TWO-CHOICE CONDITIONAL PROBABILITY LEARNING WITH MEANINGFUL AND NON-MEANINGFUL STIMULI,

An experiment using a two-choice conditional probability learning task was reported. The first-order conditional probability for the two choice tasks was held constant at .70.30. The design of the experiment was a 2 x 4 x 4 Lind quist Type III design Lindquist, 1953. A set of 200 trials on each of 2 tasks was given each S. One task was to discover the conditional re lationshipstiith meaningful stimula nouns and adjectives. The second task was to discover the same relationships with noingful stimuli Xs and Os. Four information conditions were used and four separate r plications run. Signifi cant differences among information groups within the two tasks were found. Also Ss obtained sig nificantly higher scores on the non-meaningful than on the meaningful task although they marked ly improved in each task over the 200 aiials. One puzzlang finding was that the information grou given only probability information and no cue-stimulus anformation performed significantly poorer on the meaningful set than did the no-in formation control. This was interpreted as dem onstrating the detrimental effect of testing too many hypotheses in such a task. Finally, asymp totic data were discussed in terms of matching phenomena. Author