Abstract
Ensemble coding, the brain’s ability to rapidly extract summary statistics from groups of similar items, has been demonstrated across a range of low-level (e.g., average colour), mid-level (e.g., average size), to high level (e.g., average facial expression) visual features. More recently, ensemble coding has been found for information that cannot be gleaned solely from retinal input (e.g., average lifelikeness), potentially requiring the retrieval of semantic information from long term memory (LTM) (e.g., average economic value). While ensemble coding can be applied to a range of physical and abstract impressions, the limits of this cognitive mechanism are unclear. I extended this line of abstract ensemble coding research by examining if different sensory modalities can interact during ensemble coding. Specifically, participants made judgements about the average sweetness of groups of different visually presented foods. We found that, under specific circumstances (e.g., longer durations, using sequential presentation), observers were able to incorporate all the available information when making their cross-modal ensemble judgements. However, under other viewing conditions (e.g., shorter durations, using simultaneous presentations) we observed a capacity limit for the number of items that could be incorporated when making cross-modal ensemble judgements. Interestingly, these results did not appear to depend upon individual differences in visual working memory (VWM) capacity. Overall, these results demonstrate that there is a limit to the flexibility of ensemble coding, particularly when multiple cognitive systems must interact (e.g., visual perception, taste perception, VWM, LTM) to compress sensory input into an ensemble representation.