Variations on variability: effects of display composition on same-different discrimination in pigeons

Discriminating same from different multiitem arrays can be represented as a discrimination between arrays involving low variability and arrays involving high variability. In the present investigation, we first trained pigeons with the extreme values along the variability continuum (arrays containing 16 identical items vs. 16 nonidentical items), and we later tested the birds with arrays involving intermediate levels of variability; we created these testing arrays either by manipulating the combination of same and different items (mixture testing) or by changing the number of items in the same and different arrays (number testing). According to an entropy account (Young & Wasserman, Journal of Experimental Psychology: Animal Behavior Processes 23:157-170, 1997), the particular means of changing variability should have no effect on same-different discrimination performance: Equivalent variability should yield equivalent performance. In this critical test of an entropy account, we found that entropy could explain a large portion of our data, but not the entire collection of results.