Multiple feature use in pigeons' category discrimination: The influence of stimulus set structure and the salience of stimulus differences.

Two experiments investigated what makes it more likely that pigeons' behavior will come under the control of multiple relevant visual stimulus dimensions. Experiment 1 investigated the effect of stimulus set structure, using a conditional discrimination between circles that differed in both hue and diameter. Two training conditions differed in whether hue and diameter were correlated in the same way within positive and negative stimulus sets as between sets. Transfer tests showed that all pigeons came under the control of both dimensions, regardless of stimulus set structure. Experiment 2 investigated the effect of the relative salience of the stimulus differences on three visual dimensions. Pigeons learned a multiple simultaneous discrimination between circular patches of sinusoidal gratings that differed in hue, orientation, and spatial frequency. In initial training, each stimulus only included one positive or negative feature, and the stimulus differences on the three dimensions were adjusted so that the rates of learning about the three dimensions were kept approximately equal. Transfer tests showed that all three dimensions acquired control over behavior, with no single dimension dominating consistently across pigeons. Subsequently the pigeons were trained in a polymorphous category discrimination using all three dimensions, and the level of control by the three dimensions tended to become more equal as polymorphous training continued. We conclude that the salience of the stimulus differences on different dimensions is an important factor in whether pigeons will come under the control of multiple dimensions of visual stimuli. (PsycINFO Database Record

Labeling and family resemblance in the discrimination of polymorphous categories by pigeons.

Two experiments examined whether pigeons discriminate polymorphous categories on the basis of a single highly predictive feature or overall similarity. In the first experiment, pigeons were trained to discriminate between categories of photographs of complex real objects. Within these pictures, single features had been manipulated to produce a highly salient texture cue. Either the picture or the texture provided a reliable cue for discrimination during training, but in probe tests, the picture and texture cues were put into conflict. Some pigeons showed a significant tendency to discriminate on the basis of the picture cue (overall similarity or family resemblance), whereas others appeared to rely on the manipulated texture cue. The second experiment used artificial polymorphous categories in which one dimension of the stimulus provided a completely reliable cue to category membership, whereas three other dimensions provided cues that were individually unreliable but collectively provided a completely reliable basis for discrimination. Most pigeons came under the control of the reliable cue rather than the unreliable cues. A minority, however, came under the control of single dimensions from the unreliable set. We conclude that cue salience can be more important than cue reliability in determining what features will control behavior when multiple cues are available.

A comparative analysis of the categorization of multidimensional stimuli: I. Unidimensional classification does not necessarily imply analytic processing; evidence from pigeons (Columba livia), squirrels (Sciurus carolinensis), and humans (Homo sapiens).

Pigeons (Columba livia), gray squirrels (Sciurus carolinensis), and undergraduates (Homo sapiens) learned discrimination tasks involving multiple mutually redundant dimensions. First, pigeons and undergraduates learned conditional discriminations between stimuli composed of three spatially separated dimensions, after first learning to discriminate the individual elements of the stimuli. When subsequently tested with stimuli in which one of the dimensions took an anomalous value, the majority of both species categorized test stimuli by their overall similarity to training stimuli. However some individuals of both species categorized them according to a single dimension. In a second set of experiments, squirrels, pigeons, and undergraduates learned go/no-go discriminations using multiple simultaneous presentations of stimuli composed of three spatially integrated, highly salient dimensions. The tendency to categorize test stimuli including anomalous dimension values unidimensionally was higher than in the first set of experiments and did not differ significantly between species. The authors conclude that unidimensional categorization of multidimensional stimuli is not diagnostic for analytic cognitive processing, and that any differences between human's and pigeons' behavior in such tasks are not due to special features of avian visual cognition.

A comparative analysis of the categorization of multidimensional stimuli: II. Strategic information search in humans (Homo sapiens) but not in pigeons (Columba livia).

Pigeons and undergraduates learned conditional discriminations involving multiple spatially separated stimulus dimensions. Under some conditions, the dimensions were made available sequentially. In 3 experiments, the dimensions were all perfectly valid predictors of the response that would be reinforced and mutually redundant; in 2 others, they varied in validity. In tests with stimuli in which 1 of the 3 dimensions took an anomalous value, most but not all individuals of both species categorized them in terms of single dimensions. When information was delivered as a function of the passage of time, some students, but no pigeons, waited for the most useful information, especially when the cues differed in objective validity. When the subjects could control information delivery, both species obtained information selectively. When cue validities varied, almost all students tended to choose the most valid cues, and when all cues were valid, some chose the cues by which they classified test stimuli. Only a few pigeons chose the most useful information in either situation. Despite their tendency to unidimensional categorization, the pigeons showed no evidence of rule-governed behavior, but students followed a simple "take-the-best" rule.

The logic of the stimulus.

This paper examines the contribution of stimulus processing to animal logics. In the classic functionalist S-O-R view of learning (and cognition), stimuli provide the raw material to which the organism applies its cognitive processes-its logic, which may be taxon-specific. Stimuli may contribute to the logic of the organism's response, and may do so in taxon-specific ways. Firstly, any non-trivial stimulus has an internal organization that may constrain or bias the way that the organism addresses it; since stimuli can only be defined relative to the organism's perceptual apparatus, and this apparatus is taxon-specific, such constraints or biases will often be taxon-specific. Secondly, the representation of a stimulus that the perceptual system builds, and the analysis it makes of this representation, may provide a model for the synthesis and analysis done at a more cognitive level. Such a model is plausible for evolutionary reasons: perceptual analysis was probably perfected before cognitive analysis in the evolutionary history of the vertebrates. Like stimulus-driven analysis, such perceptually modelled cognition may be taxon-specific because of the taxon-specificity of the perceptual apparatus. However, it may also be the case that different taxa are able to free themselves from the stimulus logic, and therefore apply a more abstract logic, to different extents. This thesis is defended with reference to two examples of cases where animals' cognitive logic seems to be isomorphic with perceptual logic, specifically in the case of pigeons' attention to global and local information in visual stimuli, and dogs' failure to comprehend means-end relationships in string-pulling tasks.

Why are artificial polymorphous concepts so hard for birds to learn?

Stimulus sets defined in terms of artificial polymorphous concepts have frequently been used in experiments to investigate the mechanisms of discrimination of natural concepts, both in humans and in other animals. However, such stimulus sets are frequently difficult for either animals or humans to discriminate. Properties of artificial polymorphous stimulus sets that might explain this difficulty include the complexity of the individual stimuli, the unreliable reinforcement of individual positive features, attentional load, difficulties in discriminating some stimulus dimensions, memory load, and a lack of the correlation between features that characterizes natural concepts. An experiment using chickens as subjects and complex artificial visual stimulus sets investigated these hypotheses by training the birds in discriminations that were not polymorphous but did have some of the properties listed above. Discriminations that involved unreliable reinforcement or high attentional load were found to approach the difficulty of polymorphous concept discriminations, and these two factors together were sufficient to account for the entire difficulty. The usual kind of artificial polymorphous concept may not be a good model for natural concepts as they are perceived and discriminated by birds. A RULEX account of natural concept learning may be preferable.