Cognitive Bias and Claims of Quasiperiodicity in Traditional Islamic Patterns

Abstract

In this article we discuss whether traditional Islamic geometric patterns exhibit the distinguishing structural properties of quasicrystals. We present two case studies that examine candidate examples from the western and eastern schools of Islamic ornament. In both cases, researchers made a hypothesis that the Islamic patterns have quasiperiodic features, both found that their model does not match the data, yet both found reasons to discard the errors and accept the model, rather than search for an alternative explanation. This has led to widespread confusion about the status of claims concerning quasiperiodic Islamic ornament, a situation that became clear to me during discussions at a recent conference, and which motivated me to write up these notes. In both case studies, the language and tools of crystallography are applied to analyse Islamic patterns. In each case, we shall show that the conclusions are incorrect, and also propose a different method of construction that agrees with the evidence and the wider context. The analysis of crystals is based on statistical properties of large samples, and real-world crystals display a range of defects when compared to our mathematically perfect models. We suggest that approaching the problem from a crystallographic perspective predisposes the mind to regard defects as anomalies rather than errors. This bias, together with flaws such as confirmation bias in the methodology, may explain how the researchers reached their conclusions.