Five pigeons matched element and compound samples in a symbolic matching-to-sample procedure. In Experiment 1, the sample duration varied within each session, ranging from .125 to 8 sec. Element matching accuracy was superior to compound matching accuracy, and matching accuracy improved with sample duration; but there was no evidence for the convergence of element and compound matching accuracy with long sample durations. In Experiment 2, the sample-to-test delay varied from 0 to 4 sec. The superiority of element matching accuracy over compound matching accuracy did not vary as a function of delay length. These replications of previous failures to find interactions between sample type and temporal parameters in determining matching accuracy place important constraints on explanations of element superiority in pigeon matching-to-sample performance.The present experiments were carried out in an attempt to solve a dilemma in the literature on the element superiority effect in pigeon matching-to-sample (MTS). It has been consistently reported that element matching accuracy is superior to compound matching accuracy (see, e .g. , Maki & Leith, 1973;Roberts & Grant, 1978;Santi , Grossi, & Gibson, 1982) . This has been argued to result from information overload-that is, from the additional information that is provided by a compound sample (see , e.g., . Recent evidence favors this information overload hypothesis over several competing explanations of the element superiority effect (Brown & Morrison, 1990). The favored version of the information overload view points to information uptake as the locus of information overload . That is, information overload occurs as a result of shared attention at the time of sample presentation. Attention must be distributed among the two dimensions of a compound sample, whereas it can be allocated entirely to the single dimension of an element sample. A second possible locus of information overload is memory storage (Grant, 1981). Memory load might be higher following a compound sample than following an element sample.However, several experiments, following similar logic, have failed to provide evidence for either of these versions of the information overload hypothesis. If information overload occurs at the time of sample presentation (information uptake) , then the element superiority effect should be attenuated with longer sample durations. There have been a number of reported failures to fmd this