BEHAVIORAL DIFFERENCES 291 



locus may be involved. A possible instance of this sort has been reported by Denen- 

 berg 246 who found that different sublines of C57BL/10 mice differed in rate of avoidance 

 learning even when reared under identical conditions. Genetic analysis of this pheno- 

 menon would be most interesting. We have no means of knowing how frequently such 

 cryptic mutations may occur, and how successful a systematic search with a number of 

 reliable behavioral measures may be. 



The problems of behavioral differences may also be attacked from a phenotypic 

 orientation. The behavioral patterns which can be chosen for investigation are of 

 course almost infinite, and the choice will depend upon personal interests. Behavioral 

 patterns which are important in adaptation have especial interest. Many biologists 

 and psychologists are much more interested in the heritability of important adaptive 

 patterns than in determining the effects of a rare mutant gene upon a standard 

 genetic background. One can be quite sure that findings on the inheritance of choice 

 of mate, dominance and submission, or problem-solving ability will not be trivial insofar 

 as the welfare of the species is concerned ; but the very fact that these behaviors are so 

 important makes it unlikely that they will have a simple genetic base. Thus the 

 investigator in this area must sacrifice genetic clarity in order to work with characters 

 of major importance for survival. 



The distinction between the phenotypic and genotypic orientation is, of course, 

 not absolute, but a matter of degree. Both approaches have advantages and dis- 

 advantages and both will contribute to the development of behavioral genetics. 



Statistical analysis. — One of the special problems of the analysis of the inheritance 

 of behavioral patterns is the assigment of variance to genetic and environmental com- 

 ponents. It is relatively easy to determine the heritability of a trait in any particular 

 environmental situation, but such a figure has little general significance because of the 

 responsiveness of behavior to environmental change. The method may, however, be 

 highly important in determining whether or not there is a major genetic component 

 in a behavioral trait, and Broadhurst 136 has demonstrated how it may be applied to a 

 simplified cross-breeding experiment. 



On the other hand, the detailed analysis of Mendelian mechanisms through analy- 

 sis of variance frequently fails because of complicated interaction between genetic and 

 environmental factors. Scott, Fuller, and King 1188 have demonstrated how a relatively 

 simple genetic mechanism may interact with environmental factors to produce a highly 

 complex manifestation of a phenotype. The annual seasonal breeding cycle of the 

 African basenji dog appears to be determined by a single recessive gene, but the expres- 

 sion of the character is also controlled by decreasing diurnal length. F x hybrids with 

 cocker spaniels show a tendency to run six-month nonseasonal cycles like the cocker, 

 but also to respond in part to the seasonal changes in light, so that their breeding 

 cycles are much more variable than those of either parental strain. 



With more complex genetic mechanisms, the interactions become so complex as 

 to defy analysis in most cases. Bruell 140 has suggested that it is possible through the 

 use of means and medians to determine whether or not a behavioral trait is consistent 



