We are sceptical of concerns over the opacity of algorithmic decision tools. While transparency and explainability are certainly important desiderata in algorithmic governance, we worry that automated decision-making is being held to an unrealistically high standard, possibly owing to an unrealistically high estimate of the degree of transparency attainable from human decision-makers. In this paper, we review evidence demonstrating that much human decision-making is fraught with transparency problems, show in what respects AI fares little worse or better and argue that (...) at least some regulatory proposals for explainable AI could end up setting the bar higher than is necessary or indeed helpful. The demands of practical reason require the justification of action to be pitched at the level of practical reason. Decision tools that support or supplant practical reasoning should not be expected to aim higher than this. We cast this desideratum in terms of Daniel Dennett’s theory of the “intentional stance” and argue that since the justification of action for human purposes takes the form of intentional stance explanation, the justification of algorithmic decisions should take the same form. In practice, this means that the sorts of explanations for algorithmic decisions that are analogous to intentional stance explanations should be preferred over ones that aim at the architectural innards of a decision tool. (shrink)

Many of the computing systems programmed using Machine Learning are opaque: it is difficult to know why they do what they do or how they work. Explainable Artificial Intelligence aims to develop analytic techniques that render opaque computing systems transparent, but lacks a normative framework with which to evaluate these techniques’ explanatory successes. The aim of the present discussion is to develop such a framework, paying particular attention to different stakeholders’ distinct explanatory requirements. Building on an analysis of “opacity” from (...) philosophy of science, this framework is modeled after accounts of explanation in cognitive science. The framework distinguishes between the explanation-seeking questions that are likely to be asked by different stakeholders, and specifies the general ways in which these questions should be answered so as to allow these stakeholders to perform their roles in the Machine Learning ecosystem. By applying the normative framework to recently developed techniques such as input heatmapping, feature-detector visualization, and diagnostic classification, it is possible to determine whether and to what extent techniques from Explainable Artificial Intelligence can be used to render opaque computing systems transparent and, thus, whether they can be used to solve the Black Box Problem. (shrink)

Artificial intelligence has historically been conceptualized in anthropomorphic terms. Some algorithms deploy biomimetic designs in a deliberate attempt to effect a sort of digital isomorphism of the human brain. Others leverage more general learning strategies that happen to coincide with popular theories of cognitive science and social epistemology. In this paper, I challenge the anthropomorphic credentials of the neural network algorithm, whose similarities to human cognition I argue are vastly overstated and narrowly construed. I submit that three alternative supervised learning (...) methods—namely lasso penalties, bagging, and boosting—offer subtler, more interesting analogies to human reasoning as both an individual and a social phenomenon. Despite the temptation to fall back on anthropomorphic tropes when discussing AI, however, I conclude that such rhetoric is at best misleading and at worst downright dangerous. The impulse to humanize algorithms is an obstacle to properly conceptualizing the ethical challenges posed by emerging technologies. (shrink)

ABSTRACTWhile over the last few decades AI research has largely focused on building tools and applications, recent technological developments have prompted a resurgence of interest in building a genuinely intelligent artificial agent – one that has a mind in the same sense that humans and animals do. In this paper, I offer a theoretical and methodological framework for this project of investigating “artificial minded intelligence” that can help to unify existing approaches and provide new avenues for research. I first outline (...) three desiderata that a framework for AMI research should satisfy. In Section 1, I further motivate the desiderata as well as the need for a new framework. Section 2 develops a general methodological approach, the “generative methodology,” and Section 3 develops a version of this methodology, the “Competence Framework for AI research”. (shrink)

In artificial intelligence, recent research has demonstrated the remarkable potential of Deep Convolutional Neural Networks (DCNNs), which seem to exceed state-of-the-art performance in new domains weekly, especially on the sorts of very difficult perceptual discrimination tasks that skeptics thought would remain beyond the reach of artificial intelligence. However, it has proven difficult to explain why DCNNs perform so well. In philosophy of mind, empiricists have long suggested that complex cognition is based on information derived from sensory experience, often appealing to (...) a faculty of abstraction. Rationalists have frequently complained, however, that empiricists never adequately explained how this faculty of abstraction actually works. In this paper, I tie these two questions together, to the mutual benefit of both disciplines. I argue that the architectural features that distinguish DCNNs from earlier neural networks allow them to implement a form of hierarchical processing that I call “transformational abstraction”. Transformational abstraction iteratively converts sensory-based representations of category exemplars into new formats that are increasingly tolerant to “nuisance variation” in input. Reflecting upon the way that DCNNs leverage a combination of linear and non-linear processing to efficiently accomplish this feat allows us to understand how the brain is capable of bi-directional travel between exemplars and abstractions, addressing longstanding problems in empiricist philosophy of mind. I end by considering the prospects for future research on DCNNs, arguing that rather than simply implementing 80s connectionism with more brute-force computation, transformational abstraction counts as a qualitatively distinct form of processing ripe with philosophical and psychological significance, because it is significantly better suited to depict the generic mechanism responsible for this important kind of psychological processing in the brain. (shrink)

The present century has seen renewed interest in characterizing cognition, the object of inquiry of the cognitive sciences. In this paper, I describe the problem of cognition—the absence of a positive characterization of cognition despite a felt need for one. It is widely recognized that the problem is motivated by decades of controversy among cognitive scientists over foundational questions, such as whether non-neural parts of the body or environment can realize cognitive processes, or whether plants and microbes have cognitive processes. (...) The dominant strategy for addressing the problem of cognition is to seek a dichotomous criterion that vindicates some set of controversial claims. However, I argue that the problem of cognition is also motivated by ongoing conceptual development in cognitive science, and I describe four benefits that a characterization of cognition could confer. Given these benefits, I recommend an alternative criterion of success, ecumenical extensional adequacy, on which the aim is to describe the variation in expert judgments rather than to correct this variation by taking sides in sectarian disputes. I argue that if we had an ecumenical solution to the problem of cognition, we would have achieved much of what we should want from a “mark of the cognitive”. (shrink)

In various contexts and for various reasons, writers often define cognitive processes and architectures as those involving representational states and structures. Similarly, cognitive theories are also often delineated as those that invoke representations. In this paper, I present several reasons for rejecting this way of demarcating the cognitive. Some of the reasons against defining cognition in representational terms are that doing so needlessly restricts our theorizing, it undermines the empirical status of the representational theory of mind, and it encourages wildly (...) deflationary and explanatorily vacuous conceptions of representation. After criticizing this outlook, I sketch alternative ways we might try to capture what is distinctive about cognition and cognitive theorizing. (shrink)

How should we determine the distribution of psychological traits—such as Theory of Mind, episodic memory, and metacognition—throughout the Animal kingdom? Researchers have long worried about the distorting effects of anthropomorphic bias on this comparative project. A purported corrective against this bias was offered as a cornerstone of comparative psychology by C. Lloyd Morgan in his famous “Canon”. Also dangerous, however, is a distinct bias that loads the deck against animal mentality: our tendency to tie the competence criteria for cognitive capacities (...) to an exaggerated sense of typical human performance. I dub this error “anthropofabulation”, since it combines anthropocentrism with confabulation about our own prowess. Anthropofabulation has long distorted the debate about animal minds, but it is a bias that has been little discussed and against which the Canon provides no protection. Luckily, there is a venerable corrective against anthropofabulation: a principle offered long ago by David Hume, which I call “Hume’s Dictum”. In this paper, I argue that Hume’s Dictum deserves a privileged place next to Morgan’s Canon in the methodology of comparative psychology, illustrating my point through a discussion of the debate over Theory of Mind in nonhuman animals. (shrink)

A set of hypotheses is formulated for a connectionist approach to cognitive modeling. These hypotheses are shown to be incompatible with the hypotheses underlying traditional cognitive models. The connectionist models considered are massively parallel numerical computational systems that are a kind of continuous dynamical system. The numerical variables in the system correspond semantically to fine-grained features below the level of the concepts consciously used to describe the task domain. The level of analysis is intermediate between those of symbolic cognitive models (...) and neural models. The explanations of behavior provided are like those traditional in the physical sciences, unlike the explanations provided by symbolic models. (shrink)

Artificial intelligence research has foundered on the issue of representation. When intelligence is approached in an incremental manner, with strict reliance on interfacing to the real world through perception and action, reliance on representation disappears. In this paper we outline our approach to incrementally building complete intelligent Creatures. The fundamental decomposition of the intelligent system is not into independent information processing units which must interface with each other via representations. Instead, the intelligent system is decomposed into independent and parallel activity (...) producers which all interface directly to the world through perception and action, rather than interface to each other particularly much. The notions of central and peripheral systems evaporateeverything is both central and peripheral. Based on these principles we have built a very successful series of mobile robots which operate without supervision as Creatures in standard office environments. (shrink)

This paper explores the difference between Connectionist proposals for cognitive a r c h i t e c t u r e a n d t h e s o r t s o f m o d e l s t hat have traditionally been assum e d i n c o g n i t i v e s c i e n c e . W e c l a i m t h a t t h (...) e m a j o r d i s t i n c t i o n i s t h a t , w h i l e b o t h Connectionist and Classical architectures postulate representational mental states, the latter but not the former are committed to a symbol-level of representation, or to a ‘language of thought’: i.e., to representational states that have combinatorial syntactic and semantic structure. Several arguments for combinatorial structure in mental representations are then reviewed. These include arguments based on the ‘systematicity’ of mental representation: i.e., on the fact that cognitive capacities always exhibit certain symmetries, so that the ability to entertain a given thought implies the ability to entertain thoughts with semantically related contents. We claim that such arguments make a powerful case that mind/brain architecture is not Connectionist at the cognitive level. We then consider the possibility that Connectionism may provide an account of the neural (or ‘abstract neurological’) structures in which Classical cognitive architecture is implemented. We survey a n u m b e r o f t h e s t a n d a r d a r g u m e n t s t h a t h a v e b e e n o f f e r e d i n f a v o r o f Connectionism, and conclude that they are coherent only on this interpretation. (shrink)

Let psychologism be the doctrine that whether behavior is intelligent behavior depends on the character of the internal information processing that produces it. More specifically, I mean psychologism to involve the doctrine that two systems could have actual and potential behavior _typical_ of familiar intelligent beings, that the two systems could be exactly alike in their actual and potential behavior, and in their behavioral dispositions and capacities and counterfactual behavioral properties (i.e., what behaviors, behavioral dispositions, and behavioral capacities they would (...) have exhibited had their stimuli differed)--the two systems could be alike in all these ways, yet there could be a difference in the information processing that mediates their stimuli and responses that determines that one is not at all intelligent while the other is fully intelligent. (shrink)

A surge of empirical research demonstrating flexible cognition in animals and young infants has raised interest in the possibility of rational decision‐making in the absence of language. A venerable position, which I here call “Classical Inferentialism”, holds that nonlinguistic agents are incapable of rational inferences. Against this position, I defend a model of nonlinguistic inferences that shows how they could be practically rational. This model vindicates the Lockean idea that we can intuitively grasp rational connections between thoughts by developing the (...) Davidsonian idea that practical inferences are at bottom categorization judgments. From this perspective, we can see how similarity‐based categorization processes widely studied in human and animal psychology might count as practically rational. The solution involves a novel hybrid of internalism and externalism: intuitive inferences are psychologically rational (in the explanatory sense) given the intensional sensitivity of the similarity assessment to the internal structure of the agent's reasons for acting, but epistemically rational (in the justificatory sense) given an ecological fit between the features matched by that assessment and the structure of the agent's environment. The essay concludes by exploring empirical results that show how nonlinguistic agents can be sensitive to these similarity assessments in a way that grants them control over their opaque judgments. (shrink)

A surge of empirical research demonstrating flexible cognition in animals and young infants has raised interest in the possibility of rational decision-making in the absence of language. A venerable position, which I here call “Classical Inferentialism”, holds that nonlinguistic agents are incapable of rational inferences. Against this position, I defend a model of nonlinguistic inferences that shows how they could be practically rational. This model vindicates the Lockean idea that we can intuitively grasp rational connections between thoughts by developing the (...) Davidsonian idea that practical inferences are at bottom categorization judgments. From this perspective, we can see how similarity-based categorization processes widely studied in human and animal psychology might count as practically rational. The solution involves a novel hybrid of internalism and externalism: intuitive inferences are psychologically rational (in the explanatory sense) given the intensional sensitivity of the similarity assessment to the internal structure of the agent's reasons for acting, but epistemically rational (in the justificatory sense) given an ecological fit between the features matched by that assessment and the structure of the agent's environment. The essay concludes by exploring empirical results that show how nonlinguistic agents can be sensitive to these similarity assessments in a way that grants them control over their opaque judgments. (shrink)

Ethicists have commonly appealed to science to bolster their arguments for elevating the moral status of nonhuman animals. I describe a framework within which I take many ethicists to be making such appeals. I focus on an apparent gap in this framework between those properties of animals that are part of the scientific consensus, and those to which ethicists typically appeal in their arguments. I will describe two different ways of diminishing the appearance of the gap, and argue that both (...) of them present challenges to ethicists seeking a firm scientific basis for their claims about the moral status of animals. I argue that more clarity about the role of appeals to science by applied ethicists leads to questions about the effectiveness of such appeals, and that these questions might best be pursued empirically. (shrink)

The legacy of Nisbett and Wilson’s classic article, Telling More Than We Can Know: Verbal Reports on Mental Processes , is mixed. It is perhaps the most cited article in the recent history of consciousness studies, yet no empirical research program currently exists that continues the work presented in the article. To remedy this, we have introduced an experimental paradigm we call choice blindness [Johansson, P., Hall, L., Sikström, S., & Olsson, A. . Failure to detect mismatches between intention and (...) outcome in a simple decision task. Science, 310, 116–119.]. In the choice blindness paradigm participants fail to notice mismatches between their intended choice and the outcome they are presented with, while nevertheless offering introspectively derived reasons for why they chose the way they did. In this article, we use word-frequency and latent semantic analysis to investigate a corpus of introspective reports collected within the choice blindness paradigm. We contrast the introspective reasons given in non-manipulated vs. manipulated trials, but find very few differences between these two groups of reports. (shrink)

There is a vast literature within philosophy of mind that focuses on artificial intelligence, but hardly mentions methodological questions. There is also a growing body of work in philosophy of science about modeling methodology that hardly mentions examples from cognitive science. Here these discussions are connected. Insights developed in the philosophy of science literature about the importance of idealization provide a way of understanding the neural implausibility of connectionist networks. Insights from neurocognitive science illuminate how relevant similarities between models and (...) targets are picked out, how modeling inferences are justified, and the metaphysical status of models. (shrink)

Should cognitive scientists be any more embarrassed about their lack of a discipline-fixing definition of cognition than biologists are about their inability to define “life”? My answer is “no”. Philosophers seeking a unique “mark of the cognitive” or less onerous but nevertheless categorical characterizations of cognition are working at a level of analysis upon which hangs nothing that either cognitive scientists or philosophers of cognitive science should care about. In contrast, I advocate a pluralistic stance towards uses of the term (...) ‘cognition’ that eschews the urge to treat cognition as a metaphysically well-defined “natural” kind. (shrink)

For many years the evolution of language has been seen as a disreputable topic, mired in fanciful “just so stories” about language origins. However, in the last decade a new synthesis of modern linguistics, cognitive neuroscience and neo-Darwinian evolutionary theory has begun to make important contributions to our understanding of the biology and evolution of language. I review some of this recent progress, focusing on the value of the comparative method, which uses data from animal species to draw inferences about (...) language evolution. Discussing speech first, I show how data concerning a wide variety of species, from monkeys to birds, can increase our understanding of the anatomical and neural mechanisms underlying human spoken language, and how bird and whale song provide insights into the ultimate evolutionary function of language. I discuss the “descended larynx” of humans, a peculiar adaptation for speech that has received much attention in the past, which despite earlier claims is not uniquely human. Then I will turn to the neural mechanisms underlying spoken language, pointing out the difficulties animals apparently experience in perceiving hierarchical structure in sounds, and stressing the importance of vocal imitation in the evolution of a spoken language. Turning to ultimate function, I suggest that communication among kin (especially between parents and offspring) played a crucial but neglected role in driving language evolution. Finally, I briefly discuss phylogeny, discussing hypotheses that offer plausible routes to human language from a non-linguistic chimp-like ancestor. I conclude that comparative data from living animals will be key to developing a richer, more interdisciplinary understanding of our most distinctively human trait: language. (shrink)

This paper introduces a special issue of Cognitive Science initiated on the 25th anniversary of the publication of Parallel Distributed Processing (PDP), a two-volume work that introduced the use of neural network models as vehicles for understanding cognition. The collection surveys the core commitments of the PDP framework, the key issues the framework has addressed, and the debates the framework has spawned, and presents viewpoints on the current status of these issues. The articles focus on both historical roots and contemporary (...) developments in learning, optimality theory, perception, memory, language, conceptual knowledge, cognitive control, and consciousness. Here we consider the approach more generally, reviewing the original motivations, the resulting framework, and the central tenets of the underlying theory. We then evaluate the impact of PDP both on the field at large and within specific subdomains of cognitive science and consider the current role of PDP models within the broader landscape of contemporary theoretical frameworks in cognitive science. Looking to the future, we consider the implications for cognitive science of the recent success of machine learning systems called “deep networks”—systems that build on key ideas presented in the PDP volumes. (shrink)

Fodor and Pylyshyn (1988), Fodor and McLaughlin (1990) and McLaughlin (1993) challenge connectionists to explain systematicity without simply implementing a classical architecture. In this paper I argue that what makes the challenge difficult for connectionists to meet has less to do with what is to be explained than with what is to count as an explanation. Fodor et al. are prepared to admit as explanatory, accounts of a sort that only classical models can provide. If connectionists are to meet the (...) challenge, they are going to have to insist on the propriety of changing what counts as an explanation of systematicity. Once that is done, there would seem to be as yet no reason to suppose that connectionists are unable to explain systematicity. (shrink)

Two new books—Creating Language: Integrating Evolution, Acquisition, and Processing by Morten H. Christiansen and Nick Chater, and Why Only Us: Language and Evolution by Robert C. Berwick and Noam Chomsky—present a good opportunity to assess the state of the debate about whether or not language was made possible by language-specific adaptations for syntax. Berwick and Chomsky argue yes: language was made possible by a single change to the computation Merge. Christiansen and Chater argue no: our syntactic abilities developed on the (...) back of natural selection for general-purpose sequence learning mechanisms. While Christiansen and Chater’s book testifies to impressive developments in constructivist approaches to language development, it’s not obvious that it has the resources to explain the hierarchical nature of syntactic binding. Despite this, the views have much in common. (shrink)

The charge that anthropomorphizing nonhuman animals is a fallacy is itself largely misguided and mythic. Anthropomorphism in the study of animal behavior is placed in its original, theological context. Having set the historical stage, I then discuss its relationship to a number of other, related issues: the role of anecdotal evidence, the taxonomy of related anthropomorphic claims, its relationship to the attribution of psychological states in general, and the nature of the charge of anthropomorphism as a categorical claim. I then (...) argue that the categorical reading of anthropomorphism cannot work and that it misrepresents what is being claimed when one claims that traits are shared between humans and nonhumans. We should think of such claims not as anthropomorphic per se– because that implies the trait is intrinsically human and only derivatively nonhuman. Instead, traits shared with mammals are mammalomorphic, for example, or primatomorphic when shared by primates. (shrink)

The goal of the paper is to develop and propose a general model of the state space of AI. Given the breathtaking progress in AI research and technologies in recent years, such conceptual work is of substantial theoretical interest. The present AI hype is mainly driven by the triumph of deep learning neural networks. As the distinguishing feature of such networks is the ability to self-learn, self-learning is identified as one important dimension of the AI state space. Another dimension is (...) recognized as generalization, the possibility to go over from specific to more general types of problems. A third dimension is semantic grounding. Our overall analysis connects to a number of known foundational issues in the philosophy of mind and cognition: the blockhead objection, the Turing test, the symbol grounding problem, the Chinese room argument, and use theories of meaning. It shall finally be argued that the dimension of grounding decomposes into three sub-dimensions. And the dimension of self-learning turns out as only one of a whole range of “self-x-capacities” (based on ideas of organic computing) that span the self-x-subspace of the full AI state space. (shrink)

In ‘Computing Machinery and Intelligence’, Turing, sceptical of the question ‘Can machines think?’, quickly replaces it with an experimentally verifiable test: the imitation game. I suggest that for such a move to be successful the test needs to be relevant, expansive, solvable by exemplars, unpredictable, and lead to actionable research. The Imitation Game is only partially successful in this regard and its reliance on language, whilst insightful for partially solving the problem, has put AI progress on the wrong foot, prescribing (...) a top-down approach for building thinking machines. I argue that to fix shortcomings with modern AI systems a nonverbal operationalisation is required. This is provided by the recent Animal-AI Testbed, which translates animal cognition tests for AI and provides a bottom-up research pathway for building thinking machines that create predictive models of their environment from sensory input. (shrink)