Decoding Multiple Selection Mechanisms in Visual Search using Pupillometry and the Drift-Diffusion Model

Visual attention mechanisms prioritize selected information in the context of limited cognitive resources. When multiple valid sources of information regarding a target item are present, multiple selection mechanisms may interact. My previous work (Liang & Scolari, 2020; Liang et al., 2023) implicates a dual-processing model that describes how space- and feature-based attention mechanisms synergistically facilitate target identification in a visual search task. For example, the speed with which evidence is accumulated in favor of one perceptual decision over another (drift rate) is independently modulated by space- and feature-based attention; while the amount of evidence that is accumulated before a behavioral response is generated (boundary separation) is interactively modulated by both attention mechanisms. In this dissertation, I examined pupillometry data, widely regarded for assessing several related cognitive processes, to further investigate this model. I analyzed data from one unpublished study (S1, N = 40) and two published studies (S2, N = 30; S3, N = 40), all of which used a visual search task with a pre-cue that indicated either the upcoming target’s likely location (spatial pre-cue), its likely color (feature pre-cue), or both. Additionally, I employed a computational model (Pupil Response Estimation Toolbox, or PRET) that deconvolves pupil data into linearly summed responses to sequential trial events to investigate whether and how these component responses associate with latent components of perceptual decision-making. Results showed that the net change in pupil size closely mirrored the reliability of the pre-cue object. This matched the pattern observed previously within interactive decision-making processes: boundary separation similarly tracked pre-cue reliability. Moreover, correlational analyses revealed that pupil size changes predicted boundary separation across studies. The PRET model further revealed that these effects primarily arose from the amplitude of the pupillary response to the target. In contrast, the latency of the response to the target marginally predicted drift rate. These findings offer convergent support of the dual-processing model, while providing further insight into the kinds of cognitive processes that may be tracked by pupillometry.