Numerosity differences and effects of stimulus density on pigeons’ discrimination performance

Abstract

Two experiments are described in which pigeons were trained in a simultaneous conditioning procedure to discriminate small arrays of dots that differed in numerosity. The birds successfully learned to choose the array of each pair that contained fewer dots when these choices were reinforced and choices of the array with more dots led to timeout. For the majority of numerosity values tested, discrimination performance for a fixed S+ value was better when the numerical difference between S+ and S-values was larger rather than smaller. This effect was seen in the first experiment when the numerical difference value was shifted between training trials and novel test trials. In the second experiment, too, performance level depended on the size of the numerosity difference when the birds were concurrently trained with two difference values that varied across trials within sessions. However, discrimination accuracy was influenced secondarily by variations in the density, or interdot spacing, of the stimulus arrays. In order to explain the latter finding, it is suggested that a tendency to "scan" a lowdensity array incompletely might alter the probability of accepting it as the smaller numerosity (8+) stimulus. This would increase error rates with S-arrays in which the dots are more widely spaced.

Scalar effects in the visual discrimination of numerosity by pigeons

“…So the birds did not choose "large" on the basis of larger extent of the arrays, however that is measured.) The secondary effects on performance of manipulating the spacing and arrangement of array items more closely resembled earlier findings obtained when pigeons were tested with small numbers of dots (Emmerton, 1998) than the effects reported in humans' estimation of large numbers of dots.…”

“…In an experiment on the effects of array density with small numbers of dots (Emmerton, 1998), pigeons showed effects that tended to be the opposite of those found with humans. Although the birds' discrimination performance was primarily dependent on the relative numerosities of dots in the stimulus arrays, the results indicated that the spacing of dots biased this performance in such a way that numerosity was judged to be less when the dots were spread out than when they were more closely spaced.…”

“…We suspect that the birds in our experiment most likely employed a visual search strategy of scanning localized regions of the stimulus display for individual items, along the lines suggested by Emmerton (1998), rather than integrating information across the whole array. Any bias that pigeons might have to process local stimulus components rather than a global pattern would also be enhanced by the variability in overall array configurations the birds encountered in all our experiments.…”

When in doubt, chimpanzees rely on estimates of past reward amounts

“…These aspects of performance indicate a 'fuzzy' sense of the average reward rate that matches other reports from quantity comparison tasks used with animals (e.g. Emmerton 1998;Brannon & Terrace 2000;Beran 2004Beran , 2007Judge et al 2005;Cantlon & Brannon 2006;Agrillo et al 2008). We believe that the chimpanzees attended to the outcomes of each trial and used trial-by-trial feedback of the quantity obtained to modify some representation 1 of their overall reward rate for that day.…”

Discrimination and representation of relative numerosity in a bisection task by pigeons

“…Research has studied how nonhumans bisect both visual and auditory numerical stimuli (e.g., Alsop & Honig, 1991;Fernandes & Church, 1982;Honig & Stewart, 1989;Meck & Church, 1983;Roberts & Mitchell, 1994), as well as response number (Emmerton, 1998;Emmerton & Renner, 2006;Fetterman, 1993;Rilling & McDiarmid, 1965). These have found largely positive results, showing that animals are able to respond differentially to stimuli that differ in relative numerosity, successfully categorizing them as large or small, and are able to transfer this performance to novel numerical values both within and outside the training range.…”