ABSTRACT
For cxample, just what does "cumulative effccts" or "progressive deterioration" mean in tcrms of puzzle solution times? Assume there is a way to equate puzzle difficulty or to comparc pcrformance across puzzles: Does a cumulativc effect mean that a dyad's progressively cooperative performance corresponds to the upper curve in Figure 1 and progrcssively uncooperative performance to the lower curve? This was my original expectation. The data, however, appear to be saying that a better story is told by (a curve of) the successive correlations between a dyad's composition and each puzzle in a series. That is, the correlation between dyadic characterization and performance on the first puzzle should be zero, and the magnitude of the correlation for successive puzzles should increase. Further, the curve need not be smooth, only monotonic: the change in effectiveness after a particularly satisfactory or frustrating episode could be considerable.
