The inductive category formation framework, an influential set of theories of learning in psychology and artificial intelligence, is deeply flawed. In this framework a set of necessary and sufficient features is taken to define a category. Such definitions are not functionally justified, are not used by people, and are not inducible by a learning system. Inductive theories depend on having access to all and only relevant features, which is not only impossible but begs a key question in learning. The crucial (...) roles of other cognitive processes (such as explanation and credit assignment) are ignored or oversimplified. Learning necessarily involves pragmatic considerations that can only be handled by complex cognitive processes.We provide an alternative framework for learning according to which category definitions must be based on category function. The learning system invokes other cognitive processes to accomplish difficult tasks, makes inferences, analyses and decides among potential features, and specifies how and when categories are to be generated and modified. We also examine the methodological underpinnings of the two approaches and compare their motivations. (shrink)

Previous neural network models of concept learning were mainly implemented with supervised learning schemes. However, studies of human conceptual memory have shown that concepts may be learned without a teacher who provides the category name to associate with exemplars. A modular neural network architecture that realizes concept acquisition through two functionally distinct operations, categorizing and naming, is proposed as an alternative. An unsupervised algorithm realizes the categorizing module by constructing representations of categories compatible with prototype theory. The naming module associates (...) category names to the output of the categorizing module in a supervised mode. In such a modular architecture, the interface between the modules can be conceived of as an “information relay” that encodes, constrains, and propagates important information. Five experiments were conducted to analyze the relationships among internal conceptual codes and simple conceptual and lexical development. The first two experiments show a prototype effect and illustrate some basic characteristics of the system. The third experiment presents a bottom‐up model of the narrowing down of children's early lexical categories that honors mutual exclusivity. The fourth experiment introduces top‐down constraints on conceptual coding. The fifth experiment exhibits how hierarchical relationships between concepts are learned by the architecture, and also demonstrates how a spectrum of conceptual expertise may gradually emerge as a consequence of experiencing more with certain categories than with others. (shrink)

The traditional definition per genus et differentiam is argued to be cognitively grounded in perception and in order to avoid needless argument, definitions are stipulated to assert boundaries. An analysis of the notion of perspective shows that a boundary is a composite of two distinctions: similarity that includes and difference that excludes. The concept is applied to the type-token distinction and percepts are shown to be the result of a comparison between a token as representing some phenomenon and a type (...) as representing the kind of object the phenomenon might be. The Principle of Connection is proposed as the mental architecture that gives rise to percepts. The value of a percept is shown to be one of Identical, Similar, Different or Contrary and its perceptual aspects represent the cognitive schema of Essence, Quality, Quantity and Relation. The two notions of similarity and difference are claimed to constitute the necessary and sufficient conditions for definitions. Accordingly, the qualitative elements of a definition signify the genus as representing the similarity of the defined entity in terms of others of its kind and the quantitative elements signify the differentia as representing the difference it has from everything else. Paying proper attention to both qualitative and quantitative aspects will enable formulation of definitions that are ‘good’. (shrink)

The initial stage for the discussion is the distinction between bona fide and fiat objects drawn by Barry Smith and collaborators in the context of formal ontology. This paper aims at both producing a rationale for introducing a hitherto unrecognized kind of object—here called ‘Interactive Fiat Objects’ (IFOs)—into the ontology of objects, and casting light on the relationship between embodied cognition and interactive ontology with the aid of the concepts of affordance and ad hoc category. I conclude that IFOs are (...) similar to fiat objects, affordances and ad hoc categories in a number of ways, yet they differ from these in important respects. Interaction is key to understanding the existence and peculiarities of IFOs. By adopting an embodied perspective on cognition, we can enrich our ontological typologies and highlight relevant features of our physical and symbolic environment that are otherwise overlooked. This should improve our understanding of object ontology and persuade us to include IFOs in our metaphysical inventories. (shrink)

We argue that are no such things as literal categories in human cognition. Instead, we argue that there are merely temporary coalescences of dimensions of similarity, which are brought together by context in order to create the similarity structure in mental representations appropriate for the task at hand. Fodor contends that context‐sensitive cognition cannot be realised by current computational theories of mind. We address this challenge by describing a simple computational implementation that exhibits internal knowledge representations whose similarity structure alters (...) fluidly depending on context. We explicate the processing properties that support this function and illustrate with two more complex models, one applied to the development of semantic knowledge , the second to the processing of simple metaphorical comparisons . The models firstly demonstrate how phenomena that seem problematic for literal categorisation resolve to particular cases of the contextual modulation of mental representations; and secondly prompt a new perspective on the relation between language and thought: language affords the strategic control of context on semantic knowledge, allowing information to be brought to bear in a given situation that might otherwise not be available to influence processing. This may explain one way in which human thought is creative, and distinctive from animal cognition. (shrink)

As an Introduction to the Special Issue on “Formulation, generalization, and abstraction in interaction,” this paper discusses key problems of a conversation analytic (CA) approach to semantics in interaction. Prior research in CA and Interactional Linguistics has only rarely dealt with issues of linguistic meaning in interaction. It is argued that this is a consequence of limitations of sequential analysis to capture meaning in interaction. While sequential analysis remains the encompassing methodological framework, it is suggested that it needs to be (...) complemented by analyzing semantic relationships between choices of formulation in the interaction, ethnography, and structural techniques of comparing selected options with possible alternatives. The paper describes the methodological approach taken to interactional semantics by the papers in the Special Issue, which analyse practices of generalization and abstraction in interaction as they are accomplished by formulations of prior versions of reference and description. (shrink)

Irony is an effective but challenging mode of communication that allows a speaker to express viewpoints rich in sentiment with concision, sharpness and humour. Creative irony is especially common in online documents that express subjective and deeply-felt opinions, and thus represents a significant obstacle to the accurate analysis of sentiment in web texts. In this paper we look at one commonly used framing device for linguistic irony—the simile—to show how even the most creative uses of irony are often marked in (...) ways that make them computationally feasible to detect. We conduct a very large corpus analysis of web-harvested similes to identify the most interesting characteristics of ironic comparisons, and provide an empirical evaluation of a new algorithm for separating ironic from non-ironic similes. (shrink)

The rise of appeals to intuitive theories in many areas of cognitive science must cope with a powerful fact. People understand the workings of the world around them in far less detail than they think. This illusion of knowledge depth has been uncovered in a series of recent studies and is caused by several distinctive properties of explanatory understanding not found in other forms of knowledge. Other experimental work has shown that people do have skeletal frameworks of expectations that constrain (...) richer ad hoc theory construction on the fly. These frameworks are supplemented by an ability to evaluate and rely on the division of cognitive labour in one's culture, an ability shown to be present even in young children. (shrink)

How should a practice, subservient to a public good, be regulated in order to guarantee fair access without encouraging improper claims? In the first place, a clear understanding of the goal of the practice is indispensable for knowing what criteria the regulation must contain. As to the purely formal aspect, the regulation of any practice must include both general rules and particular instances. Finally, to resolve conflicts, committees in which different kinds of expertise are represented should be installed. These three (...) theses are illustrated by the Dutch regulation for cosmetic surgery. (shrink)

This paper discusses how we understand and use a concept or the meaningof a general term to identify the objects falling under the term. There aretwo distinct approaches to research on the problems of concepts and meaningthe psychological approach and the formal (or logical) approach. My majorconcern is to consider the possibility of reconciling these two differentapproaches, and for this I propose to build a psychologically plausibleformal system of conceptualization. That is, I will develop a theory-basedaccount of concepts and propose (...) an explanation of how an agent activates aperspective (which consists of theories) in response to a situation in whichreasoning using a concept is called for. Theories are represented as sets offacts and rules, both strict and defeasible. Each theory is organized in acoherent perspective which stands for an agent's mental state or an agent'smodel of another agent's perspective. Perspectives are organized intohierarchies and the theory for a concept in one perspective may defeat thetheory for the same concept in another perspective. Which perspective issuperior is context-dependent. (shrink)

Information management systems improve the retention of information in large collections. As such they act as memory prostheses, implying an ideal basis in human memory models. Since humans process information by association, and situate it in the context of space and time, systems should maximize their effectiveness by mimicking these functions. Since human attentional capacity is limited, systems should scaffold cognitive efforts in a comprehensible manner. We propose the Principles of Mnemonic Associative Knowledge (P-MAK), which describes a framework for semantically (...) identifying, organizing, and retrieving information, and for encoding episodic events by time and stimuli. Inspired by prominent human memory models, we propose associative networks as a preferred representation. Networks are ideal for their parsimony, flexibility, and ease of inspection. Networks also possess topological properties—such as clusters, hubs, and the small world—that aid analysis and navigation in an information space. Our cognitive perspective addresses fundamental problems faced by information management systems, in particular the retrieval of related items and the representation of context. We present evidence from neuroscience and memory research in support of this approach, and discuss the implications of systems design within the constraints of P-MAK’s principles, using text documents as an illustrative semantic domain. (shrink)

Our utterances are typically if not always "situated," in the sense that they are true or false relative to unarticulated parameters of the extra-linguistic context. The problem is to explain how these parameters are determined, given that nothing in the uttered sentences indicates them. It is tempting to claim that they must be determined at the level of thought or intention. However, as many philosophers have observed, thoughts themselves are no less situated than utterances. Unarticulated parameters need not be mentally (...) represented. In this paper, I try to make precise the notion of representation at stake here. In one sense of 'representation', something is represented if it is inferentially relevant. In another, less demanding sense, something is represented if it is relevant to the construction of a context-sensitive, ad hoc concept. Ad hoc concepts act as "proxies" for cognitively more demanding representations. They "imitate" the latter's epistemic and pragmatic roles while being inferentially less sophisticated. Thus, there are two senses in which a thought can be said to be situated: (1) its truth-value is relative to a non-represented contextual parameter, (2) its truth-value is not itself relative, but it involves a context-sensitive, ad hoc concept. (shrink)