Current Controversies in Philosophy of Cognitive Science

Adam J. Lerner, Simon Cullen, Sarah-Jane Leslie, Adam J. Lerner, Simon Cullen, Sarah-Jane Leslie

• Publisher: Routledge
• Year: 2020
• eBook ISBN: 9781000063127
• Edition: 1
• Topic: Psychology
• Subtopic: Mind & Body in Philosophy
• Index: Psychology

Cognitive science is the study of minds and mental processes. Psychology, neuroscience, computer science, and philosophy, among other subdisciplines, contribute to this study. In this volume, leading researchers debate five core questions in the philosophy of cognitive science:

• 
  
  

• Is an innate Universal Grammar required to explain our linguistic capacities?

• 
  
  

• Are concepts innate or learned?

• 
  
  

• What role do our bodies play in cognition?

• 
  
  

• Can neuroscience help us understand the mind?

• 
  
  

• Can cognitive science help us understand human morality?

For each topic, the volume provides two essays, each advocating for an opposing approach. The editors provide study questions and suggested readings for each topic, helping to make the volume accessible to readers who are new to the debates.

Part I

Is There a Universal Grammar?

1Universal Grammar

Paul Pietroski and Norbert Hornstein

Pity the linguist. By tradition, the job is to describe languages in ways that extend the ancient practice of providing grammars (e.g., for Sanskrit or Latin). Modern linguists are also expected to be cognitive scientists who study some aspect of human psychology that supports the acquisition and use of languages that children readily attain. Given the tradition, linguists inherit a lot of terminology—e.g., ‘sentence’, ‘subject’, ‘word’, ‘language’, and ‘grammar’—that cannot be discreetly consigned to the attic. So old terms often get used in new ways. As a result, ‘Universal Grammar’ has meant different things for different people at different times. But the phrase does focus attention on a good idea: The diverse languages that children can naturally acquire exhibit shared features that reflect traits shared by all members of our species.

1. Laying a New Picture on an Old One

Let’s start with the obvious fact that nothing does language like humans do language. Other animals communicate. But we talk—a lot. Our distinctive loquacity is manifested along many dimensions, two of which are especially important here. First, humans are linguistically creative, in that we routinely produce and understand expressions that we have never before produced or encountered. Indeed, there seems to be no upper bound on the number of expressions that a native speaker can linguistically manage. Second, the languages that humans can naturally acquire are universally available to children. To be sure, experience matters. Native speakers are only adventitiously proficient in the languages they happen to have acquired. But there are no subgroups of humans who can only acquire languages from certain families (e.g., Indo-European). Given the right experience, any ordinary child can become a speaker of any of the languages that other children can acquire.

These facts are not subtle, and they call for explanation. Moreover, the general form of the required explanation is relatively clear. To account for linguistic creativity, we assume that a native speaker of a language has a grammar for that language. Grammars are procedures that generate expressions in certain ways. (We return to some details below.) To account for universal availability, we assume that humans have a mental capacity to acquire grammars of a certain kind, given exposure to speech that is symptomatic of those grammars. This human capacity is often called the Faculty of Language (FL). The input that FL uses, in producing particular grammars, is often called the primary linguistic data (PLD).

Once these unsubtle facts are noted, it shouldn’t be controversial that children acquire grammars by employing a shared FL, which yields various grammars given various courses of PLD. The details are and should be controversial. What kinds of rules characterize the grammars in question? What is the structure of FL? How does FL use PLD to generate any particular grammar? How much of FL is specific to language, and how much is common to other cognitive capacities? These are all legitimate (and hard) questions. But they all presuppose some FL that generates grammars in response to PLD. While it is truistic that FL exists, it takes research to find out what’s “in” FL, the PLD, and the grammars that children acquire.

In characterizing the topic this way, we follow Chomsky (1965). On this view, the subject matter of linguistics is doubly psychological: Native speakers have grammars that generate expressions; children acquire these procedures, given their idiosyncratic experience. The goal for linguists is to correctly describe the procedures and the underlying capacity to acquire them. The grammars that a linguist might inscribe on paper are taken to be models of hypothesized expression-generators, and language acquisition is viewed as a process of acquiring an expression-generator given a course of experience and a certain range of options. This suggests a picture in which various grammars can be described with a common vocabulary, making it possible to abstract a common core that can be supplemented in various ways (see Figure 1.1).

Figure 1.1Many linguists view language acquisition as a process of acquiring an expression-generator given a course of experience and a certain range of options. Thus, various grammars can be described with a common vocabulary, making it possible to abstract a common core that can be supplemented in various ways.

This picture presupposes that the child-acquirable languages are importantly alike—despite their manifest diversity—and that many universal features of the grammars that generate these languages reflect species-general features of children. This suggests a second picture, according to which internalized grammars reflect the impact of experience on the cognitive systems that support language acquisition (see Figure 1.2).

Figure 1.2Internalized grammars reflect the impact of experience on the cognitive systems that support language acquisition.

Here the idea is that children share a capacity to acquire languages of a certain sort, in response to suitable courses of experience, and acquiring a particular language is a matter of acquiring a mental state that reflects both the shared capacity and relevant experience.

Chomsky (1965, 1981, 1986) offered proposals that combined the two pictures: Acquiring a particular grammar/procedure is largely a matter of using experience to choose from a menu of ranked options, or perhaps fill in a common template for grammars, which exhibit similarities that reflect their shared representational format.¹ Linguists then face the task of capturing the similarities in terms that make it possible to describe the respects in which acquirable grammars vary. This sets the stage for investigating how particular grammars are related to the common core.

2. Unbounded but Constrained

Let’s use ‘language’ generously, allowing for talk of invented mathematical languages and the language of bee dance, along with languages like spoken French and American Sign Language (ASL). We can count something as a language if it connects signals of some kind with interpretations of some kind. Let’s use ‘Human Language’ to talk about the languages that a typical human child can acquire given an ordinary course of growth and experience. These languages connect signals of a special sort—e.g., sounds of spoken French, or gestures of ASL—with interpretations of some kind.

As noted above, Human Languages have some interesting properties that seem to reflect interesting properties of children, who can understand and use strings of words they never previously encountered. In the house that Jack built, there may be a dog that chased a cat that chased a rat, which ate some cheese and thereby led Jack’s partner to take drastic action; and so on. Some sentences are too long to say or hear. But children do not acquire languages in a way that imposes an upper bound on how many words can appear in an expression. Counting provides an obvious analogy. The procedure of “adding one” outruns our mortal ability to keep applying it. Similarly, a procedure can connect boundlessly many pronunciations with boundlessly many meanings in a rule-governed way that outruns our limited abilities to process and produce speech. And a child can acquire a procedure that generates pronunciation-meaning pairs (π-μ pairs) in an open-ended way.²

For many purposes, one can ignore the small differences among the many English procedures, and talk as if nearly a billion individuals share the same language. The more important point is that if Human Languages are generative procedures, then a proposed grammar “for English” is a hypothesis about how certain internalized procedures generate certain π-μ pairs in an open-ended way, where this way of generating π-μ pairs is one of the ways supported by the Human Language Acquisition Device, depicted above. But even if each Human Language has a lexicon of atomic expressions and at least one combinatorial principle that is somehow recursive, this is not yet a big insight. The interesting questions concern the kinds of atomic expressions and kinds of combinatorial principles that are permitted—and within those kinds, the range of variation that is possible.

Chomsky (1957) sharpened these questions by describing three kinds of computational systems that can generate boundlessly many expressions, and arguing that Human Languages generate their expressions in some other way; see Lasnik (1999) for a lucid review. This highlighted a pair of fru...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Table of Contents
• Contributors
• Introduction
• Part I Is There a Universal Grammar?
• Part II Are All Concepts Learned?
• Part III What Is the Role of the Body in Cognition?
• Part IV How Should Neuroscience Inform the Study of Cognition?
• Part V What Can Cognitive Science Teach Us About Ethics?
• Index