THE NEURAL BASIS OF LEARNING 



148; 



commonly induced (94, 95, 146-148). Presumably 

 the locations stimulated represent places where some- 

 thing important for the behavior observed is centered. 

 Through such experimentation, in part, the limbic 

 system has become strongly implicated in emotional 

 behavior (see Chapter LXIII by Brady in this vol- 

 ume) and the reticular formation in attentive func- 

 tions. We shall restrict ourselves here to the experi- 

 ments relating electrical stimulation specifically to 

 learning. 



Brain Shacks as CS and I 'S 



In Type I learning, as we have seen, two stimuli 

 appear in the animal's environment. Commonly the 

 US is a shock that activates pain receptors of the skin 

 while the CS is a tone or light that activates the ear 

 or eye. The question of whether shock to the brain can 

 replace cither US or CS or both has prompted a num- 

 ber of studies. There has been considerable contro- 

 versy about this problem and particularly about the 

 question of whether the shocks simply activate afferenl 

 fibers where they are collected together in the brain 

 instead of where they are dispersed in the periphery. 

 This point seems to have been resolved, in some cases 

 at least. 



Doty et al. (52) in cats paired shocks (CS) to the 

 cortex with shocks (US) to the paw. All their animals 

 learned within 400 trials to lift the paw to escape the 

 leg shock when the only signal of its imminence was 

 the cortical shock. It made little difference whether 

 the CS was applied to marginal, postlateral, middle 

 suprasylvian, or middle or posterior ectosylvian gyri; 

 even a point on the frontal cortex was effective. A 

 very careful series of control studies shows that neither 

 receptors in the membranes covering the brain nor 

 other sensory stimulation of the head constituted the 

 CS. This report summarizes previous experiments on 

 the problem and fully discusses the value of such 

 studies. 



Delgado et al. (48), following on the pioneering 

 efforts of Gantt and colleagues (24, 68), have used 

 shocks applied to thalamic, mesencephalic and limbic 

 structures of cats as US. Such shocks produce violent 

 'fear-like' responses which the animals presumably do 

 not enjoy. They used a Type II avoidance technique 

 with flickering light (or a 2000 cps tone) as the CS; 

 the brain shock (US) could be prevented if, within 

 5 sec. after the CS appeared, the animals rotated a 

 wheel. Alter 16 to 92 pairings, each of four cats regu- 

 larly turned the wheel and thus avoided the brain 

 shock. Control experiments include the demonstra- 



tion that, before training, the CS alone did not cause 

 wheel turnings. The possibility that collected afferent 

 tracts or normal pain or other end organs of the head 

 were stimulated was not excluded, although such ex- 

 planations seem unlikely. The authors demonstrated, 

 in addition, the important fact that shocks to the 

 motor cortex failed to serve as US in the animals in 

 which subcortical shocks succeeded, and they describe 

 three other situations in which cortical shocks proved 

 to be effective unconditioned stimuli. In one of these 

 learning was actually produced by pairing cortical 

 (CS) with subcortical (US) shocks, thus bypassing 

 all normal sensory input channels. A recent report 

 that a monkey will operate a lever to escape a brain 

 shock confirms one of the findings of this study (138). 



Much experimentation along these general lines is 

 to be found in the Russian and eastern European 

 literature. Kupalov (124), Giurgea (79-81; see also 

 124) and Rusinov (207, -'to), for instance, report use 

 of either shocks or direct currents to affect the cortex, 

 and Grastyan et al. (82) stimulated limbic (hippo- 

 campal) structures. Giurgea produced leg flexion 

 CR's by pairing occipital cortex shocks (CS) with 

 shocks to the motor cortex (US) in the dog. No EEG 

 changes accompany the conditioning process, and 

 lesions of the reticular formation and corpus callosum 

 do not influence it. 



A special word is in order about the experiment- 

 of Rusinov. These employ weak direct currents ap- 

 plied to cortex, hypothalamus and other brain re- 

 gions. A key principle in sonic Russian circles is the 

 idea of 'dominants' enunciated by Ouktomski (136, 

 190) according to which the US makes its cortical 

 area hyperexcitable, thus producing the "focus' to 

 which the effects of the CS arc attracted in the condi- 

 tioning process. Attempts to create or simulate such 

 "dominants" by direct currents and drugs are a feature 

 of many experimental designs. 



These experiments show that normal sensory events 

 are not invariably required for learning, however use- 

 ful they may ordinarily be in the process. The execu- 

 tion of the peripheral motor response is not necessary 

 either, for if the procedure that produces conditioned 

 leg flexion in a normal dog is applied to a dog which 

 cannot move the limb because its motor nerves are 

 destroyed, the conditioned leg flexion is nevertheless 

 found after the nerve regenerates (12, 137). In a 

 similar way salivation can be prevented by drugs 

 while conditioning is carried out and yet appear 

 later when, without drugs, the CS is applied (43). 

 One can therefore conclude that, in some cases at 



