We propose a non-representationalist framework for deep learning relying on a novel method computational phenomenology, a dialogue between the first-person perspective (relying on phenomenology) and the mechanisms of computational models. We thereby propose an alternative to the modern cognitivist interpretation of deep learning, according to which artificial neural networks encode representations of external entities. This interpretation mainly relies on neuro-representationalism, a position that combines a strong ontological commitment towards scientific theoretical entities and the idea that the brain operates on symbolic (...) representations of these entities. We proceed as follows: after offering a review of cognitivism and neuro-representationalism in the field of deep learning, we first elaborate a phenomenological critique of these positions; we then sketch out computational phenomenology and distinguish it from existing alternatives; finally we apply this new method to deep learning models trained on specific tasks, in order to formulate a conceptual framework of deep-learning, that allows one to think of artificial neural networks’ mechanisms in terms of lived experience. (shrink)

In this article, I focus on the role of computer simulations as exploratory strategies. I begin by establishing the non-theory-driven nature of simulations. This refers to their ability to characterize phenomena without relying on a predefined conceptual framework that is provided by an implemented mathematical model. Drawing on Steinle’s notion of exploratory experimentation and Gelfert’s work on exploratory models, I present three exploratory strategies for computer simulations: (1) starting points and continuation of scientific inquiry, (2) varying the parameters, and (3) (...) scientific prototyping. (shrink)

The representations of deep convolutional neural networks (CNNs) are formed from generalizing similarities and abstracting from differences in the manner of the empiricist theory of abstraction (Buckner, Synthese 195:5339–5372, 2018). The empiricist theory of abstraction is well understood to entail infinite regress and circularity in content constitution (Husserl, Logical Investigations. Routledge, 2001). This paper argues these entailments hold a fortiori for deep CNNs. Two theses result: deep CNNs require supplementation by Quine’s “apparatus of identity and quantification” in order to (1) (...) achieve concepts, and (2) represent objects, as opposed to “half-entities” corresponding to similarity amalgams (Quine, Quintessence, Cambridge, 2004, p. 107). Similarity amalgams are also called “approximate meaning[s]” (Marcus & Davis, Rebooting AI, Pantheon, 2019, p. 132). Although Husserl inferred the “complete abandonment of the empiricist theory of abstraction” (a fortiori deep CNNs) due to the infinite regress and circularity arguments examined in this paper, I argue that the statistical learning of deep CNNs may be incorporated into a Fodorian hybrid account that supports Quine’s “sortal predicates, negation, plurals, identity, pronouns, and quantifiers” which are representationally necessary to overcome the regress/circularity in content constitution and achieve objective (as opposed to similarity-subjective) representation (Burge, Origins of Objectivity. Oxford, 2010, p. 238). I base myself initially on Yoshimi’s (New Frontiers in Psychology, 2011) attempt to explain Husserlian phenomenology with neural networks but depart from him due to the arguments and consequently propose a two-system view which converges with Weiskopf’s proposal (“Observational Concepts.” The Conceptual Mind. MIT, 2015. 223–248). (shrink)

Why would we want to develop artificial human-like arithmetical intelligence, when computers already outperform humans in arithmetical calculations? Aside from arithmetic consisting of much more than mere calculations, one suggested reason is that AI research can help us explain the development of human arithmetical cognition. Here I argue that this question needs to be studied already in the context of basic, non-symbolic, numerical cognition. Analyzing recent machine learning research on artificial neural networks, I show how AI studies could potentially shed (...) light on the development of human numerical abilities, from the proto-arithmetical abilities of subitizing and estimating to counting procedures. Although the current results are far from conclusive and much more work is needed, I argue that AI research should be included in the interdisciplinary toolbox when we try to explain the development and character of numerical cognition and arithmetical intelligence. This makes it relevant also for the epistemology of mathematics. (shrink)

Avatars will represent us online, in virtual worlds, and in technologically supported hybrid environments. We and our avatars will stand not in an identity relation but in a proxy relation, an arrangement that is significant not least because our proxies’ actions can be counted as our own. However, this proxy relation between humans and avatars is not well understood and its consequences under-explored. In this paper I explore the relation and its potential ethical consequences.

The counterfactual approach to explainable AI (XAI) seeks to provide understanding of AI systems through the provision of counterfactual explanations. In a recent systematic review, Chou et al. (Inform Fus 81:59–83, 2022) argue that the counterfactual approach does not clearly provide causal understanding. They diagnose the problem in terms of the underlying framework within which the counterfactual approach has been developed. To date, the counterfactual approach has not been developed in concert with the approach for specifying causes developed by Pearl (...) (Causality: Models, reasoning, and inference. Cambridge University Press, 2000) and Woodward (Making things happen: A theory of causal explanation. Oxford University Press, 2003). In this paper, I build on Chou et al.’s work by applying the Pearl-Woodward approach. I argue that the standard counterfactual approach to XAI is capable of delivering causal understanding, but that there are limitations on its capacity to do so. I suggest a way to overcome these limitations. (shrink)

The debate around the notions of a priori knowledge and a posteriori knowledge has proven crucial for the development of many fields in philosophy, such as metaphysics, epistemology, metametaphysics etc. We advocate that the recent debate on the two notions is also fruitful for man-made distributed computing systems and for the epistemic analysis thereof. Following a recently proposed modal and fallibilistic account of a priori knowledge, we elaborate the corresponding concept of a priori belief: We propose a rich taxonomy of (...) types of a priori beliefs and their role for the different agents that participate in the system engineering process, which match the existing view exceedingly well and are particularly promising for explaining and dealing with unexpected behaviors in fault-tolerant distributed systems. Developing such a philosophical foundation will provide a sound basis for eventually implementing our ideas in a suitable epistemic reasoning and analysis framework and, hence, constitutes a mandatory first step for developing methods and tools to cope with the various challenges that emerge in such systems. (shrink)

The US is promoting a new vision of a “Good AI Society” through its recent AI Bill of Rights. This offers a promising vision of community-oriented equity unique amongst peer countries. However, it leaves the door open for potential rights violations. Furthermore, it may have some federal impact, but it is non-binding, and without concrete legislation, the private sector is likely to ignore it.

According to the multiple realization argument, mental states or processes can be realized in diverse and heterogeneous physical systems; and that fact implies that mental state or process kinds cannot be identified with particular kinds of physical states or processes. More specifically, mental processes cannot be identified with brain processes. Moreover, the argument provides a general model for the autonomy of the special sciences. The multiple realization argument is widely influential, but over the last thirty years it has also faced (...) serious objections. Despite those objections, most philosophers regard the fact of multiple realization and the cogency of the multiple realization argument as plainly correct. Why is that? What is it about the multiple realization argument that makes it so resilient? One reason is that the multiple realization argument is deeply intertwined with a view that minds are, in some sense, computational. But we argue that the sense in which minds are computational does not support the conclusion that they are ipso facto multiply realized. We argue that the sense in which brains compute does not imply that brains implement multiply realizable computational processes, and it does not provide a general model for the autonomy of the special sciences. (shrink)

Since the announcement and establishment of the Oversight Board (OB) by the technology company Meta as an independent institution reviewing Facebook and Instagram’s content moderation decisions, the OB has been subjected to scholarly scrutiny ranging from praise to criticism. However, there is currently no overarching framework for understanding the OB’s various strengths and weaknesses. Consequently, this article analyses, organises, and supplements academic literature, news articles, and Meta and OB documents to understand the OB’s strengths and weaknesses and how it can (...) be improved. Significant strengths include its ability to enhance the transparency of content moderation decisions and processes, to effect reform indirectly through policy recommendations, and its assertiveness in interpreting its jurisdiction and overruling Meta. Significant weaknesses include its limited jurisdiction, limited impact, Meta’s control over the OB’s precedent, and its lack of diversity. The analysis of a recent OB case in Ethiopia shows these strengths and weaknesses in practice. The OB’s relationship with Meta and governments will lead to challenges and opportunities shaping its future development. Reforms to the OB should improve the OB’s control over its precedent, apply OB precedent to currently disputed cases, and clarify the standards for invoking OB precedent. Finally, these reforms provide the foundation for an additional improvement to address the OB’s institutional weaknesses, by involving users in determining whether the OB’s precedent should be applied to decide current content moderation disputes. (shrink)

Proposals by the San Francisco Police Department (SFPD) to use bomb disposal robots for deadly force against humans have met with widespread condemnation. Media coverage of the furore has tended, incorrectly, to conflate these robots with autonomous weapon systems (AWS), the AI-based weapons used in armed conflict. These two types of systems should be treated as distinct since they have different sets of social, ethical, and legal implications. However, the conflation does raise a pressing question: what _if_ the SFPD had (...) proposed using AWS for law enforcement purposes? This article argues that current debate on AWS takes place within a ‘killing paradigm’ that leaves us ill-placed to understand the implications of using autonomous force outside of armed conflict. This is because ‘lethality’ is taken as the paradigmatic form of harm, meaning that other harms are understood in a way that is derivative of lethal harm. The article calls for more research on how other sorts of goods, such as freedom from domination, are imperilled by the use of autonomous force. The article also suggests that bringing research on AWS into dialogue with the sorts of concerns that have typically motivated political theory—e.g. power, coercion, the state—provides a fruitful starting point for addressing these issues. (shrink)

According to a mainstream position in contemporary cognitive science and philosophy, the use of abstract compositional concepts is amongst the most characteristic indicators of meaningful deliberative thought in an organism or agent. In this article, we show how the ability to develop and utilise abstract conceptual structures can be achieved by a particular kind of learning agent. More specifically, we provide and motivate a concrete operational definition of what it means for these agents to be in possession of abstract concepts, (...) before presenting an explicit example of a minimal architecture that supports this capability. We then proceed to demonstrate how the existence of abstract conceptual structures can be operationally useful in the process of employing previously acquired knowledge in the face of new experiences, thereby vindicating the natural conjecture that the cognitive functions of abstraction and generalisation are closely related. (shrink)

The Turing test has been studied and run as a controlled experiment and found to be underspecified and poorly designed. On the other hand, it has been defended and still attracts interest as a test for true artificial intelligence (AI). Scientists and philosophers regret the test’s current status, acknowledging that the situation is at odds with the intellectual standards of Turing’s works. This article refers to this as the Turing Test Dilemma, following the observation that the test has been under (...) discussion for over seventy years and still is widely seen as either too bad or too good to be a valuable experiment for AI. An argument that solves the dilemma is presented, which relies on reconstructing the Turing test as a thought experiment in the modern scientific tradition. It is argued that Turing’s exposition of the imitation game satisfies Mach’s characterization of the basic method of thought experiments and that Turing’s uses of his test satisfy Popper’s conception of the critical and heuristic uses of thought experiments and Kuhn’s association of thought experiments to conceptual change. It is emphasized how Turing methodically varied the imitation game design to address specific challenges posed to him by other thinkers and how his test illustrates a property of the phenomenon of intelligence and suggests a hypothesis on machine learning. This reconstruction of the Turing test provides a rapprochement to the conflicting views on its value in the literature. (shrink)

Kripkenstein’s challenge is usually described as being essentially about the use of a word in new kinds of cases ‒ the old kinds of cases being commonly considered as non-problematic. I show that this way of conceiving the challenge is neither true to Kripke’s intentions nor philosophically defensible: the Kripkean skeptic can question my answering “125” to the question “What is 68 plus 57?” even if that problem is one I have already encountered and answered. I then argue that once (...) the real nature of Kripkenstein’s challenge is properly appreciated, one extremely popular strategy to try to meet it, what usually goes by the name of “semantic dispositionalism”, loses much of its appeal. Along the way, I also explain that Kripkenstein’s challenge is actually two distinct challenges ‒ one concerning the mental state of meaning, or intending, something by a sign and the other concerning the meaning (referentially conceived) of linguistic expressions. (shrink)

Predictive processing (PP) and embodied cognition (EC) have emerged as two influential approaches within cognitive science in recent years. Not only have PP and EC been heralded as “revolutions” and “paradigm shifts” but they have motivated a number of new and interesting areas of research. This has prompted some to wonder how compatible the two views might be. This paper looks to weigh in on the issue of PP-EC compatibility. After outlining two recent proposals, I argue that further clarity can (...) be achieved on the issue by considering a model of scientific progress. Specifically, I suggest that Larry Laudan’s “problem solving model” can provide important insights into a number of outstanding challenges that face existing accounts of PP-EC compatibility. I conclude by outlining additional implications of the problem solving model for PP and EC more generally. (shrink)

Enactivism advances an understanding of cognition rooted in the dynamic interaction between an embodied agent and their environment, whilst new mechanism suggests that cognition is explained by uncovering the organised components underlying cognitive capacities. On the face of it, the mechanistic model’s emphasis on localisable and decomposable mechanisms, often neural in nature, runs contrary to the enactivist ethos. Despite appearances, this paper argues that mechanistic explanations of cognition, being neither narrow nor reductive, and compatible with plausible iterations of ideas like (...) emergence and downward causation, are congruent with enactivism. Attention to enactivist ideas, moreover, may serve as a heuristic for mechanistic investigations of cognition. Nevertheless, I show how enactivism and approaches that prioritise mechanistic modelling may diverge in starting assumptions about the nature of cognitive phenomena, such as where the constitutive boundaries of cognition lie. (shrink)

In this paper, we address the question of what current computers are from the point of view of human-computer interaction. In the early days of computing, the Turing machine (TM) has been the cornerstone of the understanding of computers. The TM defines what can be computed and how computation can be carried out. However, in the last decades, computers have evolved and increasingly become interactive systems, reacting in real-time to external events in an ongoing loop. We argue that the TM (...) does not provide a mechanistic explanation for interactive computing. The reason is that the fundamental phenomena relevant to interactive computing are out of the scope of classical computability theory. Part of the explanatory power of the TM relies on what we propose to call an execution model. An execution model belongs to a level of abstraction where it is possible to describe both the functional architecture and the execution in mechanistic terms. An updated execution model is warranted to provide the minimal mechanistic description for interactive computation as a counterpart of what the TM could explain regarding Church-Turing computation. It would support an explanation of the ubiquitous computing devices we know - those interacting with humans, e.g., through digital interfaces. We show that such a model is not available within interactive models of computation and that relevant abstractions and concerns are available in computer engineering but need to be identified and gathered. To fill this void, we propose to reflect on the level of abstraction required to support the mechanistic description of an interactive execution and propose some preliminary requirements. (shrink)

Organisations that design and deploy artificial intelligence (AI) systems increasingly commit themselves to high-level, ethical principles. However, there still exists a gap between principles and practices in AI ethics. One major obstacle organisations face when attempting to operationalise AI Ethics is the lack of a well-defined material scope. Put differently, the question to which systems and processes AI ethics principles ought to apply remains unanswered. Of course, there exists no universally accepted definition of AI, and different systems pose different ethical (...) challenges. Nevertheless, pragmatic problem-solving demands that things should be sorted so that their grouping will promote successful actions for some specific end. In this article, we review and compare previous attempts to classify AI systems for the purpose of implementing AI governance in practice. We find that attempts to classify AI systems proposed in previous literature use one of three mental models: _the Switch_, i.e., a binary approach according to which systems either are or are not considered AI systems depending on their characteristics; _the Ladder_, i.e., a risk-based approach that classifies systems according to the ethical risks they pose; and _the Matrix_, i.e., a multi-dimensional classification of systems that take various aspects into account, such as context, input data, and decision-model. Each of these models for classifying AI systems comes with its own set of strengths and weaknesses. By conceptualising different ways of classifying AI systems into simple mental models, we hope to provide organisations that design, deploy, or regulate AI systems with the vocabulary needed to demarcate the material scope of their AI governance frameworks. (shrink)

It is frequently demanded that AI-based Decision Support Tools (AI-DSTs) ought to be both explainable to, and trusted by, those who use them. The joint pursuit of these two principles is ordinarily believed to be uncontroversial. In fact, a common view is that AI systems should be made explainable so that they can be trusted, and in turn, accepted by decision-makers. However, the moral scope of these two principles extends far beyond this particular instrumental connection. This paper argues that if (...) we were to account for the rich and diverse moral reasons that ground the call for explainable AI, and fully consider what it means to “trust” AI in a descriptively rich sense of the term, we would uncover a deep and persistent tension between the two principles. For explainable AI to usefully serve the pursuit of normatively desirable goals, decision-makers must carefully monitor and critically reflect on the content of an AI-DST’s explanation. This entails a deliberative attitude. Conversely, calls for trust in AI-DSTs imply the disposition to put questions about their reliability out of mind. This entails an unquestioning attitude. As such, the joint pursuit of explainable and trusted AI calls on decision-makers to simultaneously adopt incompatible attitudes towards their AI-DST, which leads to an intractable implementation gap. We analyze this gap and explore its broader implications: suggesting that we may need alternate theoretical conceptualizations of what explainability and trust entail, and/or alternate decision-making arrangements that separate the requirements for trust and deliberation to different parties. (shrink)

Most, if not all, philosophers agree that computers cannot learn what words refers to from raw text alone. While many attacked Searle’s Chinese Room thought experiment, no one seemed to question this most basic assumption. For how can computers learn something that is not in the data? Emily Bender and Alexander Koller ( 2020 ) recently presented a related thought experiment—the so-called Octopus thought experiment, which replaces the rule-based interlocutor of Searle’s thought experiment with a neural language model. The Octopus (...) thought experiment was awarded a best paper prize and was widely debated in the AI community. Again, however, even its fiercest opponents accepted the premise that what a word refers to cannot be induced in the absence of direct supervision. I will argue that what a word refers to _is_ probably learnable from raw text alone. Here’s why: higher-order concept co-occurrence statistics are stable across languages and across modalities, because language use (universally) reflects the world we live in (which is relatively stable). Such statistics are sufficient to establish what words refer to. My conjecture is supported by a literature survey, a thought experiment, and an actual experiment. (shrink)

Over the past few decades, cognitive science has identified several forms of reasoning that make essential use of conceptual knowledge. Despite significant theoretical and empirical progress, there is still no unified framework for understanding how concepts are used in reasoning. This paper argues that the theory of conceptual spaces is capable of filling this gap. Our strategy is to demonstrate how various inference mechanisms which clearly rely on conceptual information—including similarity, typicality, and diagnosticity-based reasoning—can be modeled using principles derived from (...) conceptual spaces. Our first topic analyzes the role of expectations in inductive reasoning and their relation to the structure of our concepts. We examine the relationship between using generic expressions in natural language and common-sense reasoning as a second topic. We propose that the strength of a generic can be described by distances between properties and prototypes in conceptual spaces. Our third topic is category-based induction. We demonstrate that the theory of conceptual spaces can serve as a comprehensive model for this type of reasoning. The final topic is analogy. We review some proposals in this area, present a taxonomy of analogical relations, and show how to model them in terms of distances in conceptual spaces. We also briefly discuss the implications of the model for reasoning with concepts in artificial systems. (shrink)