No abstract

No abstract

The author has outlined a twoprinciple theory which predicts the effects of the CS-UCS interval in classical aversive conditioning (Jones, 1962). In brief, this theory is based upon the observation that variation in the CS-UCS interval affects the temporal relationship between (a) the CS and the UCR, and (b) the CR and the onset of the UCS. With the tentative assumption that UCS-onset is the reinforcing event, the effects of variations in the CS-UCS interval are treated in terms of the joint action of the principles of contiguity (CS-UCR) and of reinforcement (CR-UCS). In both cases a bidirectional gradient is assumed to operate such that the greater the time interval between CS and UCR or CR and UCS the less effective the learning. It is assumed that the efficacy of conditioning at a given CS-UCS interval reflects the additive combination of the overlapping gradients of contiguity and reinforcement.From this analysis it follows that any variation in experimental procedure which affects either or both of these temporal relationships will affect the function relating the CS-UCS interval to response strength.

No abstract

THE TRAINING STIMULI used in studies of stimulus generalization commonly have many attributes. However, in any given experiment the test stimuli presented have usually differed from the training stimulus in one respect only. One exception to this rule is a study by White (1958) who trained an operant response in children in the presence of a coloured stimulus and then tested in the presence of colours differing from the training stimulus in hue and value on the Munsell scale. Fewer generalized responses were given to test stimuli which differed in both dimensions than to those which differed in either of the dimensions alone. Also relevant is the experiment by Fink and Patton (1953) testing the effect of changed light, sound, and tactual stimulation on a learned drinking response in rats. They found that the greater the number of stimulus components changed, the greater the response decrement. Erikson and Hake (1955) determined discrimination accuracy by the method of absolute judgments for series of up to 20 stimuli varying in size, hue, and brightness. The discriminability for a multi-dimensional series was greater than that obtained for any of the compounding dimensions used alone and could be predicted with reasonable precision from the latter by assuming independence in the accuracy of simultaneous judgments on two or more dimensions.There are two major points of view concerning the kind of process underlying the unidimensional performance gradients which have been obtained. One is that there is a spread of excitation from the training stimulus to the test stimuli, for example, Hull, 1943. The other is that generalization is a performance phenomenon only and that it reflects the degree of failure to discriminate between the training stimulus and test stimuli, for example, Lashley and Wade, 1946; Cuttman, 1956. Guttman (1956 has suggested that a study of two-dimensional generalization gradients might illuminate the problems of interpretation involved, the two approaches yielding quite different expectations for this case. Certainly, if the concept of stimulus generalization is to be usefully employed it will be necessary to discover the laws applicable to the multidimensional as well as to the unidimensional case.