The Neuron and the Mind

Microneuronal Theory and Practice in Cognitive Neuroscience

William R. Uttal

• Publisher: Routledge
• Year: 2016
• ISBN:
  9781317208082
• Edition: 1
• Topic: Psychology
• Subtopic: Cognitive Neuroscience & Neuropsychology
• Index: Psychology

This book, a companion to William R. Uttal's earlier work on macrotheories theories of mind-brain relationships, reviews another set of theories—those based on microneuronal measurements. Microneural theories maintain the integrity of individual neurons either in isolation or as participants in the great neuronal networks that make up the physical brain. Despite an almost universal acceptance by cognitive neuroscientists that the intangible mind must, in some way, be encoded by network states, Uttal shows that the problem of how the transformation occurs is not yet supported by empirical research findings at the micro as well as at the macro levels of analysis. Theories of the neuronal network survive more as metaphors than as robust explanations. This book also places special emphasis on the technological developments that stimulate these metaphors. A major conclusion drawn in this book is that it is not at all certain that the mind-brain problem is solvable in the sense that many other grand scientific problems are.

1
Introduction¹

1.1 Explanation

Of all of the scientific mysteries confronting our inquisitive species, none is more profound or challenging than understanding how the tangible brain can give rise to intangible thought. This has traditionally been known as the mind-body problem but, as we have become more sophisticated, referring to it as the mind–brain problem seems more appropriate. Despite the enormous amount of research being conducted that deals with the brain and the mind, it is almost universally agreed that we actually have made little progress toward resolving the overarching nature of this relationship. In place of creating anything approximating the grand theory of cosmological physics, we have learned an enormous amount about each domain separately but, unfortunately, piecemeal. Fortunately, knowledge about the anatomy and physiology of the brain is succumbing to powerful new research techniques, and psychologists have learned much about the ebb and flow of mental activity by observing behavioral responses.

What we have not done is to have made any substantial progress toward building a conceptual bridge between the psychological and neurophysiological domains. The reasons that we have not are multiple and complex; as a result investigators from many fields have debated whether the problem is in principle “solvable” or “tractable.” On the one side are the optimists. To them, to “explain” how the brain makes the mind is just another scientific task, admittedly quite complicated, but still achievable within the rubric of conventional scientific inquiry. However difficult the problems may be, the obstacles to understanding them are just practical matters awaiting the development of a new method or technology. Most cognitive neuroscientists implicitly accept the ultimate achievability of the task and assume that however slow it may be in coming, eventually we will understand or explain how the brain makes the mind. To think otherwise would raise serious questions about the raison d’etre of cognitive neuroscience itself.

On the other side of the debate are the pessimists (or, if you prefer, the realists); their countervailing opinion argues that the problem is not just difficult in practice but is impossible in principle. That is, the complexity of the problem is such that there is no strategy or tool that we could adopt or invent that would ever lead to an analysis that would “explain” how the process of converting brain activity to mental activity takes place. Philosophers, combinatorial mathematicians, some behaviorally oriented psychologists, and even a few of those who would happily accept the title Cognitive Neuroscientist are now beginning to accept, implicitly if not explicitly, at least the possibility of impossibility.

The resolution of this dispute, of course, will ultimately come from empirical studies. Someday we may solve the problem and that will be that—debate over. Nothing proves the existence of a dimly seen phantom like Bigfoot better than robust evidence of its existence although absence of such evidence does not prove it does not exist. In this case, the “existence proof” for an overarching neuroreductionist theory would be the demonstration of our ability to construct high-level behavior or intelligence or sentience or cognition (or whatever it is that we may wish to call mind) from the properties of low-level neurophysiological components. Of course, even then the uncertainty of what amounts to an acceptable explanation would remain controversial.

This debate sets the stage for this book. In it, I strive to review and evaluate the current state of a particular kind of mind–brain theory—the kind that is based on the microscopic, cellular components (i.e., neurons) of the nervous system. These theories are designated as microneuronal as distinguished from those that are based on larger chunks of the brain—those designated as macroneural. I have already dealt with macroneural theories in a previous book (Uttal, 2016), to which this book is a companion, and will only deal with them casually in passing in this chapter.

To start, it would be useful to define exactly what we mean by a theory. For a number of reasons, however, categorizing existing theories is much more arbitrary than it may at first seem. Theories come in many kinds, each of which may emphasize different foundation axioms and postulates. A rough taxonomy of the major kinds of cognitive and cognitive neuroscience theories would include the following:

• Behavioral or Descriptive Theories: Theories of this kind are not meant to be reductive. That is, there is no effort to identify the cognitive or neurophysiological component parts that make up an observable behavior pattern. A pure behavioral theory, quite to the contrary, is only descriptive. It may be based on a mathematical formulation that allows us to predict the trajectory of a behavior or to represent the transformation between stimuli and responses, but there is no attempt to delve into lower-level cognitive or neurophysiological processes. Any en passant allusions to neurophysiological or inferred cognitive components are at best rough metaphors, in the absence of additional data. The mathematics itself, like behavior in a descriptive theory, is neutral with regard to specific underlying mechanisms—cognitive or neural. Thus, no matter how accurate are the predictions made by a mathematical model, unless there is additional data, it is not possible to produce a unique solution to the mind–brain problem from the formulation. There are always many alternative models that are sufficient to describe the behavior. Another way to say this is that behavioral and mathematical descriptions are underdetermined.
• Cognitive Reductive Theories: Theories of this kind are intended to be reductive but not to neurophysiological mechanisms. The underlying components are inferences implied by the behavioral data. Many theoreticians of this ilk hold the assumption that behavior is transparent (i.e., can be inferred from) the underlying cognitive process and that by careful experimental designs we should be able to parse out these inferred cognitive components. These cognitive components, modules, or faculties components have been designated as “hypothetical constructs” (MacCorquodale and Meehl, 1948), emphasizing the intrinsic difficulty of exactly defining them—a major handicap of such “top-down” approaches in psychological research. Again, these hypothetical constructs may allude in passing or by analogy to neural structures and processes, but it is impossible in principle for this approach to rigorously designate what these inferred components might be.
• Neuroreductive Theories: Because overt behavior and covert cognition are not able to lead us deductively to the underlying brain components that are necessary for mind, and because there is widespread agreement that all cognitive processes are in some ultimate sense brain processes, the search for the neural mechanisms of mental activity has motivated a vast amount of research. With the development of modern neurophysiological and neuro-anatomical techniques, the search for the brain mechanisms of mind has greatly accelerated. There are two main threads of this kind of research, which has been alternatively known as physiological psychology, psychobiology, and most recently, cognitive neuroscience. These threads are respectively known as macro-neuroreductionism and micro-neuroreductionism.
  • Macro-neuroreductionism is based on data in which the cumulative, pooled, or summed actions of many individual neurons are emphasized. The details of what individual neurons are doing are lost since their neuroelectrical and neurochemical responses are additively lumped together in the eyes of the investigator. This kind of data provides a powerful impulse to develop theories that assume a kind of gross localization of the component neural processes on or in the brain. It also stimulates ideas about specialized cognitive roles for these macro-regions of the brain, that is, phrenological and neophrenological associations of places and cognitive phenomena.
  • Micro-neuroreductionism is based on data obtained from experiments in which the activity and measurement of individual neurons is maintained either individually or in their participatory role as components of a neuronal network. There are three levels of micro-neuroreductionism within this part of this mini-taxonomy: (1) The action of single neurons has been associated with cognitive functions; (2) the action of relatively small, but computationally tractable, networks of a few neurons has been used as the basis of an intermediate level of theorizing; and (3) with the advent of supercomputers that can simulate the individual activity of billions of neurons has come a new form of micro-neuroreductionism approach in which the number of neurons involved in a theory can approximate the number of neurons in the brain. The ultimate form of a mind–brain theory, most cognitive neuroscientists probably agree, would ideally be found in the properties of such a network; however, it is still problematic whether we will be able to manipulate them in the way needed to “solve” the mind–brain problem.

A distinction should also be made between top-down and bottom-up theories in the context of this mini-taxonomy of theoretical types. The goal of top-down theories is to measure events at higher levels and then to determine by various methods the nature of the underlying neural and cognitive components that might embody the higher-level processes. The goal is, thus, to determine the underlying processes by drawing inferences from experiments on higher-level processes. This is the essence of top-down, analytical neuroreductionism. The underlying mechanisms are the unknown targets of this strategic approach.

Bottom-up theories have a different strategy; they attempt to synthesize higher-level systems from preexisting knowledge of the underlying components. This is essentially a synthetic approach to theory building. A classical example of the bottom-up approach to theory development was Watson and Crick’s (1953) reconstruction of the genetic code for DNA from what was then known about the bonds between molecular structures. Their theory was embodied in a mechanical structure that depended on laws of chemical bonding in a way that opened up a new world in biological research.

Although both top-down and bottom-up methods seek to achieve the same goal—explain higher-level processes by lower-level ones—there is a fundamental difference in their basic approach. The analytical top-down method dotes on parsing a molar process into its components, whereas the synthetic bottom-up method aims to solve the problem by concatenating known lower-level processes. Both are susceptible to errors, of course. Top-down analytical methods are prone to mistake the sufficient for the necessary. Bottom-up synthetic methods are often more robust but suffer from incomplete knowledge of how the components may be interconnected.

All theories represent hopes for the future; however, each type of theory is constrained in what it can ultimately accomplish. Not only are there differences in the level of anatomical complexity at which a theory must operate but also there are major differences in what relationships should be measured in an experiment. Thus, the exact theoretical relation between a cognitive process and a neural response is not always clear even in the most highly correlated data.

Theories also differ in their methodological origins. A “theory” may be built on any one of the following relations:

• Statistical correlation between cognitive and neural responses
• The location of a neural mechanism of a cognitive process
• A mathematical description
• A physical model or simulation
• Reproduction by reconstruction, a rough verbal statement of some perceived metaphorical relationship
• The psychoneural equivalent of a cognitive process

The psychoneural equivalent is the actual neural mechanism whose activities are supposed to be indistinguishable from the cognitive process itself. In other words, the psychoneural equivalent is the neuronal embodiment of the cognitive process. Although defined in the different languages of cognition and neurophysiology, the different words may denote exactly the same thing. Determining the nature of the psychoneural equivalent is an extremely challenging task facing many impediments and obstacles; nevertheless, it is the holy grail of cognitive neuroscience. The most powerful interpretation—literal equivalence—in all of their properties—is in the manner expressed by “identity” theorists such as Place (1956) and Feigl (1958). In identity theory, cognitive and neural activities have the strongest possible relationship—the one is the other.

The range of theoretical issues with which I have dealt in my previous work on neuroreductionism can be further clarified by the following list of assumptions, properties, postulates, restrictions, and if you wish, biases that have guided my analysis:

1. Ontologically Physicalist: The prime postulate of modern cognitive neuroscience is the physical or material origin of mental processes. That is, whatever mind is, it is a manifestation or product of laws of the singular reality that accounts for all other events, things, and processes in our real world. In other words, cognitive neuroscience is fundamentally monist; it admits of no other separable and distinguishable reality of the kind proposed by some philosophers. To accept any other (some kind of dualism) of a distinguishable reality difference between the mental and the physical would totally invalidate the science; it would require that we accept the possibility of forces that are not controlled by our experiments influencing those experiments. Such experiments, therefore, would invalidate the entire corpus of research in cognitive neuroscience.
2. Epistemologically Indeterminate: Despite the basic materialist ontological orientation of modern cognitive neuroscience, others believe that there are practical or epistemological constraints on what an empirical science can accomplish in the study of mind–brain relations. Some of these constraints have to do with the complexity of the system; some have to do with the inaccessibility of mental activity; and others arise from interpretive and logical errors. As a result of these practical problems, many proposed theories are little more than rough metaphors or prototheories rather than full-blown explanations. Thus, we may be in a situation in which despite being ontological monists, we are simultaneously doomed to be epistemological dualists and must study mind and brain separately.
3. Reductive: A reductive theory is based on the assumption that the properties of the lower level can, if appropriately manipulated, produce the properties of the higher level by concatenation. Nothing supernatural or additional happens between the two levels, just a rational flow of the processes and causes leading from one level to the other. The word “emergence” (by which is meant that new properties that have their origins in the lower level without causal connections or forces) is often used to denote the process but its use usually turns out to be a way of finessing the unknown without actually explaining anything. The radical reductionist, to the contrary, assumes that everything that is expressed at the higher level is at least implicit at the lower level. The task of cognitive neuroscience is to make explicit those properties that are implicit.
4. Bottom-Up Possibility: Bottom-up theories differ greatly from top-down ones. It is possible, in principle, knowing the properties and the rules of interaction (even if not practicable for reasons to be discussed later) to construct a bottom-up theory in which one reasons from the lower-level neurons to higher-level properties and thus to develop a “necessary,” full, complete, and uniquely accurate theoretical explanation. An example of a bottom-up theory is how the human visual sensitivity to light of different wavelengths is accounted for by the differential absorption of photons of different wavelengths by the retinal receptor photochemicals. Another example of a bottom-up theory is the suggestion that Mach Bands (edge enhancements in visual perception) are accounted for by mutual lateral inhibitory interaction among retinal neurons. Another example of a successful scientific synthesis from a totally different domain would explain how the microscopic structures of atoms account for the macroscopic behavior of chemicals.
5. Top-Down Impossibility: However, no matter how empirically robust, full, complete, and predictively accurate theoretical explanations they may be, top-down, neuroreductionist theories are not in principle robust. There are too many alternative, plausible, and possible “sufficient” explanations that can be generated from any given set of data. There is no way in which the underdetermined data of the macrocosm can be transformed into a unique “necessary” neuron-level explanatory theory. Top-down theories are useful for testing the plausibility of a prototheory but cannot distinguish between the unique “necessary” explanation and the array of “sufficient” ones that all fit the data equally well.
6. Neuronally Reductive: For the purposes of this book, the highest level of analysis is that of cognitively related behavior and the lowest level with which I deal is that of the specialized cells of the nervous system—the neurons. This instantiates a very particular postulate of the kind of neuroreductionist analysis dealt with here—namely, it is assumed, both explicitly and implicitly, throughout this discussion that any future neuroreductionist explanation of cognition is going to be framed in terms of individual neurons and their interactions at a microscopic level. It is further assumed that the identity and measures of the activity of individual neurons is necessarily preserved in any theory. This approach is distinguished from models that dote on macroneural “chunks” of the brain—a topic on which I have written extensively (e.g., Uttal, 2016)—and that have increasingly been rejected as a strategy that will lead us to the solution of the mind–brain problem. Other theoretical ...

Table of contents

• Cover
• Title
• Copyright
• Dedication
• CONTENTS
• Preface
• Acknowledgments
• 1 Introduction
• 2 Single Neuron Practice and Theory
• 3 Microneuronal Network Theories—Technique and Methods
• 4 Intermediate Level Neuronal Networks—The Role of Models
• 5 Large-Scale Computer Simulations (Theories) of Cognition
• 6 Emerging Conclusions
• Bibliography
• Index