Outline of In a Mirror, Darkly: How Superstitions Reflect Rationality

Posted on March 14, 2008


As part of my report on what I have done at the KLI over the last few months I had to write an outline of the book I’m working on. This helped me to clarify some of my ideas while making it obvious to me what other things I still need to think a bit more about. Here is the outline:

The book consists of four parts. The first part (chapters 1 and 2) introduces the main issues and presents a model of superstition. The second (chapters 3 through 6) examines the way superstitions present certain everyday occurrences as linked causally in some manner and how superstitions are related to Herbert Simon’s heuristics as well as David Hume’s habits. The Rule of Contagion serves as the central example is this part. The third part of the book (chapters 7 and 8 ) looks at the particular types of explanations offered for the supposed connections and how superstitions differ from such human practices as religion, common-sense knowledge and pseudoscience. Here, the role of the main example is played by Luck. In the final, fourth, part (chapters 9 and 10) the elements that had been discussed are brought together in order to re-evaluate the distinction between superstition and science.

Part I

Chapter 1 is the introductory chapter. In it I explain the two main aims of the book. The first of these is to provide an account of superstition as a natural phenomenon, i.e. explain within a wholly naturalist framework why people hold superstitious beliefs and engage in superstitious practices. The second is to show the relevance of several conceptual approaches to the understanding of superstition. These being: Simon’s bounded rationality programme, especially as pursued by William Wimsatt; a commitment to understanding human cognition as faculty that has arisen as the result of Darwinian evolution; a naturalist interpretation of Hume; and, finally, a generally Peircean pragmatism.

Chapter 2 examines the question of what superstitions are. After examining several proposed definitions and arguing that they are of limited way, I decide to, instead, opt for a rough model constructed on the basis of the examination of traits common to the superstitions that have been identified as common within the British Isles. The model treats superstitious beliefs as typically involving the claim that certain everyday events are causally connected (where the exact manner of the connection is often ambiguous) and explaining this connection in particular non-standard ways. I explicitly leave open the question of whether all practices and beliefs traditionally identified as superstitious can be explained in terms of this model, allowing that it may well be that superstitions may turn out not be a natural kind.

Part II

Chapter 3 turns specifically to the question of the connections superstitions claim to exist and examines the significance of uncertainty for superstition. The analysis begins with Malinowski’s observation that superstitions thrive in situations where people experience a lack of control. The insight is backed up by research that has further examined the relationship between uncertainty and superstitions. A difficulty with this research is that it is not clear if the superstitious response represents a misguided effort to obtain actual control or a way of creating a false sense of control. A Skinnerian explanation in terms of coincidental operant conditioning suggests the former and is borne out by Killeen’s work which suggests that superstitions may in part be a strategy aimed at not missing hard-to-spot correlations, an approach which can be further developed using Haselton’s error management theory. A possible problem this approach might lead to is a naïve inductivist assumption that there exists a single all-purpose belief-evaluation mechanism. That view, however, does not seem to be consistent with Hume’s criticisms of that view of reason nor with the results obtained by Seligman and others concerning the specificity of the connections which are easily made by conditioning. Finally, conditioning is too limited for understanding more complex cognitive mechanisms such as people appear to make use of, requiring a move to a more complex view of those capabilities.

Chapter 4 explores the relationship between human cognitive capabilities and our predilection for superstitions using the example of the extensive empirical work that Paul Rozin and Carol Nemeroff have done upon contagion. They apply the heuristics and biases approach developed by Kahneman and Tversky to what James Frazer identified as a rule of superstitious thinking, i.e. that what was once connected remains connected. Through a series of experiments, Rozin and Nemeroff argue that a class of superstitions can be traced back to a heuristic that is advantageous in so far as it helps people to avoid contamination by germs but which, when misapplied, leads to those superstitions. This explanation of superstition as a by-product of a generally useful heuristic is the one which is pursued throughout the book.

Chapter 5 develops the by-product explanation by applying the more general view of heuristics that has been put forward by William Wimsatt on the basis of Herbert Simon’s bounded rationality programme. Superstitions are identified as systematic errors introduced by the use of heuristics. Rozin and Nemeroff’s empirical work is understood in this context as a paradigmatic example of how these systematic errors can be used to identify the underlying heuristic in the same manner that an animal’s footprints can be used to recognise it. Of course, not all such systematic errors fall under the rubric of superstition. This makes it necessary to differentiate between superstitious and non-superstitious errors. Some non-superstitious errors can be identified as such because they do not lead to the relevant kinds of beliefs concerning the causal connections between everyday events. This leaves, however, a broad range of systematically false beliefs such as pseudoscientific claims, as well as some beliefs that belong to ‘traditional folk wisdom’. The connection between these various kinds of claims is borne out by the levels of beliefs in them being closely correlated to each other, as shown by Lindeman. Understanding the difference between superstitions and such beliefs requires, however, that the explanations provided be examined – something that is left for Part III.

Chapter 6 begins by considering the significance of dual process theories of rationality for our understanding of superstition and vice versa. On most dual process models it would appear that unlike the error-prone heuristics, the logical, analytical process should be free of superstitions. Indeed, proponents of such theories argue that the empirical evidence bears out this hypothesis. I question the dual process approach on two grounds. Firstly, I argue that the empirical evidence is better explained by the bounded rationality view. Secondly, I argue that dual process theories must either reduce to some kind of bounded rationality theory or run into Hume’s problem of induction. This leads me to a discussion of heuristics as Humean habits – the option Hume comes to accept having rejected the possibility of the traditional view of reason. As a result, the fundamental limitations originally grasped by Hume come to be seen as the ultimate explanation for the existence of superstitions.

Part III

Chapter 7 opens the discussion of the role of explanations within superstitions by examining the example of luck. Duncan Pritchard’s discussion of the recent philosophical and psychological literature on luck is of some use here, particularly his discussion of the results obtained in attribution research concerning causal explanations. The work on luck is compared with general results concerning the psychology and philosophy of explanation, the aim being to identify what is particular about the use of luck as an explanation as well as to place it within the context of explanation in general. These considerations come to play a significant role in understanding the significance of explanations in superstitions.

Chapter 8 compares superstitious and religious beliefs and practices using David Sloan Wilson’s account of religion. The problems Wilson’s account has with distinguishing religion from other human practices, including superstition, and various ways of strengthening the account are considered. In the process, the problems with identifying superstitions with falsehood, the supernatural or the sacred are discussed. It is shown that the missing element in Wilson’s account is a proper appreciation of the testability of the claims. Testability is shown to have at least three aspects, the content of the claim in question being only one. The other two are the empirical capabilities and methods that can be brought to bear upon the question and the social attitudes toward testing such claims. The supernatural nature of prototypically superstitious claims as well as the classification of such claims as dealing with sacred entities are shown to be strategies that are highly effective in lowering the effective testability of the claims.

Part IV

Chapter 9 considers the function of superstitious beliefs. The starting point for this discussion is a discussion of the relationship between testability and function. It is proposed that the lack of testability makes it easier for the latent function of a claim to become more significant due to the relative difficulty of evaluating how effectively the manifest aim is attained. Given the greater degree to which religious beliefs are cordoned off from empirical testing, this suggests that latent functions will be more significant in their case – a view in line with David Sloan Wilson’s position. In the case of superstitions, only the explanations are protected against testing, the proposed connections between mundane events being seemingly open to empirical evaluation. While this explains some of the differences between the functions of religious and superstitious beliefs and practices, it does not explain why the functions tend to be social in one case and individual in the other. An interesting exception to this rule is provided by the phenomenon of religious superstitions such as intercessory prayer. In these cases, superstitions appear to have become co-opted to effectively strengthen religious traditions. Having become co-opted into social institutions, the religious superstitions tend to be far more long-lasting than other superstitions. Importantly, by explaining how latent functions are made potentially more significant by the lack of testability, this approach opens the way to subsuming a number of what might initially have seemed to be alternative explanations of superstition in so far as they appear to consist in varying accounts of the latent functions of superstitions.

Chapter 10 concludes the examination of superstitions by comparing them to common-sense beliefs and to scientific claims in the context of the related social attitudes. The comparison is made in terms of the functions and testability of the different kinds of beliefs and practices.