Dissociation between the processing of spatial frequencies vs scale in visual word recognition

Poster Presentation: Saturday, May 17, 2025, 2:45 – 6:45 pm, Banyan Breezeway
Session: Object Recognition: Reading

Clémence Bertrand Pilon1,2, Martin Arguin1,2; 1Département de psychologie and Centre interdisciplinaire de recherche sur le cerveau et l’apprentissage, Université de Montréal, 2Centre de recherche, Institut Universitaire de Gériatrie de Montréal, Montréal, Canada.

Two consensual theories in visual recognition focus on the temporal progression of spatial processing: the “global-to-local” theory, which suggests the human visual system processes large features first, followed by smaller ones (Navon, 1991), and the “coarse-to-fine” theory, which suggests that we first process low spatial frequencies (SFs), then higher ones (Bar, 2003). While these theories are often considered equivalent, spatial scale (SS) and SF are in fact theoretically dissociable : e.g. large features can rest on high SFs. The present study investigated the temporal progression of SF and SS processing in visual word recognition using random temporal sampling. Adult readers (16 per experiment) reported a word presented for 200 ms in a four-alternative forced-choice task. Target stimuli were an additive combination of the target word and a white noise field with a randomly varying signal-to-noise ratio (SNR) over exposure duration. In Exp. 1 stimuli were filtered according to four SF conditions (central frequencies of 1.2, 2.4, 4.8, and 9.6 cycles per degree). Exp. 2 were intact word images behind an occluding mask with gaussian apertures of varying sizes (center frequencies of 1.11°, 0.55°, 0.28°, 0.14°). Classification images (CIs) were computed to show the temporal progression of SF and SS processing efficiency. In Exp. 1, initial visual processing relied on high SFs, then followed by lower SFs. Exp. 2 revealed an initial dominance of the largest SS, then progressing orderly to smaller ones. Both experiments showed complex interactions of SF/SS processing over time with the frequency spectrum of SNR oscillations . In contrast to the "coarse-to-fine" order expected for SF processing in word recognition, Exp. 1 demonstrated a "fine-to-coarse" progression. The results of Exp. 2 , revealed a "global-to-local" order. These results highlight the importance of distinguishing between SF and SS, since the two are empirically dissociable.