Transsaccadic object associations shape peripheral perception: The role of reliability.

Abstract

Faced with inhomogeneous representations, the visual system has to rely on pre- and postsaccadic processing mechanisms to assure perceptual continuity across eye movements. While postsaccadically, memorized peripheral and postsaccadic foveal information are integrated according to their reliabilities, here we investigated whether this also holds true for the presaccadic combination of peripheral input and internal associated foveal images. In three experiments, participants learned associations between objects changing transsaccadically in one feature dimension (spatial frequency in Experiment 1 and color in Experiments 2 and 3). Subsequently, participants judged the respective feature of only peripherally presented objects. Importantly, the reliability of this peripheral input was manipulated by lowering the contrast (Experiment 1) or adding color noise (Experiment 3). We hypothesized that participants’ presaccadic peripheral percepts would be biased toward the internal associated foveal image and that the biasing effect would be stronger the lower the peripheral reliability. In all experiments, perception was biased in the direction of the associated foveal image. However, the strength of the bias did not differ between reliability conditions. The presaccadic perceptual bias effect had previously not been tested with the feature color. By showing that yet another feature incorporates prior transsaccadic knowledge, our results highlight the scope of the effect. Furthermore, they point to important differences between pre- and postsaccadic processing mechanisms.

What determines location specificity or generalization of transsaccadic learning?

“…When objects are manipulated during the saccade, peripheral perception is biased in the direction of the learned transsaccadic change. This peripheral perceptual bias due to transsaccadic learning has been shown in several studies ( Bosco, Lappe, & Fattori, 2015 ; Bosco, Rifai, Wahl, Fattori, & Lappe, 2020 ; Herwig & Schneider, 2014 ; Herwig, Weiß, & Schneider, 2015 ; Herwig et al, 2018 ; Köller et al, 2020 ; Osterbrink & Herwig, 2021 ; Osterbrink, Recker, & Herwig, 2022 ; Paeye et al, 2018 ; Valsecchi & Gegenfurtner, 2016 ; Valsecchi et al, 2020 ). Whether these transsaccadic object associations are transferred to new locations is still under debate.…”

“…One could also think of the transsaccadic learning as a process of forming object-specific Bayesian priors. Learning of stimulus-specific priors has been shown previously ( Gekas, Chalk, Seitz, & Seriès, 2013 ; Kerrigan & Adams, 2013 ; Osterbrink et al, 2022 ), as well as the learning of multiple priors that are separated by spatial location ( Nagai, Suzuki, Miyazaki, & Kitazawa, 2012 ) or symbolic visual cues ( Petzschner, Maier, & Glasauer, 2012 ). Roach, Mcgraw, Whitaker, and Heron (2017) investigated how multiple highly specific priors are learned and when generalization across them occurs.…”