Partial primary reinforcement as a parameter of secondary reinforcement
Klein, Richard Milton
MetadataShow full item record
The problem of this paper is to investigate partial primary reinforcement as a possible parameter of secondary reinforcement. Although partial primary reinforcement is known to be important in many learning situations, there appears to be little systematic knowledge of its relationship to secondary reinforcement. An experiment was performed in which (1) a neutral stimulus was present on every training trial, (2) a primary reinforcer was present on only some of these trials, (3) after training was completed, a test was made for the secondary reinforcing properties of the neutral stimulus. Six independent groups of albino rats were trained in a simple runway with food as the primary reinforcer and goal box brightness as the neutral stimulus. Each group received a different number of primary reinforcements, namely, 100%, 90%, 80%, 60%, 40%, and 20%, out of one-hundred-twenty training trials. Half of the subjects were trained on a white goal box and half on a black goal box. When training was completed, the alleyway was converted to a T maze with black and white goal boxes. Neither goal box was visible to the subjects until after entrance. The animals were given twenty trials in the T maze, and the number of times they entered each goal box was tabulated. Analysis of the data revealed that the lower the percentage of reinforcement given during training, the greater were the number of entries into the training box during the test. Some characteristics of the function were: between 100% and 90% the strength of secondary reinforcement did not increase, between 90% and 80% there was a large increase, from 80% to 40% there was a further increase, and from 40% to 20% there was some decrease. It was also revealed that some subjects in the lower percentage of reinforcement groups went either to the training box or to the novel box on every test trial. Other aspects of the data were also analyzed. From this data a number of conclusions were drawn: 1. Partial primary reinforcement is a parameter of secondary reinforcement. Decrease in partial reinforcement results in an increase in secondary reinforcement various characteristics of this relationship were discussed. It was pointed out that the obtained function might be derived from two separate functions: the relationship of secondary reinforcement to the number of reinforced trials, and the relationship of secondary reinforcement to the number of non-reinforced trials. 2. The fact that some subjects went to the same box on every test trial was explained in terms of the development of strong secondary reinforcement, in the case of subjects who went to the training box, and in terms of the development of strong generalized secondary reinforcement, in the case of subjects who went to the novel box. 3. It has often been reported in the experimental literature that partially reinforced subjects show greater resistance to extinction than continuously reinforced subjects. Our findings can be applied to this phenomenon. Stimuli present during partial reinforcement are apt to acquire greater secondary reinforcing properties than those present during continuous reinforcement, and, hence, the presence of the former during extinction are able to maintain a higher frequency of responding than the presence of the latter. This hypothesis was distinguished from others offered in the literature which purport to explain the greater resistance to extinction in terms of secondary reinforcement. 4. It was pointed out that this experiment revealed a significant variable, secondary reinforcement, which might develop in studies whose training set up resembles ours. 5. Minor findings of the experiment were discussed.
Thesis (Ph.D.)--Boston University