VISUAL PERCEPTION OF THE CHICK 49 



averages for the eight subjects gives respectively for these dis- 

 crimination settings 11, 16, and 20 seconds. These averages 

 indicate that the chicks in the early stage of learning were 

 faster in choosing than in the intermediate stage. However, 

 the reliability of these averages as an index of discriminability 

 is open to question. The frequency of errors indicate that less 

 discrimination underlies the early responses. While the chicks 

 were less acquainted with the conditions of the experiment in 

 the initial training, variable o 7 + , they made many direct 

 choices without showing evidence of comparison such as fre- 

 quently occurred when the discrimination habit had become 

 well established. Less time naturally was recorded for these 

 direct choices, although errors were more frequent. Apparently 

 the averages of No. 21 as recorded in table 4 are more reliable 

 as an indication of the relative time required for each setting 

 where discrimination underlies the choices. The time of this 

 chick for each of the three variables is respectively 6, 2, and 

 9 seconds. If it were possible to obtain an average for the time 

 of those responses only in which choice was based on discrimi- 

 nation, it is probable that we would find the time greater in the 

 beginning and final stages than in the intermediate stage. Dur- 

 ing the intermediate stages the habit would be well established 

 and the discrimination settings would not be extremely difficult. 



It might be urged that this time average became relatively 

 longer towards the close of the experiments on account of the 

 physical condition of the chicks. The facts, however, seem to 

 refute such a view. Referring again to the individual record 

 of chick 21 in table 4, one finds evidence that the changes in 

 time averages are not conditioned by the health of the chicks. 

 During the course of these tests No. 21 was given open air 

 range under which conditions it throve and grew normally. 

 This chick continued in good health even at the end of 1500 

 subsequent tests in form perception. The records of table 4 

 indicate that chick 21 was slower with the variable o 7 + when 

 it was learning, the average time was not increased when the 

 variablew as changed to o 12-|- because the discrimination re- 

 mained easy, but the time was increased when the more difficult 

 discrimination, variable o 15 + , was required. 



The records of chick 21 alone cannot be accepted as reliable 

 on account of the limited number of tests, particularly in the 



