THE INTRINSIC SYSTEMS OF THE FOREBRAIN 1 329 



NORMALS 



TEMPORALS 



FRONTALS 



No. CUES in SITUATION 



_l I I I L. 



7 8 



10 



12 



FIG. 6. Graph showing the average of the total number of 

 repetitive errors made in each of the situations in the multiple 

 object experiment by each of the groups: control animals 

 {Normals); animals with posterior intrinsic sector lesions 

 (Temporals); and animals with frontal intrinsic sector lesions 

 (Fronlals). A situation is defined by the number of objects in 

 the problem and includes successions of trials. During each 

 succession the peanut is consistently placed under one of the 

 objects (cues). The succession is terminated when the monkey 

 has moved the object under which the peanut is placed on 

 five consecutive trials (criterion). (See also the legends to 

 figs. 4, 5 and 10.) A repetitive error is made by a monkey 

 when during a succession of trials he moves more than once 

 an object other than the one under which the peanut is placed. 



reflex by precunent antagonistic reflexes (such as the 

 flexion reflex). Several of Skinner's pupils (24, 46a) 

 have developed formal models. These models are 

 based on the idea that both 'learning' (or 'condition- 

 ing') and 'unlearning' (or 'extinction') involve an- 

 tagonistic response classes — that in both conditioning 

 and extinction there occurs a transfer of response 

 probabilities between response classes. This concep- 

 tion is, of course, similar to Sherrington's description 

 of the interaction of antagonistic reflexes: ". . . this 

 reflex or that reflex but not the two together." The 

 resulting equations that constitute the model contain 

 a constant which is defined as the probability of 

 sampling a particular stimulus element (46a), namely 

 the oiyect, in the discrimination experiment p'-esented 

 here. This constant is further defined (Estes) as the 

 ratio between the number of stimulus elements 

 sampled and the total number of such elements thai 

 could possibly be sampled. This definition of the 

 constant postulates that it is dependent for its deter- 

 mination upon both en\ironmental and organismic 

 factors. According to the model the rapidity of in- 



CC 

 07 



or 



(T 

 U 



NORMALS 



TEMPORALS 



FRONTALS 



No. CUES in SITUATION 



LU 

 CL 



^2 34 5678 9 10 II 12 



FIG. 7. Graph of the a\erage of the number of repetiti\e 

 errors made in the multiple object experiment by each of the 

 groups during search (see legend to fig. 6). Search trials are 

 those anteceding the first 'correct' response in a succession of 

 trials, i.e. those anteceding the movement of the object (cue) 

 under which a peanut has been placed. Note the difference 

 between the location of the "hump" in the graph of the normal 

 controls and in that of the group with posterior lesions (Tem- 

 porals). 



crease in errors in a discrimination series depends on 

 this sampling ratio — the fewer objects sampled, the 

 more delayed the peak in recorded errors. The para- 

 dox that for a portion of the experiment the group 

 with posterior lesions performs better than the control 

 group stems from the relative delay in the peak of the 

 recorded errors of the operated group.- The model 



- The actual model used to interpret the data analyzed 

 here was developed by Green (46a) and is patterned after a 

 model of discrimination learning proposed by Bush & Mostellar 

 (8). The Green model takes its roots from a parallel model 

 originated by Estes (24, 25). The general form of the model 

 is deri\ed from Estes' equations describing the conditioning 

 and extinction processes : 



pn\S — I) = I — (1 — po) II — 9))'" for conditioning to those 

 elements which constitute occasions for reinforcement, 



p„{S' — I) = pail — ffj)'" for extinction to those elements 

 which are never occasions for reinofrcement, and 



?„{/) = 



ire, 



+ ^2 



l^ne, 



TW\ 



-\- i-flc 



■po I (l 



7r9, — Jflj)" 



for the changes associated with intercept elements, i.e. 

 those present on both reinforced and unreinforced occa- 

 sions; 



uhere 



.? represents the stimulus elements (objects) which are 

 reinforced (have peanuts under them). 



