Abstract
Several studies have shown that low-level perceptual similarity does not predict interferences in memory. These studies all used protocols that promote declarative learning by the medial temporal lobe. However, we know that low-level brain regions demonstrate neuronal plasticity resulting also from rewards conveyed by the striatum, which—unlike the medial temporal lobe—receives and sends information almost everywhere in the brain, including V1. Thus, the purpose of this study was to test whether low-level visual properties (spatial frequencies and orientations) influence interferences in memory by using a task that promotes response-stimuli association by the striatum. On day 1, two subject groups (N=45) learned to discriminate two sets of 12 target faces from 20 different non-target faces (with auditory feedbacks). The two target face sets were filtered by the same log-polar checkerboards in the Fourier domain in subject group 1 while they were filtered by different, non-overlapping log-polar checkerboards in subject group 2. To promote associations between low-level properties and response, the non-target faces were also filtered by another non-overlapping log-polar checkerboard; thus making low-level properties useful for solving the task. On day 2, subjects had to discriminate between three alternatives: target face set 1, target face set 2 and novel non-target faces. We compared interferences — confusions between target face set 1 and 2 — in subject group 1 and group 2. H0 predicts no differences between subject groups, whereas H1 predicts a greater number of interferences in subject group 1 than group 2 because the group 1 target face sets produce more similar activations in V1. Bayesian analyses indicate substantial evidence (Bf01 = 3.4) in favor of H0, thereby supporting the idea that low-level visual properties do not impact interferences in memory, even when learning is based on a response-stimuli association.