One of the best known arguments against the connectionist approach to artificial intelligence and cognitive science is that neural networks are black boxes, i.e., there is no understandable account of their operation. This difficulty has impeded efforts to explain how categories arise from raw sensory data. Moreover, it has complicated investigation about the role of symbols and language in cognition. This state of things has been radically changed by recent experimental findings in artificial deep learning research. Two kinds of artificial (...) deep learning networks, namely the convolutional neural network and the generative adversarial network have been found to possess the capability to build internal states that are interpreted by humans as complex visual categories, without any specific hints or any grammatical processing. This emergent ability suggests that those categories do not depend on human knowledge or the syntactic structure of language, while they do rely on their visual context. This supports a mild form of empiricism, while it does not assume that computational functionalism is true. Some consequences are extracted regarding the debate about amodal and grounded representations in the human brain. Furthermore, new avenues for research on cognitive science are open. (shrink)

As AI systems become increasingly complex it may become unclear, even to the designer of a system, why exactly a system does what it does. This leads to a lack of trust in AI systems. To solve this, the field of explainable AI has been working on ways to produce explanations of these systems’ behaviors. Many methods in explainable AI, such as LIME, offer only a statistical argument for the validity of their explanations. However, some methods instead study the internal (...) structure of the system and try to find components which can be assigned an interpretation. I believe that these methods provide more valuable explanations than those statistical in nature. I will try to identify which explanations can be considered internal to the system using the Chomskyan notion of tacit knowledge. I argue that each explanation expresses a rule, and through the localization of this rule in the system internals, we can take a system to have tacit knowledge of the rule. I conclude that the only methods which are able to sufficiently establish this tacit knowledge are those along the lines of Olah : 4901–4911, 2017), and therefore they provide explanations with unique strengths. (shrink)

Beyond modularityattempts a synthesis of Fodor's anticonstructivist nativism and Piaget's antinativist constructivism. Contra Fodor, I argue that: (1) the study of cognitive development is essential to cognitive science, (2) the module/central processing dichotomy is too rigid, and (3) the mind does not begin with prespecified modules; rather, development involves a gradual process of “modularization.” Contra Piaget, I argue that: (1) development rarely involves stagelike domain-general change and (2) domainspecific predispositions give development a small but significant kickstart by focusing the infant's (...) attention on proprietary inputs. Development does not stop at efficient learning. A fundamental aspect of human development (“representational redescription”) is the hypothesized process by which information that isina cognitive system becomes progressively explicit knowledgetothat system. Development thus involves two complementary processes of progressive modularization and progressive “explicitation.” Empirical findings on the child as linguist, physicist, mathematician, psychologist, and notator are discussed in support of the theoretical framework. Each chapter concentrates first on the initial state of the infant mind/brain and on subsequent domain-specific learning in infancy and early childhood. It then goes on to explore data on older children's problem solving and theory building, with particular focus on evolving cognitive flexibility. Emphasis is placed throughout on the status of representations underlying different capacities and on the multiple levels at which knowledge is stored and accessible. Finally, consideration is given to the need for more formal developmental models, and a comparison is made between representational redescription and connectionist simulations of development. In conclusion, I consider what is special about human cognition by speculating on the status of representations underlying the structure of behavior in other species. (shrink)

The paper examines an alleged distinction claimed to exist by Van Gelder between two different, but equally acceptable ways of accounting for the systematicity of cognitive output (two “varieties of compositionality”): “concatenative compositionality” vs. “functional compositionality.” The second is supposed to provide an explanation alternative to the Language of Thought Hypothesis. I contend that, if the definition of “concatenative compositionality” is taken in a different way from the official one given by Van Gelder (but one suggested by some of his (...) formulations) then there is indeed a different sort of compositionality; however, the second variety is not an alternative to the language of thought in that case. On the other hand, if the concept of concatenative compositionality is taken in a different way, along the lines of Van Gelder's explicit definition, then there is no reason to think that there is an alternative way of explaining systematicity. (shrink)

Critical notice of Andy Clark's "Associative Engines".

This piece is a commentary on a precis of Maggie Boden's book "The creative mind" published in Behavioral and Brain Sciences.

Language has long been a problem‐case for subsymbolic theories of mind. The reason for this is obvious: Language seems essentially symbolic. However, recent work has developed a potential solution to this problem, arguing that linguistic symbols are public objects which augment a fundamentally subsymbolic mind, rather than components of cognitive symbol‐processing. I shall argue that this strategy cannot work, on the grounds that human language acquisition consists in projecting linguistic structure onto environmental entities, rather than extracting this structure from them.

This bold and brilliant book asks the ultimate question of the life sciences: How did the human mind acquire its incomparable power? In seeking the answer, Merlin Donald traces the evolution of human culture and cognition from primitive apes to the era of artificial intelligence, and presents an original theory of how the human mind evolved from its presymbolic form. In the emergence of modern human culture, Donald proposes, there were three radical transitions. During the first, our bipedal but still (...) apelike ancestors acquired "mimetic" skill—the ability to represent knowledge through voluntary motor acts—which made Homo erectus successful for over a million years. The second transition—to "mythic" culture —coincided with the development of spoken language. Speech allowed the large-brained Homo sapiens to evolve a complex preliterate culture that survives in many parts of the world today. In the third transition, when humans constructed elaborate symbolic systems ranging from cuneiforms, hieroglyphics, and ideograms to alphabetic languages and mathematics, human biological memory became an inadequate vehicle for storing and processing our collective knowledge. The modern mind is thus a hybrid structure built from vestiges of earlier biological stages as well as new external symbolic memory devices that have radically altered its organization. According to Donald, we are symbol-using creatures, more complex than any that went before us, and we may have not yet witnessed the final modular arrangement of the human mind. (shrink)

How does language (spoken or written) impact thought? One useful way to approach this important but elusive question may be to consider language itself as a cognition-enhancing animal-built structure. To take this perspective is to view language as a kind of self-constructed cognitive niche. These self-constructed cognitive niches play, I suggest, three distinct but deeply interlocking roles in human thought and reason. Working together, these three interlocking routines radically transform the human mind, and mark a genuine discontinuity in the space (...) of anitnal minds. (shrink)

This entry provides an overview of theories of concepts that is organized around five philosophical issues: (1) the ontology of concepts, (2) the structure of concepts, (3) empiricism and nativism about concepts, (4) concepts and natural language, and (5) concepts and conceptual analysis.

This thesis focuses on models of conceptual change in science and philosophy. In particular, I developed a new bootstrapping methodology for studying conceptual change, centered around the formalization of several popular models of conceptual change and the collective assessment of their improved formal versions via nine evaluative dimensions. Among the models of conceptual change treated in the thesis are Carnap’s explication, Lakatos’ concept-stretching, Toulmin’s conceptual populations, Waismann’s open texture, Mark Wilson’s patches and facades, Sneed’s structuralism, and Paul Thagard’s conceptual revolutions. (...) -/- In order to analyze and compare the conception of conceptual change provided by these different models, I rely on several historical reconstructions of episodes of scientific conceptual change. The historical episodes of scientific change that figure in this work include the emergence of the morphological concept of fish in biological taxonomies, the development of scientific conceptions of temperature, the Church-Turing thesis and related axiomatizations of effective calculability, the history of the concept of polyhedron in 17th and 18th century mathematics, Hamilton’s invention of the quaternions, the history of the pre-abstract group concepts in 18th and 19th century mathematics, the expansion of Newtonian mechanics to viscous fluids forces phenomena, and the chemical revolution. I will also present five different formal and informal improvements of four specific models of conceptual change. I will first present two different improvements of Carnapian explication, a formal and an informal one. My informal improvement of Carnapian explication will consist of a more fine-grained version of the procedure that adds an intermediate, third step to the two steps of Carnapian explication. I will show how this novel three-step version of explication is more suitable than its traditional two-step relative to handle complex cases of explications. My second, formal improvement of Carnapian explication will be a full explication of the concept of explication itself within the theory of conceptual spaces. By virtue of this formal improvement, the whole procedure of explication together with its application procedures and its pragmatic desiderata will be reconceptualized as a precise procedure involving topological and geometrical constraints inside the theory of conceptual spaces. My third improved model of conceptual change will consist of a formal explication of Darwinian models of conceptual change that will make vast use of Godfrey-Smith’s population-based Darwinism for targeting explicitly mathematical conceptual change. My fourth improvement will be dedicated instead to Wilson’s indeterminate model of conceptual change. I will show how Wilson’s very informal framework can be explicated within a modified version of the structuralist model-theoretic reconstructions of scientific theories. Finally, the fifth improved model of conceptual change will be a belief-revision-like logical framework that reconstructs Thagard’s model of conceptual revolution as specific revision and contraction operations that work on conceptual structures. -/- At the end of this work, a general conception of conceptual change in science and philosophy emerges, thanks to the combined action of the three layers of my methodology. This conception takes conceptual change to be a multi-faceted phenomenon centered around the dynamics of groups of concepts. According to this conception, concepts are best reconstructed as plastic and inter-subjective entities equipped with a non-trivial internal structure and subject to a certain degree of localized holism. Furthermore, conceptual dynamics can be judged from a weakly normative perspective, bound to be dependent on shared values and goals. Conceptual change is then best understood, according to this conception, as a ubiquitous phenomenon underlying all of our intellectual activities, from science to ordinary linguistic practices. As such, conceptual change does not pose any particular problem to value-laden notions of scientific progress, objectivity, and realism. At the same time, this conception prompts all our concept-driven intellectual activities, including philosophical and metaphilosophical reflections, to take into serious consideration the phenomenon of conceptual change. An important consequence of this conception, and of the analysis that generated it, is in fact that an adequate understanding of the dynamics of philosophical concepts is a prerequisite for analytic philosophy to develop a realistic and non-idealized depiction of itself and its activities. (shrink)

In psychology, increasing interest in priming has brought with it a revival of associationist views. Association seems a natural explanation for priming: simple associative links carry subcritical levels of activation from representations of the prime stimulus to representations of the target stimulus. This then facilitates use of the representation of the target. I argue that the processes responsible for priming are not associative. They are more complex. Even so, associative models do get something right about how these processes behave. As (...) a result, I argue, we should reconsider how we interpret associative models, taking them to identify regularities in the sequence of representational states in any kind of process, rather than as denoting a particular kind of process. (shrink)

The computational paradigm, which has dominated psychology and artificial intelligence since the cognitive revolution, has been a source of intense debate. Recently, several cognitive scientists have argued against this paradigm, not by objecting to computation, but rather by objecting to the notion of representation. Our analysis of these objections reveals that it is not the notion of representation per se that is causing the problem, but rather specific properties of representations as they are used in various psychological theories. Our analysis (...) suggests that all theorists accept the idea that cognitive processing involves internal information-carrying states that mediate cognitive processing. These mediating states are a superordinate category of representations. We discuss five properties that can be added to mediating states and examine their importance in various cognitive models. Finally, three methodological lessons are drawn from our analysis and discussion. (shrink)

Philosophical interest in intentional action has flourished in recent decades. Typically, action theorists propose necessary and sufficient conditions for a movement's being an action, conditions derived from a conceptual analysis of folk psychological action ascriptions. However, several key doctrinal and methodological features of contemporary action theory are troubling, in particular (i) the insistence that folk psychological kinds like beliefs and desires have neurophysiological correlates, (ii) the assumption that the concept of action is "classical" in structure (making it amenable to definition (...) in terms of necessary and sufficient conditions for its proper application), and (iii) the assumption that deferring to intuitions about the application of the concept of action amidst the context of fantastical thought experiments furnishes an effective method for judging the adequacy of proposed analyses. After consideration of these problems it is argued that action theory needs to be reoriented in a more naturalistic direction, the methods and aims of which are continuous with those of the empirical sciences. The paper concludes with a sketch (and defense) of the methodological foundations of a naturalistic approach to intentional action. (shrink)