Abstract
Performing an action enhances our perception of events related to that action (Shutz-Bosbach & Prinz, 2007). Here, we show that manual tracing of a scatterplot can improve visual trend perception in a realistic task where participants are comparing performance of 4-8 companies. Four, six, or eight scatterplots (representing companies A, B, C, D, etc.) were sequentially presented (1 sec per plot) on an iPAD. In the tracing condition, participants traced (with their index finger) what they thought indicated the line of best fit for each scatterplot. In the passive condition, participants visually estimated the linear trend for each scatterplot without tracing. After the last scatterplot was presented participants indicated which company had the fastest growth by touching the appropriate button. We measured RTs (from the disappearance of the last scatterplot to response) and accuracy (frequency of choosing the fastest growing company). When 4 or 8 scatterplots were presented, tracing did not produce a clear gain in performance; although tracing reduced RTs it increased errors. In contrast, when 6 scatterplots were presented, tracing improved performance by reducing RTs without increasing errors. Furthermore, in the tracing condition, those who traced linear trends more accurately (i.e., slopes of their traced lines being more similar to those of the actual regression lines) made fewer response errors when 4 or 6 scatterplots were presented but not when 8 scatterplots were presented. Thus, the perceptual gain from manual tracing has capacity limits. Taken together, these results suggest that (1) accurate tracing of trend lines from scatterplots improves encoding of linear trends with a capacity limit of about 6, and (2) tracing facilitates mental comparison of linear trends when working memory load is moderate (6 scatterplots). Our results suggest that action enhances visual perception of statistical trends in a realistic task that requires working memory.
Meeting abstract presented at VSS 2013