Gärdenfors (Conceptual spaces, 2000) argues that the semantic domains that natural language deals with have a geometrical structure. He gives evidence that simple natural language adjectives usually denote natural properties, where a natural property is a convex region of such a “conceptual space.” In this paper I will show that this feature of natural categories need not be stipulated as basic. In fact, it can be shown to be the result of evolutionary dynamics of communicative strategies under very general assumptions.

In this article we discuss the notion of a linguistic universal, and possible sources of such invariant properties of natural languages. In the first part, we explore the conceptual issues that arise. In the second part of the paper, we focus on the explanatory potential of horizontal evolution. We particularly focus on two case studies, concerning Zipf's Law and universal properties of color terms, respectively. We show how computer simulations can be employed to study the large scale, emergent, consequences of (...) psychologically and psychologically motivated assumptions about the working of horizontal language transmission. (shrink)

In this article we discuss the notion of a linguistic universal, and possible sources of such invariant properties of natural languages. In the first part, we explore the conceptual issues that arise. In the second part of the paper, we focus on the explanatory potential of horizontal evolution. We particularly focus on two case studies, concerning Zipf’s Law and universal properties of color terms, respectively. We show how computer simulations can be employed to study the large scale, emergent, consequences of (...) psychologically and psychologically motivated assumptions about the working of horizontal language transmission. (shrink)

This paper deals with Gärdenfors’ theory of conceptual spaces. Let \({\mathcal {S}}\) be a conceptual space consisting of 2-type fuzzy sets equipped with several kinds of metrics. Let a finite set of prototypes \(\tilde{P}_1,\ldots,\tilde{P}_n\in \mathcal {S}\) be given. Our main result is the construction of a classification algorithm. That is, given an element \({\tilde{A}}\in \mathcal {S},\) our algorithm classifies it into the conceptual field determined by one of the given prototypes \(\tilde{P}_i.\) The construction of our algorithm uses some physical analogies (...) and the Newton potential plays a significant role here. Importantly, the resulting conceptual fields are not convex in the Euclidean sense, which we believe is a reasonable departure from the assumptions of Gardenfors’ original definition of the conceptual space. A partitioning algorithm of the space \(\mathcal {S}\) is also considered in the paper. In the application section, we test our classification algorithm on real data and obtain very satisfactory results. Moreover, the example we consider is another argument against requiring convexity of conceptual fields. (shrink)

Few philosophers believe in the existence of so-called negative properties. Indeed, many find it mind-boggling just to imagine such entities. By contrast, I believe not only that negative properties are quite conceivable, but also that there are good reasons for thinking that some such properties actually exist. In this paper, I would like to explicate a concept of negative properties which I think avoids the logical absurdities commonly believed to frustrate theories of negative existences. To do this, I shall deploy (...) a conceptual approach to ontology: I first evaluate the ontological commitments of ordinary negative statements, before I argue that, from the perspective of conceptual analysis, negative properties are no more suspicious than positive properties. In the second part of this paper, I probe the extent to which one can explain the distinction between positive and negative properties in terms of facts about logico-semantical entailment. (shrink)

The notion of conceptual space, proposed by Gärdenfors as a framework for the representation of concepts and knowledge, has been highly influential over the last decade or so. One of the main theses involved in this approach is that the conceptual regions associated with properties, concepts, verbs, etc. are convex. The aim of this paper is to show that such a constraint—that of the convexity of the geometry of conceptual regions—is problematic; both from a theoretical perspective and with regard to (...) the inner workings of the theory itself. On the one hand, all the arguments provided in favor of the convexity of conceptual regions rest on controversial assumptions. Additionally, his argument in support of a Euclidean metric, based on the integral character of conceptual dimensions, is weak, and under non-Euclidean metrics the structure of regions may be non-convex. Furthermore, even if the metric were Euclidean, the convexity constraint could be not satisfied if concepts were differentially weighted. On the other hand, Gärdenfors’ convexity constraint is brought into question by the own inner workings of conceptual spaces because: some of the allegedly convex properties of concepts are not convex; the conceptual regions resulting from the combination of convex properties can be non-convex; convex regions may co-vary in non-convex ways; and his definition of verbs is incompatible with a definition of properties in terms of convex regions. Therefore, the mandatory character of the convexity requirement for regions in a conceptual space theory should be reconsidered. (shrink)

Points of view are a central phenomenon in human cognition. Although the concept of point of view is ambiguous, there exist common elements in different notions. A point of view is a certain way to look at things around us. In conceptual points of view, things are looked at or interpreted through conceptual lenses. Conceptual points of view are important for epistemology, cognitive science, and philosophy of science. In this article, a new method to formalize conceptual points of view is (...) introduced. It is based on the conceptual space approach, where concepts are regions of multi-dimensional quality spaces. Points of view, as defined in this article, consist of a selection of relevant dimensions, referred to here as determinables, and of a certain supposition, referred to here as a theory, about the subject content. After considering some early efforts to formalize points of view, the notion of conceptual space is defined and explored. One concept of point of view is defined and developed in the framework of conceptual space, and a new logic for points of view is also outlined. The problem of mind–body correlation is discussed as an example of the application of the notion of points of view. To conclude, the meaning and applications of the new concepts and tools developed in the article are discussed. (shrink)

A number of recent attempts to bridge Husserlian phenomenology of time consciousness and contemporary tools and results from cognitive science or computational neuroscience are described and critiqued. An alternate proposal is outlined that lacks the weaknesses of existing accounts.

A venerable view holds that a border between perception and cognition is built into our cognitive architecture and that this imposes limits on the way information can flow between them. While the deliverances of perception are freely available for use in reasoning and inference, there are strict constraints on information flow in the opposite direction. Despite its plausibility, this approach to the perception-cognition border has faced criticism in recent years. This article develops an updated version of the architectural approach, which (...) I call the dimension restriction hypothesis (DRH). According to DRH, perceptual processes are constrained to compute over a bounded range of dimensions, while cognitive processes are not. This view allows that perception is cognitively penetrable, but places strict limits on the varieties of penetration that can occur. The article argues that DRH enjoys both theoretical and empirical support, and also defends the view against several objections. (shrink)

We present a model of how counting is learned based on the ability to perform a series of specific steps. The steps require conceptual knowledge of three components: numerosity as a property of collections; numerals; and one-to-one mappings between numerals and collections. We argue that establishing one-to-one mappings is the central feature of counting. In the literature, the so-called cardinality principle has been in focus when studying the development of counting. We submit that identifying the procedural ability to count with (...) the cardinality principle is not sufficient, but only one of the several steps in the counting process. Moreover, we suggest that some of these steps may be facilitated by the external organization of the counting situation. Using the methods of situated cognition, we analyze how the balance between external and internal representations will imply different loads on the working memory and attention of the counting individual. This analysis will show that even if the counter can competently use the cardinality principle, counting will vary in difficulty depending on the physical properties of the elements of collection and on their special arrangement. The upshot is that situated factors will influence counting performance. (shrink)

This paper offers a novel way of reconstructing conceptual change in empirical theories. Changes occur in terms of the structure of the dimensions—that is to say, the conceptual spaces—underlying the conceptual framework within which a given theory is formulated. Five types of changes are identified: (1) addition or deletion of special laws, (2) change in scale or metric, (3) change in the importance of dimensions, (4) change in the separability of dimensions, and (5) addition or deletion of dimensions. Given this (...) classification, the conceptual development of empirical theories becomes more gradual and rationalizable. Only the most extreme type—replacement of dimensions—comes close to a revolution. The five types are exemplified and applied in a case study on the development within physics from the original Newtonian mechanics to special relativity theory. (shrink)

This article presents two learning processes in order to explain how children at an early age can transform a complex sensory input to concepts and categories. The first process constructs the perceptual structures that emerge in children’s cognitive development by detecting invariants in the sensory input. The invariant structures involve a reduction in dimensionality of the sensory information. It is argued that this process generates the primary domains of space, objects and actions and that these domains can be represented as (...) conceptual spaces. Once the primary domains have been established, the second process utilizes covariances between different dimensions of the domains in order to identify natural clusters of entities. The clusters are then are used to determine concepts as regions in the spaces. As an application, the processes are used to resolve the so-called ‘complex first paradox’ that emerges from the fact that children, in general, learn nouns earlier than adjectives, even though nouns are semantically more complex than adjectives. (shrink)

Within analytic philosophy, induction has been seen as a problem concerning inferences that have been analysed as relations between sentences. In this article, we argue that induction does not primarily concern relations between sentences, but between properties and categories. We outline a new approach to induction that is based on two theses. The first thesis is epistemological. We submit that there is not only knowledge-how and knowledge-that, but also knowledge-what. Knowledge-what concerns relations between properties and categories and we argue that (...) it cannot be reduced to knowledge-that. We support the partition of knowledge by mapping it onto the long-term memory systems: procedural, semantic and episodic memory. The second thesis is that the role of inductive reasoning is to generate knowledge-what. We use conceptual spaces to model knowledge-what and the relations between properties and categories involved in induction. (shrink)

Within analytic philosophy, induction has been seen as a problem concerning inferences that have been analysed as relations between sentences. In this article, we argue that induction does not primarily concern relations between sentences, but between properties and categories. We outline a new approach to induction that is based on two theses. The first thesis is epistemological. We submit that there is not only knowledge-how and knowledge-that, but also knowledge-what. Knowledge-what concerns relations between properties and categories and we argue that (...) it cannot be reduced to knowledge-that. We support the partition of knowledge by mapping it onto the long-term memory systems: procedural, semantic and episodic memory. The second thesis is that the role of inductive reasoning is to generate knowledge-what. We use conceptual spaces to model knowledge-what and the relations between properties and categories involved in induction. (shrink)

The meanings of words are not permanent but change over time. Some changes of meaning are quick, such as when a pronoun changes its reference; some are slower, as when two speakers find out that they are using the same word in different senses; and some are very slow, such as when the meaning of a word changes over historical time. A theory of semantics should account for these different time scales. In order to describe these different types of meaning (...) changes, I present an analysis of three levels of communication: instruction, coordination of common ground and coordination of meaning. My first aim is to show that these levels must be considered when discussing lexical semantics. A second aim is to use the levels to identify the communicative roles of some of the main word classes, in particular nouns, adjectives, verbs, indexicals and quantifiers. I argue that the existence of word classes can, to a large extent, be explained by the communicative needs that arise on the different levels. (shrink)

Over the past few decades, cognitive science has identified several forms of reasoning that make essential use of conceptual knowledge. Despite significant theoretical and empirical progress, there is still no unified framework for understanding how concepts are used in reasoning. This paper argues that the theory of conceptual spaces is capable of filling this gap. Our strategy is to demonstrate how various inference mechanisms which clearly rely on conceptual information—including similarity, typicality, and diagnosticity-based reasoning—can be modeled using principles derived from (...) conceptual spaces. Our first topic analyzes the role of expectations in inductive reasoning and their relation to the structure of our concepts. We examine the relationship between using generic expressions in natural language and common-sense reasoning as a second topic. We propose that the strength of a generic can be described by distances between properties and prototypes in conceptual spaces. Our third topic is category-based induction. We demonstrate that the theory of conceptual spaces can serve as a comprehensive model for this type of reasoning. The final topic is analogy. We review some proposals in this area, present a taxonomy of analogical relations, and show how to model them in terms of distances in conceptual spaces. We also briefly discuss the implications of the model for reasoning with concepts in artificial systems. (shrink)

We present a model of how counting is learned based on the ability to perform a series of specific steps. The steps require conceptual knowledge of three components: numerosity as a property of collections; numerals; and one-to-one mappings between numerals and collections. We argue that establishing one-to-one mappings is the central feature of counting. In the literature, the so-called cardinality principle has been in focus when studying the development of counting. We submit that identifying the procedural ability to count with (...) the cardinality principle is not sufficient, but only one of the several steps in the counting process. Moreover, we suggest that some of these steps may be facilitated by the external organization of the counting situation. Using the methods of situated cognition, we analyze how the balance between external and internal representations will imply different loads on the working memory and attention of the counting individual. This analysis will show that even if the counter can competently use the cardinality principle, counting will vary in difficulty depending on the physical properties of the elements of collection and on their special arrangement. The upshot is that situated factors will influence counting performance. (shrink)

This paper aims to contribute to the existing literature on normative cognition by providing a lineage explanation of human social norm psychology. This approach builds upon theories of goal-directed behavioral control in the reinforcement learning and control literature, arguing that this form of control defines an important class of intentional normative mental states that are instrumental in nature. I defend the view that great ape capacities for instrumental reasoning and our capacity (or family of capacities) for shared intentionality coadapted to (...) each other and argue that the evolution of this capacity has allowed the representation of social norms and the emergence of our capacity for normative guidance. (shrink)

Structure–activity relationship and quantitative SAR are modeling methods largely used in assessing biological properties of chemical substances. QSAR is based on the hypothesis that the chemical structure is responsible for the activity; it follows that similar molecules are expected to have similar properties. Similarity plays an important role in read across, which categorizes molecules primarily on the basis of similarity. Similarity, and chemical similarity too, is a property differently perceived by humans. The various proposed metrics often disagree with human judgment, (...) and no a unique metric for chemical similarity is universally adopted. Researchers argued that categorization is not only explained by similarity but depends as well on abstract knowledge and the task to accomplish. Moreover, similarity cannot be the unique explanation of a categorization, as different perceptual processes take place in category formation. Assuming that similarity judgments are deeply rooted in human knowledge and perception, cognitive sciences contributions are as important as the mathematical considerations of the classical theories. After an excursus in the many views of similarity in philosophy, mathematics, and cognitive science, the paper explores how connectionist systems, which loosely mimic the human cognitive system, could improve similarity-based choices. A case study on building SARs using connectionism and deep neural networks shows the role of similarity in building and explaining those models. A discussion about deep learning for QSAR and as a modeling tool for science concludes the presentation. (shrink)

Natural languages vary in their quantity expressions, but the variation seems to be constrained by general properties, so-calleduniversals. Their explanations have been sought among constraints of human cognition, communication, complexity, and pragmatics. In this article, we apply a state-of-the-art language coordination model to the semantic domain of quantities to examine whether two quantity universals—monotonicity and convexity—arise as a result of coordination. Assuming precise number perception by the agents, we evolve communicatively usable quantity terminologies in two separate conditions: a numeric-based condition (...) in which agents communicate about a number of objects and a quotient-based condition in which agents communicate about the proportions. We find out that both universals take off in all conditions but only convexity almost entirely dominates the emergent languages. Additionally, we examine whether the perceptual constraints of the agents can contribute to the further development of universals. We compare the degrees of convexity and monotonicity of languages evolving in populations of agents with precise and approximate number sense. The results suggest that approximate number sense significantly reinforces monotonicity and leads to further enhancement of convexity. Last but not least, we show that the properties of the evolved quantifiers match certain invariance properties from generalized quantifier theory. (shrink)

When we imaginatively picture what might happen, we may take what we imagine to be either realistic or fantastic. A wine glass falling to the floor and shattering is realistic. A wine glass falling and morphing into a bird and flying away is fantastic. What does the distinction consist in? Two important necessary conditions are here defined. The first is a condition on the realistic representation of spatial configuration, grounded in an account of the imagistic representation of spatial configuration. The (...) second is a condition on the manner in which realistic courses of mental imagery may be grounded in remembered perceptions. This is defined in terms of an account of the representation of comparative similarity. This is an open access publication. (shrink)

A similarity space is a hyperspace in which the dimensions represent various dimensions on which objects may differ. The similarity space theory of concepts is the thesis that concepts are regions of similarity spaces that are somehow realized in the brain. Proponents of such a theory of concepts include Paul Churchland and Peter Gärdenfors. This paper argues that the similarity space theory of concepts is mistaken because regions of similarity spaces cannot serve as the components of judgments. It emerges that (...) although similarity spaces cannot model concepts, they may model a kind of nonconceptual representation. (shrink)

This paper argues in favor of the ambiguity in the interpretive process and mechanisms that can lead to stop the so-called unlimited semiosis. As we shall see, our way of approaching the problem can be used in other contexts, and the resulting conclusions can be applied to text, artwork, spoken messages, and to any sign to be interpreted.

Signalling games are popular models for studying the evolution of meaning, but typical approaches do not incorporate vagueness as a feature of successful signalling. Complementing recent like-minded models, we describe an aggregate population-level dynamic that describes a process of imitation of successful behaviour under imprecise perception and realization of similar stimuli. Applying this new dynamic to a generalization of Lewis’s signalling games, we show that stochastic imprecision leads to vague, yet by-and-large efficient signal use, and, moreover, that it unifies evolutionary (...) outcomes and helps avoid sub-optimal categorization. The upshot of this is that we see ‘as-if’-generalization at an aggregate level, without agents actually generalizing. _1_ Introduction _2_ Background _2.1_ Sim-max games and conceptual spaces _2.2_ Vagueness in sim-max games and conceptual spaces _2.3_ Vagueness, functional pressure, and transmission biases _3_ Imprecise Imitation _3.1_ Replicator dynamic in behavioural strategies _3.2_ Noise-perturbed conditional imitation _4_ Exploring Imprecise Imitation _4.1_ Setting the stage _4.2_ Simulation set-up _4.3_ Measures of interest _4.4_ Results _5_ Discussion _5.1_ Levels of vagueness _5.2_ Evolutionary benefits of imprecision _5.3_ Related work _6_ Conclusion Appendix. (shrink)

We propose a logic for reasoning about metric spaces with the induced topologies. It combines the 'qualitative' interior and closure operators with 'quantitative' operators 'somewhere in the sphere of radius r.' including or excluding the boundary. We supply the logic with both the intended metric space semantics and a natural relational semantics, and show that the latter (i) provides finite partial representations of (in general) infinite metric models and (ii) reduces the standard '∈-definitions' of closure and interior to simple constraints (...) on relations. These features of the relational semantics suggest a finite axiomatisation of the logic and provide means to prove its EXPTIME-completeness (even if the rational numerical parameters are coded in binary). An extension with metric variables satisfying linear rational (in)equalities is proved to be decidable as well. Our logic can be regarded as a 'well-behaved' common denominator of logical systems constructed in temporal, spatial, and similarity-based quantitative and qualitative representation and reasoning. Interpreted on the real line (with its Euclidean metric), it is a natural fragment of decidable temporal logics for specification and verification of real-time systems. On the real plane, it is closely related to quantitative and qualitative formalisms for spatial representation and reasoning, but this time the logic becomes undecidable. (shrink)

Both early analytic philosophy and the branch of mathematics now known as topology were gestated and born in the early part of the 20th century. It is not well recognized that there was early interaction between the communities practicing and developing these fields. We trace the history of how topological ideas entered into analytic philosophy through two migrations, an earlier one conceiving of topology geometrically and a later one conceiving of topology algebraically. This allows us to reassess the influence and (...) significance of topological methods for philosophy, including the possible fruitfulness of a third conception of topology as a structure determining similarity. (shrink)

This essay will be divided into three parts. In the first part, we discuss the case of infintesimals seen as a bridge between the discrete and the continuous. This leads in the second part to a discussion of the nature of numbers. In the last part, we follow up with some observations on the obvious applicability of mathematics.

The concept of representation has been a key element in the scientific study of mental processes, ever since such studies commenced. However, usage of the term has been all but too liberal—if one were to adhere to common use it remains unclear if there are examples of physical systems which cannot be construed in terms of representation. The problem is considered afresh, taking as the starting point the notion of activity spaces—spaces of spatiotemporal events produced by dynamical systems. It is (...) argued that representation can be analyzed in terms of the geometrical and topological properties of such spaces. Several attributes and processes associated with conceptual domains, such as logical structure, generalization and learning are considered, and given analogues in structural facets of activity spaces, as are misrepresentation and states of arousal. Based on this analysis, representational systems are defined, as is a key concept associated with such systems, the notion of representational capacity. According to the proposed theory, rather than being an all or none phenomenon, representation is in fact a matter of degree—that is can be associated with measurable quantities, as is behooving of a putative naturalistic construct. (shrink)

Psychology has benefited from an enormous wealth of knowledge about processes of cognition in relation to how the brain organizes information. Within the categorization literature, this behavior is often explained through theories of memory construction called exemplar theory and prototype theory which are typically based on similarity or rule functions as explanations of how categories emerge. Although these theories work well at modeling highly controlled stimuli in laboratory settings, they often perform less well outside of these settings, such as explaining (...) the emergence of background knowledge processes. In order to explain background knowledge, we present a non-similarity-based post-Skinnerian theory of human language called Relational Frame Theory which is rooted in a philosophical world view called functional contextualism. This theory offers a very different interpretation of how categories emerge through the functions of behavior and through contextual cues, which may be of some benefit to existing categorization theories. Specifically, RFT may be able to offer a novel explanation of how background knowledge arises, and we provide some mathematical considerations in order to identify a formal model. Finally, we discuss much of this work within the broader context of general semantic knowledge and artificial intelligence research. (shrink)

Current cognitive architectures are either working at the abstract, symbolic level, or the low, emergent level related to neural modeling. The best way to understand phenomena is to see, or imagine them, hence the need for a geometric model of mental processes. Geometric models should be based on an intermediate level of modeling that describe mental states in terms of features relevant from the first-person perspective but also linked to neural events. Concepts should be represented as geometrical objects that have (...) sufficiently rich structures to show their properties and their relations to other concepts. The best way to create such geometrical representations of concepts is through the approximate description of the physical states of neural networks. The evolution of brain states is then represented as a trajectory linking successful concepts, and topological constraints on the shape of such trajectory define grammar and logic. (shrink)

Edgington has proposed a solution to the sorites paradox in terms of ‘verities’, which she defines as degrees of closeness to clear truth. Central to her solution is the assumption that verities are formally probabilities. She is silent on what verities might derive from and on why they should be probabilities. This paper places Edgington’s solution in the framework of a spatial approach to conceptualization, arguing that verities may be conceived of as deriving from how our concepts relate to each (...) other. Building on work by Kamp and Partee, this paper further shows how verities, thus conceived of, may plausibly be assumed to have probabilistic structure. The new interpretation of verities is argued to also help answer the question of what the verities of indicative conditionals are, a question which Edgington leaves open. Finally, the question of how to accommodate higher-order vagueness, given this interpretation, is addressed. (shrink)

Edgington has proposed a degree-theoretic account of vagueness that yields a highly elegant solution to the sorites paradox. This paper applies her account to the paradox of Theseus’ ship, which is generally classified among the paradoxes of material constitution and not as a sorites paradox.

The conceptual spaces approach has recently emerged as a novel account of concepts. Its guiding idea is that concepts can be represented geometrically, by means of metrical spaces. While it is generally recognized that many of our concepts are vague, the question of how to model vagueness in the conceptual spaces approach has not been addressed so far, even though the answer is far from straightforward. The present paper aims to fill this lacuna.

Conceptual spaces have become an increasingly popular modeling tool in cognitive psychology. The core idea of the conceptual spaces approach is that concepts can be represented as regions in similarity spaces. While it is generally acknowledged that not every region in such a space represents a natural concept, it is still an open question what distinguishes those regions that represent natural concepts from those that do not. The central claim of this paper is that natural concepts are represented by the (...) cells of an optimally designed similarity space. (shrink)

There is an ongoing debate about which rule we ought to use for scoring probability estimates. Much of this debate has been premised on scoring-rule monism, according to which there is exactly one best scoring rule. In previous work, I have argued against this position. The argument given there was based on purely a priori considerations, notably the intuition that scoring rules should be sensitive to truthlikeness relations if, and only if, such relations are present among whichever hypotheses are at (...) issue. The present paper offers a new, quasi-empirical argument against scoring-rule monism. This argument uses computational simulations to show that different scoring rules can have different economical consequences, depending on the context of use. (shrink)

This paper considers Kamp and Partee's account of graded membership within a conceptual spaces framework and puts the account to the test in the domain of colors. Three experiments are reported that are meant to determine, on the one hand, the regions in color space where the typical instances of blue and green are located and, on the other hand, the degrees of blueness/greenness of various shades in the blue–green region as judged by human observers. From the locations of the (...) typical blue and typical green regions in conjunction with Kamp and Partee's account follow degrees of blueness/greenness for the color shades we are interested in. These predicted degrees are compared with the judged degrees, as obtained in the experiments. The results of the comparison support the account of graded membership at issue. (shrink)

The standard approach to the so-called paradoxes of identity has been to argue that these paradoxes do not essentially concern the notion of identity but rather betray misconceptions on our part regarding other metaphysical notions, like that of an object or a property. This paper proposes a different approach by pointing to an ambiguity in the identity predicate and arguing that the concept of identity that figures in many ordinary identity claims, including those that appear in the paradoxes, is not (...) the traditional philosophical concept but one that can be defined in terms of relevant similarity. This approach to the paradoxes will be argued to be superior to the standard one. (shrink)

Comparative concepts such as greener than or higher than are ways of ordering objects. They are fundamental to our grasp of gradable concepts, that is, the type of meanings expressed by gradable general terms, such as "is green" or "is high", which are embeddable in comparative constructions in natural language. Some comparative concepts seem natural, whereas others seem gerrymandered. The aim of this paper is to outline a theoretical approach to comparative concepts that bears both on the account of naturalness (...) for comparative concepts and on the theory of gradable concepts. The approach is novel in that it carries some basic assumptions from Peter Gärdenfors' conceptual spaces account of categorical concepts over to comparative concepts. The offered approach is more general than Gärdenfors' account in that it supplies a framework of graded categorisation that includes his categorisation rule as a limiting case. Importantly, it provides also a new argument for adopting Gärdenfors' particular model of categorisation. (shrink)