We compared the ability of three different contextual models of lexical semantic memory and of a simple associative model to predict the properties of semantic networks derived from word association norms. None of the semantic models were able to accurately predict all of the network properties. All three contextual models over-predicted clustering in the norms, whereas the associative model under-predicted clustering. Only a hybrid model that assumed that some of the responses were based on a contextual model and others on (...) an associative network successfully predicted all of the network properties and predicted a word's top five associates as well as or better than the better of the two constituent models. The results suggest that participants switch between a contextual representation and an associative network when generating free associations. We discuss the role that each of these representations may play in lexical semantic memory. Concordant with recent multicomponent theories of semantic memory, the associative network may encode coordinate relations between concepts, and contextual representations may be used to process information about more abstract concepts. (shrink)

We investigate the family of concepts that an agent comes to know through a set of defining features, and examine the role played by these features in the process of categorization. In a qualitative framework, categorial membership is evaluated through an order relation among the objects at hand, which translates the fact that an object may fall more than another under a given concept. For concepts defined by their features, this global membership order depends on the degree with which each (...) feature applies to the objects of the universe. The passage from these individual membership degrees to a global membership order poses a problem analogous to vote aggregation in social choice theory. This similarity leads to an original solution that is particularly well-adapted to the framework of cognitive psychology. The resulting membership order extends to compound concepts, and provides a good description of the guppy paradox and the conjunction effect. (shrink)

Models of category learning have been extensively studied in cognitive science and primarily tested on perceptual abstractions or artificial stimuli. In this paper, we focus on categories acquired from natural language stimuli, that is, words. We present a Bayesian model that, unlike previous work, learns both categories and their features in a single process. We model category induction as two interrelated subproblems: the acquisition of features that discriminate among categories, and the grouping of concepts into categories based on those features. (...) Our model learns categories incrementally using particle filters, a sequential Monte Carlo method commonly used for approximate probabilistic inference that sequentially integrates newly observed data and can be viewed as a plausible mechanism for human learning. Experimental results show that our incremental learner obtains meaningful categories which yield a closer fit to behavioral data compared to related models while at the same time acquiring features which characterize the learned categories. (shrink)

This paper presents a general computational treatment of how mammals are able to deal with visual objects and environments. The model tries to cover the entire range from behavior and phenomenological experience to detailed neural encodings in crude but computationally plausible reductive steps. The problems addressed include perceptual constancies, eye movements and the stable visual world, object descriptions, perceptual generalizations, and the representation of extrapersonal space.The entire development is based on an action-oriented notion of perception. The observer is assumed to (...) be continuously sampling the ambient light for information of current value. The central problem of vision is taken to be categorizing and locating objects in the environment. The critical step in this process is the linking of visual information to symbolic object descriptions; this is calledindexing, from the analogy of identifying a book from index terms. The system must also identifysituationsand use this knowledge to guide movement and other actions in the environment. The treatment focuses on the different representations of information used in the visual system.The four representational frames capture information in the following forms: retinotopic, head-based, symbolic, and allocentric. The functional roles of the four frames, the communication among them, and their suggested neurophysiological realization constitute the core of the paper. The model is perforce crude, but appears to be consistent with all relevant findings. (shrink)

Much of the progress in the fields constituting cognitive science has been based upon the use of explicit information processing models, almost exclusively patterned after conventional serial computers. An extension of these ideas to massively parallel, connectionist models appears to offer a number of advantages. After a preliminary discussion, this paper introduces a general connectionist model and considers how it might be used in cognitive science. Among the issues addressed are: stability and noise‐sensitivity, distributed decision‐making, time and sequence problems, and (...) the representation of complex concepts. (shrink)

Semantic memory encompasses one's knowledge about the world. Distributional semantic models, which construct vector spaces with embedded words, are a proposed framework for understanding the representational structure of human semantic knowledge. Unlike some classic semantic models, distributional semantic models lack a mechanism for specifying the properties of concepts, which raises questions regarding their utility for a general theory of semantic knowledge. Here, we develop a computational model of a binary semantic classification task, in which participants judged target words for the (...) referent's size or animacy. We created a family of models, evaluating multiple distributional semantic models, and mechanisms for performing the classification. The most successful model constructed two composite representations for each extreme of the decision axis (e.g., one averaging together representations of characteristically big things and another of characteristically small things). Next, the target item was compared to each composite representation, allowing the model to classify more than 1,500 words with human‐range performance and to predict response times. We propose that when making a decision on a binary semantic classification task, humans use task prompts to retrieve instances representative of the extremes on that semantic dimension and compare the probe to those instances. This proposal is consistent with the principles of the instance theory of semantic memory. (shrink)

The syntagmatic paradigmatic model is a distributed, memory‐based account of verbal processing. Built on a Bayesian interpretation of string edit theory, it characterizes the control of verbal cognition as the retrieval of sets of syntagmatic and paradigmatic constraints from sequential and relational long‐term memory and the resolution of these constraints in working memory. Lexical information is extracted directly from text using a version of the expectation maximization algorithm. In this article, the model is described and then illustrated on a number (...) of phenomena, including sentence processing, semantic categorization and rating, short‐term serial recall, and analogical and logical inference. Subsequently, the model is used to answer questions about a corpus of tennis news articles taken from the Internet. The model's success demonstrates that it is possible to extract propositional information from naturally occurring text without employing a grammar, defining a set of heuristics, or specifying a priori a set of semantic roles. (shrink)

A taxonomy of part‐whole or meronymic relations is developed to explain the ordinary English‐speaker's use of the term “part of” and its cognates. The resulting classification yields six types of meronymic relations: 1. component‐integral object (pedal‐bike), 2. member‐collection (ship‐fleet), 3. portion‐mass (slice‐pie), 4. stuff‐object (steel‐car), 5. feature‐activity (paying‐shopping), and 6. place‐area (Everglades‐Florida). Meronymic relations ore further distinguished from other inclusion relations, such as spatial inclusion, and class inclusion, and from several other semantic relations: attribution, attachment, and ownership. This taxonomy is (...) then used to explain cases of apparent intransitivity in merological syllogisms, and standard form syllogisms whose premises express different inclusion relations. The data suggest that intransitivities arise due to equivocations between different types of semantic relations. These results are then explained by means of the relation element theory which accounts for the character and behavior of semantic relations in terms of more primitive relational elements. The inferential phenomena observed are then explained by means of a single principle of element matching. (shrink)

Investigating instances where lexical selection fails can lead to deeper insights into the cognitive machinery and architecture supporting successful word retrieval and speech production. In this paper, we used a multiplex lexical network approach that combines semantic and phonological similarities among words to model the structure of the mental lexicon. Network measures at different levels of analysis (degree, network distance, and closeness centrality) were used to investigate the influence of network structure on picture naming accuracy and errors by people with (...) Anomic, Broca's, Conduction, and Wernicke's aphasia. Our results reveal that word retrieval is influenced by the multiplex lexical network structure in at least two ways—(a) the accuracy of production and error type on incorrect productions were influenced by the degree and closeness centrality of the target word, and (b) error type also varied in terms of network distance between the target word and produced error word. Taken together, the analyses demonstrate that network science techniques, particularly the use of the multiplex lexical network to simultaneously represent semantic and phonological relationships among words, reveal how the structure of the mental lexicon influences language processes beyond traditionally examined psycholinguistic variables. We propose a framework for how the multiplex lexical network approach allows for understanding the influence of mental lexicon structure on word retrieval processes, with an eye toward a better understanding of the nature of clinical impairments, like aphasia. (shrink)

Similarity‐based theories of concepts have a broad intuitive appeal and have been successful in accounting for various phenomena related to the formation and application of concepts. Their adequacy as theories of concepts has been questioned, however, as similarity is often taken as too flexible, too unconstrained, to be explanatory of categorization. In this article, I propose an account of similarity that takes the “foil” against which the target items are measured as integral to the process of comparison, making the similarity (...) relation a fundamentally triadic one. I argue that this account delivers more internal constraints on the process of comparison than the traditional, dyadic account of similarity does. Finally, I propose that this account is advantageous for a similarity‐based theory of concepts, as it easily integrates with the notion of conceptual taxonomy and uses it to deliver additional constraints on categorization. On such a theory of concepts, the foil, whose variability is in large part responsible for the flexible nature of similarity judgements, is turned within the framework of conceptual taxonomy to the very element that constrains similarity judgements in categorization. Therefore, this notion of similarity mitigates the worry that similarity is too flexible to be explanatory of categorization. (shrink)

A growing consensus in the philosophy and psychology of concepts is that while theories such as the prototype, exemplar, and theory theories successfully account for some instances of concept formation and application, none of them successfully accounts for all such instances. I argue against this ‘new consensus’ and show that the problem is, in fact, more severe: the explanatory force of each of these theories is limited even with respect to the phenomena often cited to support it, as each fails (...) to satisfy an important explanatory desideratum with respect to these phenomena. I argue that these explanatory shortcomings arise from a shared assumption on the part of these theories, namely, they take similarity judgements and application of causal knowledge to be discrete elements in a theory of concepts. I further propose that the same assumption carries over into alternative theories offered by proponents of the new consensus: pluralism, eliminativism, and hybrid theories. I put forth a sketch of an integrated model of concept formation and application, which rejects this shared assumption and satisfies the explanatory desiderata I discuss. I suggest that this model undermines the motivation for hybrid, pluralist, and eliminativist accounts of concepts. _1_ Introduction _2_ The Similarity-Based Approach and the Importance of Theory _2.1_ The similarity-based approach _2.2_ The selection desideratum _2.3_ Causal knowledge as satisfying the selection desideratum _3_ The Theory-Based Approach and the Importance of Similarity _3.1_ The theory-based approach _3.2_ The range desideratum _3.3_ Similarity as satisfying the range desideratum _4_ An Integrated Approach to Concepts _4.1_ An integrated model _4.2_ The integrated theory versus hybrid theories of concepts _5_ Conclusion. (shrink)

This paper proposes the ‘AGENCY model’ of conscious state attribution, according to which an entity's displaying certain relatively simple features (e.g. eyes, distinctive motions, interactive behavior) automatically triggers a disposition to attribute conscious states to that entity. To test the model's predictions, participants completed a speeded object/attribution task, in which they responded positively or negatively to attributions of mental properties (including conscious and non-conscious states) to different sorts of entities (insects, plants, artifacts, etc.). As predicted, participants responded positively to conscious (...) state attributions only for those entities that typically display the simple features identified in the AGENCY model (eyes, distinctive motion, interaction) and took longer to deny conscious states to those entities. (shrink)

This paper reviews the uneven history of the relationship between Anthropology and Cognitive Science over the past 30 years, from its promising beginnings, followed by a period of disaffection, on up to the current context, which may lay the groundwork for reconsidering what Anthropology and (the rest of) Cognitive Science have to offer each other. We think that this history has important lessons to teach and has implications for contemporary efforts to restore Anthropology to its proper place within Cognitive Science. (...) The recent upsurge of interest in the ways that thought may shape and be shaped by action, gesture, cultural experience, and language sets the stage for, but so far has not fully accomplished, the inclusion of Anthropology as an equal partner. (shrink)

The current consensus is that most natural categories are not organized around strict definitions (a list of singly necessary and jointly sufficient features) but rather according to a family resemblance (FR) principle: Objects belong to the same category because they are similar to each other and dissimilar to objects in contrast categories. A number of computational models of category construction have been developed to provide an account of how and why people create FR categories (Anderson, 1990; Fisher, 1987). Surprisingly, however, (...) only a few experiments on category construction or free sorting have been run and they suggest that people do not sort examples by the FR principle. We report several new experiments and a two‐stage model for category construction. This model is contrasted with a variety of other models with respect to their ability to account for when FR sorting will and will not occur. The experiments serve to identify one basis for FR sorting and to support the two‐stage model. The distinctive property of the two‐stage model is that it assumes that people impose more structure than the examples support in the first stage and that the second stage adjusts for this difference between preferred and perceived structure. We speculate that people do not simply assimilate probabilistic structures but rather organize them in terms of discrete structures plus noise. (shrink)

Seven features which in practice seem to differentiate belief systems from knowledge systems are discussed. These are: nonconsensuality, “existence beliefs,” alternative worlds, evaluative components, episodic material, unboundedness, and variable credences. Each of these features gives rise to challenging representation problems. Progress on any of these problems within artificial intelligence would be helpful in the study of knowledge systems as well as belief systems, inasmuch as the distinction between the two types of systems is not absolute.

This thesis studies the role of conceptual content in inference and reasoning. The first two chapters offer a theoretical and historical overview of the relation between inference and meaning in philosophy and psychology. In particular, a critical analysis of the formality thesis, i.e., the idea that rational inference is a rule-based and topic-neutral mechanism, is advanced. The origins of this idea in logic and its influence in philosophy and cognitive psychology are discussed. Chapter 3 consists of an analysis of the (...) relationship between inference and representation. It is argued that inference has to be studied from a pluralistic per- spective due to its dependence on different formats of representing information. The following four chapters apply conceptual spaces, a formal theory of concepts within cognitive semantics, to three concept-based inference-types. First, an explication of Sellars notion of material inference is advanced. Later, the model is extended to account for nonmonotonic inference by studying the role expectations in reasoning. Finally, a conceptual space-model of category-based induction is presented. This model predicts most of the empirical properties of this psychological phenomenon and subsumes some of the previous theories in psychology. It is stated that the explanatory fruitfulness of this new approach is evidence for the failure of the formality thesis and calls for a unified model of rational inference that puts semantics at center stage. The last chapter of the thesis discusses how inference and concepts interact in scientific reasoning, which makes constant use of hybrid symbolic structures for representing conceptual information. Stephen Toulmin’s notions of method of representation and inferential technique are developed and applied in a case study about the emergence of the notion of instantaneous speed during the passage from geometrical physics to analytical mechanics. It is claimed that this analysis provides support to the pluralistic perspective for theorizing about reasoning. (shrink)

This thesis is a contribution toward an integral cognitive theory of memory and technology. It, furthermore, develops a theory and prototype for technologically extending mind (memory and thinking).

Given the fundamental role that concepts play in theories of cognition, philosophers and cognitive scientists have a common interest in concepts. Nonetheless, there is a great deal of controversy regarding what kinds of things concepts are, how they are structured, and how they are acquired. This chapter offers a detailed high-level overview and critical evaluation of the main theories of concepts and their motivations. Taking into account the various challenges that each theory faces, the chapter also presents a novel approach (...) to concepts that is organized around two ideas. The first is a pluralistic view of differing types of conceptual structure. The second is a model that treats concepts as atomic representations that are linked to various types of conceptual structures. (shrink)

Beginning in the late 1960s, psychologists began to challenge the view the definitional theory of concepts. According to that theory a concept is a mental representation comprising representations of properties (or “features”) that are individually necessary and jointly sufficient for membership in a category. In place of the definitional view, psychologists initially put forward the prototype theory of concept, according to which concepts comprise representations of features that are typical, salient, and diagnostic for category membership, but not necessarily necessary. The (...) prototype theory gained considerable support in the 1970s, but came under attack in the 1980s. One objection, most forcefully advanced by Jerry Fodor, is that prototypes do not combine compositionally. Compositionality is said to be an adequacy condition on a theory of concepts. If prototypes don’t compose, then prototypes are not concepts. Or so the argument goes. (shrink)

Recently, Casasanto and Lupyan have asserted that there are no context-independent concepts: all concepts are constructed ad hoc when they are instantiated. My aim is to show that the ad hoc cognition framework can be characterized by a similarity-based theory of concepts, and that two different notions of concept should be distinguished —which may be identified with two distinct stages of their life cycle. This approach brings together virtues from opposing views: invariantist: stored concepts are stable enough to collect new (...) information; contextualist: instantiated concepts are context-dependent, what explains our adaptive ability to changing environments. (shrink)

Here is an overview of what is to come. In Sections I and II, I will sketch two of the projects frequently pursued by moral philosophers, and the methods typically invoked in those projects. I will argue that these projects presuppose (or at least suggest) a particular sort of account of the mental representation of human value systems, since the methods make sense only if we assume a certain kind of story about how the human mind stores information about values. (...) The burden of my argument in Section III will be that while the jury is still out, there is some evidence suggesting that this account of mental representation is mistaken. If it is mistaken, it follows that two of the central methods of moral philosophy have to be substantially modified, or perhaps abandoned, and that the goals philosophers have sought to achieve with these methods may themselves be misguided. I fear that many of my philosophical colleagues will find this a quite radical suggestion. But if anything is clear in this area, it is that the methods we will be considering have not been conspicuously successful, though it certainly has not been for want of trying. So perhaps it is time for some radical, empirically informed rethinking of goals and methods in these parts of moral philosophy. (shrink)