Metacognition in Putative Magno- and Parvocellular Vision

Abstract

A major distinction in early visual processing is the magnocellular (MC) and parvocellular (PC) pathways. The MC pathway preferentially processes motion, transient events, and low spatial frequencies, while the PC pathway preferentially processes color, sustained events, and high spatial frequencies. Prior work has theorized that the PC pathway more strongly contributes to conscious object recognition via projections to the ventral "what" visual pathway, whereas the MC pathway underlies non-conscious, action-oriented motion and localization processing via the dorsal stream "where/how" pathway. This invites the question: Are we equally aware of activity in both pathways? And if not, do task demands interact with which pathway is more accessible to awareness? We investigated this question in a set of two studies measuring participant's metacognition for stimuli biased towards MC or PC processing. The "Steady/Pulsed Paradigm" presents brief stimuli under two conditions thought to favor either pathway. In the "pulsed" condition, the target appears atop a strong luminance pedestal which theoretically saturates the transient MC response and leaves the PC pathway to process the stimulus. In the "steady" condition, the stimulus is identical except the luminance pedestal is constant throughout the trial, rather than flashed alongside the target. This theoretically adapts the PC neurons and leaves MC for processing. Experiment 1 was a spatial localization task thought to rely on information relayed from the MC pathway. Using both a model-based and model-free approach to quantify participants' metacognitive sensitivity to their own task performance, we found greater metacognition in the steady (MC-biased) condition. Experiment 2 was a fine-grained orientation-discrimination task more reliant on PC pathway information. Our results show an abolishment of the MC pathway advantage seen in Experiment 1 and suggest that the metacognitive advantage for MC processing may hold for stimulus localization tasks only. More generally, our results highlight the need to consider the possibility of differential access to low-level stimulus properties in studies of visual metacognition.