Laws and Hypotheses for Behavior 279 
except for their connection-systems, each being the aver- 
age condition of the animal in question. 
Call S; and S, two external states of affairs, each being 
near the indifference point for the animal in question, — 
that is, being one which the animal does little to either 
avoid or secure. 
Call G, and G, two responses which result in O, and O, the 
optima or most satisfying state of affairs for 1 and 2. 
Call I, and I, two responses which result in the continua- 
tion of S, and S.. 
The only responses possible for 1 are G, and Ij. 
The only responses possible for 2 are G, and I,. 
Animal 1 upon the recurrence of S; and C;, is little or no 
more likely to respond by G, than he was before. 
Animal 2 upon the recurrence of S, and C, is far more 
likely to respond by G, than he was before. 
The fact thus outlined might conceivably be due to an 
intrinsic inequality between O, and Og, the power of equally 
satisfying optima to influence, their antecedents being iden- 
tical. This is not the case in the evolution of learning, 
however. For even if, instead of O,, we had only a moder- 
ately satisfying state of affairs, such as the company of 
other chicks to (2) a 15-day-old chick, while O, was the 
optimum of darkness, dampness, coolness, etc., for (1) an 
earthworm, 2 would learn far, far more rapidly than 1. 
The fact is due, of course, to the unequal power of equally 
satisfying conditions to influence their antecedents. The 
same argument holds good for the influence of discomfort. 
The ability to learn, — that is, the possession of a con- 
nection-system subject to the laws of exercise and effect, 
—has been found in animals as ‘low’ as the starfish and 
perhaps in the protozoa. It is hard to tell whether the 
changed responses observed in Stentor by Jennings and in 
