In 1995, Wasserman, Hugart, and Kirkpatrick-Steger reported that pigeons learned to discriminate 16-item arrays of identical black-and-white computer icons from 16-item arrays of nonidentical computer icons (Figure 1). The pigeons not only discriminated same from different arrays that were created from the set of 16 training icons, but they also discriminated additionalsame and different arrays that were created from a set of 16 novel testing icons. These data suggest that the pigeons had acquired an abstract concept that depended on the relationship among the display items, and not on the particular items themselves.But the same and the different arrays that are depicted in Figure 1 also differ in their spatial orderliness; the same arrays possess horizontal and vertical regularities that the different arrays lack. This could be considered a perceptual rather than a conceptual distinction.In order to determine whether spatial orderliness is necessary for pigeonsto discriminate same from differentarrays, Young and Wasserman (1997, Experiment 1) trained pigeons to discriminate 16-item same arrays from 16-item different arrays, in which the items were randomly placed in a subset of the 25 locations of a 5´5 grid (see Figure 2). The pigeons readily acquired this discrimination,and they showed excellent transfer to additional arrays that were created from a novel set of icons.Although placing the 16 icons into the 25 grid locations did disrupt some of the spatial orderliness of the displays, other regularities remained, particularlywhen adjacent items appeared in the rows and columns of the same arrays. Still, our prior work suggested that the pigeons may not use spatial orderliness as the basis of their same-different discrimination. Two pieces of evidence are relevant.First, in a transfer test to arrays of novel icons (Young & Wasserman, 1997), we found not only that our pigeons evidenced good discriminative transfer, but also that there was a statistically significant drop in accuracy from the familiar to the novel icon arrays from 93% to 79% (see also Wasserman et al., 1995). This generalization decrement suggests that the pigeons did attend to the individualicons in the arrays; if the pigeons had attended solely to spatial orderliness, changes in icon identity should have had no effect on performance. It is nevertheless possible that the pigeons may have attended to both icon identity and spatial orderliness.Second, we conducted another study in which we showed pigeons lists of same and different icons on a one-at-a-time basis (a successive same-different discrimination; Young,
677Copyright 2001 Psychonomic Society, Inc.This research was supported by Research Grant IBN 99-04569 from the National Science Foundation. We thank Brian Nolan and Jessie Peissig for their assistance in the collection of the data. Correspondence concerning this article should be addressed to M. E. Young, Department of Psychology, Southern Illinois University, Carbondale, IL 62901 (e-mail: meyoung@siu.edu).Evidence for a conceptual accou...