ON THE SIGN OF SLOPE IN THE LEARNING OF LINEAR FUNCTIONS

Three groups of subjects were presented with pairs of values ( E and S ) of three different linear functions E = a+bS. In a test phase they had to indicate the E -values that corresponded to the S -values. The values were represented as distances marked on straight lines. Two groups, one a control group, had a positive slope, the third group a negative. The squared sum of the S-E differences was kept constant. All groups had roughly the same performance as measured by the squared sum of deviations from empirical and correct E -values. The learning curves indicated, however, that some learning had taken place in the group with the negative slope. It was concluded that a principle rather than a function was learned.