Purchase this article with an account.
Madison Elliott, Ronald Rensink; Interference in the Perception of Two-Population Scatterplots. Journal of Vision 2015;15(12):893. doi: 10.1167/15.12.893.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The visual system represents correlation in much the same way that it represents simple visual quantities such as density or brightness (Rensink 2014). For example, just noticeable differences (jnds) show linear behavior closely approximating Weber’s law. But until recently, testing was done using only single populations of data. However, displaying two or more populations in a single graph is a common practice. Participants completed a two-condition correlation discrimination task, which was counterbalanced to control for order effects. The first condition closely resembled the original task from Rensink & Baldridge (2010); observers viewed two side-by-side scatterplots, each containing a single data population. They were instructed to select the plot with the higher correlation. The second condition had two data populations, each of a different color. As in the first condition, observers chose the higher correlation of the first (target) population, but were now required to ignore the second (distractor) population. Results from the first condition replicated earlier findings (Rensink, 2014). However, results from the second condition showed that when a distractor population was present, strong violations of Weber’s law appeared. Jnds were now larger overall, and deviated from the linear pattern most when the correlations of target and distractor populations were the same. This suggests that the perception of correlation in scatterplots with two populations differs from the process underlying the perception a single population. These findings also stand in contrast to assertions that color feature selection, even with distinct colors, aids tasks such as visual search and target discrimination.
Meeting abstract presented at VSS 2015
This PDF is available to Subscribers Only