Cognitive Science Theory Research Articles

Representation in Cognitive Science: Replies

In their constructive reviews, Frances Egan, Randy Gallistel and Steven Gross have raised some important problems for the account of content advanced by Nicholas Shea in Representation in Cognitive Science. Here the author addresses their main challenges. Egan argues that the account includes an unrecognised pragmatic element; and that it makes contents explanatorily otiose. Gallistel raises questions about homomorphism and correlational information. Gross puts the account to work to resolve a dispute about probabilistic contents in perception, but argues that a question remains about whether probabilities are found in the content or instead in the manner of representation.

Content is pragmatic: Comments on Nicholas Shea'sRepresentation in cognitive science

Nicholas Shea offers what he takes to be a naturalistic account of representational content in cognitive science. I argue that the account secures determinate content only by appeal to pragmatic considerations, and so it fails to respect naturalism. But that is fine, because representational content is not, strictly speaking, necessary for explanation in cognitive science. Even in Shea's own account, content serves only a variety of heuristic functions.

Probabilistic representations in perception: Are there any, and what would they be?

Nick Shea's Representation in cognitive science commits him to representations in perceptual processing that are about probabilities. This commentary concerns how to adjudicate between this view and an alternative that locates the probabilities rather in the representational states’ associated “attitudes.” As background and motivation, evidence for probabilistic representations in perceptual processing is adduced, and it is shown how, on either conception, one can address a specific challenge Ned Block has raised to this evidence.

Legal (post)realism and prescriptive legal science: Can legal realism provide normative guidance?

One of the objections against both historical and current legal realism is that it is a merely descriptive approach and that it focuses only on attorneys and parties to a dispute. It is claimed that it provides no guidance for judges or other decision-making authorities on how to proceed in solving legal disputes and in applying law onto the facts. Should the objection be correct, legal realism can either give up on these ambitions completely and leave them for traditional legal theory, or it can instead move to a new, post-realist phase, and try to develop a theory of law which will be able to prescribe to real actors how to behave in a real-life situation. Economy denotes such a theory by the notion of “prescriptive” theory. This undertaking would, nevertheless, require to focus on modern theories of decision-making, behavioral economics and cognitive science, combined with (neo)institutionalist and discursive theories of law.

Improved Visual-Semantic Alignment for Zero-Shot Object Detection

Zero-shot object detection is an emerging research topic that aims to recognize and localize previously ‘unseen’ objects. This setting gives rise to several unique challenges, e.g., highly imbalanced positive vs. negative instance ratio, proper alignment between visual and semantic concepts and the ambiguity between background and unseen classes. Here, we propose an end-to-end deep learning framework underpinned by a novel loss function that handles class-imbalance and seeks to properly align the visual and semantic cues for improved zero-shot learning. We call our objective the ‘Polarity loss’ because it explicitly maximizes the gap between positive and negative predictions. Such a margin maximizing formulation is not only important for visual-semantic alignment but it also resolves the ambiguity between background and unseen objects. Further, the semantic representations of objects are noisy, thus complicating the alignment between visual and semantic domains. To this end, we perform metric learning using a ‘Semantic vocabulary’ of related concepts that refines the noisy semantic embeddings and establishes a better synergy between visual and semantic domains. Our approach is inspired by the embodiment theories in cognitive science, that claim human semantic understanding to be grounded in past experiences (seen objects), related linguistic concepts (word vocabulary) and the visual perception (seen/unseen object images). Our extensive results on MS-COCO and Pascal VOC datasets show significant improvements over state of the art.1

Explainable Reinforcement Learning through a Causal Lens

Prominent theories in cognitive science propose that humans understand and represent the knowledge of the world through causal relationships. In making sense of the world, we build causal models in our mind to encode cause-effect relations of events and use these to explain why new events happen by referring to counterfactuals — things that did not happen. In this paper, we use causal models to derive causal explanations of the behaviour of model-free reinforcement learning agents. We present an approach that learns a structural causal model during reinforcement learning and encodes causal relationships between variables of interest. This model is then used to generate explanations of behaviour based on counterfactual analysis of the causal model. We computationally evaluate the model in 6 domains and measure performance and task prediction accuracy. We report on a study with 120 participants who observe agents playing a real-time strategy game (Starcraft II) and then receive explanations of the agents' behaviour. We investigate: 1) participants' understanding gained by explanations through task prediction; 2) explanation satisfaction and 3) trust. Our results show that causal model explanations perform better on these measures compared to two other baseline explanation models.

« Comment peut-on être cochon ? » : darwinisme et conscience chez Darrieussecq

Les liens de plus en plus étroits entre les sciences cognitives et la théorie darwinienne ont permis de faire des découvertes fascinantes au cours des dernières décennies, souvent adoptées par les sciences humaines avec enthousiasme, et on a vu émerger de cette confluence la naissance de nouvelles disciplines comme la psychologie évolutionniste et la neuro-phénoménologie. L’intérêt de Darrieussecq pour la cognition humaine et animale fait de son oeuvre un site d’exploration de plusieurs questions liées à ces domaines de recherche. Ses personnages s’interrogent sur la vie intérieure et l’univers phénoménologique des chiens, des chats et des insectes. La focalisation narrative de son histoire saute brusquement d’une conscience humaine à celle d’un lion de mer ou d’un requin pèlerin. Notamment, dans le premier roman de Darrieussecq, le lecteur suit la subjectivité de la narratrice à travers sa métamorphose entre être humain et truie, y compris un statut incertain entre les deux, en se rendant compte petit à petit que ce n’est pas que la forme physique de la narratrice qui se transforme mais aussi son univers mental. Cet article démontre que l’exploration de la conscience animale et humaine chez Darrieussecq ne sert pas seulement de métaphore de l’intersubjectivité et de la difficulté à comprendre la conscience de l’autre, mais offre aussi une méditation sur ce que c’est que l'humanité et sur sa place dans un spectre darwinien de cerveaux différemment adaptés.

Multiple conceptions of resource rationality.

Resource rationality holds great promise as a unifying principle across theories in neuroscience, cognitive science, and economics. The target article clearly lays out this potential for unification. However, resource-rational models are more diverse and less easily unified than might appear from the target article. Here, we explore some of that diversity.

Historical Institutionalism the Cognitive Foundations of Cooperation

This essay argues that in order to understand how institutions shape political choices and history we should go further toward understanding the interactive relationships between institutions and the cognitive mind. The article explores the significant body of research and literature developing in social and evolutionary psychology, cognitive science and decision theory. This literature has gone beyond the observation humans are not the individual utility maximizers stylized in early institutionalist theorizing. A massive body of experimental and empirical research clearly demonstrates, for example, that a) individuals only rarely have stable and hierarchical preferences, b) are generally quite unlikely to go through the cognitive effort implied in a ‘rational’ choice decision matrix, c) humans are efficient decision makers, but we have remarkable capacities to post-hoc justify choices and restructure our memories so that we can view our decisions as coherent. In the end the paper argues that public administration scholars need to adopt a non-reductionist approach to studying human motivation.

Representation in Cognitive Science, by Nicholas Shea. Oxford: Oxford University Press, 2018. Pp. 292

<i>Representation in Cognitive Science</i>, by Nicholas Shea. Oxford: Oxford University Press, 2018. Pp. 292

生成认知：理论基础与实践走向

<p id="C2">There are different theoretical views in the embodied cognition camp, and there is sometimes even conflict among them. Enactive cognition is a new version of recently developed embodied cognition approaches. It claims that the explanation of human cognitive processes across the board should not make any appeal to internal representational or computational states. According to the traditional computation-representation theory of mind, cognitive processes are mechanistically realized in computational processes of building, storing and manipulating detached and abstract internal representations. This cognitivist paradigm has dominated mainstream cognitive psychology for decades. The key assumptions that characterize this representation- centered theory of mind are including that cognition might be understood as computation over mental representations, and that models of cognition should take into account only the inner states of a cognitive individual. But recently this framework has been challenged and criticized. Out of this challenge and criticism emerged the beginnings to an enactive cognition paradigm. The enactive approach to cognitive science proposed a new set of theoretical assumptions for understanding what cognition is and how it works that aims to break the tight conceptual connection between cognition and representation. It takes as its starting point that cognition must not be understood as a capacity for getting an internal representation of a corresponding external reality, which in turn would provide a foundation for supporting thinking, learning, and problem solving. Instead, cognitive processes are deeply entangled in action. Cognition is thus best understood as “enactive”; that is, as a form of practice itself. From the point view of enactive cognition, cognition comes from bodily action and serves bodily action, that is, cognition is embodied action. The key postulate of enactive Cognition is action-related and action-oriented, with the capacity to generate environmental structure by action. Cognition is thus best understood as “enactive”; that is, as embedded action it comes from bodily action and serves bodily action. At the same time, advocates of enactivism state that despite the emphasis on the function of the organism’s action of the mind, the model cannot be equated with behaviorism; action, in contrast to behavior, is purposeful and has a cognitive component. Key aspects of enactive cognition are: (1) perception consists in perceptually guided action and (2) cognitive structures emerge from the recurrent sensorimotor patterns that allow action to be perceptually guided. There are three enactive theories in cognitive science. Autopoietic enactivism emphasizes the deep continuity between life and mind; sensorimotor enactivism focuses on analyzing perceptual consciousness in terms of sensorimotor contingency, and radical enactivism focuses on rejecting representationalism in favour of explanatory strategies that emphasize patterns of embodied interaction. These perspectives on enactive cognition are more informed by phenomenology and pragmatism than were earlier versions of embodied cognition. Therefore, we can say that phenomenology and pragmatism constitute the theoretical origin of enactive cognition. The enactive cognition view is not only theoretically viable, but also supported by substantial experimental evidence demonstrating that cognitive processes can be reinterpreted using this new conceptual framework. The theoretical premises of enactive cognition open up new prospects for improving theoretical research and the practical application of cognitive science in the future.

Lyra: Simulating Believable Opinionated Virtual Characters

Creating believable simulations of large populations of characters in virtual worlds represents a grand challenge for interactive artificial intelligence, requiring reasoning about social intelligence. In this paper, we focus on one aspect of this challenge: the dynamics of opinion change for virtual characters and its relationship with social affinity. We developed a simulated population of characters that debate politically-charged topics, called Lyra. Characters’ knowledge, opinions, and biases spread through this society based on existing cognitive models and social science theories. Our simulation generates outlines of group conversations that portray the system’s evolution, and clusters characters into affinity groups based on the outcome of the debates. We conducted a human-subjects study to evaluate these generated conversations and affinity groups for their believability and to inform future iterations of the simulation. We believe successful simulation of opinion change in social dynamics provides a foundation for computational recognition, prediction, and interfacing with humans.

Review to Daniel Hutto and Erik Myin’s Evolving Enactivism: Basic Minds Meet Content. MIT Press, 2017.

In this review, Hutto and Myin’s new book “Evolving Enactivism: Basic Minds Meet Content” (2017) is critically presented. Although they do not provide a detailed cognitive science theory based on their Radical Enactive approach, one may say that Hutto and Myin originally address the perennial philosophical issue about our nature as human beings giving an impossible-to-neglect enactivist contribution to the current state-of-art in the discussion concerning embodied cognition.&#x0D;

The Rhetoric and Reality of Anthropomorphism in Artificial Intelligence

Artificial intelligence (AI) 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.

Choice Set Formation in Residential Mobility and Its Implications for Segregation Dynamics.

We develop and estimate a statistical model of neighborhood choice that draws on insights from cognitive science and decision theory as well as qualitative studies of housing search. The model allows for a sequential decision process and the possibility that people consider a small and selective subset of all potential destinations. When combined with data from the Los Angeles Family and Neighborhood Survey, our model reveals that affordability constraints and households' tendency toward short-distance moves lead blacks and Hispanics to have racially stratified choice sets in which their own group is disproportionately represented. We use an agent-based model to assess how racially stratified choice sets contribute to segregation outcomes. Our results show that cognitive decision strategies can amplify patterns of segregation and inequality.

On savvy reading

ABSTRACT This essay uses the metaphor of “savvy travelling” to discuss the limitations and problems associated with three current “best practice” approaches to the reading of literature in upper grade levels, particularly in the United States: close reading, response-based reading, and disciplinary literacy. A “savvy” approach to reading and reading instruction combines close attention to the text, readers’ intuitive responses, and concepts from cognitive science, cultural studies, and literary theory in a narrative approach that identifies and compares the story in the text to the story of the text. Two examples from classroom practice are presented and analyzed to identify elements of savoir faire in each. Identification of these elements leads to a method of savvy reading whose goal is to produce students who will become savvy world travellers of texts on their own: readers who are autonomous, flexible, and able to read carefully and responsibly across a wide range of contexts.

The significance of shape in cartographic communication

The significance of shape in cartographic communication

Ecological Representations

Representations feature heavily in cognitive science theories about our behavioral repertoire. Their critical feature is its ability to designate (stand in for) spatially or temporally distant properties, so that organizing our behavior with respect to mental/neural representations means organizing our behavior with respect to the otherwise unavailable property they designate. Representations are a powerful tool, but serious problems (grounding, system-detectable error) remain unsolved. Ecological explanations reject representations. However, this has left us without a straightforward vocabulary to engage with “representation-hungry” problems involving spatial or temporal distance, nor the role of the nervous system in cognition. To develop such a vocabulary, here we show that ecological information functions to designate the ecologically scaled dynamical world to an organism. We then show that this designation analysis of information leads to an ecological conceptualization of the neural activity caused by information, and finally we argue these together can support intentional behavior with respect to spatially and temporally distal properties. The way they do so (via designation) does mean information and the related neural activity can be conceptualized as representations; but they do so in a grounded way that remains true to key ecological ontological commitments. We advocate this path for expanding the ecological approach.

Radical views on cognition and the dynamics of scientific change

Radical views on cognition are generally defined by a cluster of features including non-representationalism and vehicle-externalism. In this paper, I concentrate on the way radical views on cognition define themselves as revolutionary theories in cognitive science. These theories often use the Kuhnian concepts of “paradigm” and “paradigm shift” for describing their ambitions and the current situation in cognitive science. I examine whether the use of Kuhn’s theory of science is appropriate here. There might be good reasons to think that cognitive science is in a situation of foundational crisis, but that does not entail that the classical paradigm (computationalism) is currently displaced to the benefit of a new paradigm. Larry Laudan’s theory of research traditions is more enlightening than Kuhn’s theory for describing the scope and ambitions of radical views on cognition, and their relations with an anti-intellectualist tradition in philosophy.

Possible Selves, Body Schemas, and Sādhana: Using Cognitive Science and Neuroscience in the Study of Medieval Vaiṣṇava Sahajiyā Hindu Tantric Texts

In recent decades, historians of religions have turned to, and developed, entirely new methodologies for the study of religion and human consciousness. Foremost among these are a collection of approaches often termed the “cognitive science of religion” (CSR), typically drawing on cognitive science, neuroscience, linguistics, and contemporary metaphor theory. Although we are still “early” in this enterprise, I hope to show how a meaningful dialogue between religious studies and contemporary neuroscience and cognitive science can help us to better understand some intriguing mystical texts and practices from a tradition of medieval South Asian Hinduism. Known collectively as the Vaiṣṇava Sahajiyās, these followers of transgressive and antinomian Tantric Yoga provide excellent examples for exploring the nature of religion, ritual, consciousness, embodiment, identity, gender, emotions and sexuality. This paper will show how the study of these rich materials from seventeenth through eighteenth century Bengal in northeastern South Asia can be enhanced using insights from the philosopher Shaun Gallagher and the neurologist Patrick McNamara.