Concepts and the Appeal to Cognitive Science

Samuel D. Taylor

• Publisher: Düsseldorf University Press
• Year: 2021
• ISBN:
  9783110708226
• Edition: 1
• Topic: Languages & Linguistics
• Subtopic: Linguistics
• Index: Languages & Linguistics

1 Introduction

1.1 Concepts: The Standard View

Our concepts enable us to entertain thoughts about a rich variety of objects, events, and states of affairs. By employing concepts we are able to ponder the beginnings of the universe, the treatment of disease, the nature of infinity, and the organisation of society. It is through concept use that we are able to rationally deliberate and discuss, and so come to a greater understanding of our surroundings and our place within them. Or so the story goes.

This story, however, is not particularly helpful until we can say what a concept is and what a concept is not. Without an answer in this regard, it is unclear how concepts come to be employed by concept users in the first place. A neutral—one might even say pre-scientific—characterisation of concepts takes them to be the “materials of reason and knowledge” (Locke, 1690). A modern rendering of this view is that concepts are constituents of thought. In this way, it is said that my thought that Eric Cantona played for Manchester United contains or, perhaps, makes use of at least two concepts: Eric Cantona and Manchester United.¹

The problem with this view is that it says nothing about what thoughts are or what it takes to be a constituent of thought. As a result, one might think that the question is put back, because in order to characterise concepts as constituents of thought we first require an account of thoughts. Typically, however, the threat of regress is circumvented by arguing that it is concepts that explain what thoughts are and how they behave. So, on this line of reasoning, concepts are taken to be those things that explain thought and thought is that thing we commit ourselves to in order to explain how organisms like us interact with and negotiate our environments. Accordingly, many philosophers have argued that it is our commitment to the kind CONCEPT—as the set of concepts—that helps us to formulate viable explanations of cognitive behaviour (cf. Margolis and Laurence, 1999).²

However, in order for concepts to satisfactorily play the explanatory role required of them, a theory of the kind CONCEPT must be developed to explain how all concepts—as members of this kind—operate as the constituents of thought. To achieve this end, a theory of CONCEPT would need to satisfy four desiderata—(1) Scope, (2) Intentional/Cognitive Content, (3) Acquisition, and (4) Compositionality—, which I spell-out in detail in chapter two. According to the “standard view,” a theory of CONCEPT will be best able to satisfy these desiderata if it takes concepts to be a single kind of “mental representation” storing a single kind of information (see the definition below).³

Definition: The Standard View

A theory of CONCEPT that accounts for the properties of concepts will be a theory of a single kind of mental representation storing a single kind of information.

But the standard view is not without its problems. The main problem is that saying that objects of the kind CONCEPT are a single kind of mental representation is not all that informative. The question that naturally follows such a claim is: what are mental representations? Mental representations are mental objects that “are said to have “intentionality”—they are about or refer to things, and may be evaluated with respect to properties like consistency, truth, appropriateness and accuracy” (Pitt, 2018). For example, if I have a mental representation of Eric Cantona, then this mental representations will refer to Eric Cantona and can be evaluated with respect to how well it consistently, truthfully, accurately etc. represents Eric Cantona (does it represent him as a man, a footballer, etc.).

The idea of the standard view, then, is that mental representations—as concepts—make possible thoughts about, say, Eric Cantona and Manchester United. But even if this is correct, a further question can be asked: what kind of mental representations are concepts? By asking this question, one is forced headways into a debate about the structure of concepts. Here one finds a number of competing accounts of conceptual structure that each purport to have one explanatory advantage or another in virtue of better accounting for desiderata (1)−(4). Charting one’s way through this debate is deemed necessary if one is to motivate and defend a theory of the representational structure of the members of the kind CONCEPT.

Many adhering to the standard view have taken up the task of developing theories of CONCEPT that give an account of the representational structure of concepts as storing one kind of information. For thirty years or more, different theories have been developed and debate has raged about which of these theories is best able to satisfy desiderata (1)−(4). This debate has thrown into relief the difficulties of developing a theory of CONCEPT that satisfies all of the desiderata (1)−(4). In fact, some are now coming to realise that although different theories of the representational structure of concepts will better satisfy different desiderata, none are likely to satisfy them all.⁴

1.2 The Appeal to Cognitive Science

In chapter three, I follow recent developments in the literature—e.g. Machery (2009) and Weiskopf (2009)—and argue that the relative satisfaction of the desiderata (1)−(4) cannot help us to decide between different theories of the representational structure of concepts. It follows that a new methodology must be found to decide between available theories of CONCEPT. For many, the solution is an appeal to empirical research. Therefore, many now assume that we can evaluate and compare theories of CONCEPT be considering how consistent each theory is with the explanatory results of cognitive science, and that such a process will enable us to determine the “best” theory of CONCEPT (see the definition below).

Definition: The Appeal to Cognitive Science

Deciding on the best theory of concept by comparing different theories of concept with respect to how consistent they are with the explanatory results of cognitive science.

Cognitive science is the interdisciplinary study of mind and intelligence, embracing philosophy, psychology, artificial intelligence, neuroscience, linguistics, and anthropology (Thagard, 2014). Although it is widely accepted that these disciplines share unifying theoretical ideas, there are clearly big differences in their outlooks, methods, and results. For example, designing, building, and experimenting with computational models is the central method of artificial intelligence (AI), but does not play a central role in cognitive anthropology. Analogously, (experimental) methods differ both between (e.g. between cognitive psychology and neuroscience) and within disciplines (e.g. between computational linguistics and semantics).

Therefore, one problem presents itself straight away: that it is unclear what counts as the “explanatory results of cognitive science.” One could simply view the explanatory results of cognitive science as the sum total all of explanations in cognitive scientific disciplines (e.g. linguistics, neuroscience etc.); where inter-disciplinarity is defined in terms of “theoretical and experimental convergence on conclusions about the nature of mind” (Thagard, 2005). From this perspective, the “explanatory results of cognitive science” would turn out to be the set of all explanations put forward by practitioners of the disciplines mentioned above. This characterisation, as rough and ready as it is, may be the best we can hope for given the diversity of disciplines tasked with explaining the mind.

But even if we endorse such a characterisation a further problem cannot be so easily dismissed; namely, that cognitive science is replete with many different kinds of competing explanations aligning with many different perspectives on the mind as an explanandum. For example, we have mechanistic, psychological, and dynamicist explanations aligning with perspectives of the mind as mechanistic, functional, or dynamic respectively. Advocates of different kinds of explanations often challenge the reality of the explanatory postulates and explanatory force of the other kinds of explanation. For instance, some have argued that “psychological explanations” do not explain at all, but only provide “sketches” of neurally-implemented mechanisms (cf. Piccinini and Craver, 2011).

In chapters four and five, I consider whether or not the tension between these different kinds of explanations can be resolved. If it cannot, then it seems difficult to make sense of how, exactly, we are to bring empirical research to bear on debates about theories of CONCEPT. But even if it can, it remains unclear what effect an appeal to cognitive science will have on long-standing ideas about how we are to theorise about CONCEPT; for instance, the standard view. The appeal to cognitive science, therefore, is an oft promoted, but little understood, methodology in debates about concepts. My central aim in this thesis is to closely examine the commitments and implications of the appeal to cognitive science, and to consider how—if at all—it supports a naturalistic approach to theorisation about CONCEPT.

1.3 Overview of the Book

In order to make progress, I first undertake a negative project, which undermines both the standard view and the appeal to cognitive science as it has been implemented thus far. In chapter two, I argue that the relative satisfaction of desiderata (1)−(4) cannot help us to decide between different theories of CONCEPT adhering to the standard view, because there are a number of different, competing theories of CONCEPT that satisfy different desiderata respectively. More specifically, I consider the imagist, definitionist, prototype, exemplar, and theory theories of CONCEPT, and conclude that none of these theories of CONCEPT adequately satisfy all or even most of the desiderata (1)−(4). The upshot is that none of these theories of CONCEPT is to be preferred to any other.

In chapter three, I argue that an appeal to cognitive science to help us decide between the theories of CONCEPT brings into question the standard view itself, because different theories of CONCEPT, satisfying different desiderata, are required to do the explanatory work in cognitive science. I then consider three “new” theories of CONCEPT, which respond to this “explanatory challenge to the standard view”: concept eliminativism, concept pluralism, and concept hybridism. However, I argue that we cannot appeal to cognitive science to decide between these “new” theories of CONCEPT either, because each theory presupposes different specifications of the explananda of cognitive science and so affords the kind CONCEPT a different explanatory roles. I call this the “meta-explanatory challenge to new theories of CONCEPT.”

In chapter four, I consider the first of two problems for any attempt to overcome the “meta-explanatory challenge to new theories of CONCEPT”: the ambiguity about what counts as a cognitive scientific explanation. I consider three different kinds of explanations in cognitive science: mechanistic explanations, psychological explanations, and dynamicist explanations. Then, I examine long-standing disagreements about whether certain explanations have autonomy (e.g. psychological explanations) or whether we can find a reductive explanatory format that does all the explanatory work required of cognitive science (e.g. mechanistic explanations). This leaves open the question of whether or not “new” theories of CONCEPT agree about the explanatory relevance of different kinds of cognitive scientific explanation.

In chapter five, I consider the second problem for any attempt to respond to the “meta-explanatory challenge to new theories of CONCEPT”: the fact that different specifications of the explananda of cognitive science bias one in favour of different cognitive scientific explanations. Connecting to my claims in chapter four, I argue that at least two different kinds of explanatory integration can be identified in cognitive science: integrations of mechanistic explanations and cross-explanatory integrations of mechanistic, dynamicist, and psychological explanation. I go on to argue that one’s choice between these two kinds of explanatory integration depends on the attitude one has about the explananda of cognitive science. By then arguing that different “new” theories of CONCEPT will endorse different kinds of explanatory integration in cognitive science, I argue that the meta-explanatory challenge is sealed.

As a result of my arguments in chapter two to five, one may think that theorisation about CONCEPT should be cut off from cognitive science altogether. In chapter six, I resist this conclusion by developing a radically different view on what theorisation about CONCEPT entails. On my account, theorisation about CONCEPT should be thought of a process “internal” to, as opposed to “external” to, cognitive science, which establishes working hypotheses about what concepts are in order to facilitate the investigation and explanation of cognitive competences. From this perspective, it is not necessary or possible to appeal to cognitive science in the traditional sense to decide on the best theory of CONCEPT, because theorisation about CONCEPT is re-conceived as a constitutive part of the diverse explanatory practices of different cognitive sciences. However, it remains possible to appeal to the relative success or failure of many diffe...

Table of contents

• Title Page
• Copyright
• Contents
• Acknowledgements
• List of Figures
• List of Tables
• 1 Introduction
• 2 Standard View Theories of CONCEPT
• 3 The Appeal to Cognitive Science
• 4 Problem 1: Explanatory Ambiguity
• 5 Problem 2: Explananda Ambiguity
• 6 CONCEPT as a Working Hypothesis
• 7 Why Appeals to Cognitive Science Fail
• 8 Appendix