Persistent resampling of external information despite twenty-five repetitions of the same search templates

We commonly load visual working memory minimally when to-be-remembered information remains available in the external world. In visual search, this is characterised by both encoding novel (externally available) templates during search and revisiting previously encoded templates. When all search templates have been encoded, subsequent rehearsal should lead to even stronger representations in memory, eventually making the (re)inspection of templates redundant. To test whether repetition indeed leads to decreased external sampling, participants (n=15) searched for 1, 2, and 4 search templates which remained available throughout trials. Critically, each unique set of search templates was repeated 25 trials consecutively. When presented with a novel template set, participants primarily encoded and searched templates one-by-one. Although the number of inspections and inspection durations initially decreased significantly as template sets were repeated, behaviour largely stabilised between the tenth and last repetition and participants still frequently resampled templates. In Experiment 2, participants (n=14) performed the same task, but could not inspect templates in the last 10 repetitions. Strikingly, accuracy remained high when templates could not be resampled, although the number of inspections decreased faster than in Experiment 1. This suggests that participants prepared for template removal by encoding templates more elaborately before the cut-off point. Across both experiments, resampling was linked to increased response times and no benefit to accuracy. Furthermore, participants could draw from long-term memory for the search task, as indicated by above-chance level recognition of template sets ten minutes after the main experiment finished. We thus argue that external sampling is not only used to offload effortful working memory maintenance, but also to boost confidence in existing (long-term) memory content. These findings demonstrate the persistence with which we sample information from the external world rather than rely on memory – even when we repeatedly need that same information.