Barresi & Moore's account has at least two implications for moral psychology. First, it appears to provide support for cognitive theories of moral competence. Second, their claim that the development of social understanding depends upondomain-generalchanges in cognitive ability appears to oppose the idea that moral competence is under-pinned by a moral module.

This chapter examines the scope and limits of the tractability argument. It argues for two claims. First, that when explored with appropriate care and attention, it becomes clear that the argument provides no good reason to prefer massive modularity to the more traditional rationalist alternative. Second, while it is denied that tractability considerations support massive modularity per se, this does not mean that they show nothing whatsoever. Careful analysis of tractability considerations suggests a range of characteristics that any plausible version (...) of psychological rationalism is likely to possess. The chapter proceeds as follows: Section 1 outlines and clarifies the general form of the tractability argument. Section 2 explains how massive modularity is supposed to resolve intractability worries. Sections 3 to 7 highlight the deficiencies of the main extant arguments for claiming that nonmodular mechanisms are intractable. Section 8 concludes by sketching some of the general characteristics that a plausible rationalist alternative to massive modularity — one capable of subserving tractable cognitive processes — is likely to possess. (shrink)

This paper reviews the recent history of a subset of research in dynamical cognitive science, in particular that subset that allies itself with the sciences of complexity and casts cognitive systems as interaction dominant, noncomputational, and nonmodular. I look at this history in the light of C.S. Peirce's understanding of scientific reasoning as progressing from abduction to deduction to induction. In particular, I examine the development of a controversy concerning the use of the interaction dominance of human cognitive systems (...) as an explanation of the ubiquitous 1/f noise, multifractality, and complexity matching in human behavior. (shrink)

This chapter examines the scope and limits of the tractability argument. It argues for two claims. First, that when explored with appropriate care and attention, it becomes clear that the argument provides no good reason to prefer massive modularity to the more traditional rationalist alternative. Second, while it is denied that tractability considerations support massive modularity per se, this does not mean that they show nothing whatsoever. Careful analysis of tractability considerations suggests a range of characteristics that any plausible version (...) of psychological rationalism is likely to possess. The chapter proceeds as follows: Section 1 outlines and clarifies the general form of the tractability argument. Section 2 explains how massive modularity is supposed to resolve intractability worries. Sections 3 to 7 highlight the deficiencies of the main extant arguments for claiming that nonmodular mechanisms are intractable. Section 8 concludes by sketching some of the general characteristics that a plausible rationalist alternative to massive modularity — one capable of subserving tractable cognitive processes — is likely to possess. (shrink)

Many explanations in molecular biology, neuroscience, and other fields of experimental biology describe mechanisms underlying phenomena of interest. These mechanistic explanations account for higher-level phenomena in terms of causally active parts and their spatiotemporal organization. What makes such a mechanistic description explanatory? The best-developed answer, Craver's causal-mechanical account, has several weaknesses. It does not fully explicate the target of explanation, interlevel relation, or interactive nonmodular character of many biological mechanisms as we understand them. An alternative account of MEx, emphasizing (...) interdependence among a mechanism's components, remedies these difficulties. (shrink)

Modules are widely held to play a central role in explaining mental development and in accounts of the mind generally. But there is much disagreement about what modules are, which shows that we do not adequately understand modularity. This paper outlines a Fodoresque approach to understanding one type of modularity. It suggests that we can distinguish modular from nonmodular cognition by reference to the kinds of process involved, and that modular cognition differs from nonmodular forms of cognition in (...) being a special kind of computational process. The paper concludes by considering implications for the role of modules in explaining mental development. (shrink)

We integrate information from distinct domains, especially in metaphor. What sort of cognitive architecture underlies this kind of integration? Fodor (1983) argued that it involves nonmodular mechanisms. He also contended that the nonmodular mechanisms evolved from modular ones through a process of demodularization, a position elaborated by Mithen (1996). In this article, I defend Fodor and Mithen from criticisms offered by Sperber (1994). Sperber suggested that nonmodular mechanisms are unlikely to have evolved because an increasingly large database (...) would incapacitate the cognitive system due to combinatorial explosion and the frame problem. Moreover, he argued that nonmodular mechanisms are not required to explain our ability to integrate information. I argue that proponents of demodularization can respond to Sperber's criticisms by assuming that either an expanding database co-evolved with increasingly sophisticated search mechanisms or the cognitive system evolved along connectionist lines. Moreover, I argue that Sperber's modular account of conceptual integration fails to satisfy the criteria for modularity-namely, information encapsulation. (shrink)

Downey and Lempp 1215–1240) have shown that the contiguous computably enumerable degrees, i.e. the c.e. Turing degrees containing only one c.e. weak truth-table degree, can be characterized by a local distributivity property. Here we extend their result by showing that a c.e. degree a is noncontiguous if and only if there is an embedding of the nonmodular 5-element lattice N5 into the c.e. degrees which maps the top to the degree a. In particular, this shows that local nondistributivity coincides (...) with local nonmodularity in the computably enumerable degrees. (shrink)

The article examines the question of how learning multiple tasks interacts with neural architectures and the flow of information through those architectures. It approaches the question by using the idealization of an artificial neural network where it is possible to ask more precise questions about the effects of modular versus nonmodular architectures as well as the effects of sequential versus simultaneous learning of tasks. A prior work has demonstrated a clear advantage of modular architectures when the two tasks must (...) be learned at the same time from the start, but this advantage may disappear when one task is first learned to a criterion before the second task is undertaken. Indeed, in some cases of sequential learning, nonmodular networks achieve success levels comparable to those of modular networks. In particular, if a nonmodular network is to learn two tasks of different difficulty and the more difficult task is presented first and learned to a criterion, then the network will learn the second, easier one without permanent degradation of the first one. In contrast, if the easier task is learned first, a nonmodular task may perform significantly less well than a modular one. It seems that the reason for this difference has to do with the fact that the sequential presentation of the more difficult task first minimizes interference between the two tasks. More broadly, the studies summarized in this article seem to imply that no single learning architecture is optimal for all situations. (shrink)

In this article we question the utility of the distinction between conceptual and nonconceptual content in cognitive science, and in particular, in the empirical study of visual perception. First, we individuate some difficulties in characterizing the notion of “concept” itself both in the philosophy of mind and cognitive science. Then we stress the heterogeneous nature of the notion of nonconceptual content and outline the complex and ambiguous relations that exist between the conceptual/nonconceptual duality and other pairs of notions, such as (...) top–down/bottom-up and modular/nonmodular. Finally we look in greater detail at the proposal developed by Jacob and Jeannerod (Ways of seeing. The scopes and limits of visual cognition. Oxford, UK: Oxford University Press, 2003 ), who apply the notion of nonconceptual content to empirical research on visual perception. After reconstructing their point of view on concepts, we try to reject their major arguments in support of the conceptual/nonconceptual distinction, i.e. the compositionality of thought and the fineness of grain of percepts. (shrink)

Following the “modularity” orthodoxy of some years ago, it has traditionally been assumed that there is a clear and obvious separation between perception and cognition. Close examination of this concept, however, fails to reveal the join. Ballard et al.'s contention that the two “cannot be easily separated” is consistent with nonmodular views of the way that symbol grounding might be achieved in situated systems. Indeed, the traditional separation is viewed as unhelpful.

Based on multiDEVS formalism, we introduce multiPDEVS, a parallel and nonmodular formalism for discrete event system specification. This formalism provides combined advantages of PDEVS and multiDEVS approaches, such as excellent simulation capabilities for simultaneously scheduled events and components able to influence each other using exclusively their state transitions. We next show the soundness of the formalism by giving a construction showing that any multiPDEVS model is equivalent to a PDEVS atomic model. We then present the simulation procedure associated, usually (...) called abstract simulator. As a well-adapted formalism to express cellular automata, we finally propose to compare an implementation of multiPDEVS formalism with a more classical Cell-DEVS implementation through a fire spread application. (shrink)

We study Bell inequalities and EPR states in the context of Jordan algebras. We show that the set of states violating Bell inequalities across two operator commuting nonmodular Jordan Banach algebras is norm dense in the global state space. It generalizes hitherto known results in quantum field theory in several directions. We propose new Jordan quantity for incommensurable observables in a given state, introduce the concept of EPR state for Jordan structures, and study relationship between EPR states and Bell (...) correlated states. Our analysis shows crucial role of spin factors and Pauli spin matrices for studying noncommutative properties of states and observables. (shrink)

In this paper, we offer an argument in favor of the modular character of mind, based on a more detailed proof of the modular character of the linguistic capacity: in comparing the properties of different components of grammar in a specific area we will draw general consequences about the properties of the cognitive system. More specifically, we analyze and compare the properties, in logical form (LF)and in phonology, of “empty eIements” - eIements that are “visible” or “full” at some level (...) of representation, but not at another level.Our analysis shows that empty eIements are governed at LF by principIes on variables. binding conditions, and thematic principles, that relate eIements at a distance. In phonology, on the other hand, empty elements, i.e., empty segments and empty properties, can be obtaines by deletion and are governed by very different principles (underspecification, principles relating to recoverability of deletion effects), andthese principles relates eIements under strict adjacency. In addition, when a principle seems to be operating in a similar fashion in both components, as in the case of the Projection Principle, it involves lexical properties, which are properties affecting all components. We conclude, hence, that grammar is a modular system, with different components following different principles but related of course, among other things, by the properties of lexical structure.It follows from this analysis that a theory that views language as part of a homogeneous cognitive capacity cannot be mantained: cognition cannot be nonmodular in a strict sense, since a subpart of it, the linguistic capacity is modular itself. We conclude also that the linguistic capacity being modular, we ean only expect that, when studied in more detail, other cognitive capacities will be shown to differ intheir structure still more from the language system. (shrink)

The article examines the question of how learning multiple tasks interacts with neural architectures and the flow of information through those architectures. It approaches the question by using the idealization of an artificial neural network where it is possible to ask more precise questions about the effects of modular versus nonmodular architectures as well as the effects of sequential versus simultaneous learning of tasks. A prior work has demonstrated a clear advantage of modular architectures when the two tasks must (...) be learned at the same time from the start, but this advantage may disappear when one task is first learned to a criterion before the second task is undertaken. Indeed, in some cases of sequential learning, nonmodular networks achieve success levels comparable to those of modular networks. In particular, if a nonmodular network is to learn two tasks of different difficulty and the more difficult task is presented first and learned to a criterion, then the network will learn the second, easier one without permanent degradation of the first one. In contrast, if the easier task is learned first, a nonmodular task may perform significantly less well than a modular one. It seems that the reason for this difference has to do with the fact that the sequential presentation of the more difficult task first minimizes interference between the two tasks. More broadly, the studies summarized in this article seem to imply that no single learning architecture is optimal for all situations. (shrink)

We argue that atomistic learning?learning that requires training only on a novel item to be learned?is problematic for networks in which every weight is available for change in every learning situation. This is potentially significant because atomistic learning appears to be commonplace in humans and most non-human animals. We briefly review various proposed fixes, concluding that the most promising strategy to date involves training on pseudo-patterns along with novel items, a form of learning that is not strictly atomistic, but which (...) looks very much like it ?from the outside? (shrink)