Two sources of proactive interference in spatial working memory: Multiple effects of repeated trials on radial maze performance by rats

Rats were trained in 8- and 12-arm radial mazes. Each trial began with a study phase (forced choices of 4 arms). The trial ended after a 2-h delay in a test phase consisting of free choices among 8 arms; choices of the 4 arms not yet visited were correct (rewarded). Proactive interference (PI) was induced by an interference phase that occurred on some days 2 or 3 h prior to the study phase. In the PI-repetition condition, the interference phase consisted of forced choices of the 8 arms that were later presented in the study and test phases; in the PI-nonrepetition condition, the interference consisted of forced choices of the 4 arms that were correct during the test phase. Test-phase performance was most accurate in the No PI (single-trial) condition and least accurate in the PI-repetition condition. A second experiment showed that repetitions per se were not responsible for the PI; when the interference phase consisted only of choices of the same 4 arms later presented in the study phase, no PI was observed. These findings suggest two sources of PI. One source, measured by the difference between No PI and PI-nonrepetition conditions, appears to be a difficulty in discriminating the temporal order of visits to arms in the interference and study phases. The other source of PI, measured by the difference between the nonrepetition and repetition conditions, remains to be identified; some possibilities are discussed.