[ Pobierz całość w formacie PDF ]
.In 1926, ina discussion of an experiment involving the speed of movement ofwhite mice in a maze, Washburn noted that only the speed of runningresulting from hunger was associated with the elimination of errors.She wrote,  The hardest thing to explain about the drive.is itsbackward direction, the fact that food at the end of the maze makesan animal readier to perform a turning at the beginning of a maze. 21She offered a mechanistic solution to the problem, saying that as a re­sult of past learning, hunger would be associated with running, andthat, furthermore, because drive was a continuous stimulus, it wouldbe persistently associated with running.Moreover, the degree of rein­forcement of any act would be negatively related to the distance be-tween the goal and the point at which the act was performed.As work on the goal gradient proceeded, Hull and his associatesbegan to differentiate between stimulus-response and response-rein-forcer gradients.In the former, one is concerned with the operationsof incentives that  drive an animal toward a particular goal.In thelatter, one is concerned with the effect that reinforcers have in enforc­ing the process of learning.Washburn was, fairly clearly, concentrat-108 | The Behaviorist as Research Managering on the incentive aspect.It would seem, then, that Hull initiallyconflated a stimulus-response (incentive) interpretation, derived fromWashburn, with a response-reinforcer (associationistic) interpreta­tion, drawn from Thorndike.22 The Thorndikian concept that mostresembled the goal gradient was the spread of effect, which Thorndikefirst reported in 1933.23 Thorndike concentrated on response-rein-forcer relationships, asserting that if a stimulus-response connectionwas rewarded or punished, the effect would spread automatically totemporally adjacent connections.Hull s conceptual confusion arose inpart from his own uncertainty regarding the role of reinforcement inlearning.He was inclined to concentrate on the incentive interpreta­tion during the 1930s because he was thinking hard about such no­tions as  the pure stimulus act and goal-gaining anticipatory re­sponses.He set that aspect of his theory aside late in the fall of 1939,once he started to work on the Principles in earnest.24When Hull wrote his first paper on the goal gradient hypothesis,published in 1932, he did not subject the rather tentative notions putforward by Washburn to conceptual analysis.25 Instead, characteristi­cally more interested in quantification, he asked what sort of mathe­matical function would emerge if one plotted some measure of thestrength of a learned reaction against distance from a goal.He chosea negatively accelerated curve derived from a logarithmic function,basing that function on Weber s law, and using the results of an ex­periment by Yoshioka as support.In 1929 Yoshioka had shown thatit was the ratio between the lengths of long and short paths that de­termined ease of discrimination between such paths in maze learn-ing.26 In their experiments Hull and his coworkers tried to verify pre-dictions based on the goal gradient hypothesis in three areas (thespeed of locomotion gradient in the straight alley, the choice of short,as opposed to long, paths in the maze, and error patterns associatedwith the elimination of blind alleys in the maze).Their important1934 study on the speed of locomotion gradient in rats is typical oftheir work.27 They showed that once the rats were habituated to themaze, they would accelerate their speed of running from start box togoal box.Furthermore, as trials proceeded, the point in the runway atwhich animals reached top speed moved progressively toward thestart box.But it was also true that the rats would slow down a littlejust before the goal box.Hull noted that the gradient tended to flat-ten with increasing practice, but he did not discuss the point.The Behaviorist as Research Manager | 109The 1934 study yielded successful predictions, but a close readingyields some puzzling points.It seemed that, as demanded by the hy­pothesis, reward exerted a progressively greater effect on performanceas training proceeded.In order to interpret his results Hull replottedone of the learning curves he had obtained as a linear function andcompared it with a plot of the electrical stimulation required to stim­ulate nerve-muscle preparations as a function of the amount of priorstimulation [ Pobierz caÅ‚ość w formacie PDF ]