Sample frequency and sample duration as sources of stimulus control in delayed matching to sample

Three experiments examined sample duration and sample presentation frequency (SPF) on choice in a two-alternative, delayed matching-to-sample task. In Experiment 1, using a behavioural-detection approach, we demonstrated bias toward the more frequent sample, despite the conditional probability of reinforcement for a correct match on any particular trial remaining at 1.0 for both stimuli. Although retention-interval duration influenced both discriminability and bias in Experiment 1, bias was independent of retention interval in two subsequent experiments. Experiments 2 and 3 replicated the effects of SPF on bias, and demonstrated that discriminability of the stimuli was not influenced by the SPF manipulation. Experiment 2 also investigated the effect of a within-session variation in sample duration on discriminability and bias measures, both with and without unequal SPFs. Discriminability was enhanced to both the short and long samples of the unequal sample-durations' condition relative to discrimination in conditions where samples were presented for equal durations. Bias generated by varying SPF was independent of sample-duration effects and of retention-interval duration. Taken together, these data suggest independent and qualitatively different effects of sample frequency and sample duration on matching behaviour: Sample frequency has its effect on bias measures, while sample duration influences discriminability. We suggest that sample frequency is a global task factor influencing reference memory, and sample duration is a trial-specific, conditional discrimination factor, involving short-term or working memory processing.