Abstract
Vision science includes a notion of the limited precision of our representations, e.g., internal noise. Here, I present a re-understanding of the contents of visual analog magnitude representations (e.g., approximate duration, distance, density, number). As my main example, I consider the Approximate Number System (ANS), which supports numerical representations that are widely described as fuzzy, noisy, and limited in their representational precision. I contend that these characterizations are largely based on misunderstandings of psychophysical theory. Specifically, I propose that what has been called "noise" and "fuzziness" in these representations (e.g., approximately 7) is actually an important epistemic signal of confidence in my estimate of the value (e.g., think 7, with confidence intervals). As such, I suggest that our analog magnitude representations have precise content that is subject to epistemic limitations. Throughout, I focus on visual magnitude representations (i.e., analog magnitudes) including demos engaging representations of approximate duration (e.g., what does a 1.5 second flash feel like?), approximate distance (e.g., how far does it look to be between me and the wall?), and approximate number (e.g., around how many dots are on the screen?). I describe a standard model for these representations, discuss a confusion we may fall prey to when theorizing about them, present demonstrations of failings of such notions, and present a positive proposal for replacing these notions with the notion of epistemic limitation. The major result of this re-analysis is the proposal that we all share a commitment that there is one true value for any experienced visual magnitude in the world – even if our analog magnitude systems are incredibly limited in their ability to discern what value this is. Visual analog magnitudes do not represent "fuzzy" or "noisy" estimates – rather they represent precise estimates that are subject to epistemic limitations.
Meeting abstract presented at VSS 2018