Psychomythics

Sources of Artifacts and Misconceptions in Scientific Psychology

William R. Uttal

• Publisher: Psychology Press
• Year: 2014
• ISBN:
  9781135623715
• Edition: 1
• Topic: Psychology
• Subtopic: History & Theory in Psychology
• Index: Psychology

Psychology deals with the most complex subject matter of any science. As such, it is subject to misunderstandings, artifacts, and just simple errors of data, logic, and interpretation. This book teases out the details of some of the sources of these errors. It considers errors in psychological data and theories that arise from confusing endogenous and exogenous causal forces in perceptual research, misinterpreting the effects of inevitable natural laws as psychological phenomena, improper application of statistics and measurement, and flawed assumptions. Examples of each of these sources of error are presented and discussed. Finally, the book concludes that a return to a revitalized kind of behaviorism is preferred, rather than continuing on the current cognitive path.

Chapter 1
Introduction

1.1 Purpose and Goals

Psychology, like any other natural science, is dedicated to the goal of analyzing, describing, and understanding the nature of a particular subject matter. Although it can legitimately be debated whether the subject matter of this science is externally observable behavior or inferred internal cognitive processes, we usually are able to identify a psychologist by what is being done in the lab or what is being written. That is, psychologists by experimenting, hypothesizing, theorizing, professing in oral or written form, and to be completely candid, often wildly speculating, seek to define, and to the extent possible, “explain” cognition.¹ There are many ways to explain something. One is to describe it thoroughly. Another is to hypothesize, infer, or speculate about what kind of internal, unobserved mechanism could account for an observed behavior. Another is to determine what are its primary causes. That is, to identify the properties of the environmental stimulus or of the organism that cause, trigger, or lead to a particular behavioral response. Over the years, many “phenomena” have been attributed to functional properties of the mind-brain that represent and transform incoming stimulus information.

However, in many instances and for many phenomena, it turns out that what we believe to be cognitive in origin is not and what we believe to be an adequate theory is totally incorrect, not just in its details, but also in terms of its most fundamental axioms and assumptions. The purpose of this book is to demonstrate that the variety of misinterpretations and misconceptions of the causes of psychological phenomena is substantial. That is, a large number of observations and findings are incorrectly attributed to information transformations and processes within the nervous system in place of the true antecedent causal conditions. Logical and conceptual errors of many different kinds provide the basis for the invention of psychomythical interpretations, theories, and sometimes deeply held convictions about cognitive activity. Expanding upon the definition presented in the preface, by psychomyths I am referring to the generally accepted, but erroneous, gamut of reductive and descriptive theories, metaphors, and conclusions that do not hold up under close scrutiny. This book examines this aspect of the conceptual foundations of psychological explanation to winnow out the valid from the fanciful.

1.2 A Proposed Taxonomy

Given that much of the discussion in this book deals with an interacting set of topics and issues, it is imperative that I make an explicit effort to categorize and organize the subject matter that is considered. This section develops a personal taxonomy of the sources of what I believe are some of the main driving forces behind psychomyths. At the outset, it is clear that such a classification system, of necessity, will be quite arbitrary and there are likely to be many overlapping and redundant categories. Nevertheless, I propose the following outline of the various categories of sources of psychomyths:

1. Confusion of the exogenous with the exogenous
2. Inevitable natural laws
3. Superpowerful mathematics
4. Self-organizing systems
5. Misconceptions about measurement
6. Miraculous graphs
7. Misleading statistics
8. Erroneous assumptions and conceptual errors
9. Nonillusions
10. Persistent mysteries

In the remainder of this section, the nature and significance of each of these categories is discussed.

The Endogenous-Exogenous Confusion. Chapter 2 begins the discussion of the origins of psychomyths by examining the role that the external environment plays in defining the information content of our perceptual experiences. I argue here that, like the rainbow, many other illusions are actually defined by transformations or events occurring long before the physical stimulus is coded into neural activity. Therefore, what has often been identified as a distortion or illusion produced by a neural or cognitive transformation of the information carried by a stimulus by the nervous system is, in fact, a veridical response to the stimulus.² I also emphasize that this is not a new debate but one with an immensely long history in which the controversy became incorrectly framed in the antagonistic terms of a conflict between rationalism and empiricism. As I argue, this false dichotomy has to be eventually resolved in favor of an eclectic compromise.

The point here is that some processes that produce particular perceptual phenomena are due to simple geometrical effects such as diminishment of retinal size with distance. Others, furthermore, are attributable to the inevitable outcome of much more complex natural processes. I argue that many “illusions” previously attributed to endogenous psychoneural transformations are actually caused by exogenous events occurring in the external environment.

Inevitable Natural Laws. Throughout the history of science there have been some relationships that are “facts of nature” that seem immutable and constant. The most obvious ones are to be found in the physical sciences where a relative organizational simplicity obtains. Newton’s laws of motion (e.g., force = mass x acceleration) have been replicated again and again and are generally accepted as universal laws of physical action at macroscopic scales of measurement. The precision with which we launched and controlled our spacecraft during the last four decades is clear evidence of their applicability and generality.

In contrast, there are relatively few such persistent and universal laws of psychological activity that hold across all individuals and situations; observations of human behavior are notoriously variable and especially sensitive to what may often seem to be random influences. In the large, however, some relationships that reflect an average description of the behavior of individuals have been observed sufficiently often to justify generalization to groups. Even within this limited range of psychological laws, there is considerable variability and, thus, limits on the generality of their application. Some, for example, the Law of Effect (i.e., reinforcement increases or produces learning) are not intended to be exact in the same sense that the quantitatively precise laws of physics are. Although this famous psychophysical principle (Thorndike, 1913) and a close modern corollary—Hebb’s (1949) principle of synaptic usage—express general tendencies, different situations, and different individuals could and presumably do apply the general law in different specific ways. Thus, the roles of effect or usage were and are probably intended to be qualitative generalizations that had to be fleshed out by specific experiments. Although these and related principles appear in quite different forms in a variety of contexts, no one expects them to be precise predictors of individual behavior.

In other, even subtler, cases, similar processes, subject to mathematical or statistical laws and solutions, occur regardless of the specific properties of the system being observed. For example, linguists are familiar with a relation called Zipf’s law (Zipf, 1935/1965), which asserts that the relation between the frequency (F) with which a word occurs and its rank order (r) in terms of the number of times it appears in a piece of English prose is closely approximated by an inverse power law. It was originally assumed that Zipf’s law reflected a psychobiological property of human linguistic activity. That is, that it was a putative law describing a property, attribute, or transformational characteristics of human linguistic skills and, therefore, of the mind-brain. Indeed, Zipf’s initial tome was actually entitled The Psycho-Biology of Languages (Zipf, 1935/1965).

However, there is an enormous complication to the simple suggestion that Zipf’s law is a valid psychological law or description of an attribute of human linguistic information process. Zipf’s law works as well for many other rank order-frequency relations as well as it does for language! There is, therefore, an emerging implication that however good the fit between the law and the behavior of some natural system, there is something more general implicit in it that transcends the particular aspects of human linguistic behavior. Topics of this kind are discussed in chapter 3 along with an examination of how superpowerful mathematics often lead to the error of attributing properties of the mathematics to the mind-brain.

Superpowerful Mathematics. Cutting (1986) summed it up exceedingly well when he said: “Mathematics is too powerful to provide constraints on information: it models truth and drivel with equal felicity” (p. xi). The problem is that, in point of fundamental fact, mathematics is neutral with regard to underlying processes. It is the tool par excellence of description but contrary to the view of many mathematical psychologists it cannot, for reasons of deepest principle, delve deeply into underlying structure. For example, some kinds of mathematics (e.g., Fourier analysis) are capable of providing analyses of complex processes in terms of simple fictional or hypothetical components. The output of a totally nonsinusoidal machine can be analyzed into sinusoidal components that have no underlying physical reality. The point is that mathematical methods have properties of their own and on occasion these properties have been inadvertently attributed to the system being described. In short, as Cutting highlighted, mathematics is too powerful to be uncritically applied. Chapter 3 also details how some kinds of mathematical models can give rise to psychomyths.

Self-Organizing Systems. Biology is replete with examples of self-organizing systems. For example, chemical affinities and repulsions between the lipid molecules that make up the membrane of a cell cause them to organize themselves into regular linear structures. Biological clocks of one kind or another are omnipresent in organic tissue. Other physical and chemical forces control the development of patterns of growth in embryological development. Even the growth of the coloring patterns of fish or of the pattern of sulci on the surface of the cerebral cortex are now known to be at least partially dependent on self-organizing properties once the basic genetic code has been expressed. Resonant effects in which a low level stimulus energy produces a high level of response as a result of a special sensitivity or specialized tuning to some aspect of the stimulus is another example of how a system’s self-organizing properties can lead inevitably to responses that are too easily misinterpreted as active cognitive transformations.

However, there is also a caveat to this theoretical approach. Such an argument parallels discussions of the emergence of macroproperties from the microproperties of a system. It is often asked: Can the whole be more than the sum of the parts? The only really plausible answer to this question is, No, if you include the rules of interaction between them in the property list of the components! No mysterious or supernatural emergence need be postulated. Individual parts of a complex system have definable properties that are only exhibited when they interact with their fellows. These properties are the ones that govern the rules of interaction and when included among the properties of the parts, in principle all is predictable.

There is an enormously important caveat embedded in this last assertion: What can be done in principle cannot always be done in practice. Simple numerousness precludes knowing the effect of all individual interactions, particularly in nonlinear systems in which there are a huge variety of feedbacks, feedforwards, and other kinds of contingent interactions. From this practical epistemological limitation arises totally unjustified ideas of mystical “emergence” that are supposedly outside the normal rules of science. Such subtle, complex, and inexplicable self-organizing interactions can sometimes be invoked without adding anything to our knowledge of how the mind-brain system is performing. This topic is also considered in chapter 3.

Misconceptions About Measurement There is perhaps no greater of psychomyths than those arising from misunderstanding about the role of measurement in psychology. Some new ideas challenge some of the most widely held assumptions of the role measurement including what can and cannot be measured. The problem of whether or not psychological processes can be measured in the same sense as physical dimensions is not yet resolved. In the event, that the answer to this question is negative, many putative explanations of cognitive processes may also turn out to be psycho-myths. Chapter 4 delves deeply into the role of measurement in scientific psychology.

Miraculous Graphs. The necessity for displaying data in an easily digestible form often provides a ferule seedbed for the kind of erroneous conclusions I have designated as psychomyths. If a set of data extends over very wide ranges, it is often necessary to nonlinearly compress it in order to see the details, particularly at the lower values of the range. For example the dynamic range of brightness and loudness perception are usually compressed so that details of the relationship between stimuli and responses are not obscured. Such a compression (e.g., when a logarithmic coordinate system is used) can, in some cases, produce an illusion of order when disorder is actually a better description of what is happening. In some cases, highly processed data of this kind can produce graphical relationships that may appear to reflect something about the psychology of the observer but actually hide important microdetails of the functional relationship. Graphs and inevitable laws can strongly interact to mislead the unwary and create psychomyths. Chapter 4 also considers the roles of graphics in the study of cognition.

Misleading Statistics. No other branch of mathematics has been of persistent or profound utility to psychology as statistics. The reasons for its particular value in this kind of research are obvious. Psychology deals with responses and individuals and situations in which there is an enormous opportunity for variance. The multidimensional determination of psychophysical responses generates enormous variability from person to person and from day to day, indeed very often from trial to trial. No single instance or sample typically tells us anything useful about the transforms carried on a stimulus by an entire population. This is why anecdotes (very often those from neuropsychological clinics) are so mischievously deceiving—they obscure the actual variability between individuals. For these and related reasons, psychologists have been led to the use of statistical estimates of the central tendency or variability of a set of responses. Indeed, given the variable nature of human behavior, they have no other choice.

On the other hand, no other branch of mathematics or analysis of any kind offers greater opportunities for erroneous conclusions and interpretive mischief than does statistical analysis. Naïve applications of “cookbook” statistical methods may provide what appear to be highly precise measures of human performance while at the same time leading the researcher wildly astray to draw erroneous conclusions. Even as straightforward a process as a “simple” test of significance can mislead an unwary or unlucky scholar. Nor does one have to use intricate statistical analysis to expose oneself to misleading inferences. Even a process as simple as averaging is likely to hide the details of individual performance in ways that may lead to incorrect conclusions. Furthermore, most statistical analysis techniques are based on hidden assumptions about the distribution of data that often do not hold (or cannot be shown to hold) in real situations. It is surprising, as we see later, how far back this chain of logic goes and how often the limitations of the logic are ignored in psychological research. The roles of measurement, graphs, and statistics in generating psychomyths are also discussed in chapter 4.

Erroneous Assumptions and Conceptual Errors. Perhaps the most egregious source of mistaken theoretical conclusions—psychomyths—is a fuzzy or careless interpretation of the foundation assumptions on which a theory or research project is built. It is a truism that in the everyday efforts to “produce” new knowle...

Table of contents

• Cover
• Half Title
• Title
• Copyright
• Dedication
• Contents
• Preface
• 1. INTRODUCTION
• 2. ENDOGENOUS AND EXOGENOUS CAUSAL FORCES IN PERCEPTION
• 3. INEVITABLE NATURAL LAWS AND SUPERPOWERFUL MATHEMATICS
• 4. MEASUREMENT, COUNTING, MAGICAL GRAPHS, AND SOME STATISTICAL CURIOSITIES
• 5. ERRONEOUS ASSUMPTIONS AND CONCEPTUAL ERRORS
• 6. FINAL CONCLUSIONS AND SUMMARIES
• REFERENCES
• AUTHOR INDEX
• SUBJECT INDEX