How Do Ideas Become General in their Signification?

Abstraction is one of the central notions in philosophy and cognitive science. Though its origins are often traced to Locke, various senses of abstraction have been developed in fields as diverse as philosophy, psychology, cognitive science, artificial intelligence, and computer science (e.g. Barsalou 2005). The notion of abstraction on which I am focusing here is as that of a process of similarities recognition across instances of a given kind involving progressive exclusion of instance details. As such, abstraction plays a major role in concept-formation and learning.ion plays a major role in concept-formation and learning. Traditionally, abstraction models have been deemed circular (e.g. Berkeley 1710/1957), while in recent years abstraction models have also come under fire for being incoherent (e.g. HendriksJansen 1996), requiring large conceptual resources in order to operate, and so forth. Here, I flesh out a psychological process on the basis of which general ideas are formed out of representations of particulars. The main characteristics of the suggested view are that abstract representations are structured entities with general representational powers, while a major role is given to top-down effects in perception. I argue that the challenges that both traditional and modern abstraction accounts faced are avoided in the light of the present suggestion. How Do Ideas Become General 2 1. TRADITIONAL ABSTRACTION MODELS The question of how we perceive the world has long figured prominently in the agenda of philosophers and psychologists alike. In contemporary philosophy and psychology it is widely accepted that on perceiving the world representations of our environment are formed in our minds (e.g. Findlay & Gilchrist 2003; Gazzaniga et al. 1998; Biederman 1987; Hochberg 1999; Goldstone 1994; Smith & Heise 1992). Furthermore, it is argued that these perceptual representations are often used as raw materials in the process of concept formation in one way or another (e.g. citealtPrinz02,Barsalou99). Concepts, in turn, have an extension in the world and they are thought to pick out all members of a given category. Given that we only have perceptual experiences with particulars though, how is it that we can form mental entities, like concepts, of greater generality or ideas about general properties? General ideas are critical for categorization, which in turn is one of the fundamental aspects of reasoning, decision-making, linguistic inferences and so forth. For Locke in particular the process of abstraction is of considerable importance to human knowledge, as he thinks that most words we use are general (1690/1975: III, I. 1. p., 409). It is worth clarifying that Locke does not imply that it is only general or sortal ideas that can serve in a classificatory scheme (Uzgalis 2012). Furthermore, abstraction is often seen as a gateway to structured ideas, which in turn are often seen as a hallmark of human cognition that any theory must explain (e.g. Fodor & Pylyshyn 1988). Furthermore, having an account of how general ideas are formed plays a key role in accounting for learning and concept acquisition. For concepts are precisely mental particulars with general representational powers — they pick out all members of a given category. Admittedly, an account of general ideas is not the only option for accounting for the origins of our conceptual repertoire. However, the alternative options are not without problems. The first option is to argue that there are no representations that exhibit this kind of generality. Instead, ideas of particular instances play that role when used in the appropriate kind of way (functional role). For instance, an idea of a particular triangle could be used as a stand-in for all triangles. The second option is to argue that concepts do not actually derive from perceptual experiences with particular instances. Rather they are part of our genetic endowVol. 9: Perception and Concepts