John Dorsch

In the context of AI decision support systems (AI-DSS), we argue that meeting the demands of ethical and explainable AI (XAI) is about developing AI-DSS to provide human decision-makers with three types of human-grounded explanations: reasons, counterfactuals, and confidence, an approach we refer to as the RCC approach. We begin by reviewing current empirical XAI literature that investigates the relationship between various methods for generating model explanations (e.g., LIME, SHAP, Anchors), the perceived trustworthiness of the model, and end-user accuracy. We demonstrate how current theories about what constitutes good human-grounded reasons either do not adequately explain this evidence or do not offer sound ethical advice for development. Thus, we offer a novel theory of human-machine interaction: the theory of epistemic quasi-partnerships (EQP). Finally, we motivate adopting EQP and demonstrate how it explains the empirical evidence, offers sound ethical advice, and entails adopting the RCC approach.

“Know thyself.” Once carved into the stone at Delphi and popularized by Socrates as the guiding maxim of philosophy, this command captures the enduring human struggle to examine and justify our own beliefs. This book asks how such a capacity for self-knowledge could have arisen—our ability to reflect on and regulate what we believe in light of epistemic norms like justification and truth. To answer that question, it weaves together insights from comparative and developmental psychology, cognitive and affective neuroscience, and socio-cultural theories of cognition, including the 4E approach, which holds that cognition is extended across tools and environments, embodied in the biological and affective processes of the body, embedded within socio-cultural contexts, and enacted through active engagement with the world. Thus, at the heart of this inquiry lies the concept of 4E metacognition: the idea that self-knowledge originates in bodily mechanisms that generate emotional and dynamic feedback signals, which drive information-seeking behavior and manifest in outwardly observable expressions, scaffolding epistemic regulation within social communities. Yet the story of self-knowledge is not only about origins; it also has a future. Toward its conclusion, this book turns to how self-knowledge might evolve in our digital age. Technologies such as large language models and decision-support systems are not merely tools but infrastructures for cognition—new extensions of epistemic agency that shape how we reason, justify, and believe. This book explores whether such technologies will erode or enhance our capacity for justificatory reasoning, and asks a provocative question: if the conditions that give rise to epistemic agency in humans can be identified and replicated, could an artificial agent—such as a socially embedded robot—one day possess something akin to self-knowledge? In tracing both the evolutionary roots and the technological horizons of self-knowledge, this book offers a new account of what it means to “know thyself” in a world where the boundaries of cognition are rapidly shifting.

This study explores whether labeling AI as either “trustworthy” or “reliable” influences user perceptions and acceptance of automotive AI technologies. Utilizing a one-way between-subjects design, the research presented online participants (N = 478) with a text presenting guidelines for either trustworthy or reliable AI, before asking them to evaluate 3 vignette scenarios and fill in a modified version of the Technology Acceptance Model which covers different variables, such as perceived ease of use, human-like trust, and overall attitude. While labeling AI as “trustworthy” did not significantly influence people’s judgements on specific scenarios, it increased perceived ease of use and human-like trust, namely benevolence, suggesting a facilitating influence on usability and an anthropomorphic effect on user perceptions. The study provides insights into how specific labels affect adopting certain perceptions of AI technology.

This paper presents a systematic review of explainable reinforcement learning methodologies with an emphasis on human-centered evaluation frameworks. Drawing from literature between 2017 and 2025, we apply and extend the Reasons, Confidence, and Counterfactuals (RCC) framework—originally designed for supervised learning—to reinforcement learning contexts. Our analysis reveals two predominant explanatory strategies: constructive, where explicit explanations are generated, and supportive, where users must infer reasoning from provided visual or textual cues. Our review also emphasizes human factor considerations, like task complexity, explanation formats, and evaluation methodologies. Particularly, for the latter, our analysis shows that improvement of the quality of decision is rarely measured.

How did humans come to know themselves—not just to have thoughts, but to reflect on those thoughts and ask whether they’re true or justified? And as our world slowly integrates a new kind of intelligence—non-human, non-animal, and increasingly entangled with our thinking—what might this mean for the future of self-knowledge? What is the relationship between artificial intelligence and our capacity to justify what we believe? Are AI-curated information environments enhancing this capacity—or hijacking it? Could machines themselves ever acquire self-knowledge one day, making it no longer a uniquely human achievement? This book explores the origins and future of self-knowledge, arguing that our ability to evaluate beliefs in light of justification and truth depends on embodied feelings—like confidence and doubt—that guide how we monitor our thinking. Far from being private or internal, these feelings are shaped by social interactions and play a key role in how we learn to reason, justify, and take responsibility for our beliefs. At the heart of this story is the idea that self-knowledge isn’t something we’re born with, nor something we achieve in isolation. It emerges through social scaffolding: the shared practices, expectations, and forms of accountability through which communities cultivate epistemic agency. Drawing from neuroscience, comparative and developmental psychology, philosophy of mind, and the ethics of technology, the book traces how these processes make self-knowledge possible—and where they might be leading us next: toward a more intellectually virtuous or ignorant society, or perhaps even toward a world in which machines come to know themselves.

This chapter examines possible ramifications of mindshaping a social robot. It explores how such an agent might learn to represent psychological states, align its behavior with evolving societal norms, and develop capacities for self-directed mindreading and normative self-knowledge. Integrating perspectives from cultural evolution and naturalized intentionality, this approach suggests that social robots could achieve a level of norm-based self-regulation typically reserved for humans, fulfilling criteria for moral and legal personhood. However, this possibility raises ethical concerns: creating a self-knowing agent would tax care-giving resources as we would need to provide AI welfare, thus undermining our capacity to act responsibly toward humans, non-human animals, and the environment, to whom our moral consideration is already owed and in desperate need. Thus, this chapter concludes by urging caution, warning that attempts to cultivate moral responsibility in artificial agents may have destabilizing consequences for moral practices.

The first premise of Schellenberg’s particularity argument reads, “If a subject S perceives a particular α, then S discriminates and singles out α” (2018: 25). But this is false if seeing a ganzfeld is possible (i.e., a homogeneous field without any particulars to discriminate). In response, Schellenberg argues that seeing a ganzfeld is impossible by appealing to the ganzfeld effect (viz. hallucinatory experiences caused by ganzfeld exposure) exclusively as a ‘sense of blindness’. I present two challenges for this line of reasoning. The first concerns the need to account for the long delay before blindness is reported (if it is reported at all) and the second concerns the need to be consistent with potential causes of ganzfeld hallucinations (in general, not exclusive to the sense of blindness). I show how Schellenberg’s particularity argument can meet the first challenge but struggles to meet the second. I present evidence based on considerations related to sensory deprivation and that follow from the predictive processing theory of cognition, which describes seeing a ganzfeld as the cause of ganzfeld hallucinations. If correct, this means that you can see a homogeneous field, and thus we ought to be skeptical of the soundness of Schellenberg’s particularity argument. I conclude with a possible strategy for rehabilitating the particularity argument that reconceptualizes perceptual experience as fundamentally multimodal, constituted by both exteroceptive and interoceptive sensation.

Many policies and ethical guidelines recommend developing “trustworthy AI”. We argue that developing morally trustworthy AI is not only unethical, as it promotes trust in an entity that cannot be trustworthy, but it is also unnecessary for optimal calibration. Instead, we show that reliability, exclusive of moral trust, entails the appropriate normative constraints that enable optimal calibration and mitigate the vulnerability that arises in high-stakes hybrid decision-making environments, without also demanding, as moral trust would, the anthropomorphization of AI and thus epistemically dubious behavior. The normative demands of reliability for inter-agential action are argued to be met by an analogue to procedural metacognitive competence (i.e., the ability to evaluate the quality of one’s own informational states to regulate subsequent action). Drawing on recent empirical findings that suggest providing reliability scores (e.g., F1-scores) to human decision-makers improves calibration in the AI system, we argue that reliability scores provide a good index of competence and enable humans to determine how much they wish to rely on the system.

One routinely undergoes a noetic feeling (also called “metacognitive feeling” or “epistemic feeling”), the so-called “feeling of knowing”, whenever trying to recall a person’s name. One feels the name is known despite being unable to recall it. Other experiences also fall under this category, e.g., the tip-of-the-tongue experience, the feeling of confidence. A distinguishing characteristic of noetic feelings is how they are crucially related to the facts we know, so much so that the activation of semantic memory can easily result in the production of noetic feelings – a regularity that memory research has often exploited. And yet little is known about the mechanism that produces noetic feelings. Is it solely brain-based or does it depend upon the extracerebral body for its production of feelings? To arrive at an answer, various studies in metamemory research will be analyzed to determine what ought to be made of the mechanism responsible for noetic feelings. I argue that evidence suggests that it relies upon extracerebral processes, in particular cardiovascular processes, the result being support for an embodied view of metacognition.

The aim of this project to expose the reasons behind the pandemic of misinformation (henceforth, PofM) by examining the enabling conditions of epistemic agency and the emerging technologies that threaten it. I plan to research the emotional origin of epistemic agency, i.e. on the origin of our capacity to acquire justification for belief, as well as on the significance this emotional origin has for our lives as epistemic agents in our so-called Misinformation Age. This project has three objectives. First, I plan to expose the degree to which epistemic agency is made possible by an under-researched species of emotion called epistemic feelings. Perhaps, the most epistemically significant is the feeling of confidence. In particular, epistemic feelings make epistemic agency possible by making errors in reasoning salient or the potential lack thereof. Second, in order to diagnose the reasons for PofM, I will analyze the emotional basis of epistemic agency in the context of emerging technologies. Epistemic feelings ought to be construed as motivators of epistemic acts, specifically acts following exposure to misinformation spread by social media. For example, a recent study found that subjects on YouTube have a 6.3% probability in five clicks to go from watching innocuous videos to misogynistic videos. Accordingly, one prominent view holds that social media algorithms lead subjects down the proverbial rabbit hole, producing content that elicits strong emotional reactions. But under-researched is the degree to which these emotions are intertwined with epistemic feelings, so that subjects are likely to misjudge the information as correct due to being primed to do so by processing the AI's suggested media. Thus, when this technology is used in epistemic practices, the result is invariably epistemically bad behavior because of how this manner of producing content motivates epistemic acts through the production of erroneous feelings of confidence, which result from the perceived ease of cognitive processing: the content is easy to process and so, the subject confidently misjudges, it is true. As of yet, an account of this cognitive and emotional decision-making process has not been used to analyze how emerging technologies are threatening epistemic agency, and thus our wellbeing as knowers. Finally, I plan to develop therapeutics for PofM by outlining how to reform our collective epistemic practices. Thus, this project complements existing research in virtue epistemology, specifically by articulating the challenges to cultivating epistemic virtues and warding off the vices of the mind. In summation, the project aims to account for how emerging technologies are committing a form of epistemic injustice through the exploitation of cognitive biases and the production of erroneous epistemic feelings, with the overarching goal being a framework for convalescing from the pandemic of misinformation plaguing our age.

The aim of the present work is to make a plausible case for the phylogenetic origin of self-knowledge, one which is compatible with a prevalent view about its ontogenetic origin, the social-scaffolding view. Essentially, the phylogenetic origin is generally argued to be evaluative metacognition, i.e. a system of cognitive control mechanisms, while the ontogenetic origin is generally argued to be mindreading, i.e. cognitive capacities supporting mental state attribution. So put simply, the present work aims to provide a plausible solution to the problem of how to establish metacognition as a significant precursor to mindreading. Previous attempts to establish metacognition as an evolutionary precursor to mindreading have struggled to account for the exact relevance that metacognition has for the origin of self-knowledge. This present work analyses this struggle as first having to solve ‘the problem of triviality’, explain how metacognitive mechanisms are more sophisticated than mechanisms enabling basic cognitive feats. Once this problem is solved, however, the problem develops into having to show that the added layers of sophistication warrant construing metacognition as a significant precursor to self-knowledge. This latter problem is analysed as the ultimate problem, referred to here as simply ‘the problem of the origin of self-knowledge’. Thus, the present work seeks to offer a plausible solution to both of these problems. The solution to these problems is argued to be embodied metacognition, cognitive control mechanisms that enable conscious self-control of information-seeking behavior through the production of strongly embodied feedback signals. These feedback signals, often called ‘noetic feelings’, are theorized to motivate, demotivate, and guide cognitive acts to be more accurate. Here it is argued that these signals are strongly embodied, characterizing conscious emotional states constituted by psychological imperatives, states with external manifestations in the form of bodily expressions that can be detected, monitored, and regulated by individuals in the subject’s socio-cultural environment. To argue for this, I augment evaluative metacognition in light of evidence from recent interoception-based research in metacognition which reveals the degree to which noetic feelings are strongly embodied. They are thus argued to be cerebrally generated feedback signals produced by cognitive comparators that become intricately intertwined with bodily processes. Recent evidence suggests these signals come to fulfill their role in cognition through the conscious perception (the interoception) of internal bodily changes, specifically cardiovascular changes, and hence the subtitle of the present work, “How We Feel Our Hearts to Know Our Minds”. Reconceptualizing evaluative metacognition as embodied metacognition provides us with the tools for solving the two problems above. Concerning the first, mechanisms enabling conscious self-control of cognition are clearly more sophisticated than mechanisms that enable basic feats of cognition, such as reinforcement learning. Moreover, appealing to strongly embodied noetic feelings offers the possibility of solving the first problem in such a manner that the second problem is also solved. As conscious expressions of emotions, noetic feelings demand robust information-seeking behaviours and manifest feelings of confidence and uncertainty externally, thus they are well-suited to link metacognition to the social-scaffolding view about the origin of self-knowledge.

On the basis of Descartes’s account of the passions of the soul, we argue that current interoception-based theories of emotions cannot account for the hallmark of a passion of the soul, i.e., that its effects are felt as being in the soul itself. We also pay attention to the epistemic functions of the passions and to Descartes’s category of emotions that are caused and occur in the soul alone. Certain passions of the soul and certain internal emotions are similar to what are today called ‘epistemic feelings’ and ‘epistemic emotions.’ Descartes’s work reflects another challenge for contemporary embodied cognition: how might epistemic affect be embodied? Since the signature of embodiment is increasingly understood as interoceptive, the challenge to interoceptive research is demonstrating the degree to which affect results from interoception. This challenge also implies that the locus of emotional experience is taken into account.

Upon first hearing sinewaves, all that can be discerned are beeps and whistles. But after hearing the original speech, the beeps and whistles sound like speech. The difference between these two episodes undoubtedly involves an alteration in phenomenal character. O’Callaghan (2011) argues that this alteration is non-sensory, but he leaves open the possibility of attributing it to some other source, e.g. cognition. I discuss whether the alteration in phenomenal character involved in sinewave speech provides evidence for cognitive phenomenology. I defend both the existence of cognitive phenomenology and the phenomenal contrast method, as each concerns the case presented here.

Elijah Chudnoff’s case for irreducible cognitive phenomenology hinges on seeming to see the truth of a mathematical proposition (Chudnoff 2015). In the following, I develop an augmented version of Chudnoff’s case, not based on seeming to see, or intuition, but based on being in a state with presentational phenomenology of high-level content. In contrast to other cases for cognitive phenomenology, those based on Strawson’s case (Strawson 2011), I argue that the case presented here is able to withstand counterarguments, which attempt to reduce cognitive phenomenology to sensory phenomenology. To support my argument, I present findings from Bowden and Jung-Beeman’s experiments with the Aha! Experience (Bowden & Jung-Beeman 2004), and argue that the Aha! Experience is a species of the experience of understanding presented here. I interpret the results of these experiments to provide further evidence for irreducible cognitive phenomenology.