Abstract
When identifying the number of objects in a visual scene, observers frequently mistake patches of dots that are denser, larger, or longer as more numerous. Such "incongruency effects" have supported claims that the visual system doesn't represent number directly, but instead encodes number by combining several non-numeric features, including density, convex hull, and contour length (e.g., Gebuis et al., 2011; Dakin et al., 2011). However, incongruency effects may instead be a product of a response conflict between number and other dimensions, much like in a Stroop task. To test whether incongruency emerges due to encoding or response conflict, participants completed two tasks: a Discrimination task, indicating whether more of the dots are blue or yellow, and an Estimation task, estimating the number of blue or yellow dots. Non-numeric cues, including contour length, convex hull, and density, were made incongruent with number (Fig. 2). Importantly, while the Discrimination task creates a response conflict – the "blue" and "yellow" response for number competes with the response for the non-numeric dimensions – the Estimation task eliminates it, as participants are forced to use number words. In Experiment 1, dots were arranged along lines, inducing a conflict between number and contour length. While we observed the incongruency effect in the Discrimination task – dots along the longer line were identified as more numerous – participants showed no bias in the Estimation task (Fig. 2). In Experiment 2, we replicate these findings when inducing an incongruency between number and convex hull. Finally, in Experiment 3, we found that density – unlike contour length and convex hull – produces an incongruency in both Discrimination and Estimation (Fig. 3). These results provide a method for testing encoding vs. response conflict components in number encoding, and show that only some dimensions affect number when response conflicts are controlled for.
Meeting abstract presented at VSS 2016