A key challenge in visual object recognition is to generate stable recognition performance under changing viewing conditions. Past attempts to decipher the neural foundation of this ability have mainly focused on the primate ventral stream which creates a risk of confounding general mechanistic insights with layout-specific correlates of object invariance. To promote a synthesis of data beyond the mammalian clade, this study will explore neuronal correlates of object invariance in of pigeons (Columba livia) and jackdaws (Corvus monedula). The visual system of birds offers ideal conditions to explore the generality of neuronal principles as it displays an organization radically different from the ventral stream. In line with a protocol adopted from primate research (Kell et al., 2023), subjects will have to discriminate between four naturalistic objects presented on cluttered backgrounds across various identity-preserving transformations. Due to the high variability among the tested images, the design of this study has characteristics complementary to a previous project of our group in which we assessed object recognition based on relatively simple images with a lower number of manipulated parameters. For our primary behavioral analysis, we will extract correlations between image-level performances of primates and representatives of the avian clade to shed light on the extent to which these two phylogenetic lineages have evolved common mechanisms for visual processing. As pigeons and jackdaws failed to reach a primate-like level of object invariance in our previous study, we expect none of the avian species to produce image-by-image performance strongly correlated with marmosets. However, rather than pigeon and jackdaw behavior being indistinguishable, we predict their respective performance vectors to be correlated only moderately as jackdaws were previously found to generalize more successfully to novel images than pigeons following extensive training. To assess neuronal correlates of object recognition, we aim to obtain representational similarity measures from the tectofugal visual areas ENTO, MVL, NI, and the executive area NCL.
Annika Verfers (Wed,) studied this question.