Abstract
Ensemble coding is the brain’s ability to rapidly extract summary statistics from groups of similar items (e.g., average colour of leaves on a tree). Ensemble coding has also been found for information that cannot be gleaned solely from retinal input (e.g., average animacy of objects). We extended this line of ensemble coding research by examining if observers were sensitive to two different features that require access to stored information in long-term memory. Specifically, participants made judgements about the average weight or real-world size of groups of objects. We found that participants were unable to integrate information from multiple items to produce accurate summary statistics for either judgement. Next, we examined how learning and memory may influence ensemble coding for weight and real-world size. Could associative learning (e.g., learning which items belong in an ensemble via their relationship to other items) aid in ensemble coding? Specifically, we examined if repeated exposures to the different items within an ensemble would enable people to form summary statistics more efficiently. To test this, participants made judgements about the average weight or average real-world size of groups of object stimuli, and we manipulated how frequently people were exposed to certain subsets and combinations of the stimuli in a given ensemble. We found that repeated presentation of the ensemble stimuli leads to improved performance on the ensemble tasks for both the weight and size judgements. In summary, while observers were unable to extract average weight and real-world size information from objects when stimuli were not repeatedly presented, they were able to produce accurate summary statistics with multiple stimulus exposures. We speculate that this may be due to mechanisms governing associative statistical learning and memory for repeatedly encountered visual information.