Recall and anticipation methods in probabilistic associative learning1

Six S-R₁ S-R2 modified paired-associates were presented for 60 trials. A 2 by 6 factorial design was employed in which method, recall or anticipation, and the S-R₁, S-R₂ probability-ratio were varied. The S-R₁ S-R₂ ratios were: 1.00-0.00, .90-.10, 80-.20, .70-30. .60-.40, and .50-.50. The results indicated that the recall procedure yielded superior performance in early acquisition. In later stages, recall yielded results approaching matching, whereas the anticipation procedure yielded overmatching of Rt anticipations and undermatching of R₂ anticipations. The results were interpreted as showing that as probability-ratio varies from .50-.50 to 1.00-0.00, and hence, as the relative difference of response strengths of S-R₁, and S-R₂ increases, the selection factor in the anticipation procedure yields a heavy weighting of the stronger response, R,, and the acquisition of the weaker association, S-R₂, occurs slowly; for the recall paradigm, however, the data suggest that R₁, is inhibited and S-R₂ is more readily acquired. The results also were interpreted as showing that one difference in the standard paired-associate recall and anticipation paradigms is that in the recall method. strong error tendencies may be inhibited, but in the anticipation method, strong error tendencies are weighted heavily.