Ioan Muntean

K. Stanford (2006, 2009) has offered an antirealist argument (the “problem of unconceived alternatives”, PUA) based on the argument that scientists are not able to grasp alternatives to a current scientific theory T. According to PUA, the mere existence of some epistemically inaccessible alternatives (T’, T”, …) weakens our trust in T and shakes the foundations of mainstream scientific realism. The realist may entertain the inkling that inter-theoretical relations (both existing and expected or ‘hoped-for’) play a role in accepting or rejecting PUA. The most celebrated intertheoretical relations, such as consilience, reduction, realization, emergence, equivalence, or approximation—whether prospective, expected, or realized—bear relevance to the conceivability of their alternatives. This paper presents an ‘eliminative inference’ based on an ‘unification posit’ that weakens the PUA. We employ first a minimal model of inter-theoretical unification couched in terms of the ‘term identification’ of the theoretical terms of two initially different theories, T1 and T2. Then we rethink unification as an ‘ideological identification’ where predicates in different theories are identified. Finally, we can envision a more sophisticated unification as entailment relations among T1 and T2 and their empirical grounds. In all these cases, we propose scenarios of inconceivability based on a minimal consistency requirement run against the “syntactic view” of scientific theories. The upshot of this mechanism is that some alternatives to T1, which remain unconceived within the conceptual and ideological space of T1, can be eliminated because they are inconsistent with empirical constraints on T2. The overall space of ‘serious’ alternatives to both theories is ‘thinned.’ Consistency is a requirement that conditions inter-theoretical relations, mainly when the overlapping evidence supports theories. This argument illustrates in what sense PUA is lessened when scientists or scientific communities operate based on theoretical posits.

The current project focuses on models of “artificial moral learning” (as a type of moral cognition) and “moral coherentism.” It clarifies how artificial moral agency sheds light on some meta-ethical questions in the coherentism framework (Brink, Dorsey, Lynch, Sayre-McCord). In the current approach, data of artificial moral cognition is divided into two subspaces (representing facts and values, respectively) and contains complex, mixed machine-learnable patterns. Inspired by Lynch's “moral concordance,” some schematic models of this type of two-dimensional data are proposed and assessed. The last, more comprehensive model is premised on the theoretical concept of “distributed concordance” over a population of artificial moral agents. The paper concludes that coherentism, when generalized to machine ethics and artificial moral learning, has some advantages over foundationalist or reliabilist approaches in meta-ethics.

Trust is defined as a belief of a human H (‘the trustor’) about the ability of an agent A (the ‘trustee’) to perform future action(s). We adopt here dispositionalism and internalism about trust: H trusts A iff A has some internal dispositions as competences. The dispositional competences of A are high-level metacognitive requirements, in the line of a naturalized virtue epistemology. (Sosa, Carter) We advance a Bayesian model of two (i) confidence in the decision and (ii) model uncertainty. To trust A, H demands A to be self-assertive about confidence and able to self-correct its own models. In the Bayesian approach trust can be applied not only to humans, but to artificial agents (e.g. Machine Learning algorithms). We explain the advantage the metacognitive trust when compared to mainstream approaches and how it relates to virtue epistemology. The metacognitive ethics of trust is swiftly discussed.

This paper proposes a model of the Artificial Autonomous Moral Agent (AAMA), discusses a standard of moral cognition for AAMA, and compares it with other models of artificial normative agency. It is argued here that artificial morality is possible within the framework of a “moral dispositional functionalism.” This AAMA is able to “read” the behavior of human actors, available as collected data, and to categorize their moral behavior based on moral patterns herein. The present model is based on several analogies among artificial cognition, human cognition, and moral action. It is premised on the idea that moral agents should not be based on rule-following procedures, but on learning patterns from data. This idea is rarely implemented in AAMA models, albeit it has been suggested in the machine ethics literature (W. Wallach, C. Allen, J. Gips and especially M. Guarini). As an agent-based model, this AAMA constitutes an alternative to the mainstream action-centric models proposed by K. Abney, M. Anderson and S. Anderson, R. Arkin, T. Powers, W. Wallach, i.a. Moral learning and moral development of dispositional traits play here the fundamental role in cognition. By using a combination of neural networks and evolutionary computation, called “soft computing” (H. Adeli, N. Siddique, S. Mitra, L. Zadeh), the present model reaches a certain level of autonomy and complexity, which illustrates well “moral particularism” and a form of virtue ethics for machines, grounded in active learning. An example based on the “lifeboat metaphor” (G. Hardin) and the extension of this model to the NEAT architecture (K. Stanley, R. Miikkulainen, i.a.) are briefly assessed.

Some philosophers of science have suggested that contemporary science should be the source of inspiration to the new analytic metaphysics (A. Chakra vartty, C. Callender, S. French, J. Ladyman, T. Maudlin, etc.). This paper explores the prospect of a string metaphysics: a research program in analytic metaphysics based on string theory. Different forms of fundamentalism and pluralism are discussed in this context. The paper focus on string metaphysics with S-dualities (a relation between models of string theory at different coupling regimes) and argues that fundamentality and compositionality have to be reconceptualized. String metaphysics with dualities is better couched in terms of metaphysical pluralism. Grounding, as well as a sketch of a string modality, are briefly discussed. The paper concludes with a suggestion for future work: the metaphysician may find a productive ground for research in discussing other dualities (especially the T-dualities, or the AdS/CFT duality), the emergence of spacetime, the concept of time in string theory, the multiverse etc.

In this chapter I discuss the attempts by Theodor Kaluza [Kaluza, T., 1921. Zum Unitätproblem der Physik. Sitzungsber. der K. Ak. der Wiss. zu Berlin, 966–972] and by Oskar Klein [Klein, O., 1926a. Quantentheorie und fünfdimensionale Relativitätstheorie. Zeitschrift für Physik 37 (12), 895–906; Klein, O., 1926b. The atomicity of electricity as a quantum theory law. Nature 118, 516], respectively, to unify electromagnetism and general relativity within a five-dimensional Riemannian manifold. I critically compare Kaluza's results to Klein's. Klein's theory possesses more explanatory power and unificatory strength and uses less types of brute facts than Kaluza's. The characteristic feature of Klein's theory is that it relies on an extrinsic element of unification, i.e. the wavefunction behavior, which is not intrinsic to EM or GR. Finally, I compare and discuss Kaluza's and Klein's theories in the context of Tim Maudlin's [Maudlin, T., 1996. On the unification of physics. Journal of Philosophy 93 (3), 129–144] ranking of unification and I clarify in what sense they constitute counterexamples to some of Margaret Morrison's [Morrison, M., 2000. Unifying Scientific Theories: Physical Concepts and Mathematical Structures. Cambridge University Press] assertions about unification.

This paper proposes a model for an artificial autonomous moral agent (AAMA), which is parsimonious in its ontology and minimal in its ethical assumptions. Starting from a set of moral data, this AAMA is able to learn and develop a form of moral competency. It resembles an “optimizing predictive mind,” which uses moral data (describing typical behavior of humans) and a set of dispositional traits to learn how to classify different actions (given a given background knowledge) as morally right, wrong, or neutral. When confronted with a new situation, this AAMA is supposedly able to predict a behavior consistent with the training set. This paper argues that a promising computational tool that fits our model is “neuroevolution,” i.e. evolving artificial neural networks.

We argue that the concepts of mechanism and autonomy appear to be antagonistic when autonomy is conflated with agency. Once these concepts are disentangled, it becomes clearer how autonomy emerges from complex forms of control. Subsequently, current biomimetic strategies tend to focus on homeostatic regulatory systems; we propose that research in AI and robotics would do well to incorporate biomimetic strategies that instead invoke models of allostatic mechanisms as a way of understanding how to enhance autonomy in artificial systems.