Abstracts from Selected Articles by Dr. Robert G. Cook
Pictorial Same-Different Categorical Learning and Discrimination in Pigeons
Two experiments examined the range of conditions over which five pigeons previously demonstrated to have a generalized same-different concept would transfer this behavior. Using two-alternative choice task that required discrimination of odd-item Different displays, in which contrasting elements were present, from Same displays, in which all of the elements were identical, these birds were transfer tested to novel display types that were outside the stimulus range of their established discriminative behavior. Experiment 1 found evidence of positive transfer for a majority of the birds to semi-realistic gray scale pictures of objects from four categories (birds, flowers, fish, & humans). Experiment 2 found evidence of positive transfer for a majority of the pigeons to realistic gray scale and color photographs of objects from six categories (birds, flowers, cars, cats, dogs & buildings). The results suggest these pigeons possessed a generalized rule that could be applied to novel stimuli from outside their direct experience, and add to the growing evidence that pigeons can form broadly-defined relational same-different concepts.
Three experiments examined pigeon discrimination of computer-generated 3D projections of cube and pyramid objects. Four pigeons were tested using a go/no-go procedure involving static and dynamically rotating presentations of these stimuli. Transfer tests with different types of rotational and featural transformations suggested the pigeons may have used a three dimensional representation of the objects as their primary means of performing the discrimination. The comparative implications for object and motion perception in animals are considered.
Two experiments examined the acquisition and transfer of a complex Same-Different discrimination by pigeons. Using a two-alternative choice task, five pigeons were reinforced for discriminating odd-item Different displays, in which a contrasting target was present, from Same displays where all the elements were identical. Four different types of Same-Different displays were concurrently tested. The display types differed in their configuration (texture versus visual search organization), the nature of their elements (small and large colored shapes; pictures of birds, flowers, fish, and humans), and the processing demands required by their global/local element arrangement. Despite these differences, the birds learned to discriminate all four display types at the same rate and showed positive discrimination transfer to novel examples of each type suggesting that a single generalized rule was used to discriminate all display types. These results provide some of the strongest evidence yet that pigeons, like many primates, can learn an abstract visually-mediated "Same-Different" concept.
Publication List | Home Page | Article's web site
The perception and discrimination of rapidly changing texture stimuli by pigeons was examined in a target localization task. Five experienced pigeons were rewarded for finding and pecking at a randomly placed odd target block of small repeated elements embedded in a larger rectangular array of contrasting distractor elements. Target/distractor contrasts were formed by element differences in color or shape. On dynamic color test trials, the color of the target, distractor, or both of these regions changed at rates of either 100, 250, 500, or 1000 ms per frame. The number of colors appearing within such trials also varied. Pigeons performed well above chance in all test conditions, with target associated color changes producing the best discrimination. The results suggest: 1) global relational information can exclusively guide target localization behavior, 2) pigeons can perceptually group and segregate colored textured differences quite rapidly (<= 100 ms) and, 3) may possess automatic search control processes that can be Acaptured@ by stimulus-driven changes in the display.
Publication List | Home Page| Article's web site
The choice behavior of six pigeons performing a multidimensional Same-Different texture discrimination was examined. On each trial, they had to choose among two "choice" hoppers depending on whether a color, shape, or redundant (color & shape) target signal was present or not in a textured stimulus. ROC curves were produced by variations in the a priori signal presentation probabilities across conditions. Quantitative analyses of these ROC curves were used to evaluate different competing theories of discrimination (signal detection versus high threshold/default response models) and information integration (independent observations, additive integration, unidimensional models). The results suggested the structure of the pigeons' choice behavior in the Same-Different discrimination was best described by an unequal variance signal detection model involving a unidimensional evidence variable (e.g., degree of difference).
Publication List | Home Page | Article's web site
Three experiments using a new reference memory procedure examined how male rats search for consistently located food in a cue-controlled spatial environment. The animals searched the tops of 24 poles for six hidden baits in an enclosed circular arena containing a fixed configuration of six object landmarks. In Experiment 1, acquisition was faster and overall performance superior for the Consistent group (10 rats), in which the six baited poles were fixed relative to the landmarks for each session, than for the Random group (4 rats), in which baited poles were randomly configured. Cue-control tests and computer simulations suggested that the Consistent group relied on the landmarks to directly go to the baited poles, while the Random group used them to employ a response strategy for searching the arena. Experiments 2 and 3 revealed that the number, identity, and geometric configuration of the landmarks were important to the Consistent group's search performance. Overall, these results are most consistent with the use of a geometric representation by male rats which includes information about both the identity and relative geometry of discrete landmarks in the surrounding spatial environment.
Three tests investigated how the geometric relation between object/landmarks and goals influenced spatial choice behavior in rats. Two groups searched for hidden food in an object-filled circular arena containing 24 small poles. For the Proximal group, four distinct objects in a square configuration were placed adjacently to four baited poles. For the Distal group, the identical configuration of objects was rotated 45° relative to the poles containing the hidden food. The Proximal group learned to locate the baited poles more quickly than the Distal group. Tests with removed and rearranged landmarks indicated that both groups learned to use the objects differently. The results suggested that the close proximity of objects to the goals encouraged their use as beacons, while more distal spatial relations resulted in the global encoding of the geometric properties of the arena and the use of the objects as landmarks.
The acquisition and transfer of a "same/different" conditional discrimination using multidimensional visual texture stimuli was investigated in pigeons. Using a choice task, four pigeons were reinforced for discriminating different displays, created from aggregated differences in element color or shape, from uniform displays, in which all elements were identical. Discrimination of these two display types was readily acquired by the birds when required to locate and peck the contrasting target region of the different displays. The birds showed high levels of discrimination transfer to novel texture stimuli both during acquisition and in two subsequent transfer tests. The results suggest pigeons may be able to learn a generalized "same/different" concept when promoted by the use of large numbers of multi-element stimuli during training.
Three experiments examined the influence of dimensional organization on pigeon texture perception in a simultaneous conditional discrimination procedure. Six experienced pigeons were reinforced for pecking at a small block of target elements randomly located within a larger array of distractor elements in each texture stimulus. Target/distractor differences in color, size, orientation, and combinations of these dimensions were examined. Experiment 1 investigated how target/distractor similarity influenced performance with different forms of unidimensional and conjunctively-organized texture stimuli made of two and three dimensions. Targets in feature displays, in which the two regions consistently differed along a single dimension, were located more accurately than in conjunctive displays, where a combination of values from all dimensions defined each region. Experiment 2 found a trade-off between response speed and accuracy in the pigeons' processing of conjunctive displays. Experiment 3 found that the number of distractors differentially influenced the localization of feature and conjunctive targets. Overall, the pigeons' reactions to these feature and conjunctive stimuli paralleled those of humans, suggesting that functionally equivalent mechanisms may mediate the perceptual grouping, search, and discrimination of textured multidimensional stimuli in both species.
Revised 08/25/99