Abstract
Identification thresholds and corresponding efficiencies (ideal/human thresholds) are typically computed by collapsing data across an entire stimulus set within a given task in order to obtain a ‘multiple-item’ summary measure of performance. However, some individual stimuli may be processed more efficiently than others, and such differences are not captured by conventional multiple-item threshold measurements. Here, we present a simple technique for measuring ‘single-item’ identification efficiencies. The technique involves measuring identification performance for both human and ideal observers at a range of contrast levels using the method of constant stimuli. Each stimulus is shown the same number of times at a fixed set of contrast levels, randomly permuted across trials. The data are sorted conditional upon stimulus identity, allowing a single-item psychometric function to be computed for each individual stimulus. Human and ideal single-item thresholds are then compared for each stimulus in order to obtain single-item efficiencies. The resulting single-item efficiencies describe the ability of the human observer to make use of the information provided by a single stimulus item within the context of the larger set of stimuli. We applied this technique to the identification of several different classes of complex patterns embedded in noise, including 3-D rendered objects and human faces. Our results show that efficiency can vary markedly across stimuli within a given task, demonstrating that single-item efficiency measures can reveal important information lost by conventional multiple-item efficiency measures.