When Summary Statistics Clash: Competing summary statistics modulate the attentional prioritization of ensemble representations

Ensemble processing allows the visual system to condense visual information into useful summary statistics (e.g., the average size of a group of objects). We investigated whether such representations receive priority in the attentional system by merging a typical size-based ensemble-processing task with a variation of an attentional-cueing task. Participants saw a display of eight randomly sized ovals and then a second display of two ovals. In the second display, one of the ovals was the average size of the previous eight ovals (target), while the other was a differently sized oval (distractor). On most trials, participants were instructed to report the average size of the ensemble by choosing one of the two ovals in the second display. On critical trials, a probe dot was presented on either the target or distractor oval, and participants were instructed to make a keypress to localize the probe as quickly as possible. We hypothesized that if attention prioritizes ensemble representations, reaction times (RTs) would be faster when the probe appeared on the target compared with the distractor object. Experiment 1 examined this under conditions of low interference, wherein the size of the distractor was outside of the range of size values encountered in the previously seen ensemble. Here, we found faster RTs when the probe appeared on the target compared with the distractor object. In contrast, Experiment 2 used a high-interference ensemble task, where the size of the distractor was within the range of size values in the ensemble, and, interestingly, no difference in RTs was found when the probe was on the target compared with the distractor object. Together, these results demonstrate that ensemble representations can receive priority by the attentional system (Exp1), but this effect is attenuated when multiple summary statistics (i.e., mean, range) compete for attentional priority (Exp2).