| Cognitive Science
as the Computational Neuroethology of Intelligent Behavior: Why biological facts are important for explaining intelligent behavior Brian L. Keeley |
| DISSERTATION
ABSTRACT Are the details and facts of biology important to Cognitive Science (CogSci)? Whether biological facts, for example, neurobiological facts, are important to CogSci is a controversial issue. On one side of this debate are those (e.g., Fodor, Putnam, Pylyshyn) who argue that psychology is independent of neural considerations. They contend that arguments for functionalism in the philosophy of mind show that higher level, "special" sciences, such as psychology, are independent of implementation level sciences, such as neurobiology. On the other side of this debate are those (e.g., the Churchlands, Andy Clark, Lycan) who argue that psychology and neurobiology must coevolve; these sciences proceed through a process of mutual epistemic interaction. They argue that psychology should be open to revision on the basis of neurobiological discoveries, and vice versa. This thesis takes the second side of this debate. I defend a philosophical position--taken from a short discussion by Andy Clark--called microfunctionalism. Microfunctionalism attempts to show how multiple levels of science can interact in a mutually beneficial fashion. Microfunctionalism is a species of traditional, philosophy of mind functionalism, but one that avoids traditional functionalist conclusions concerning the autonomy of higher level sciences. In its strongest form, microfunctionalism holds that sciences at all levels--psychological, social, neural--are necessary for proper theorizing within CogSci, but that none on its own is sufficient. Multiple levels of inquiry produce a large breadth of evidence that acts in a mutually constraining fashion. I augment the philosophical discussion and arguments for microfunctionalism with concrete examples and lessons drawn from the science of neuroethology. Neuroethology is the subfield of biology concerned with the neural basis of naturally occurring animal behavior. Paradigmatic neuroethological models include echolocation in bats, vocalization in primates, and the electrical sensory modality of certain teleost fish and sharks. Neuroethology represents a highly successful approach to understanding animal behavior that makes simultaneous use of data generated at a variety of levels of investigation. I show that microfunctionalism is consistent with the kinds of interdisciplinary interactions that occur within neuroethology. |
| CHAPTER ONE ABSTRACT Introduction: The Advantage of Two Hands Chapter One has no abstract. See text of Chapter One for details. |
| CHAPTER TWO ABSTRACT Comparison and Cognition This chapter defends the reconstrual of cognitive science in terms of another, structurally similar field: computational neuroethology. Whereas cognitive science combines psychology, neuroscience, linguistics, artificial intelligence, and philosophy for the joint study of cognitive behavior (primarily in humans), computational neuroethology combines ethology, neuroscience, evolutionary biology, philosophy, and computational modeling for the joint study of behavior (primarily in non-humans). After arguing for the value of a comparative approach to cognitive science (the most significant difference between the approaches of traditional cognitive science and neuroethology), I identify six principles of a neuroethological approach to a science of the mind. Finally, I turn to a relatively new approach to cognitive science, the new robotics primarily associated with Rodney Brooks, and show how this area of research embodies the principles of a neuroethological approach to cognitive science. The philosophical framework discussed here is then used in defense of this new approach. |
| CHAPTER THREE
ABSTRACT Fixing Content and Function in Neurobiological Systems: The neuroethology of electroreception Are attributions of content and function determinate, or is there no fact of the matter to be fixed? Daniel Dennett has argued in favor of indeterminacy and concludes that, in practice, content and function cannot be fixed. The discovery of an electrical modality in vertebrates offers one concrete instance of attributions of function and content that are supported by a strong scientific consensus. A century ago, electroreception was unimagined, whereas today it is widely believed that many species of bony fish, amphibians, sharks, skates, and rays possess this non-human sensory modality. A look at the history of science related to this iscovery reveals a highly interdisciplinary endeavor, encompassing ethology, behavioral analysis, neuroscience, and evolutionary biology. While each area provides important evidence, each is insufficient on its own to fix content and function. Instead, it is argued that an interdisciplinary, neuroethological approach is required to carry out such a determination. Further, a detailed look at biological research suggests that while content and function claims are empirically underdetermined and uncertain, there is insufficient reason to believe that there is an additional problem resulting from indeterminism. In particular, Dennett's indeterminism arises from a research methodology--logical adaptationism--that generates evidence from just one of the areas of neuroethology, and that logical adaptationism does not reflect adaptationism as it practiced in contemporary biology. I conclude that Dennett is faced with a dilemma: On the one hand, he can hold to logical adaptationism and the indeterminism that results from it, while giving up the relevance of his arguments to biological practice. On the other, he can embrace a more accurate version of adaptationism--one which plays a role in a larger neuroethological framework--but from which no strong indeterminacy claims follow. |
| CHAPTER FOUR
ABSTRACT A Scientific Intermission: Large, slow changes of electric organ discharge frequency in the weakly electric fish, Eigenmannia virescens According to previous reports on the electric organ discharge (EOD) of the weakly electric fish, Eigenmannia virescens, it is maintained at the frequency characteristic of the individual (range: 250-500 Hz for the species) for hours and days within >1 Hz, except for the jamming avoidance response and brief interruptions that act as a communication signal in certain social contexts. Individuals with a difference in EOD frequency of 2-10 Hz will shift their EOD frequencies 10-20 Hz away from one another in a behavior known as the jamming avoidance response (JAR) in a matter seconds. EOD frequencies also change during the course of maturation. The work reported here focuses on changes that occur on the time-scale of days to months. On these times scales, individual Eigenmannia virescens can either maintain a very steady frequency (changing less than 10% over the period of several months) or can change their EOD frequency (more than 80%--100+ Hz--in less than a week. These EOD frequency changes occur as a function of social context, even when the social interaction occurs among fish whose frequency differences were significantly larger than needed to elicit a JAR. Further, it was observed that this effect is larger in females than in males. This evidence demonstrates that social context has an effect on the distribution of EOD frequencies in a population of fish, and may be interpreted as evidence for fitness advertisement in female Eigenmannia virescens, as the higher the frequency maintained by an individual the higher the metabolic load on that individual. |
| CHAPTER FIVE
ABSTRACT Dogs, Dogfish, and Dretske's Externalist Theory of Experience In Fred Dretske's Naturalizing the Mind, an Externalist theory of conscious experience-- particularly, of qualia--is defended as part of a naturalized, representational theory of mind. Dretske defends Representational Naturalism, the position that the mind, including the phenomenon of experience, is best understood in representational terms. On an Externalist account of experience, the nature of this phenomenon is in part determined by facts that lie outside of the mind/brain. On Dretske's particular account, what it is to experience something in a particular way is in part determined by what that experience has been selected to represent. After setting out a significant portion of Dretske's representational theory of mind, I will focus on Dretske's argument for Externalism with respect to experience. He argues that once one accepts Externalism with respect to mental content, one is left without resources for rejecting Externalism with respect to experience. I argue that upon further reflection, Internalists with respect to experience are not as defenseless as Dretske would have us believe. In particular, Dretske overlooks the important consequence of his own theory that internal facts, e.g., neurobiological facts, are necessary for understanding the mind. |
| CHAPTER SIX ABSTRACT Historical Considerations There is a tension within the philosophy of Cognitive Science (CogSci) between those who argue that the facts of biology are largely irrelevant to the study of cognition and those who stress its ineliminable role. One goal of this dissertation is to dissolve this tension and come to an understanding of how and why biology is important to CogSci. As an introduction to the issue, in this chapter, I take a look at the issue's history. I concentrate primarily on the early debate between Logical Behaviorism, Identity Theory, and Functionalism--a debate commonly seen as having been won by Functionalism. I offer a explanations for the failure and continued contemporary unpopularity of Logical Behaviorism and Identity Theory. I then identify some problems with traditional Functionalism, particularly those associated with the phenomenon of multiple realization. This conceptual geography of philosophical positions then allows me to place the central doctrine to be argued in this dissertation--Microfunctionalism-- in context. Microfunctionalism shares points with each of the other three positions, but also rejects key claims of each. |
| CHAPTER SEVEN
ABSTRACT Microfunctionalism I defend a theory of the relationship between levels of scientific explanation, called "microfunctionalism" (MF). Taken from a short discussion in Andy Clark's Microcognition, MF can be characterized roughly as traditional functionalism without strong claims for the disunity of science entailed by Fodor's multiple realizibility thesis (MRT). This chapter lays out and defends four principles, which I argue constitute MF. First, it is argued that all physicalist explanation is functional in nature; the theoretical entities posited at a given level of explanation are treated dispositionally, i.e., as "black boxes". Second, specific levels of explanation are identified by the presence of lawlike regularities among the theoretical entities posited at that level. Third, the theoretical content of a levelthe particular taxonomy posited to explain the patterns discernible at that levelmust cohere with the taxonomies of other levels. Fourth, whereas functional entities are often multiply realizable, such sets of realizing entities either 1) constitute natural kinds, or 2) can be subsumed under a teleological mechanism that itself can be explained in uniform physical terms. This positive characterization of MF is drawn from a concrete scientific example: the neuroethology of electric fish. The example is discussed in detail and provides a basis for the theoretical framework of MF. MF suggests that scientific explanation is unified in that all explanation is either physicalist or requires an explanation of how higher level phenomena "emerge" out of lower level phenomena. Higher level sciences are not fully autonomous in the ways which some (e.g., Fodor) claim, but still describe legitimate phenomena not directly reducible to descriptions at lower levels. |
Last Modified: 24 March, 2006.