraints on freedom. I contend that the incompatibilist's argument is best seen as showing that, given divine foreknowledge, something makes one unfree – and that this something is most plausibly identified, not with the foreknowledge itself, but with the causally deterministic factors that would have to be in place in order for there to be infallible foreknowledge in the first place. (shrink)

This paper defines the form of prior knowledge that is required for sound inferences by analogy and single-instance generalizations, in both logical and probabilistic reasoning. In the logical case, the first order determination rule defined in Davies (1985) is shown to solve both the justification and non-redundancy problems for analogical inference. The statistical analogue of determination that is put forward is termed 'uniformity'. Based on the semantics of determination and uniformity, a third notion of "relevance" is defined, both logically and (...) probabilistically. The statistical relevance of one function in determining another is put forward as a way of defining the value of information: The statistical relevance of a function F to a function G is the absolute value of the change in one's information about the value of G afforded by specifying the value of F. This theory provides normative justifications for conclusions projected by analogy from one case to another, and for generalization from an instance to a rule. The soundness of such conclusions, in either the logical or the probabilistic case, can be identified with the extent to which the corresponding criteria (determination and uniformity) actually hold for the features being related. (shrink)

It is more or less uncontroversial that when we harm someone through wrongful conduct we incur an obligation to compensate her. But sometimes compensation is impossible: when the victim is killed, for example. Other times, only partial compensation is possible. In this article, I take some initial steps towards exploring this largely ignored issue. I argue that the perpetrator of a wrongful harm incurs a duty to promote the impartial good in proportion to the amount of harm that cannot be (...) made up for by compensating her victim. (shrink)

Inspired by Peter van Inwagen’s “simulacra model” of the resurrection, I investigate whether it could be reasonable to adopt an analogous approach to the problem of evil. Empirically Skeptical Theism, as I call it, is the hypothesis that God shields our lives from irredeemable evils surreptitiously (just as van Inwagen proposes that God shields our bodies from destruction surreptitiously). I argue that EST compares favorably with traditional skeptical theism and with eschatological theodicies, and that EST does not have the negative (...) moral consequences we might suppose. (shrink)

Representationalists have routinely expressed skepticism about the idea that inflexible responses to stimuli are to be explained in representational terms. Representations are supposed to be more than just causal mediators in the chain of events stretching from stimulus to response, and it is difficult to see how the sensory states driving reflexes are doing more than playing the role of causal intermediaries. One popular strategy for distinguishing representations from mere causal mediators is to require that representations are decoupled from specific (...) stimulus conditions. I believe this requirement on representation is mistaken and at odds with explanatory practices in sensory ecology. Even when sensory states have the job of coordinating a specific output with a specific input, we can still find them doing the work of representations, carrying information needed for organisms to respond successfully to environmental conditions. We can uncover information at work by intervening specifically on the information conveyed by sensory states, leaving their causal role undisturbed. (shrink)

Many people believe that the wrongness of killing a person does not depend on factors like her age, condition, or how much she has to lose by dying – a view Jeff McMahan calls the ‘Equal Wrongness Thesis’. This article argues that we should reject the Equal Wrongness Thesis on the basis of the moral equivalence between killing a person and knocking her unconscious.

In this paper, I articulate an argument for incompatibilism about moral responsibility and determinism. My argument comes in the form of an extended story, modeled loosely on Peter van Inwagen’s “rollback argument” scenario. I thus call it “the replication argument.” As I aim to bring out, though the argument is inspired by so-called “manipulation” and “original design” arguments, the argument is not a version of either such argument—and plausibly has advantages over both. The result, I believe, is a more convincing (...) incompatibilist argument than those we have considered previously. (shrink)

The plane was going to crash, but it didn't. Johnny was going to bleed to death, but he didn't. Geach sees here a changing future. In this paper, I develop Geach's primary argument for the (almost universally rejected) thesis that the future is mutable (an argument from the nature of prevention), respond to the most serious objections such a view faces, and consider how Geach's view bears on traditional debates concerning divine foreknowledge and human freedom. As I hope to show, (...) Geach's view constitutes a radically new view on the logic of future contingents, and deserves the status of a theoretical contender in these debates. (shrink)

This Introduction has three sections, on "logical fatalism," "theological fatalism," and the problem of future contingents, respectively. In the first two sections, we focus on the crucial idea of "dependence" and the role it plays it fatalistic arguments. Arguably, the primary response to the problems of logical and theological fatalism invokes the claim that the relevant past truths or divine beliefs depend on what we do, and therefore needn't be held fixed when evaluating what we can do. We call the (...) sort of dependence needed for this response to be successful "dependence with a capital 'd'": Dependence. We consider different accounts of Dependence, especially the account implicit in the so-called "Ockhamist" response to the fatalistic arguments. Finally, we present the problem of future contingents: what could "ground" truths about the undetermined future? On the other hand, how could all such propositions fail to be true? (shrink)

Everyone agrees that we can’t change the past. But what about the future? Though the thought that we can change the future is familiar from popular discourse, it enjoys virtually no support from philosophers, contemporary or otherwise. In this paper, I argue that the thesis that the future is mutable has far more going for it than anyone has yet realized. The view, I hope to show, gains support from the nature of prevention, can provide a new way of responding (...) to arguments for fatalism, can account for the utility of total knowledge of the future, and can help in providing an account of the semantics of the English progressive. On the view developed, the future is mutable in the following sense: perhaps, once, it was true that you would never come to read a paper defending the mutability of the future. And then the future changed. And now you will. (shrink)

Incompatibilism about responsibility and determinism is sometimes directly construed as the thesis that if we found out that determinism is true, we would have to give up the reactive attitudes. Call this "the consequence". I argue that this is a mistake: the strict modal thesis does not entail the consequence. First, some incompatibilists (who are also libertarians) may be what we might call *resolute responsibility theorists* (or "flip-floppers"). On this view, if we found out that determinism is true, this would (...) not be to find out that no one is responsible; instead, what this would suggest is that incompatibilism is after all false. Second, some incompatibilists may instead deny the claim that if we found out that no one deserves the reactiveattitudes, we would have to give up the reactive attitudes. Call this position *innocent incompatibilism*. I explore several different reasons why the fact that no one deserves the reactive attitudes likely does not translate into a requirement for us to give up the reactive attitudes. The upshot: incompatibilism may be practically irrelevant, even if true. (shrink)

I want to suggest that it would be bad if compatibilism were true, and that this gives us good reason to think that it isn't. This is, you might think, an outlandish argument, and the considerable burden of this paper is to convince you otherwise. There are two key elements at stake in this argument. The first is that it would be ‐ in a distinctive sense to be explained ‐ bad if compatibilism were true. The thought here is that (...) compatibilism ultimately presents us with a picture on which, in principle, powerful manipulators can effectively guarantee that finite moral agents should become blameworthy. To my mind, this isn't just false ‐ though I think that it is ‐ it is also such that it would be bad (unfortunate, undesirable…) if it were true. The second is that the fact that it would be ‐ in this sense ‐ bad if true gives us reason to think that it isn't. It may be bad that there is no afterlife. But that, in itself, hardly gives us reason to think that there is an afterlife. That is true, but ‐ as others before me have suggested ‐ when the object of the relevant badness is morality itself, the inference seems secure. A more general aim of the paper is to investigate the nature of this very form of argument in itself, and I compare my argument (inter alia) to a recent argument from Sayre–McCord against the possibility of genuine moral dilemmas. (shrink)

A common objection to representationalism is that a representationalist view of phenomenal character cannot accommodate the effects that shifts in covert attention have on visual phenomenology: covert attention can make items more visually prominent than they would otherwise be without altering the content of visual experience. Recent empirical work on attention casts doubt on previous attempts to advance this type of objection to representationalism and it also points the way to an alternative development of the objection.

It is generally assumed that existing artificial systems are not phenomenally conscious, and that the construction of phenomenally conscious artificial systems would require significant technological progress if it is possible at all. We challenge this assumption by arguing that if Global Workspace Theory (GWT) — a leading scientific theory of phenomenal consciousness — is correct, then instances of one widely implemented AI architecture, the artificial language agent, might easily be made phenomenally conscious if they are not already. Along the way, (...) we articulate an explicit methodology for thinking about how to apply scientific theories of consciousness to artificial systems and employ this methodology to arrive at a set of necessary and sufficient conditions for phenomenal consciousness according to GWT. (shrink)

These are my materials (a short precis, and replies to John MacFarlane, Mitchell Green, Ryan Wasserman, and Anthony Bigg and Kristie Miller) for a symposium on my book, _The Open Future: Why Future Contingents are All False_ (OUP, 2021) in *Analytic Philosophy*. [The contribution from MacFarlane is available on his website, those from Wasserman and Green are on their Academia profiles, and the contribution from Bigg and Miller is on Miller's PhilPapers profile.].

A central question, if not the central question, of philosophy of perception is whether sensory states have a nature similar to thoughts about the world, whether they are essentially representational. According to the content view, at least some of our sensory states are, at their core, representations with contents that are either accurate or inaccurate. Tyler Burge’s Origins of Objectivity is the most sustained and sophisticated defense of the content view to date. His defense of the view is problematic in (...) several ways. The most significant problem is that his approach does not sit well with mainstream perceptual psychology. (shrink)

LLMs have dramatically improved in capabilities in recent years. This raises the question of whether LLMs could become genuine agents with beliefs and desires. This paper demonstrates an in principle limit to LLM agency, based on their architecture. LLMs are next word predictors: given a string of text, they calculate the probability that various words can come next. LLMs produce outputs that reflect these probabilities. I show that next word predictors are exploitable. If LLMs are prompted to make probabilistic predictions (...) about the world, these predictions are guaranteed to be incoherent, and so Dutch bookable. If LLMs are prompted to make choices over actions, their preferences are guaranteed to be intransitive, and so money pumpable. In short, the problem is that selecting an action based on its potential value is structurally different then selecting the description of an action that is most likely given a prompt: probability cannot be forced into the shape of expected value. The in principle exploitability of LLMs raises doubts about how agential they can become. This exploitability also offers an opportunity for humanity to safely control such AI systems. (shrink)

This paper examines the question of whether Large Language Models (LLMs) like ChatGPT possess minds, focusing specifically on whether they have a genuine folk psychology encompassing beliefs, desires, and intentions. We approach this question by investigating two key aspects: internal representations and dispositions to act. First, we survey various philosophical theories of representation, including informational, causal, structural, and teleosemantic accounts, arguing that LLMs satisfy key conditions proposed by each. We draw on recent interpretability research in machine learning to support these (...) claims. Second, we explore whether LLMs exhibit robust dispositions to perform actions, a necessary component of folk psychology. We consider two prominent philosophical traditions, interpretationism and representationalism, to assess LLM action dispositions. While we find evidence suggesting LLMs may satisfy some criteria for having a mind, particularly in game-theoretic environments, we conclude that the data remains inconclusive. Additionally, we reply to several skeptical challenges to LLM folk psychology, including issues of sensory grounding, the "stochastic parrots" argument, and concerns about memorization. Our paper has three main upshots. First, LLMs do have robust internal representations. Second, there is an open question to answer about whether LLMs have robust action dispositions. Third, existing skeptical challenges to LLM representation do not survive philosophical scrutiny. (shrink)

Existing indexing systems used to arrange philosophical works have been shown to misrepresent the discipline in ways that reflect and perpetuate exclusionary attitudes within it. In recent years, there has been a great deal of effort to challenge those attitudes and to revise them. But as the discipline moves toward greater equality and inclusivity, the way it has indexed its work has unfortunately not. To course correct, we identify in this article some of the specific changes that are needed within (...) current indexing systems and propose a new model that could embody them. We use the Diversity Reading List in Philosophy as a case study and PhilPapers as a basis for comparison. The model we propose not only represents the discipline in a more inclusive and fair way, but it is also efficient, easy to use or implement, and adaptable for a variety of contexts. (shrink)

There is widely thought to be a proportionality constraint on harming others in self-defense, such that an act of defensive force can be impermissible because the harm it would inflict on an attacker is too great relative to the harm to the victim it would prevent. But little attention has been given to whether a corresponding constraint exists in the ethics of compensation, and, if so, what the nature of that constraint is. This article explores the issue of proportionality as (...) it applies to the liability to compensate. The view that some perpetrators are not liable to pay full compensation because doing so would be disproportionately burdensome is clarified and defended, and it is asked what view we should adopt instead. A key step in that inquiry is an argument that someone is liable to bear the cost of compensating for an injury if and only if she would have been liable to bear that same cost in defense against that same injury ex ante. (shrink)

Teleological accounts of sensory normativity treat normal functioning for a species as a standard: sensory error involves departure from normal functioning for the species, i.e. sensory malfunction. Straightforward reflection on sensory trade-offs reveals that normal functioning for a species can exhibit failures of accuracy. Acknowledging these failures of accuracy is central to understanding the adaptations of a species. To make room for these errors we have to go beyond the teleological framework and invoke the notion of an ideal observer from (...) vision science. The notion of an ideal observer also sheds light on the important distinction between sensory malfunction and sensory limitation. (shrink)

A letter replying to Bayne et al. "Tests for consciousness in humans and beyond", 2024, arguing that the search for consciousness "beyond" healthy adult humans should begin with other animals.

This paper answers two interpretive questions surrounding belief in God in Thomas Reid’s philosophy, the status question and the detachability question. The former has to do with the type of justification Reid assigns to belief in God – immediate or mediate. The later question is whether anything philosophically significant depends on his belief in God. I argue that, for Reid, belief in God is immediately justified and integral to some parts of his system. Reid’s response to skepticism about God is (...) more complicated and more interesting than many of the contemporary philosophers who, citing Reid as inspiration, also hold that belief in God enjoys immediate justification. In Reid’s hands the approach to belief in God does not compete with inferential justification, does not rely on a special faculty, and foregrounds the developmental character of his epistemology. (shrink)

This paper offers the first careful analysis of the possibility that AI and humanity will go to war. The paper focuses on the case of artificial general intelligence, AI with broadly human capabilities. The paper uses a bargaining model of war to apply standard causes of war to the special case of AI/human conflict. The paper argues that information failures and commitment problems are especially likely in AI/human conflict. Information failures would be driven by the difficulty of measuring AI capabilities, (...) by the uninterpretability of AI systems, and by differences in how AIs and humans analyze information. Commitment problems would make it difficult for AIs and humans to strike credible bargains. Commitment problems could arise from power shifts, rapid and discontinuous increases in AI capabilities. Commitment problems could also arise from missing focal points, where AIs and humans fail to effectively coordinate on policies to limit war. In the face of this heightened chance of war, the paper proposes several interventions. War can be made less likely by improving the measurement of AI capabilities, capping improvements in AI capabilities, designing AI systems to be similar to humans, and by allowing AI systems to participate in democratic political institutions. (shrink)

In this paper, we clarify, unpack, and ultimately resist what is perhaps the most prominent argument for the compatibility of free will and determinism: the epistemic argument for compatibilism. We focus on one such argument as articulated by David Lewis: (i) we know we are free, (ii) for all we know everything is predetermined, (iii) if we know we are free but for all we know everything is predetermined, then for all we know we are free but everything is predetermined, (...) (iv) if for all we know we are free but predetermined, then it is really possible that we are, so (v) compatibilism. We uncover how the crucial epistemic modality underlying (iv) must be understood, and contend that, understood this way, the libertarian can resist (iv). Importantly, however, resisting the argument does commit the libertarian to what has been called “flip-flopping”—but we argue that this is perfectly coherent. We conclude by articulating two crucially ways the libertarian can resist the argument, by saying that we can know that determinism is false "from the armchair". (shrink)

At the most general level, "manipulation" refers one of many ways of influencing behavior, along with (but to be distinguished from) other such ways, such as coercion and rational persuasion. Like these other ways of influencing behavior, manipulation is of crucial importance in various ethical contexts. First, there are important questions concerning the moral status of manipulation itself; manipulation seems to be mor- ally problematic in ways in which (say) rational persuasion does not. Why is this so? Furthermore, the notion (...) of manipulation has played an increasingly central role in debates about free will and moral responsibility. Despite its significance in these (and other) contexts, however, the notion of manipulation itself remains deeply vexed. I would say notoriously vexed, but in fact direct philosophical treatments of the notion of manipulation are few and far between, and those that do exist are nota- ble for the sometimes widely divergent conclusions they reach concerning what it is. I begin by addressing (though certainly not resolving) the conceptual issue of how to distinguish manipulation from other ways of influencing behavior. Along the way, I also briefly address the (intimately related) question of the moral status of manipulation: what, if anything, makes it morally problematic? Then I discuss the controversial ways in which the notion of manipulation has been employed in contemporary debates about free will and moral responsibility. (shrink)

A person is biased toward the future when she prefers, other things being equal, bad events to be in her past rather than her future or good ones to be in her future rather than her past. In this paper, I explain why both critics and defenders of future bias have failed to consider the best version of the view. I distinguish external time from personal time, and show that future bias is best construed in terms of the latter. This (...) understanding of future bias avoids several standard objections. I then consider a new justification of future bias which is consistent with that construal. My discussion points to a new position regarding the basic relation that grounds rational egoistic concern over time, according to which that relation is asymmetric between person-stages. I also explain how this way of justifying future bias would resolve the apparent tension between the future bias we display in our own case and our relative indifference toward the timing of the good and bad things that happen to other people. (shrink)

The words "violence" and "nonviolence" are increasingly misleading translations for the Sanskrit words hiṃsā and ahiṃsā -- which were used by Gandhi as the basis for his philosophy of satyāgraha. I argue for re-reading hiṃsā as “maleficence” and ahiṃsā as “beneficence.” These two more mind-referring English words – associated with religiously contextualized discourse of the past -- capture the primacy of intention implied by Gandhi’s core principles, better than “violence” and “nonviolence” do. Reflecting a political turn in moral accountability detectable (...) through linguistic data, both the scope and the usage of the word “violence” have expanded dramatically. The expanded scope of “violence” reflects greater consciousness of the various forms that serious harm can take, but also makes it harder to convincingly characterize people and actions as “nonviolent.” New translations could clarify the distinction between hiṃsā and ahiṃsā, and thereby prevent some misunderstandings of Gandhi. Training in beneficence would reflect Gandhi’s psychological path to reducing avoidable harm: detachment from the ego, learning to love universally, and seeking truth by experiment. (shrink)

A central component of the cognitive revolution is a commitment to explaining behaviour by reference to internal representations of the world. This core aspect.