1330 



HANDBOOK OF PHYSIOLOGY 



NEUROPH^■SIOLOGY II 



FIG. 8. Graph of the average of the number 

 of repetitive errors made in the multiple 

 object experiment during those search trials 

 in each situation when the additional, i.e. the 

 novel, cue is first added. Note that the peaks 

 in errors shown in fig. 7 are accounted for by 

 the monkeys confusion between novel and 

 familiar objects as graphed here. 



UJ 



a. 



NORMALS 



TEMPORALS 



FRONTALS 



[y- Transform) 



No. CUES in SITUATION 



10 II 



12 





■100- 



< 

 CO 



CO 

 UJ 



o 



50 



LJ 



o 



tr 



UJ 

 CL 



— 1 r- 



No. CUES in SITUATION 



3 4 5 6 7 8 9 



NORMALS 



TEMPORALS 



FRONTALS 



10 



FIG. 9. Graph of the average of the per 

 cent of the total number of oVjjrcts (cues) that 

 are sampled by each of the groups in each of 

 the situations (see legend to fig. 6). To sample, 

 a monkey had to move an object imtil the 

 content or lack of content of the food well was 

 clearly visible to the experimenter. As was 

 predicted (see texti, during the first half of 

 the experiment the curve representing the 

 sampling ratio of the posteriorly lesioned 

 group differs significantly from the others at 

 the .024 level [according to the nonpara- 

 mctric Mann-Whitney U procedure (84)]. 



predicts, therefore, that this operated group has 

 sampled fewer objects during the early portions of 

 the experiment. This prediction is tested as shown in 

 the graph of figure 9. 



The prediction is confirmed. The posterior intrinsic 

 sector is thus estabhshed as one of the organismic 

 variables that determine the constant of the model. As 

 postulated by the model, the ratio of objects sampled 



S' represents those stimulus elements which are not rein- 

 forced, 

 / represents the overlap between S and S' which expresses 



confusion when reinforcement is shifted from one to 



another object, 

 TT represents the relative frequency of reinforced trials in the 



stimulus series, 

 TT represents the relative frequency of nonreinforced trials 



in the stimulus series, 

 p„ represents the mean probability of response on the nth 



trial, 

 ^0 represents the initial probability of response (operant 



level), 

 $1 and $■, represent the sampling ratios for reinforced and 

 nonreinforced stimulus sets respectively, and 

 n denotes the number of trials. 



It is assumed that the above equations are weighted directly 

 as a function of the proportion of elements within the intercept 

 and nonintcrcept subsets, such that 



P„iS') = k'p„iS' - /) -f (I - k')pM) 

 In these experiments, then, 



S' is the set of unrcinforced stimulus elements (objects under 

 which no peanut is located), 



/ includes among the subset of elements common to both 

 reinforced and imreinforced trials those objects which 're- 

 cently' have had a peanut under them, 



k' is the proportion of stimulus elements not common to 

 both reinforced and unrcinforced trials, and 



p„(S') is the mean probability of response on nonreinforced 

 trials (probability of error responses) on the ;:th trial. 



In the present experiment only the objects with no peanuts 

 under them are considered since only one object at a time had 

 a peanut under it. Thus the set of reinforced objects reduces to 

 one, and the sampling ratio associated with it fli is maximized 

 with respect to the sampling ratio associated uith the un- 

 rcinforced sets, 9-.. 



