Selective attention in dairy cattle

T.L Blackmore,W Temple,T.M Foster

doi:10.1016/j.beproc.2016.06.001

Abstract

In a replication of Reynolds (1961), two cows learned to discriminate between compound stimuli in a forced choice procedure where pushing through a one-way gate marked with a red cross (S+) gave access to food. Pushing through a one-way gate marked with a yellow triangle (S−) gave no access to food. To investigate whether shape or colour was controlling behaviour, probe tests varied either the shape or the colour of the stimuli (e.g., a red vs. a yellow cross, and a red cross vs. a red triangle). Results suggested control by colour rather than shape, as the gate marked with the red stimulus was chosen more than the gate marked with the yellow stimulus regardless of stimulus shape, and when two shapes of the same colour (either red or yellow) were presented, cows chose both equally. Further probe tests with painted red, white, and yellow stimuli showed that the cows had learned to avoid yellow rather than to approach red, suggesting discriminative behaviour was controlled by the colour of the negative stimulus and not by either aspect of the positive stimulus. It is not clear why the negative stimulus was more salient, but it may reflect a tendency for cows to learn to avoid farm handling practices which involve mainly negative stimuli.

Highlights

U-value is a measure of uncertainty first reported in a paper on the analysis of serial dependencies in response chains by Miller and Frick (1949)
Analysis 1 showed stereotypical ways of responding that resulted in extremely high U-values; Analysis 2 and 3 showed comparisons of U-values need to take into account the number of the available options that were used
The U-value should be used with caution as a measure of variability in responses and consideration should be given to reporting alternative measures of variability

Summary

Introduction

U-value is a measure of uncertainty first reported in a paper on the analysis of serial dependencies in response chains by Miller and Frick (1949). Providing reinforcement based on the previous use of these categories could possibly prevent the stereotypical responding; high U-values would be less likely resulted from stereotypical response patterns shown in Analysis 1.

Results

Conclusion