Abstract: This paper describes current work on the causal analysis of behaviour systems. It is noted that while causal work investigating the neural, hormonal, and genetic bases of behaviour is flourishing, work being conducted at a strictly behavioural level of analysis has declined greatly over the past 40 years. Nonetheless, most recent research on animal cognition and applied ethology is still being carried out at a behavioural level of analysis and examples of both types of research are presented: memory mechanisms of food-storing birds and decisions of spider-eating jumping spiders, as well as feather pecking in fowl and animal welfare issues, are all briefly discussed. Finally, I discuss the similarities between neural network modelling and early ethological models of motivation, and then show how a modern version of Lorenz's model of motivation can account for current research findings on dustbathing in chickens and sleep in humans. I conclude that valuable information can still be obtained by research at a behavioural level of analysis.

Abstract: The four questions that Niko Tinbergen identified for behavioural biology ? evolution, function, development and causation ? are all important and should be studied in their own right. Recently, there has been a debate as to whether these four questions should be investigated separately or whether they should be integrated. Integration of the four questions has been attempted in novel research disciplines such as cognitive ecology, evolutionary psychology and neuroecology. Euan Macphail and I have criticised these integrative approaches, suggesting that they are fundamentally flawed as they confound function and mechanism. Investigating the function or evolutionary history of a behaviour or cognitive system is important and entirely legitimate. However, such investigations cannot provide us with answers to questions about the mechanisms underlying behaviour or cognition. At most, functional or evolutionary considerations can provide clues that may be useful for a causal analysis of the underlying mechanisms. However, these clues can be misleading and are often wrong, as is illustrated with examples from song learning and food storing in birds. After summarising the main issues in the neuroecology debate, I discuss some misunderstandings that were apparent in the responses to our critique, as well as some recent relevant data. Recent results do not support the neuroecological approach. Finally, I suggest that the way forward is a cautious and critical use of functional and evolutionary clues in the study of the mechanisms of behaviour.

Abstract: The results of a standardised radiological examination of 5928 Hanoverian Warmblood horses selected for sale at auction were used to predict relative breeding values (RBV) in the 23,662 horses included in the last four generations of the probands' pedigrees. The distribution of the RBV was investigated in the whole four-generation pedigree, in the probands and in those stallions that contributed at least three offspring to the probands. The RBV of the probands' sires were further compared with the officially published performance-based relative breeding values, i.e., total indices dressage (TID) and jumping (TIJ). The sires' level of dressage indices was considerably higher (mean TID=110) than of the jumping indices (mean TIJ=98). Total indices radiographic findings (TIR) were calculated for the sires with varying weighting given to the individual RBV. In each case, this resulted in a mean TIR of 99. Finally, total indices were derived from TIR and TID and/or TIJ in order to develop different selection schemes for all-purpose breeding and for breeding focused on dressage and show jumping. All breeding values under consideration increased by between 1% and 19% when radiographic findings were weighted with between 30% and 60% complementary to the corresponding performance parameters, and when only sires were selected with above average total indices. At the same time, the prevalences of the radiographic findings investigated here were lowered by up to 10% each. When only one radiographic finding was considered at a time, the maximum attainable response to selection was a 16-23% increase in the relative breeding values and a relative decrease in prevalences of radiographic findings of between 31% and 52%. The results of this study indicate that it is possible in horse breeding to consider simultaneously health and performance traits. Medical data should be included in the prediction of breeding values in order to improve the radiological status of today's riding horses.