THE NEURAL BASIS OF LEARNING 



'473 



Pavlov observed that the CS (the bell) came to evoke 

 the UR (salivation) which he now called the 'con- 

 ditioned response' (CR). This Pavlovian method of 

 studying learning has in recent years been called 

 'classical conditioning' or conditioning of the 'first 

 type.' We shall refer to it as Type I conditioning. 



Type I conditioning procedures were subsequently 

 modified by several investigators so as to employ a 

 noxious stimulus, usually a shock applied somewhere 

 to the skin of an animal, as the US. In this case an 

 animal is presented with a CS (light or bell) followed 

 by a shock (US), but only if the animal fails to make 

 some specified response like lifting its leg (CR). In the 

 United States, this conditioning procedure is usually 

 called 'avoidance conditioning,' while others refer to 

 it as defensive conditioning or conditioning of the 

 'second type.' Along with many others we shall call it 

 Type II conditioning. Since what the subject does in 

 the Type II procedure is instrumental in avoiding 

 shock, it falls into the more general category of 

 'instrumental conditioning' — a term suggested by 

 Hilgard & Marquis (98). There is evidence, however, 

 that avoidance conditioning also involves the classical 

 conditioning of autonomic reactions and consequently 

 is not purely instrumental. Skinner's 'operant con- 

 ditioning," described in the paragraph below, is a 

 better example of instrumental conditioning. Pavlov's 

 classical conditioning is not instrumental because the 

 animal's response (salivation! has nothing to do with 

 whether or not the unconditioned stimulus (food) is 

 presented. 



Many years later, Konorski (121, p. 418) and 

 Skinner (220) applied the term conditioning to a still 

 different learning procedure. They put hungry rats in 

 a box and delivered a pellet to them each time they 

 pushed a lever. In this case, food was the US and 

 pushing the lever became the CR. This situation 

 differs, however, from both Pavlovian and avoidance 

 conditioning in that: a) it contains no specifiable CS 

 and h) the CR (pushing the lever) is not originally a 

 UR to the food (US). For this reason, Skinner called 

 it 'operant conditioning' in contrast to the first two 

 kinds of conditioning, which he called 'respondent 

 conditioning.' However, since the animal's response 

 is instrumental in obtaining the food (US), this 

 learning procedure must also be classified as a form 

 of instrumental or Type II conditioning. 



Trwl-and-Error Learning 



Skinner's Type II procedure, however, does not 

 differ in principle from other learning methods that 



have been called 'trial and error.' Thorndike's cat in 

 a puzzle box (235) indeed must do practically the 

 same thing as Skinner's rat, namely push a latch to 

 get out of a box to reach food. Similarly, animals 

 required to run through a maze to receive a food 

 reward must learn to make a number of turns correctly 

 to reach the goal of food. Although the animal must 

 make a series of responses, rather than just one, these 

 are still instrumental responses which, like the rat in 

 Skinner\s box, are not themselves UR's and are not 

 evoked by any identifiable CS's. Hence, it is only a 

 matter of convention, not one of fundamental differ- 

 ence, to call maze learning and learning in puzzle 

 boxes trial-and-error learning while calling Skinner- 

 box learning conditioning. 



Perceptual Learning 



In addition to the learning studied in Type I and 

 Type II procedures, organisms also can learn some- 

 thing about the relation of stimuli and of objects in 

 their environment without necessarily making any 

 overt responses. The general name for such learning 

 is 'perceptual learning' because it invokes a change 

 in the perception of the environment. For our purposes 

 two varieties of ^uch learning may be distinguished, 

 although there are others that need not be mentioned. 



One is 'sensory-sensory learning.' In this case, one 

 stimulus becomes associated with another by being 

 regularly paired with it. An auditory stimulus, for 

 example, might be presented immediately preceding 

 a visual stimulus. The learning that occurs in such a 

 procedure is not immediately available for scientific 

 study because it does not involve a response 1 [owever, 

 l>v conditioning some response to one of the paired 

 stimuli and then testing with the other stimulus, one 

 can establish that sensory-sensory learning actually 

 takes place because the tesl stimulus evokes a response 

 with which it has not previously been paired (23). By 

 use of electrical methods of recording brain changes 

 during learning;, which will be described later, it is 

 possible to measure changes in sensory-sensory 

 learning directlv. 



Another kind of perceptual learning is that which 

 is commonly called "insight.' Its central feature is that 

 the organism learns suddenly and without trial-and- 

 error responses, as emphasized by Kohler (119). By- 

 attending to a problem and perceiving the relation- 

 ships in it, the organism gains insight into its solution. 

 Since we do not yet have any appreciable information 



