Matching to compound samples by monkeys (Cebus apella): Shared attention or generalization decrement?

Abstract

In Experiments 1 and 2, cebus monkeys matched element comparison stimuli to either element or color-form compound samples, with retention intervals ranging from ,5 to 32 sec. In both experiments, performance on compound sample trials was substantially below performance on element sample trials at the .5-sec delay, and the retention gradients were essentially parallel. These results were viewed as supporting the generalization-decrement rather than the shared-attention interpretation of how animals process compound samples. Experiment 3 employed form-form rather th'an color-form compounds and obtained the same results with simultaneous matching and delays ranging to 128 sec. Experiment 4 verified the findings of Experiment 3 for both form-form and color-form compounds and showed that as predicted by the generalization-decrement view, when the compound sample served as the correct comparison stimulus, performance was closely comparable to that achieved on element trials. The present results, taken in conjunction with earlier work with pigeons, suggest that pigeons and monkeys do not readily abstract the elements of a compound sample, which seems at odds with the pigeon's reported ability to rapidly form concepts based on natural and artificial categories.In an interesting series of experiments, sion (red vs. green, or vertical vs. horizontal Maki and his associates investigated shared lines), matching behavior was poorer than attention in the pigeon in the context of when the sample stimulus consisted of a sinmatching to sample with element and com-gle element (e.g., red or vertical lines). This pound sample stimuli (Maki & Leith, 1973; difference in performance, which survived Maki & Leuin, 1972; Maki, Riley, & Leith, extensive training with compound and ele-1976). The major result of these studies was ment samples, can be interpreted in a numthat when presented with a compound sam-ber of ways. That favored by Maki and his pie stimulus (e.g., vertical lines superinv associates is that when presented with a composed on a red background) followed by a pound sample of relatively brief duration, choice between the elements of one dimen-the bird must share its attention between the , two available elements and, due to a limited This research, based in part on a master's thesis and information-processing system, neither ele-PhD dissertation of j. K. Cox, was supported by grants ment is processed to the same degree as when from the National Science Foundation. j t occurs alone. These investigators ruled out D'&e^ the mor « ? b y ious and Parsimonious expla-

A search for the locus of information overload in pigeon compound matching-to-sample performance

“…Taken together with the recent data of Brown and Momson (1990), the present results place severe constraints on the interpretation of the element superiority effect. Althoughprevious data have indicated a role for both generalization decrement (Cox & D'Amato, 1982;Roberts & Grant, 1978)and training history (Grant & MacDonald, 1986), Brown and Morrison's (1990)data rule out theuse of these as sufficientexplanations for theelementsuperiorityeffectfoundin thesubjectsin thepresent experiments. If infonnationoverloadoccursat thetimeof sampleencoding, this would seem to predict rather strongly an interaction between sampletype and sampleduration.The factthat, evenunderthese improved DELAY (Sec.…”

“…Likewise, if information overload occurs during the storage of information between sample presentation and test response, then the element superiority effect should increase when a delay is inserted between the sample and presentation of the test stimuli. This effect has not been found (Cox & D'Amato, 1982 ;Roberts & Grant, 1978). Thus, while on the one hand information overload is currently the favored explanation of the element superiority effect, on the other hand, attempts to specify the locus of that overload have been unsuccessful.…”

Matching visual stimuli on the basis of global and local features by chimpanzees (Pan troglodytes) and rhesus monkeys (Macaca mulatta)

“…Specifically, several investigators have examined the role of attention and memory in the processing of perceptual elements that comprise compound stimuli (Maki and Leuin 1972;Cox and D'Amato 1982;Riley 1984;Lamb 1988Lamb ,1991. For example, in a matching-to-sample paradigm, a red circle with a white horizontal line is presented as the sample stimulus.…”

“…Results indicate that pigeons and monkeys perform significantly better on trials in which the stimuli are derived of a single perceptual element compared to compound stimulus trials. There remains some debate over the interpretation of these results, with some suggesting that the poorer performance on the compound test trials is due to stimulus generalization decrements (Cox and D'Amato 1982;D'Amato and Salmon 1984;Grant and MacDonald 1986) while others have suggested that there is a "limited capacity" for the amount of information that can be encoded within a given time frame by pigeons and monkeys (Riley and Roitblat 1978).…”

Generalization gradients following training on a symbolic matching-to-sample task

“…A further manipulation might involve increasing the number of stimuli contained in the sample. It is a common finding that matching is better when the sample contains only one element, and when an extra element is added, matching accuracy declines (Cox & D'Amato, 1982;Richards & Bowers, 1985;Santi, Grossi, & Gibson, 1982). It has been suggested (Leith & Maki, 1975;Maki & Leith, 1973;Maki & Leuin, 1972;Maki, Riley, & Leith, 1976;Roberts & Grant, 1978) that dual-element samples may require the subject to divide its attention among the separate elements, thereby taxing its attentional resources (for further discussion, see Riley & Roitblat, 1978).…”