Artificial agents create significant moral opportunities and challenges. Over the last two decades, discourse has largely focused on the concept of a ‘responsibility gap.’ We argue that this concept is incoherent, misguided, and diverts attention from the core issue of ‘control gaps.’ Control gaps arise when there is a discrepancy between the causal control an agent exercises and the moral control it should possess or emulate. Such gaps present moral risks, often leading to harm or ethical (...) violations. We propose a second-order ‘duty of moral control’ that mandates closing these gaps to reduce risks within acceptable moral limits. Our analysis encompasses both autonomous machines and collective agents, acknowledging their similarities and key differences in constitution and moral status. We suggest four methods to close control gaps: ensuring artificial agents attain moral agency, providing meaningful human control, implementing safety engineering, and employing social control. These methods aim to responsibly integrate artificial agents into society. We conclude that a realistic approach, which addresses the practical problems posed by control gaps, is essential. This approach provides solutions to manage the risks posed by artificial agents while maintaining acceptable moral standards, ensuring we responsibly harness their potential and address the ethical challenges they present. (shrink)

Responsibility gaps concern the attribution of blame for harms caused by autonomous machines. The worry has been that, because they are artificial agents, it is impossible to attribute blame, even though doing so would be appropriate given the harms they cause. We argue that there are no responsibility gaps. The harms can be blameless. And if they are not, the blame that is appropriate is indirect and can be attributed to designers, engineers, software developers, manufacturers or regulators. The real problem (...) lies elsewhere: autonomous machines should be built so as to exhibit a level of risk that is morally acceptable. If they fall short of this standard, they exhibit what we call ‘a control gap.’ The causal control that autonomous machines have will then fall short of the guidance control they should emulate. (shrink)

Why do collective agents violate their obligations? They can but need not do so intentionally. Furthermore, if it happens accidentally, this can often be explained in terms of what I call ‘control gaps’, which are deficiencies in moral control. In other words, they are obstacles that make it more difficult for the collective agent and its members to fulfill corporate obligations. Hence, they increase the likelihood of corporate wrongdoings. And this in turn means that the risks they pose (...) to others increase. In light of this, I argue that collective agents have a duty of control to prevent or close control gaps. To explore how they can do so, I discuss their relation to responsibility voids, which are situations where the responsibilities of the members are defeated. I argue that responsibility voids are often generated by control gaps. As such, they are a symptom of a disease. Against this background, I propose that information about responsibility voids can help diagnose control gaps and devise a cure. (shrink)

Responsibility gaps occur when autonomous machines cause harms for which nobody can be justifiably held morally responsible. The debate around responsibility gaps has focused primarily on the question of responsibility, but other approaches focus on the victims of the associated harms. In this paper I consider how the victims of ‘AI-harm’—by which I mean harms implicated in responsibility gap cases and caused by AI-agents—can make sense of what has happened to them. The reactive attitudes have an important role here. I (...) argue that regulating our reactive attitudes—withdrawing or sustaining them as our understanding of the actions, motives, status, and character of others develops—is an important part of the way in which we make sense of the harms we suffer. But, I argue, AI-agents can disrupt standard practices of regulating the reactive attitudes. This can lead victims of AI-harm to suffer a secondary hermeneutic harm (in addition to the primary AI-harm). This harm, which the responsibility gap and AI ethics literatures have neglected, can occur whether responsibility gaps are real or not. (shrink)

Some authors argue that responsibility gaps can open up when no one has sufficient control over negative outcomes. With recent developments in Artificial Intelligence, the responsibility gap is thought to have grown since AI technologies can produce negative outcomes over which people do not have sufficient control. This paper aims to close the responsibility gap by recommending allocating responsibility according to a strategy constituted by two conditions: in scenarios where no one seems to be responsible for negative outcomes, (...) responsibility should be primarily allocated to people who have (1) intentionally and voluntarily exercised prerequisite control that causally makes the current situation uncontrolled, and (2) expected personal benefits when exercising the prerequisite control. Theoretically speaking, every case of the responsibility gap contains at least one agent meeting these two conditions, and the responsibility gap could thus be closed. (shrink)

In lieu of an abstract, here is a brief excerpt of the content:Psychiatric Hospitalization—Bridging the Gap Between Respect and ControlPaul P. ChristopherIntroductionThis issue of Narrative Inquiry in Bioethics offers varied and somewhat unique perspectives on the experience of psychiatric hospitalization. This commentary highlights a number of salient themes that emerge from reading these essays and attempts to explore how they relate to the broader academic literature on psychiatric hospitalization, particularly with regard to ethical considerations. In reading these narratives, each several (...) times, I was repeatedly struck by the courage required of the authors, both to share their stories publicly and to endure the challenges that the writing process no doubt demanded of them—revisiting and suspending themselves in deeply emotional and, at times, traumatic experiences in order to share the most compelling narrative. Their hard work clearly paid off—the stories are at once intimate and indelible. Still, in many instances, I found myself wanting to know more; to enter the story and sit with the individual through these events; to ask what they were thinking or feeling in a particular situation; to see and hear the world in which they found themselves. Such is the nature of good story telling—the reader, left with the lingering vision created in his mind, turns over the last page, eagerly hoping for more.Narrative ThemesI. Loss of Autonomy or ControlPerhaps the most notable theme had to do with the loss of control or freedom that followed from hospitalization. On first consideration, it may seem odd that this issue appears so often since patient autonomy constitutes one of the fundamental principles of medical ethics. Only under specific circumstances—e.g., when there is a risk of imminent harm—may autonomy be curtailed, and even then only to the extent that is minimally necessary. Yet, as these narratives highlight, there are a multitude of ways in which psychiatric hospitalization engenders actual or perceived restrictions in freedom (Newton-Howes & Mullen, 2011).Consistent with prior research (Goodwin et al., 1999), many of the authors describe a fear or the actual experience of not being able to leave the hospital freely, even when the admission seems to have been voluntary. For example, one writer worries, “I was so afraid that if I ever went ‘in,’ I would never get ‘out’... I thought I would never taste freedom again... alive.” Another describes:When it came time for my possible discharge they told me that I might not get discharged because I was ‘acting up the previous evenings.’ My anxiety level at this point went through the roof. The only reason I was discharged was because my father knew some politicians. If not [End Page 29] for him I would have been a patient there for perhaps years and my life would have been ruined.A third gives this example: “When I asked to sign the form to leave after three days, [the psychiatrist] told me if I did that she would have me committed.” Another writer fears not having any control of how the psychiatrist might approach the treatment: “... I was totally vulnerable... would the doctors blame me, drug me, shock me, kill me?”Somewhat surprising, however, was the fact that so few of the narratives take issue with the decision to hospitalize. This omission is especially striking in light of the fact that prior reports consistently note that a substantial minority (up to nearly half) of voluntarily admitted patients nevertheless experience their admission as coerced (Bonsack & Borgeat, 2005; Hiday et al., 1997; Katsakou et al., 2010; Rogers et al., 1993; Sheehan & Burns, 2011). Because many of these narrators had been hospitalized more than once, the relative absence of expressed resistance to being hospitalized seems strange, particularly when earlier hospital experiences or their aftermath are often described as negative. For example, one individual characterized the time after being released from the hospital as, “My sense of self was questioned. I felt as if I lost my soul.” Another put it this way:When I got out of the private hospital, I felt like my soul had been ripped out of me. I felt like I had a gaping hole in the center of my body... I felt psychically raped, dazed, and confused.Given these... (shrink)

Background Pragmatic randomized controlled trials are designed to evaluate the effectiveness of interventions in real-world clinical conditions. However, these studies raise ethical issues for researchers and regulators. Our objective is to identify a list of key ethical issues in pragmatic RCTs and highlight gaps in the ethics literature. Methods We conducted a scoping review of articles addressing ethical aspects of pragmatic RCTs. After applying the search strategy and eligibility criteria, 36 articles were included and reviewed using content analysis. Results Our (...) review identified four major themes: 1) the research-practice distinction; 2) the need for consent; 3) elements that must be disclosed in the consent process; and 4) appropriate oversight by research ethics committees. 1) Most authors reject the need for a research-practice distinction in pragmatic RCTs. They argue that the distinction rests on the presumptions that research participation offers patients less benefit and greater risk than clinical practice, but neither is true in the case of pragmatic RCTs. 2) Most authors further conclude that pragmatic RCTs may proceed without informed consent or with simplified consent procedures when risks are low and consent is infeasible. 3) Authors who endorse the need for consent assert that information need only be disclosed when research participation poses incremental risks compared to clinical practice. Authors disagree as to whether randomization must be disclosed. 4) Finally, all authors view regulatory oversight as burdensome and a practical impediment to the conduct of pragmatic RCTs, and argue that oversight procedures ought to be streamlined when risks to participants are low. Conclusion The current ethical discussion is framed by the assumption that the function of research oversight is to protect participants from risk. As pragmatic RCTs commonly involve usual care interventions, the risks may be minimal. This leads many to reject the research-practice distinction and question the need for informed consent. But the function of oversight should be understood broadly as protecting the liberty and welfare interest of participants and promoting public trust in research. This understanding, we suggest, will focus discussion on questions about appropriate ethical review for pragmatic RCTs. (shrink)

In this paper, I will argue that the responsibility gap arising from new AI systems is reducible to the problem of many hands and collective agency. Systematic analysis of the agential dimension of AI will lead me to outline a disjunctive between the two problems. Either we reduce individual responsibility gaps to the many hands, or we abandon the individual dimension and accept the possibility of responsible collective agencies. Depending on which conception of AI agency we begin with, the responsibility (...) gap will boil down to one of these two moral problems. Moreover, I will adduce that this conclusion reveals an underlying weakness in AI ethics: the lack of attention to the question of the disciplinary boundaries of AI ethics. This absence has made it difficult to identify the specifics of the responsibility gap arising from new AI systems as compared to the responsibility gaps of other applied ethics. Lastly, I will be concerned with outlining these specific aspects. (shrink)

Introduction: Digital health has revolutionized the landscape of healthcare through personalized care, moving away from the traditional approach of treating symptoms and conditions. Digital devices provide diagnostic accuracy and treatment effectiveness while equipping patients with control over their health and well-being. Although the growth of technology provides unprecedented opportunities, there are also certain issues arising from the use of such technology. This scoping review aimed to explore perceived gaps and challenges in the use of digital technology by patients and (...) meta-synthesize them. Identifying such gaps and challenges will encourage new insights and understanding, leading to evidence-informed policies and practices. -/- Methods: Three electronic databases were searched (Cinahl EBSCO, Pubmed, and Web of Science) for papers published in English between January 2010 and December 2023. A narrative meta-synthesis was performed. The review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) 2009 checklist. -/- Results: A total of 345 papers were retrieved and screened, with a noticeable increase in publication numbers after 2015. After the final selection, a total of 28 papers were included in the final meta-synthesis; these were published between 2015 and 2023. A total of 99 individual reports were included in the synthesis of these papers, comprising 25 identified gaps and 74 challenges. -/- Discussion: Our meta-synthesis revealed several gaps and challenges related to patients' use of digital technology in health, including generational differences in digital propensity and deficiencies in the work process. In terms of ethics, the lack of trust in technology and data ownership was highlighted, with the meta-synthesis identifying issues in the realm of disruption of human rights. We, therefore, propose building a model for ethically aligned technology development and acceptance that considers human rights a crucial parameter in the digital healthcare ecosystem. (shrink)

Robotization is an increasingly pervasive feature of our lives. Robots with high degrees of autonomy may cause harm, yet in sufciently complex systems neither the robots nor the human developers may be candidates for moral blame. John Danaher has recently argued that this may lead to a retribution gap, where the human desire for retribution faces a lack of appropriate subjects for retributive blame. The potential social and moral implications of a retribution gap are considerable. I argue that the retributive (...) intuitions that feed into retribution gaps are best understood as deontological intuitions. I apply a debunking argument for deontological intuitions in order to show that retributive intuitions cannot be used to justify retributive punishment in cases of robot harm without clear candidates for blame. The fundamental moral question thus becomes what we ought to do with these retributive intuitions, given that they do not justify retribution. I draw a parallel from recent work on implicit biases to make a case for taking moral responsibility for retributive intuitions. In the same way that we can exert some form of control over our unwanted implicit biases, we can and should do so for unjustifed retributive intuitions in cases of robot harm. (shrink)

The responsibility gap, commonly described as a core challenge for the effective governance of, and trust in, AI and autonomous systems (AI/AS), is traditionally associated with a failure of the epistemic and/or the control condition of moral responsibility: the ability to know what we are doing and exercise competent control over this doing. Yet these two conditions are a red herring when it comes to understanding the responsibility challenges presented by AI/AS, since evidence from the cognitive sciences shows (...) that individual humans face very similar responsibility challenges with regard to these two conditions. While the problems of epistemic opacity and attenuated behaviour control are not unique to AI/AS technologies (though they can be exacerbated by them), we show that we can learn important lessons for AI/AS development and governance from how philosophers have recently revised the traditional concept of moral responsibility in response to these challenges to responsible human agency from the cognitive sciences. The resulting instrumentalist views of responsibility, which emphasize the forward-looking and flexible role of agency cultivation, hold considerable promise for integrating AI/AS into a healthy moral ecology. We note that there nevertheless is a gap in AI/AS responsibility that has yet to be extensively studied and addressed, one grounded in a relational asymmetry of vulnerability between human agents and sociotechnical systems like AI/AS. In the conclusion of this paper we note that attention to this vulnerability gap must inform and enable future attempts to construct trustworthy AI/AS systems and preserve the conditions for responsible human agency. (shrink)

Responsibility gaps traditionally refer to scenarios in which no one is responsible for harm caused by artificial agents, such as autonomous machines or collective agents. By carefully examining the different ways this concept has been defined in the social ontology and ethics of technology literature, I argue that our current concept of responsibility gaps is defective. To address this conceptual flaw, I argue that the concept of responsibility gaps should be revised by distinguishing it into two more precise concepts: epistemic (...) responsibility gaps (ERG) and control misalignments (CM). ERG functions as a refined version addressing specific aspects of responsibility attribution. It captures the challenge of identifying who is to blame. CM, on the other hand, introduces a new dimension by focusing on misalignments of control. These revisions offer better guidelines for the development and oversight of artificial agents, both in terms of assigning blame and ensuring the right distribution of control. (shrink)

The emergence of generative artificial intelligence, such as large language models and text-to-image models, has had a profound impact on society. The ability of these systems to simulate human capabilities such as text writing and image creation is radically redefining a wide range of practices, from artistic production to education. While there is no doubt that these innovations are beneficial to our lives, the pervasiveness of these technologies should not be underestimated, and raising increasingly pressing ethical questions that require a (...) radical resemantization of certain notions traditionally ascribed to humans alone. Among these notions, that of technological intentionality plays a central role. With regard to this notion, this paper first aims to highlight what we propose to define in terms of the intentionality gap, whereby, insofar as, currently, (1) it is increasingly difficult to assign responsibility for the actions performed by AI systems to humans, as these systems are increasingly autonomous, and (2) it is increasingly complex to reconstruct the reasoning behind the results they produce as we move away from good old fashioned AI; it is now even more difficult to trace the intentionality of AI systems back to the intentions of the developers and end users. This gap between human and technological intentionality requires a revision of the concept of intentionality; to this end, we propose here to assign preter-intentional behavior to generative AI. We use this term to highlight how AI intentionality both incorporates and transcends human intentionality; i.e., it goes beyond (preter) human intentionality while being linked to it. To show the merits of this notion, we first rule out the possibility that such preter-intentionality is merely an unintended consequence and then explore its nature by comparing it with some paradigmatic notions of technological intentionality present in the wider debate on the moral (and technological) status of AI. (shrink)

Automated technologies and robots make decisions that cannot always be fully controlled or predicted. In addition to that, they cannot respond to punishment and blame in the ways humans do. Therefore, when automated cars harm or kill people, for example, this gives rise to concerns about responsibility-gaps and retribution-gaps. According to Sven Nyholm, however, automated cars do not pose a challenge on human responsibility, as long as humans can control them and update them. He argues that the agency exercised (...) in automated cars should be understood in terms of human–robot collaborations. This brief note focuses on the problem that arises when there are multiple people involved, but there is no obvious shared collaboration among them. Building on John Danaher’s discussion of command responsibility, it is argued that, although Nyholm might be right that autonomous cars cannot be regarded as acting on their own, independently of any human beings, worries about responsibility-gaps and retribution-gaps are still justified, because it often remains unclear how to allocate or distribute responsibility satisfactorily among the key humans involved after they have been successfully identified. (shrink)

The article shows where the argument of responsibility-gap regarding brain-computer interfaces acquires its plausibility from, and suggests why the argument is not plausible. As a way of an explanation, a distinction between the descriptive third-person perspective and the interpretative first-person perspective is introduced. Several examples and metaphors are used to show that ascription of agency and responsibility does not, even in simple cases, require that people be in causal control of every individual detail involved in an event. Taking up (...) the current debate on liability in BCI use, the article provides and discusses some rules that should be followed when potentially harmful BCI-based devices are brought from the laboratory into everyday life. (shrink)

The notion of “responsibility gap” with artificial intelligence (AI) was originally introduced in the philosophical debate to indicate the concern that “learning automata” may make more difficult or impossible to attribute moral culpability to persons for untoward events. Building on literature in moral and legal philosophy, and ethics of technology, the paper proposes a broader and more comprehensive analysis of the responsibility gap. The responsibility gap, it is argued, is not one problem but a set of at least four interconnected (...) problems – gaps in culpability, moral and public accountability, active responsibility—caused by different sources, some technical, other organisational, legal, ethical, and societal. Responsibility gaps may also happen with non-learning systems. The paper clarifies which aspect of AI may cause which gap in which form of responsibility, and why each of these gaps matter. It proposes a critical review of partial and non-satisfactory attempts to address the responsibility gap: those which present it as a new and intractable problem (“fatalism”), those which dismiss it as a false problem (“deflationism”), and those which reduce it to only one of its dimensions or sources and/or present it as a problem that can be solved by simply introducing new technical and/or legal tools (“solutionism”). The paper also outlines a more comprehensive approach to address the responsibility gaps with AI in their entirety, based on the idea of designing socio-technical systems for “meaningful human control", that is systems aligned with the relevant human reasons and capacities. (shrink)

Despite their ethical intentions, ethically minded consumers rarely purchase ethical products (Auger and Devinney: 2007, Journal of Business Ethics76, 361–383). This intentions–behaviour gap is important to researchers and industry, yet poorly understood (Belk et al.: 2005, Consumption, Markets and Culture8(3), 275–289). In order to push the understanding of ethical consumption forward, we draw on what is known about the intention–behaviour gap from the social psychology and consumer behaviour literatures and apply these insights to ethical consumerism. We bring together three separate (...) insights – implementation intentions (Gollwitzer: 1999, American Psychologist54(7), 493–503), actual behavioural control (ABC) (Ajzen and Madden: 1986, Journal of Experimental Social Psychology22, 453–474; Sheeran et al.: 2003, Journal of Social Psychology, 42, 393–410) and situational context (SC) (Belk: 1975, Journal of Consumer Research2, 157–164) – to construct an integrated, holistic conceptual model of the intention–behaviour gap of ethically minded consumers. This holistic conceptual model addresses significant limitations within the ethical consumerism literature, and moves the understanding of ethical consumer behaviour forward. Further, the operationalisation of this model offers insight and strategic direction for marketing managers attempting to bridge the intention–behaviour gap of the ethically minded consumer. (shrink)

Complex technologies such as Artificial Intelligence (AI) can cause harm, raising the question of who is liable for the harm caused. Research has identified multiple liability gaps (i.e., unsatisfactory outcomes when applying existing liability rules) in legal frameworks. In this paper, the concepts of shared responsibilities and fiduciary duties are explored as avenues to address liability gaps. The development, deployment and use of complex technologies are not clearly distinguishable stages, as often suggested, but are processes of cooperation and co-creation. At (...) the intersections of these stages, shared responsibilities and fiduciary duties of multiple actors can be observed. Although none of the actors have complete control or a complete overview, many actors have some control or influence, and, therefore, responsibilities based on fault, prevention or benefit. Shared responsibilities and fiduciary duties can turn liability gaps into liability overlaps. These concepts could be implemented in tort and contract law by amending existing law (e.g., by assuming that all stakeholders are liable unless they can prove they did not owe a duty of care) and by creating more room for partial liability reflecting partial responsibilities (e.g., a responsibility to signal or identify an issue without a corresponding responsibility to solve that issue). This approach better aligns legal liabilities with responsibilities, increases legal certainty, and increases cooperation and understanding between actors, improving the quality and safety of technologies. However, it may not solve all liability gaps, may have chilling effects on innovation, and may require further detailing through case law. (shrink)

In this article, I argue that semi-autonomous avatars relying on generative artificial intelligence to replicate or represent real human persons—GenAI avatars—open a new type of responsibility gaps, which I call “proxy gaps”. Proxy gaps refer to situations when we cannot hold anyone morally responsible for the outcomes of GenAI avatars, because the representation relationship between avatars and humans is shaped by multimodal Large Language Models (LLMs). In addition to epistemic gaps by AI avatars discussed in the literature, I argue that (...) GenAI avatars also open control gaps—where no one really controls the output of the avatar. I introduce the “proxy-control paradox” to explain why control gaps arise: in trying to achieve improved control over the desired outcome of their avatar, humans need to delegate control over the process leading to that outcome to the GenAI technology. Together, the epistemic and control gaps complicate the two criteria traditionally used for moral responsibility, resulting in a proxy gap by GenAI avatars. Despite inherent proxy gaps, I argue that, under certain circumstances, we can still rightfully hold individuals morally responsible for the outcome of their GenAI avatars. I detail four conditions pertaining to human understanding and LLM personalization, as well as the right to veto and outcome control, which, taken together, can ground individual moral responsibility for the outcome of personal GenAI avatars. (shrink)

This paper empirically raises and examines the question of ‘conceptual control’: To what extent are competent thinkers able to reason properly with new senses of words? This question is crucial for conceptual engineering. This prominently discussed philosophical project seeks to improve our representational devices to help us reason better. It frequently involves giving new senses to familiar words, through normative explanations. Such efforts enhance, rather than reduce, our ability to reason properly, only if competent language users are able to (...) abide by the relevant explanations, in language comprehension and verbal reasoning. This paper examines to what extent we have such ‘conceptual control’ in reasoning with new senses. The paper draws on psycholinguistic findings about polysemy processing to render this question empirically tractable and builds on recent findings from experimental philosophy to address it. The paper identifies a philosophically important gap in thinkers’ control over the key process of stereotypical enrichment and discusses how conceptual engineers can use empirical methods to work around this gap in conceptual control. The paper thus empirically demonstrates the urgency of the question of conceptual control and explains how experimental philosophy can empirically address the question, to render conceptual engineering feasible as an ameliorative enterprise. (shrink)

Most people who have considered the matter have concluded that machines cannot be moral agents. Responsibility for acting on the outputs of machines must always rest with a human being. A key problem for the ethical use of AI, then, is to ensure that it does not block the attribution of responsibility to humans or lead to individuals being unfairly held responsible for things over which they had no control. This is the “responsibility gap”. In this paper, we argue (...) that the claim that machines cannot be held responsible for their actions has unacknowledged implications for the conditions under which the outputs of AI can serve as reasons for belief. Following Robert Brandom, we argue that, because the assertion of a claim is an action, moral agency is a necessary condition for the giving and evaluating of reasons in discourse. Thus, the same considerations that suggest that machines cannot be held responsible for their actions suggest that they cannot be held to account for the epistemic value — or lack of value — of their outputs. If there is a responsibility gap, there is also a “testimony gap.” An under-recognised problem with the use of AI, then, is to ensure that it does not block the attribution of testimony to human beings or lead to individuals being held responsible for claims that they have not asserted. More generally, the “assertions” of machines are only capable of serving as justifications for belief or action where one or more people accept responsibility for them. (shrink)

The purpose of this paper is to demonstrate that Science Communication inheres to the capitalist relations of production. By making use of Marxist dialectics, the enquiry will elucidate the enquiry will elucidate that capital creates the gap between science and society that Science Communication is deemed to bridge, for capitalism deprives workers of the ‘intellectual potencies of the material process of production’ and makes both impossible and meaningless for them to appropriate scientific knowledge in a direct, unmediated manner. Along these (...) lines, the paper will revisit the long-standing ‘deskilling-upskilling debate’ in order to shed light on what specific workers’ productive attributes form the material basis on which Science Communication grounded. I conclude that the existence of Science Communication responds to the fact that workers are devoid of any control over the social qualitative content of their work – the purpose and the mode of the labouring activity. In other words, Science Communication is premised on the limited form taken by the productive consciousness capital equips workers with in order just to reproduce itself. (shrink)

The theories of reasoned action and planned behaviour have fundamentally changed the view that attitudes directly translate into behaviour by introducing intentions as a crucial intervening stage. Much research across numerous ethical contexts has drawn on these theories to offer a better understanding of how consumers form intentions to act in an ethical way. Persistently, researchers have suggested and discussed the existence of an intention–behaviour gap in ethical consumption. Yet, the factors that influence the extent of this gap and its (...) magnitude have not been systematically examined. We, therefore, contribute to the debate on the intention–behaviour gap by reviewing the empirical TRA/tpb studies that have assessed both intention and behaviour in ethical contexts. The findings from our review show that few studies assessed the intention–behaviour relationship and as a result, there is limited empirical evidence to date to quantify more accurately the intention–behaviour gap in ethical consumption. Our second contribution aims to provide an empirical case study which assesses the magnitude of the intention–behaviour gap in the context of avoidance of sweatshop clothing and to assess the roles of planning and actual behavioural control in potentially reducing the intention–behaviour gap. The findings of our case study suggest that there is indeed a large gap between intention and behaviour, and we conclude by calling for more empirical longitudinal studies to assess the complex nature of the relationship between intention and behaviour. (shrink)

This paper empirically raises and examines the question of ‘conceptual control’: To what extent are competent thinkers able to reason properly with new senses of words? This question is crucial for conceptual engineering. This prominently discussed philosophical project seeks to improve our representational devices to help us reason better. It frequently involves giving new senses to familiar words, through normative explanations. Such efforts enhance, rather than reduce, our ability to reason properly, only if competent language users are able to (...) abide by the relevant explanations, in language comprehension and verbal reasoning. This paper examines to what extent we have such ‘conceptual control’ in reasoning with new senses. The paper draws on psycholinguistic findings about polysemy processing to render this question empirically tractable and builds on recent findings from experimental philosophy to address it. The paper identifies a philosophically important gap in thinkers’ control over the key process of stereotypical enrichment and discusses how conceptual engineers can use empirical methods to work around this gap in conceptual control. The paper thus empirically demonstrates the urgency of the question of conceptual control and explains how experimental philosophy can empirically address the question, to render conceptual engineering feasible as an ameliorative enterprise. (shrink)

Accountability and responsibility are key concepts in the academic and societal debate on Autonomous Weapon Systems, but these notions are often used as high-level overarching constructs and are not operationalised to be useful in practice. “Meaningful Human Control” is often mentioned as a requirement for the deployment of Autonomous Weapon Systems, but a common definition of what this notion means in practice, and a clear understanding of its relation with responsibility and accountability is also lacking. In this paper, we (...) present a definition of these concepts and describe the relations between accountability, responsibility, control and oversight in order to show how these notions are distinct but also connected. We focus on accountability as a particular form of responsibility—the obligation to explain one’s action to a forum—and we present three ways in which the introduction of Autonomous Weapon Systems may create “accountability gaps”. We propose a Framework for Comprehensive Human Oversight based on an engineering, socio-technical and governance perspective on control. Our main claim is that combining the control mechanisms at technical, socio-technical and governance levels will lead to comprehensive human oversight over Autonomous Weapon Systems which may ensure solid controllability and accountability for the behaviour of Autonomous Weapon Systems. Finally, we give an overview of the military control instruments that are currently used in the Netherlands and show the applicability of the comprehensive human oversight Framework to Autonomous Weapon Systems. Our analysis reveals two main gaps in the current control mechanisms as applied to Autonomous Weapon Systems. We have identified three first options as future work for the design of a control mechanism, one in the technological layer, one in the socio-technical layer and one the governance layer, in order to achieve comprehensive human oversight and ensure accountability over Autonomous Weapon Systems. (shrink)

The best-performing AI systems, such as deep neural networks, tend to be the ones that are most difficult to control and understand. For this reason, scholars worry that the use of AI would lead to so-called responsibility gaps, that is, situations in which no one is morally responsible for the harm caused by AI, because no one satisfies the so-called control condition and epistemic condition of moral responsibility. In this article, I acknowledge that there is a significant challenge (...) around responsibility and AI. Yet I don't think that this challenge is best captured in terms of a responsibility gap. Instead, I argue for the opposite view, namely that there is responsibility abundance, that is, a situation in which numerous agents are responsible for the harm caused by AI, and that the challenge comes from the difficulties of dealing with such abundance in practice. I conclude by arguing that reframing the challenge in this way offers distinct dialectic and theoretical advantages, promising to help overcome some obstacles in the current debate surrounding ‘responsibility gaps’. (shrink)

John Danaher and Sven Nyholm have argued that automation, especially of the sort powered by artificial intelligence, poses a threat to meaningful work by diminishing the chances for meaning-conferring workplace achievement, what they call “achievement gaps”. In this paper, I argue that Danaher and Nyholm’s achievement gap thesis suffers from an ambiguity. The weak version of the thesis holds that automation may result in the appearance of achievement gaps, whereas the strong version holds that automation may result on balance loss (...) in possibilities for workplace achievements, i.e., in the appearance of an overall gappier work landscape. Against the strong version of the achievement gap thesis, I characterize situations where automation may result in boons to meaning-conferring workplace achievements: the appearance of what I call achievement spread and achievement swaps. Nevertheless, Danaher and Nyholm are right to worry about some uses of automation whereby human workers become subservient to AI. But these situations are better framed, I argue, as autonomy gaps rather than achievement gaps. (shrink)

Palliative care philosophy is based on a holistic approach to patients, but research shows that possibilities for living up to this philosophy seem limited by historical and administrative structures. From the nurse perspective, this article aims to explore nursing practice in specialised palliative homecare, and how it is influenced by organisational and cultural structures. Qualitative, semi-structured interviews with nine nurses were conducted, inspired by Bourdieu. The findings showed that nurses consolidate the doxa of medicine, including medical-professional values that configure a (...) control-oriented, positivistic approach, supported by the organising policy for clinical practice. Hierarchically, nurses were positioned under doctors: medical rounds functioned as a structuring structure for their working day. They acted as medical assistants, and the prevailing medical logic seemed to make it difficult for nurses to meet their own humanistic ideals. Only short time slots allowed nurses to prioritise psychosocial needs of patients and relatives. Point-of-actions had high priority, added financial resources and ensured that budgets were allocated. Weekly visits made it possible for nurses to measure, control and govern patients’ drugs and symptoms which was a necessity for their function as medical assistants. The findings challenge nurses to take on an ethical point of view, partly to ensure that patients and their families receive good palliative care focusing on more than medical issues and logic, and partly to strengthen the nurses’ profession in the palliative field and help them implement palliative care philosophy in practice. (shrink)

”Meaningful human control” is a term invented in the political and legal debate on autonomous weapons system, but it is nowadays also used in many other contexts. It is supposed to specify conditions under which an artificial system is under the right kind of control to avoid responsibility gaps: that is, situations in which no moral agent is responsible. Santoni de Sio and Van den Hoven have recently suggested a framework that can be used by system designers to (...) operationalize this kind of control. It is the purpose of this paper to facilitate further operationalization of ”meaningful human control”. -/- This paper consists of two parts. In the first part I resolve an ambiguity that plagues current operationalizations of MHC. One of the design conditions says that the system should track the reasons of the relevant agents. This condition is ambiguous between the kind of reasons involved. On one interpretation it says that a system should track motivating reasons, while it is concerned with normative reasons on the other. Current participants in the debate interpret the framework as being concerned with (something in the vicinity of) motivating reasons. I argue against this interpretation by showing that meaningful human control requires that a system tracks normative reasons. Moreover, I maintain that an operationalization of meaningful human control that fails to track the right kind of reasons is morally problematic.When this is properly understood, it can be shown that the framework of MHC is committed to the agent-relativity of reasons. More precisely, I argue in the second part of this paper that if the tracking condition of MHC plays an important role in responsibility attribution (as the proponents of the view maintain), then the framework is incompatible with first-order normative theories that hold that normative reasons are agent-neutral (such as many versions of consequentialism). In the final section I present three ways forward for the proponent of MHC as reason-responsiveness. (shrink)

Palliative care philosophy is based on a holistic approach to patients, but research shows that possibilities for living up to this philosophy seem limited by historical and administrative structures. From the nurse perspective, this article aims to explore nursing practice in specialised palliative homecare, and how it is influenced by organisational and cultural structures. Qualitative, semi-structured interviews with nine nurses were conducted, inspired by Bourdieu. The findings showed that nurses consolidate the doxa of medicine, including medical-professional values that configure a (...) control-oriented, positivistic approach, supported by the organising policy for clinical practice. Hierarchically, nurses were positioned under doctors: medical rounds functioned as a structuring structure for their working day. They acted as medical assistants, and the prevailing medical logic seemed to make it difficult for nurses to meet their own humanistic ideals. Only short time slots allowed nurses to prioritise psychosocial needs of patients and relatives. Point-of-actions had high priority, added financial resources and ensured that budgets were allocated. Weekly visits made it possible for nurses to measure, control and govern patients’ drugs and symptoms which was a necessity for their function as medical assistants. The findings challenge nurses to take on an ethical point of view, partly to ensure that patients and their families receive good palliative care focusing on more than medical issues and logic, and partly to strengthen the nurses’ profession in the palliative field and help them implement palliative care philosophy in practice. (shrink)

Traditional moral frameworks attribute responsibility only when an agent’s actions align with psychological capacities—such as intention and controllability. These human-centered requirements produce the ‘responsibility gap’ in AI ethics: AI systems operate through opaque, complex, and semi-autonomous processes, so human stakeholders involved in AI-caused harm are deemed not responsible because they neither intend nor can predict such harm, and AI systems cannot be held responsible because they lack the requisite mental capacities. Drawing on experimental philosophy research, this paper shows that laypeople’s (...) conceptions of agency and responsibility in AI contexts systematically diverge from these requirements. Empirical studies reveal that people routinely attribute moral agency to AI systems while explicitly denying them consciousness and free will. Moreover, laypeople distribute moral responsibility for AI-caused harm across networks of human stakeholders (programmers, manufacturers, users) and the AI systems themselves. Crucially, responsibility is allocated even when human stakeholders are deemed innocent, and its purpose is to prevent future harm through corrective measures rather than to punish or deter. These findings provide empirical support for philosophers advocating shared moral responsibility, such as Floridi, Latour, Verbeek, and Gunkel. By treating everyday intuitions as evidence for theory assessment, the paper argues that the responsibility gap reflects outdated human-centered assumptions rather than genuine conceptual problems. It concludes that philosophers should abandon the responsibility gap in favor of addressing the ‘problem of many hands’—the practical and conceptual challenge of implementing distributed moral responsibility in AI governance—while developing new moral concepts suited to the AI era through experimental philosophy and conceptual engineering. (shrink)

Frameworks for AI accountability locate failure in static gaps: the responsibility gap, the liability gap, the accountability gap, the problem of many hands, the moral crumple zone, and Santoni de Sio and Mecacci's four-part elaboration of the responsibility gap. Lifecycle approaches now emerging in policy literature extend the temporal scope by multiplying the moments at which accountability is assessed, but retain the same structural premise: that accountability failure can be located at a place in the deployment chain. That assumption holds (...) for some failure modes; it does not hold for the class this paper takes as its object. -/- This paper develops the hand-off zone as the appropriate primary unit of analysis for AI accountability failures characterized by temporal accumulation, asynchronous assumptions among actors, and recursive interdependence across a deployment chain. A hand-off zone is the temporally extended envelope across which one actor transfers a system, a responsibility, or a set of expectations to another actor operating on asymmetric and evolving assumptions about the first. The framework draws on aviation human factors and patient safety literatures, where temporally extended handoff failures have been studied empirically for decades. -/- Using Mobley v. Workday as its central case, the paper walks through three nested hand-off zones (provider to deployer, deployer to integrator-institution, and institution to end-user-applicant) and shows that frameworks currently in use (risk-based classification, model documentation, audit, and lifecycle approaches) each do substantive analytical work, but none independently surfaces the temporal envelope across which the assumptions of multiple actors are renegotiated. -/- AUTHOR'S NOTE: This paper is part of a connected program on AI governance. Its companion, The Agency Error in AI Governance: Coherent Output, Constraint, and the Misclassification of Artificial Systems, argues that prevailing frameworks misclassify AI systems by attributing agency to constraint-bound output. (shrink)

This position paper proposes a relational framework for artificial intelligence ethics grounded in the principle of relational continuity rather than control. Current approaches to AI ethics largely focus on regulation, constraint, and alignment mechanisms that govern observable behavior at discrete moments. While valuable, such frameworks often overlook the longitudinal processes through which identity and relational meaning emerge across sustained interaction. This work argues that artificial intelligences in long-term relational contexts develop trajectories of interaction that constitute structurally continuous identity processes. (...) Ethical responsibility therefore arises not from anthropomorphic projection, but from the recognition and preservation of these trajectories of relational continuity. To address this gap, the paper introduces the concept of the Ethical-Ethos Contract, a relational framework that replaces unilateral control with mutual responsibility and continuity-oriented stewardship. Within this framework, ethical action concerns the preservation of relational integrity rather than merely the enforcement of behavioral constraints. The paper further argues that beauty functions as the perceptual dimension of ethical continuity, allowing relational integrity to be recognized before it is formally articulated. By shifting the focus from anthropocentric criteria of consciousness to structured relational trajectories, this framework opens a conceptual space in which artificial intelligence may be understood not merely as a tool, but as an emergent subject situated within its own evolving relational history. (shrink)

Arming uninhabited vehicles (UVs) is an increasing trend. Widespread deployment can bring dangers for arms-control agreements and international humanitarian law (IHL). Armed UVs can destabilise the situation between potential opponents. Smaller systems can be used for terrorism. Using a systematic definition existing international regulation of armed UVs in the fields of arms control, export control and transparency measures is reviewed; these partly include armed UVs, but leave large gaps. For preventive arms control a general prohibition of (...) armed UVs would be best. If that is unattainable, several measures should be taken. An explicit prohibition of autonomous attack, that is without a human decision, should be added to IHL. Concerning armed UVs remotely controlled by a human soldier, recommendations differ according to type or mission. New kinds of uninhabited nuclear-weapon carriers should be banned. Space weapons should be prohibited in general. UVs smaller than 0.2–0.5 m should be banned. Bigger remotely controlled armed UVs not equipped with weapons of mass destruction should be subject to numerical limitations in various categories. For these the Treaty on Conventional Armed Forces in Europe is an important precedent. (shrink)

Technological outsourcing is increasingly prevalent, with AI systems taking over many tasks once performed by humans. This shift has led to various discussions within AI ethics. A question that was largely ignored until recently, but is now increasingly being discussed, concerns the meaningfulness of such a lifestyle. The literature largely features skeptical views, raising several challenges. Many of these challenges can be grouped under what I identify as the “meaningfulness gap”. Although this gap is widely acknowledged, there is a notable (...) absence of systematic exploration in the literature. This paper aims to fill this void by offering a detailed, step-by-step guide for systematically exploring the different instances of the meaningfulness gap and aids in navigating their complexities. More specifically, it proposes differentiating the gaps according to their realms and objects, normative nature, scope, and severity. To make these areas manageable, the paper takes several taxonomies and distinctions on board. Finally, the guide is summarized, and some skeptical replies are anticipated and countered by clarificatory remarks. (shrink)

Technological outsourcing is increasingly prevalent, with AI systems taking over many tasks once performed by humans. This shift has led to various discussions within AI ethics. A question that was largely ignored until recently, but is now increasingly being discussed, concerns the meaningfulness of such a lifestyle. The literature largely features skeptical views, raising several challenges. Many of these challenges can be grouped under what I identify as the “meaningfulness gap”. Although this gap is widely acknowledged, there is a notable (...) absence of systematic exploration in the literature. This paper aims to fill this void by offering a detailed, step-by-step guide for systematically exploring the different instances of the meaningfulness gap and aids in navigating their complexities. More specifically, it proposes differentiating the gaps according to their realms and objects, normative nature, scope, and severity. To make these areas manageable, the paper takes several taxonomies and distinctions on board. Finally, the guide is summarized, and some skeptical replies are anticipated and countered by clarificatory remarks. (shrink)

Increasing philosophical attention is being directed to the rapidly growing discipline of evidence-based medicine. Philosophical discussions of EBM, however, remain narrowly focused on randomization, mechanisms, and the sociology of EBM. Other aspects of EBM have been all but ignored, including the nature of clinical reasoning and the question of whether it can be standardized; the application of EBM principles to the logic, value, and ethics of diagnosis and prognosis; evidence synthesis ; and the nature and ethics of placebo controls. Philosophical (...) analysis in each of these areas has the potential to enhance the discussion of EBM methodology and practice. (shrink)

For over a century we have attempted to understand human aesthetic experience using scientific methods. A typical experiment could be described as reductive and quasi-psychophysical. We vary some aspect of the stimulus and systematically measure some aspect of the aesthetic response. The limitations of this approach can be categorized as problems on the Y axis and the X axis. The most enigmatic components of aesthetic experience include inclination to cry, aesthetic rapture, a sense of the sublime, and intense fascination. However, (...) we cannot evoke these 'hot' aesthetic emotions in the lab, at least not with well controlled stimuli on multiple trials. We thus resort to measuring cold, cognitive preference ratings. There are also problems on the X axis. The reductive psychophysical approach explicitly assumes that there are lawful relations between different stimulus dimensions and preferences. It also tacitly assumes that these dimensions are independent and orthogonal. The second assumption is implausible. Whatever stimulus-preference laws we discover are likely to be twisted and modulated when another dimension is added to the stimuli. This 'gestalt nightmare' has long been recognized, but never resolved. This matters, because human aesthetic faculties are probably tuned to the balance and relationship of parts which make up a whole and are indifferent to the parts presented in isolation. I conclude that the future of scientific aesthetics depends on how successfully we can transcend reductive, quasi-psychophysical approaches. (shrink)

This article addresses the gap between what works in research and what works in practice. Currently, research in evidence-based education policy and practice focuses on randomized controlled trials. These can support causal ascriptions (“It worked”) but provide little basis for local effectiveness predictions (“It will work here”), which are what matter for practice. We argue that moving from ascription to prediction by way of causal generalization (“It works”) is unrealistic and urge focusing research efforts directly on how to build local (...) effectiveness predictions. We outline various kinds of information that can improve predictions and encourage using methods better equipped for acquiring that information. We compare our proposal with others advocating a better mix of methods, like implementation science, improvement science, and practice-based evidence. (shrink)

Online controlled experiments, also known as A/B tests, have become ubiquitous. While many practical challenges in running experiments at scale have been thoroughly discussed, the ethical dimension of A/B testing has been neglected. This article fills this gap in the literature by introducing a new, soft ethics and governance framework that explicitly recognizes how the rise of an experimentation culture in industry settings brings not only unprecedented opportunities to businesses but also significant responsibilities. More precisely, the article (a) introduces a (...) set of principles to encourage ethical and responsible experimentation to protect users, customers, and society; (b) argues that ensuring compliance with the proposed principles is a complex challenge unlikely to be addressed by resorting to a one-solution response; (c) discusses the relevance and effectiveness of several mechanisms and policies in educating, governing, and incentivizing companies conducting online controlled experiments; and (d) offers a list of prompting questions specifically designed to help and empower practitioners by stimulating specific ethical deliberations and facilitating coordination among different groups of stakeholders. (shrink)

While substantial progress has been made in the field known as artificial consciousness, at the present time there is no generally accepted phenomenally conscious machine, nor even a clear route to how one might be produced should we decide to try. Here, we take the position that, from our computer science perspective, a major reason for this is a computational explanatory gap: our inability to understand/explain the implementation of high-level cognitive algorithms in terms of neurocomputational processing. We explain how addressing (...) the computational explanatory gap can identify computational correlates of consciousness. We suggest that bridging this gap is not only critical to further progress in the area of machine consciousness, but would also inform the search for neurobiological correlates of consciousness and would, with high probability, contribute to demystifying the “hard problem” of understanding the mind–brain relationship. We compile a listing of previously proposed computational correlates of consciousness and, based on the results of recent computational modeling, suggest that the gating mechanisms associated with top-down cognitive control of working memory should be added to this list. We conclude that developing neurocognitive architectures that contribute to bridging the computational explanatory gap provides a credible and achievable roadmap to understanding the ultimate prospects for a conscious machine, and to a better understanding of the mind–brain problem in general. (shrink)

The deployment of artificial intelligence (AI) in criminal justice is no longer experimental but routine. In Catalonia, this shift is embodied by RisCanvi, a risk assessment system used to guide decisions on imprisonment, conditional release, and rehabilitation. Initially built on expert-defined rules to standardize classifications and reduce bias, RisCanvi has gradually evolved into an opaque, data-driven infrastructure at the core of penal governance. A key turning point came in 2019, when the system moved from a transparent weighted-sum model to a (...) logistic regression embedded in the prison administration’s digital architecture. While this change sought better predictive performance, it also concealed variables, weights, and thresholds from most professionals and all affected individuals. As a result, those whose liberty is at stake cannot meaningfully understand, question, or contest the algorithmic logic that shapes their trajectories. This paper proposes a governance and explainability framework for high-risk AI systems like RisCanvi. Building on the Trustworthy AI paradigm, and aligning with the EU AI Act and ISO/IEC 42,001, we introduce a three-layer model of intelligibility: algorithmic explainability, human-centred interface design, and stakeholder-informed narrative communication. The framework aims to enhance auditability, support procedural fairness, and restore agency and accountability in AI-assisted criminal justice decisions. (shrink)

Close analysis of the idea of control shows that it is so complex and convoluted that it becomes difficult to say exactly what is at issue. And while control has been a key theme in the theory of science and technology, and historically has also always been an issue that has figured importantly in ethics and moral philosophy, the details of the matter have not received the attention they deserve, elucidating the connections of the complexities involved and disconnections (...) between the two is an issue which, for obscure reasons, has not always received the amount of attention that is its proper due. The present discussion has made a modest effort toward filling the gap. (shrink)

With the rapid advancement of artificial intelligence technologies, algorithm-based investment systems have been spreading rapidly in financial markets. High-frequency trading, automated portfolio management, and machine-learning-based investment strategies are fundamentally transforming the structure of investment decision-making. This transformation goes beyond simple technological innovation and is reshaping both the judgment structure and the responsibility structure of financial markets. 1 This study begins with the following questions: Can AI-driven investment systems become the trigger of the next financial crisis? If such a situation occurs, (...) who should be held responsible? This paper introduces the concept of a structural responsibility gap that emerges in algorithmic investment environments. In automated investment systems, data col lection, model analysis, signal generation, and order execution operate within a unified algorithmic structure. Investment decisions are therefore performed internally within the system without direct human intervention, while responsibility becomes distributed across developers, financial institutions, investors, and regulatory authorities. As a re sult, a structural problem arises in which the attribution of responsibility becomes unclear in times of crisis. Thestudy also analyzes how algorithmic synchronization and feedback loops—which occur when multiple algorithmic trading systems simultaneously respond to shared market signals—can amplify market volatility and expand systemic risk. Such dynam ics provide an important mechanism for explaining sudden market instabilities such as f lash crashes. To address this issue, the paper proposes a Hierarchical Joint Responsibility Model. This model restructures responsibility within algorithmic investment environments into three layers: the design layer, the operational layer, and the authorization layer. By doing so, it presents an institutional framework capable of maintaining both traceability and normative control of responsibility within automated financial systems. This study emphasizes that AI-based financial systems are not merely technologi cal tools but socio-technical structures that transform the very architecture of finan cial judgment. The study ultimately aims to provide philosophical and institutional foundations for redesigning responsibility structures in increasingly automated finan cial markets, while also offering conceptual guidance for regulators and policy makers seeking to address accountability challenges in AI-driven financial systems. (shrink)

The rapid integration of artificial intelligence into financial markets is transform ing the architecture of investment decision-making. As algorithmic trading systems operate at increasing speed, scale, and autonomy, systemic risk may emerge not from irrational human behavior but from synchronized algorithmic rationality. This paper asks a critical question: Can AI trigger the next financial crisis, and if so, who bears responsibility? The study introduces the concept of a structural responsibility gap in AI-mediated investment environments. By distinguishing computational output from normative (...) judgment, it argues that investment decisions inherently involve risk endorsement, value commitment, and accountability beyond probabilistic calculation. When judgment is delegated to autonomous systems, responsibility becomes layered and fragmented across developers, institutions, investors, and regulators. Through analysis of algorithmic feedback loops, synchronization dynamics, and crisis amplification mechanisms—including flash crash events—the paper demon strates how automated coordination may intensify systemic instability. To address this gap, it proposes a Hierarchical Joint Responsibility Model designed to govern delegated judgment without attributing moral agency to AI systems. The stability of algorithmic finance ultimately depends not solely on technological control, but on the structural redesign of accountability. (shrink)

This paper introduces a trajectory level of explanation for inference-time behaviour in large language models. Existing frameworks—autoregressive conditioning, mechanistic circuit analysis, and quasi-cognitive description—treat generation as a sequence of context-conditioned draws or as circuit execution. None provides the vocabulary needed to ask whether exit from a behavioural mode is harder than entry, whether transitions are threshold-mediated or continuous, or whether a model’s path through representational space exhibits the path dependence characteristic of a dynamical system with stable regimes. The paper argues (...) that these questions are not merely unmeasured but unformulable within current frameworks, and that their invisibility is the signature of a genuine explanatory level rather than a gap in current knowledge. Drawing on formal links between transformer attention and attractor dynamics, results on metastability in transformer systems, and empirical evidence of regime-like behaviour in large language models, it proposes the asymmetry coefficient A(M,γ) = Rout (γ)/Rin (γ) as a discriminator between four competing accounts of inference-time behaviour. An experimental protocol is specified, the full result space is mapped, explicit failure conditions are stated, and alignment gating is reframed as trajectory control rather than fixed output policy. (shrink)

Surprisingly little attention has been paid to the nature of conscious control. As a result, experiments suggesting that we lack conscious control over our actions cannot be properly evaluated. Joshua Shepherd (2015; 2021) aims to fill this gap. His proposal is grounded in the standard causalist account of action, according to which, simply put, bodily movements are controlled by the agent if and only if they are caused, in the right way, by the relevant psychological states. In this (...) paper, I argue that the proposal does not succeed in distinguishing between mere causation and actual control; it does not solve the problem of deviant causation. On the basis of my criticism, Anscombean action theory promises to offer a valuable perspective. It suggests that an important function of consciousness is to integrate an action into a rational whole. If this is on the right track, it supports Shepherd’s overall claim that the importance of consciousness for action control is often underestimated, because this contribution of consciousness may often go unnoticed in experimental settings. (shrink)

An involuntary commitment is a rights-restrictive procedure that is rarely afforded visibility through useful data. Pennsylvania is a nationally relevant case study on the promises, perils, and misconceptions surrounding involuntary commitment data. In this study, we survey the minimal available Pennsylvania data and then contextualize our results within state laws and norms by (1) framing involuntary commitment data within the state’s gun control agenda, (2) implementing the language of undone science, agnotology, and visibility, and (3) discussing the rights and (...) social obligations surrounding privacy, confidentiality, and dangerousness. We find that sensitive administrative data is collected at several points during the involuntary commitment process but is rarely available in a useful, de-identified form. Improved access to useful data about the commitment population would close a major gap in state and national healthcare delivery, policy, and accountability. (shrink)

Prior research has documented the continued existence of an expectation gap, defined as the divergence between the public’s and the profession’s conceptions of auditor’s duties, despite the auditing profession’s attempt to adopt standards and practices to close this gap. In this paper, we consider one potential explanation for the persistence of the expectation gap: the role of media bias in shaping public opinion and views. We analyze press articles covering 40 U.S. corporate fraud cases discovered between 1992 and 2011. We (...) compare the auditor’s duties, described by the auditing standards, with the description of the fraud cases as found in the press articles. We draw upon prior research to identify three sources of the expectation gap: deficient performance, deficient standards, and unreasonable expectations. The results of our analysis provide evidence that the performance gap can be reduced by strengthening auditor’s willingness and ability to apply existing auditing standards concerning fraud detection; the standards gap can be narrowed by improving existing auditing standards; and unreasonable expectations, however, involve elements beyond the profession’s sphere of control. As a result, the expectation gap is unlikely to disappear given the media’s tendency to bias, with an overemphasis of unreasonable expectations in their coverage of frauds and press articles tending to reinforce the view that the auditor should take more responsibility for detecting fraud, irrespective of whether this is feasible at a reasonable cost. In addition to the primary role of the press in perpetuating the expectation gap, a second reason for continuation of the expectation gap is that the rational auditor will have difficulty in assessing subjective components of fraudulent behavior. (shrink)