Abstract
Do we perceive fine details in the visual periphery, or do we overestimate how much we see outside the center of the visual field? While some researchers claim the former (Haun, Tononi, Koch, & Tsuchiya, 2017; Kaunitz, Rowe, & Tsuchiya, 2016; Vandenbroucke et al., 2014), phenomena such as inattentional blindness and change blindness provide evidence for the latter. Interestingly, recent work has consistently replicated one finding which relates to this question: observers tend to use liberal perceptual criteria when detecting items at unattended or peripheral locations (Rahnev et al., 2011; Solovey et al., 2015), with a tendency to report seeing items that were never presented. These experiments used very simple stimuli in an artificial environment; here, we asked whether observers would exhibit similar liberal detection criteria when making judgments about visual items in a more natural environment. Using an innovative game-building engine, we created a task where subjects had to drive a car down a city street and make judgments about attributes of pedestrians' clothing in the visual periphery. Results from our first experiment demonstrated than when participants were asked about whether one specific color was presented on an individual in the periphery, they used liberal detection criteria and exhibited relatively high numbers of false alarms, similar to previous investigations. Following this first experiment, we conducted a second study which varied the color that subjects were asked to detect on every trial. Results from this experiment showed that when the color to be detected changed across trials, observers actually exhibited relatively conservative detection criteria, with few numbers of false alarms, but higher numbers of misses. These results can be quantitatively characterized using a detection theoretic formal model, and provide evidence of interactions between task demands and the criteria subjects use to make perceptual judgments.
Meeting abstract presented at VSS 2018