In this lucid portrayal of human behavior, Fred Dretske provides an original account of the way reasons function in the causal explanation of behavior.

The doyen of living English philosophers, by these reflections, took hold of and changed the outlook of a good many other philosophers, if not quite enough. He did so, essentially, by assuming that talk of freedom and responsibility is talk not of facts or truths, in a certain sense, but of our attitudes. His more explicit concern was to look again at the question of whether determinism and freedom are consistent with one another -- by shifting attention to certain personal (...) rather than moral attitudes, first of all gratitude and resentment. In the end, he arrived at a kind of Compatibilist or, as he says, Optimist conclusion. That is no doubt a recommendation but not the largest recommendation of this splendidly rich piece of philosophy. (shrink)

This book examines some of the deepest questions in philosophy: What is involved in judging a belief, action, or feeling to be rational?

What is the relation between a reason and an action when the reason explains the action by giving the agent's reason for doing what he did? We may call such explanations rationalizations, and say that the reason rationalizes the action. In this paper I want to defend the ancient - and common-sense - position that rationalization is a species of ordinary causal explanation. The defense no doubt requires some redeployment, but not more or less complete abandonment of the position, as (...) urged by many recent writers. (shrink)

Introduction This book is about the nature of the content of psychological states. Examples of psychological states with content are: believing today is a ...

This paper presents a partial analysis of perceptual knowledge, an analysis that will, I hope, lay a foundation for a general theory of knowing. Like an earlier theory I proposed, the envisaged theory would seek to explicate the concept of knowledge by reference to the causal processes that produce (or sustain) belief. Unlike the earlier theory, however, it would abandon the requirement that a knower's belief that p be causally connected with the fact, or state of affairs, that p.

How do rational minds make contact with the world? The empiricist tradition sees a gap between mind and world, and takes sensory experience, fallible as it is, to provide our only bridge across that gap. In its crudest form, for example, the traditional idea is that our minds consult an inner realm of sensory experience, which provides us with evidence about the nature of external reality. Notoriously, however, it turns out to be far from clear that there is any viable (...) conception of experience which allows it to do the job. The original problem is to show that thought is rationally constrained by external reality. If sensory experience is to provide the solution--in particular, if it is to provide the answer to sceptical challenges--it must therefore meet two criteria. First, it must itself be `receptive'--i.e., appropriately constrained by external reality. Second, it must be the kind of thing that can enter into a logical or rational relationship with belief--it must already be `conceptual,' in other words. In arguing against the idea that anything could serve both roles, Wilfred Sellars termed this conception of experience "the Myth of the Given.". (shrink)

"Why should we care about having true beliefs? And why do demonstrably false beliefs persist and spread despite consequences for the people who hold them? Philosophers of science Cailin O’Connor and James Weatherall argue that social factors, rather than individual psychology, are what’s essential to understanding the spread and persistence of false belief. It might seem that there’s an obvious reason that true beliefs matter: false beliefs will hurt you. But if that’s right, then why is it irrelevant to many (...) people whether they believe true things or not? In an age riven by "fake news," "alternative facts," and disputes over the validity of everything from climate change to the size of inauguration crowds, the authors argue that social factors, not individual psychology, are what’s essential to understanding the persistence of false belief and that we must know how those social forces work in order to fight misinformation effectively."–Publisher’s description. (shrink)

Artificial agents (AAs), particularly but not only those in Cyberspace, extend the class of entities that can be involved in moral situations. For they can be conceived of as moral patients (as entities that can be acted upon for good or evil) and also as moral agents (as entities that can perform actions, again for good or evil). In this paper, we clarify the concept of agent and go on to separate the concerns of morality and responsibility of agents (most (...) interestingly for us, of AAs). We conclude that there is substantial and important scope, particularly in Computer Ethics, for the concept of moral agent not necessarily exhibiting free will, mental states or responsibility. This complements the more traditional approach, common at least since Montaigne and Descartes, which considers whether or not (artificial) agents have mental states, feelings, emotions and so on. By focussing directly on mind-less morality we are able to avoid that question and also many of the concerns of Artificial Intelligence. A vital component in our approach is the Method of Abstraction for analysing the level of abstraction (LoA) at which an agent is considered to act. The LoA is determined by the way in which one chooses to describe, analyse and discuss a system and its context. The Method of Abstraction is explained in terms of an interface or set of features or observables at a given LoA. Agenthood, and in particular moral agenthood, depends on a LoA. Our guidelines for agenthood are: interactivity (response to stimulus by change of state), autonomy (ability to change state without stimulus) and adaptability (ability to change the transition rules by which state is changed) at a given LoA. Morality may be thought of as a threshold defined on the observables in the interface determining the LoA under consideration. An agent is morally good if its actions all respect that threshold; and it is morally evil if some action violates it. That view is particularly informative when the agent constitutes a software or digital system, and the observables are numerical. Finally we review the consequences for Computer Ethics of our approach. In conclusion, this approach facilitates the discussion of the morality of agents not only in Cyberspace but also in the biosphere, where animals can be considered moral agents without their having to display free will, emotions or mental states, and in social contexts, where systems like organizations can play the role of moral agents. The primary cost of this facility is the extension of the class of agents and moral agents to embrace AAs. (shrink)

Argues that the key distinction between human and nonhuman social cognition consists in our complex, diverse and flexible capacities to shape each other's minds in ways that make them easier to interpret.

Does metacognition--the capacity to self-evaluate one's cognitive performance--derive from a mindreading capacity, or does it rely on informational processes? Joëlle Proust draws on psychology and neuroscience to defend the second claim. She argues that metacognition need not involve metarepresentations, and is essentially related to mental agency.

One of the main difficulties in assessing artificial intelligence (AI) is the tendency for people to anthropomorphise it. This becomes particularly problematic when we attach human moral activities to AI. For example, the European Commission’s High-level Expert Group on AI (HLEG) have adopted the position that we should establish a relationship of trust with AI and should cultivate trustworthy AI (HLEG AI Ethics guidelines for trustworthy AI, 2019, p. 35). Trust is one of the most important and defining activities in (...) human relationships, so proposing that AI should be trusted, is a very serious claim. This paper will show that AI cannot be something that has the capacity to be trusted according to the most prevalent definitions of trust because it does not possess emotive states or can be held responsible for their actions—requirements of the affective and normative accounts of trust. While AI meets all of the requirements of the rational account of trust, it will be shown that this is not actually a type of trust at all, but is instead, a form of reliance. Ultimately, even complex machines such as AI should not be viewed as trustworthy as this undermines the value of interpersonal trust, anthropomorphises AI, and diverts responsibility from those developing and using them. (shrink)

Argues that the key distinction between human and nonhuman social cognition consists in our complex, diverse and flexible capacities to shape each other's minds in ways that make them easier to interpret.

In this paper I consider fairness of blaming a wrongdoer. In particular, I consider the claim that blaming a wrongdoer can be unfair because blame has a certain characteristic force, a force which is not fairly imposed upon the wrongdoer unless certain conditions are met--unless, e.g., the wrongdoer could have done otherwise, or unless she is someone capable of having done right, or unless she is able to control her behavior by the light of moral reasons. While agreeing that blame (...) has a characteristic force, I am skeptical of this charge of unfairness. My skepticism concerns itself less with the particular conditions of fairness proposed than with the idea that blame can be rendered unfair by its characteristic force. If to blame a person were simply to perform certain intentional actions, then, as we will see, blame could be rendered unfair by its force. But to blame a person is not just to act in certain ways. It is, at least in large part, to make certain judgments or adopt certain attitudes. However, it is unclear how these attitudes or judgments carry "force"? and also unclear whether they can be rendered unfair by their force. Examining these issues, I will suggest that much of the force of blame is found in a set of judgments--most centrally, the judgment that one person failed to show proper regard for others. But, I will argue, once it is granted that such judgments are true, their characteristic force cannot render them unfair. (shrink)

We argue that artificial networks are explainable and offer a novel theory of interpretability. Two sets of conceptual questions are prominent in theoretical engagements with artificial neural networks, especially in the context of medical artificial intelligence: Are networks explainable, and if so, what does it mean to explain the output of a network? And what does it mean for a network to be interpretable? We argue that accounts of “explanation” tailored specifically to neural networks have ineffectively reinvented the wheel. In (...) response to, we show how four familiar accounts of explanation apply to neural networks as they would to any scientific phenomenon. We diagnose the confusion about explaining neural networks within the machine learning literature as an equivocation on “explainability,” “understandability” and “interpretability.” To remedy this, we distinguish between these notions, and answer by offering a theory and typology of interpretation in machine learning. Interpretation is something one does to an explanation with the aim of producing another, more understandable, explanation. As with explanation, there are various concepts and methods involved in interpretation: Total or Partial, Global or Local, and Approximative or Isomorphic. Our account of “interpretability” is consistent with uses in the machine learning literature, in keeping with the philosophy of explanation and understanding, and pays special attention to medical artificial intelligence systems. (shrink)

This paper considers the connection between automaticity, control and agency. Indeed, recent philosophical and psychological works play up the incompatibility of automaticity and agency. Specifically, there is a threat of automaticity, for automaticity eliminates agency. Such conclusions stem from a tension between two thoughts: that automaticity pervades agency and yet automaticity rules out control. I provide an analysis of the notions of automaticity and control that maintains a simple connection: automaticity entails the absence of control. An appropriate analysis, however, shows (...) that actions are forms of control and pervasively automatic even if automaticity implies the absence of control. Consequences are drawn for the theory of mental agency and the psychological concepts of automaticity and control. (shrink)

Can we trust robots? Responding to the literature on trust and e-trust, this paper asks if the question of trust is applicable to robots, discusses different approaches to trust, and analyses some preconditions for trust. In the course of the paper a phenomenological-social approach to trust is articulated, which provides a way of thinking about trust that puts less emphasis on individual choice and control than the contractarian-individualist approach. In addition, the argument is made that while robots are neither human (...) nor mere tools, we have sufficient functional, agency-based, appearance-based, social-relational, and existential criteria left to evaluate trust in robots. It is also argued that such evaluations must be sensitive to cultural differences, which impact on how we interpret the criteria and how we think of trust in robots. Finally, it is suggested that when it comes to shaping conditions under which humans can trust robots, fine-tuning human expectations and robotic appearances is advisable. (shrink)

As robots are deployed in a widening range of situations, it is necessary to develop a clearer position about whether or not they can be trusted to make good moral decisions. In this paper, we take a realistic look at recent attempts to program and to train robots to develop some form of moral competence. Examples of implemented robot behaviours that have been described as 'ethical', or 'minimally ethical' are considered, although they are found to only operate in quite constrained (...) and limited application domains. There is a general recognition that current robots cannot be described as full moral agents, but it is less clear whether will always be the case. Concerns are raised about the insufficiently justified use of terms such as 'moral' and 'ethical' to describe the behaviours of robots that are often more related to safety considerations than to moral ones. Given the current state of the art, two possible responses are identified. The first involves continued efforts to develop robots that are capable of ethical behaviour. The second is to argue against, and to attempt to avoid, placing robots in situations that demand moral competence and an understanding of the surrounding social situation. There is something to be gained from both responses, but it is argued here that the second is the more responsible choice. (shrink)

Generative AI programs can produce high-quality written and visual content that may be used for good or ill. We argue that a credit–blame asymmetry arises for assigning responsibility for these outputs and discuss urgent ethical and policy implications focused on large-scale language models.

What does it mean to "trust?" What makes us feel secure enough to place our confidence—even at times our welfare—in the hands of other people? Is it possible to "trust" an institution? What exactly do people mean when they claim to "distrust" their governments? As difficult as it may be to define, trust is essential to the formation and maintenance of a civil society. In Trust and Trustworthiness political scientist Russell Hardin addresses the standard theories of trust and articulates his (...) own new and compelling idea: that much of what we call trust can be best described as "encapsulated interest." Research into the roles of trust in our society has offered a broad range of often conflicting theories. Some theorists maintain that trust is a social virtue that cannot be reduced to strategic self-interest; others claim that trusting another person is ultimately a rational calculation based on information about that person and his or her incentives and motivations. Hardin argues that we place our trust in persons whom we believe to have strong reasons to act in our best interests. He claims that we are correct when we assume that the main incentive of those whom we trust is to maintain a relationship with us—whether it be for reasons of economic benefit or for love and friendship. Hardin articulates his theory using examples from a broad array of personal and social relationships, paying particular attention to explanations of the development of trusting relationships. He also examines trustworthiness and seeks to understand why people may behave in ways that violate their own self-interest in order to honor commitments they have made to others. The book also draws important distinctions between vernacular uses of "trust" and "trustworthiness," contrasting, for example, the type of trust (or distrust) we place in individuals with the trust we place in institutions Trust and Trustworthiness represents the culmination of important new research into the roles of trust in our society; it offers a challenging new voice in the current discourse about the origins of cooperative behavior and its consequences for social and civic life. A Volume in the Russell Sage Foundation Series on Trust. (shrink)

Trustful interaction serves the interests of those involved. Thus, one could reason that trust itself may be analyzed as part of rational, goaloriented action. In contrast, common sense tells us that trust is an emotion and is, therefore, independent of rational deliberation to some extent. I will argue that we are right in trusting our common sense. My argument is conceptual in nature, referring to the common distinction between trust and pure reliance. An emotional attitude may be understood as some (...) general pattern in the way the world or some part of the world is perceived by an individual. Trust may be characterized by such a pattern. I shall focus on two central features of a trusting attitude. First, trust involves a participant attitude (Strawson) toward the person being trusted. Second, a situation of trust is perceived by a trusting person as one in which shared values or norms motivate both his own actions as well as those of the person being trusted. As an emotional attitude, trust is, to some extent, independent of objective information. It determines what a trusting person will believe and how various outcomes are evaluated. Hence, trust is quite different from rational belief and the problem with trust is not adequately met in minimizing risk by supplying extensive information or some mechanism of sanctioning. Trust is an attitude that enables us to cope with risk in a certain way. If we want to promote trustful interaction, we must form our institutions in ways that allow individuals to experience their interest and values as shared and, thus, to develop a trusting attitude. (shrink)

Trust is pervasive in our lives. Both our simplest actions--like buying a coffee, or crossing the street--as well as the functions of large collective institutions--like those of corporations and nation states--wouldn't be possible without it. Yet, only in the last several decades has trust started to receive focused attention from philosophers as a specific topic of investigation. The Routledge Handbook of Trust and Philosophybrings together XX never before published essays, accessible for both students and researchers, created to cover the most (...) salient topics in the various theories of trust. The Handbookis broken up into four sections: a first on basic foundational and conceptual issues; a second on when trust is warranted; a third on the epistemology of trust; and a fourth on trust and science and technology. Each section is preceded by a short introduction, and each chapter includes a bibliography and cross-references to other entries in the volume. (shrink)

Metacognition – the ability to represent, monitor and control ongoing cognitive processes – helps us perform many tasks, both when acting alone and when working with others. While metacognition is adaptive, and found in other animals, we should not assume that all human forms of metacognition are gene-based adaptations. Instead, some forms may have a social origin, including the discrimination, interpretation, and broadcasting of metacognitive representations. There is evidence that each of these abilities depends on cultural learning and therefore that (...) cultural selectionmight shape humanmetacognition. The cultural origins hypothesis is a plausible and testable alternative that directs us towards a substantial new programme of research. (shrink)

In the past decade, the fields of machine learning and artificial intelligence (AI) have seen unprecedented developments that raise human-machine interactions (HMI) to the next level.Smart machines, i.e., machines endowed with artificially intelligent systems, have lost their character as mere instruments. This, at least, seems to be the case if one considers how humans experience their interactions with them. Smart machines are construed to serve complex functions involving increasing degrees of freedom, and they generate solutions not fully anticipated by humans. (...) Consequently, their performances show a touch of action and even autonomy. HMI is therefore often described as a sort of “cooperation” rather than as a mere application of a tool. Some authors even go as far as subsuming cooperation with smart machines under the label ofpartnership, akin to cooperation between human agents sharing a common goal. In this paper, we explore how far the notion of shared agency and partnership can take us in our understanding of human interaction with smart machines. Discussing different topoi related to partnerships in general, we suggest that different kinds of “partnership” depending on the form of interaction between agents need to be kept apart. Building upon these discussions, we propose a tentative taxonomy of different kinds of HMI distinguishing coordination, collaboration, cooperation, and social partnership. (shrink)

Despite facing significant criticism for assigning human-like characteristics to artificial intelligence, phrases like “trustworthy AI” are still commonly used in official documents and ethical guidelines. It is essential to consider why institutions continue to use these phrases, even though they are controversial. This article critically evaluates various reasons for using these terms, including ontological, legal, communicative, and psychological arguments. All these justifications share the common feature of trying to justify the official use of terms like “trustworthy AI” by appealing to (...) the need to reflect pre-existing facts, be it the ontological status, ways of representing AI or legal categories. The article challenges the justifications for these linguistic practices observed in the field of AI ethics and AI science communication. In particular, it takes aim at two main arguments. The first is the notion that ethical discourse can move forward without the need for philosophical clarification, bypassing existing debates. The second justification argues that it’s acceptable to use anthropomorphic terms because they are consistent with the common concepts of AI held by non-experts—exaggerating this time the existing evidence and ignoring the possibility that folk beliefs about AI are not consistent and come closer to semi-propositional beliefs. The article sounds a strong warning against the use of human-centric language when discussing AI, both in terms of principle and the potential consequences. It argues that the use of such terminology risks shaping public opinion in ways that could have negative outcomes. (shrink)

I discuss an influential argument put forward by Hatherley in theJournal of Medical Ethics. Drawing on influential philosophical accounts of interpersonal trust, Hatherley claims that medical artificial intelligence is capable of being reliable, but not trustworthy. Furthermore, Hatherley argues that trust generates moral obligations on behalf of the trustee. For instance, when a patient trusts a clinician, it generates certain moral obligations on behalf of the clinician for her to do what she is entrusted to do. I make three objections (...) to Hatherley’s claims: (1) At least one philosophical account of interagent trust implies that medical AI is capable of being trustworthy. (2) Even if this account should ultimately be rejected, it does not matter much because what we care mostly about is that medical AI is reliable. (3) It is false that trust in itself generates moral obligations on behalf of the trustee. (shrink)

After discussing the distinction between artifacts and natural entities, and the distinction between artifacts and technology, the conditions of the traditional account of moral agency are identified. While computer system behavior meets four of the five conditions, it does not and cannot meet a key condition. Computer systems do not have mental states, and even if they could be construed as having mental states, they do not have intendings to act, which arise from an agent’s freedom. On the other hand, (...) computer systems have intentionality, and because of this, they should not be dismissed from the realm of morality in the same way that natural objects are dismissed. Natural objects behave from necessity; computer systems and other artifacts behave from necessity after they are created and deployed, but, unlike natural objects, they are intentionally created and deployed. Failure to recognize the intentionality of computer systems and their connection to human intentionality and action hides the moral character of computer systems. Computer systems are components in human moral action. When humans act with artifacts, their actions are constituted by the intentionality and efficacy of the artifact which, in turn, has been constituted by the intentionality and efficacy of the artifact designer. All three components – artifact designer, artifact, and artifact user – are at work when there is an action and all three should be the focus of moral evaluation. (shrink)

We argue that the notion of trust, as it figures in an ethical context, can be illuminated by examining research in artificial intelligence on multi-agent systems in which commitment and trust are modeled. We begin with an analysis of a philosophical model of trust based on Richard Holton’s interpretation of P. F. Strawson’s writings on freedom and resentment, and we show why this account of trust is difficult to extend to artificial agents (AAs) as well as to other non-human entities. (...) We then examine Margaret Urban Walker’s notions of “default trust” and “default, diffuse trust” to see how these concepts can inform our analysis of trust in the context of AAs. In the final section, we show how ethicists can improve their understanding of important features in the trust relationship by examining data resulting from a classic experiment involving AAs. (shrink)