Measuring behaviour into the twenty-first century

Abstract

Comparison with `direct behavioural observation' is considered essential to validate automated measures. However, `direct behavioural observation' is not without problems. Such observations are usually encoded as a series of actions and postures, called `behavioural elements', and listed in so-called `ethograms'. However, the term `element' is deceptive. Such `elements' are not elements in the same way that amino acids are elements in protein chemistry. There are no intrinsic defining characteristics of what should constitute an element, nor what would limit it in time or what kind of variance is allowed within each category. Ethograms might conceal intuitive concepts and hidden interpretations. An ethogram rather is a set of actions and postures chosen to suit convention and convenience in a certain type of experiment. There is little standardization, even between labs that reportedly use the same ethogram. Already in 1985, the author proposed an exchange of video records between groups for comparison. Recently, Spruijt took up the same idea, with discouraging results, due to differences in perceptions and definitions.As the meeting progressed, it became clear that the bottle-neck in achieving automated, reliable, non-arbitrary registration does not lie primarily in the development of better hardware and software. For some time to come, the main stumbling block will remain the advancement of better behavioural concepts. It is clear that direct behavioural observation is not the golden standard to validate automated detection procedures. Moreover, striving towards a perfect match with direct behavioural observations entails the risk that weak concepts from classical approaches will be passed uncritically on into `fuzzy' novel software procedures. However, a more optimistic perspective is that the stern logic required by automation will force behavioural scientists to reassess critically the intuitive criteria on which classical behavioural distinctions are based, to the benefit of behavioural neuroscience. In his concluding remarks the chairman of the meeting called this prospect `inevitable and fascinating'.