SAME-Ysame
adjective
alike
aforementioned, aforesaid, carbon*, carbon copy, clone, coequal, comparable, compatible, corresponding, dead ringer, ditto*, double, dupe*, duplicate, equal, equivalent, identical, indistinguishable, interchangeable, like, likewise, look-alike, related, ringer, same difference, selfsame, similar, similarly, synonymous, tantamount, twin, very, xerox
another, different
Roget's New Millennium™ Thesaurus, First Edition (v 1.2.1)
Copyright © 2006 by Lexico Publishing Group, LLC. All rights reserved.
* = informal or slang
― lf (lfam), Tuesday, 2 May 2006 06:04 (seventeen years ago) link
http://www.pigeon.psy.tufts.edu/jep/images/sdmsc_1.gifThe major finding of the present experiments concerns the relative ease and flexibility demonstrated by these pigeons in performing a complex and demanding Same-Different discrimination. This discrimination required them to concurrently process very large numbers of highly variable, often ill-defined, multidimensional elements configured in four different ways. In Experiment 1, the birds showed little difficulty in acquiring these different display type discriminations, learning to classify the Same and Different displays of all four display types at the same rate and in the same way. In Experiment 2, they readily transferred this discrimination to new exemplars of each of the four display types. These results expand considerably on those reported by Cook et al. (1995), which were limited to investigations of only the texture display type. Of most importance in this regard, is that these new training and transfer results suggest that low level perceptual factors are not critical to producing the type of discrimination behavior observed in multi-element Same-Different experiments (see Young, Wasserman, & Dalrymple (in press) for a similar conclusion based a temporally-based manipulation of Wasserman et al.=s (1995) Same-Different discrimination). If not based exclusively on simple perceptual factors, what then is the abstract basis for this complex discrimination by the pigeons?
Cook & Wixted (1997) recently tested the applicability of using a signal detection framework to better understanding pigeon choice behavior in Cook et al.=s (1995) textured Same-Different procedure. Of direct relevance to the present discussion, Cook and Wixted=s (1997) signal detection analyses strongly suggested the pigeons discriminated Same texture displays from shape, color, and redundant Different displays using only a single type of information or evidence. That is, regardless of what dimension (shape, color, or redundantly from both dimensions) the Different display=s contrast was made from, the birds seemed to base their choices on only a single common unidimensional encoding of the target information in the displays. To acknowledge the possibility that either perceptual or conceptual interpretations of this common code were possible in this texture-only context, they neutrally labeled this unidimensional evidence variable as Adegree of difference.@ While the same type of signal detection analyses remains to be extended to the current testing context, Cook & Wixted=s (1997) unidimensional interpretation is highly consistent with the proposed hypothesis that the current birds deployed only a single rule to discriminate the four sets of stimuli tested here. One straight forward interpretation of this similarity is that the birds in both cases were employing a singular abstract Same-Different rule. But before accepting this conclusion, however, at least one other unidimensional alternative should be considered.
Young & Wasserman (1997), following up on Wasserman et al.=s (1995) observations, recently proposed a new unidimensional alternative of what the pigeons might be processing in this type of choice task. Instead of employing an abstract Same-Different concept, these authors present evidence suggesting their pigeons were responding to the perceived entropy in their icon-based Same and Different displays. Entropy is an information-theoretic concept which measures the amount of variability present among a display=s component elements. A display in which all of the elements are identical (i.e., a Same display) has an entropy of zero, for example. In contrast, a display in which every single element is different from every other one (the kind of Different display tested by Wasserman et al., 1995) has the maximal possible entropy for that particular organization. In a series of experiments, Young and Wasserman (1997; see also Young, Wasserman, & Garner 1997) systematically varied the number and nature of the elements used to create different types of ASame@ and ADifferent@ displays. They found that the amount of variability in these displays as described by entropy correlated quite highly with the proportion of Different and Same responses made by the birds. Given this, could this alternative unidimensional hypothesis account for the discriminations tested in Cook et al. (1995), Cook & Wixted (1997), and the present experiments? That is, could these birds have learned to treat the two choice hoppers as representing Alow@ and Ahigh@ values of display entropy rather than as ASame@ and ADifferent@ choice alternatives? Because they only used a single display type, an entropy-based account of Cook et al. (1995) and Cook & Wixted (1997) is certainly not at odds with any of their findings.
An entropy-based account of the present experiment=s results, however, is considerably more problematic. Young and Wasserman=s method for computing entropy is keyed to the number of different types of elements present in the display (Equation 1 in Young & Wasserman, 1997). While this works well for describing their particular icon-based displays, it creates some problems when applied to separate display types tested here. Using their formula, for instance, it turns out that the computed entropy for the Same feature displays (entropy value=1) is actually larger than for any of the Different displays of the other three display types (values between .59 and .65). This greater entropy in the feature displays is directly due to the presence of the irrelevant local variation in its elements. As such, if the birds were learning to respond simply based on the entropy of component elements in the displays, then the resulting incompatible mapping of the feature display=s entropy values relative to the other three display types should have made this former display much harder for the birds to learn about. This clearly was not the case.
Far more problematic for an entropy-based analysis of the present task are subsequent observations collected from these same birds when we varied the number of distractor elements in the object and geometric displays. In this experiment we introduced and tested two new display organizations. Besides the standard 3 x 2 organization, we tested 2 x 2 and 3 x 1 stimulus arrays of the geometric or object elements. Like before, the difference between Same and Different displays consisted of the presence and absence of a single odd target item. For all three of these organizations the entropy of their Same displays is zero.
But for the Different displays, the entropy is greatest in the 3 x 1 array (1 target & 2 distractors; entropy =.91), intermediate for the 2 x 2 array (1 target & 3 distractors; entropy =.81), and smallest for the 3 x 2 array (1 target & 5 distractors, entropy =.65). Thus, if the birds were responding only to entropy, the proportion of different responses should be highest to the 3 x 1 arrays, followed by 2 x 2, and then the 3 x 2 arrays. The figure below shows the mean results for 38 sessions testing these three display organizations. They were collected about six months after the completion of Experiment 2. The figure below shows that the birds continued to response accurately to the Same displays regardless of the number of elements, but showed a systematic decline in Different responding as the number of distractors in the array decreased. This difference was confirmed by a significant Trial Type x Display Organization interaction, F(2,8)=6.8, in these data. This outcome is directly opposite of the one predicted by an entropy-based account of the discrimination. Rather this pattern suggests that the present birds were being more strongly influenced by the relative oddity of the target in these displays (see Blough, 1989 for a similar result).
Thus, this collective pattern of results suggests that neither an entropy-based nor a perceptually-based account can easily accommodate the entire set of Same-Different findings reported here. We propose that the most parsimonious unidimensional interpretation is that the pigeons employed a single abstract Same-Different rule in processing each of the four display types. The acceptance of this conclusion carries the further implication that the birds in Wasserman et al. (1995) and Cook et al. (1995) may have learned different rules for dealing with what looked like otherwise quite comparable tasks. Perhaps because Wasserman et al. (1995) employed Different displays containing the largest possible number of contrasting icons (16 different icons in a 4 x 4 array), their birds were more sensitized to the variability or entropy dimension in their displays. In contrast, Cook et al. (1995) and the current study employed dimensional differences that resulted in generally smaller entropy differences between the Same and Different displays, and which were always embodied as spatially localized contrasts, both factors that may have promoted a more oddity-based evaluation of the identity relations in these displays.
In sum, the present acquisition and transfer results provide some of the strongest evidence yet collected for the existence of the ability in pigeons to learn a generalized ASame-Different@ concept, at least as mediated by global differences in the color and shape dimensions. As well, the present multiple class Same-Different task will also provide an excellent vehicle for the study of many fundamental questions about avian visual cognition, such as the relation between visual grouping and visual search, the identification of fundamental visual features, their multidimensional integration, and how these integrated features eventually become perceived as the objects that appear to control avian behavior in the wild. The answers to such questions will form an important advance towards a unified comparative theory of visual cognition in human and non-human animals.
http://www.pigeon.psy.tufts.edu/jep/images/fig2.gif
````````````````````````````````````````````
http://www.pigeon.psy.tufts.edu/jep/sdmodel/images/3x1.gif http://www.pigeon.psy.tufts.edu/jep/sdclass/images/msc_fig7.gif http://www.pigeon.psy.tufts.edu/jep/sdmodel/images/2x2.gif
― lf (lfam), Tuesday, 2 May 2006 06:19 (seventeen years ago) link