Abstract
The visual system appears to make a fundamental distinction between continuous substances and discrete objects. Visual search can operate over a continuous space of features, but there is also evidence that those features can be clumped into 'preattentive object files' (Wolfe & Bennett, 1997). Visual tracking is possible for discrete objects, but is more difficult for flowing substances (Scholl & vanMarle, 2003). Infants similarly struggle with tracking substances (Hespos, Dora, Rips & Christie, 2012). Some argue that visual number is processed as a distinct feature based on discrete objects, in contrast to judgments of continuous properties such as textural density (Burr, Turi & Anobile, 2010, c.f. Durgin, 2008). We will review how this perceptual distinction between object and continuous substances affects visual thinking abilities in a real-world context: understanding statistical distributions. Distributions may be represented continuously (as line graphs that represent density) or discretely (as icons to represent count). In both continuous and discrete distributions, frequency is usually plotted on the Y axis against observations on the X axis. While the continuous representation is more prevalent in statistics courses and even lay-viewer presentations, the visual processing of frequency over observations as a single continuous value may substantially interfere with a viewer's understanding. In contrast, although rarely used, a discrete representation of stacked objects may trigger a form of visual processing more congruent with counts. We review distinctions between the perception of substances and objects (Rips & Hespos, 2015; vanMarle & Scholl 2003; Soja, Carey & Spelke, 1991; Hespos, Dora, Rips & Christie, 2012). From this review, we identify implications of this substance/object distinction on data visualization, as a case study for extending knowledge in vision science to translational domains.
Meeting abstract presented at VSS 2018