SIGNALING SYSTEMS FOR INDIVIDUAL RECOGNITION - AN INFORMATION-THEORY APPROACH
SIGNALING SYSTEMS FOR INDIVIDUAL RECOGNITION - AN INFORMATION-THEORY APPROACH
复制标题
DOI：
10.1016/s0003-3472(89)80087-9
复制
发表时间：
1989-08-01
期刊：
影响因子：
2.5
通讯作者：
BEECHER, MD
中科院分区：
文献类型：
--
作者：
BEECHER, MD
本文提出了一种定量的方法来测量动物的“签名系统”的信息容量，即一组线索，通过它来识别个体。信息测度（Hs）是通过将连续变量中的信息的香农测度应用于简单线性模型而得到的。该模型本质上是方差分析模型II（随机效应），并隐含在许多方差分析和判别函数分析中，这些分析已经对动物的特征系统进行了研究。对于多变量测量，数据的主成分变换允许将独立成分中的信息相加以给出总信息。对说明性数据集的分析揭示了Hs与通过判别函数分析获得的个体的正确分类的概率（P）之间的密切相关性。然而，Hs的优点是它是一个总体估计，而P的值与样本中的个体数有关。信息分析的方法可能被证明是有价值的比较分析，进化假说预测一个物种比另一个有一个更好的发达的签名系统。
This paper develops a quantitative method for measuring the information capacity of an animal''s ''signature system'', i.e. the set of cues by which individuals are didentified. The information measure (Hs) is derived by applying Shannon''s measure for the information in a continuous variable to a simple linear model. The model is essentially the analysis of variance model II (random effects), and is implicit in the many ANOVAs and discriminant function analyses that have been done on the signature systems of animals. For multivariate measurements, a principal components transformation of the data permits the information in the independent components to be added to give the total information. An analysis of illustrative data sets reveals a close correlation between Hs and the probability of a correct classification of an individual (P) obtained by discriminant function analysis. Hs has the advantage, however, that it is a population estimate whereas the value of P is tied to the number of individuals in the sample. The informaion analysis approach may prove valuable for comparative analyses where evolutionary hypotheses predict one species to have a better developed signature system than another.