Scene categorization is performed extremely rapidly suggesting that an efficient, feedforward coding system is in place. Converging evidence from behavioral, neural, and computational investigations indicate that this system may rely on the extraction of simple features and does not require the segmentation of these features into coherent objects, or surfaces. This is consistent with the idea that image segmentation is a computationally expensive process requiring feedback. However, are there truly no grouping and image segmentation processes occurring during the fast feedforward processing?
In a series of three experiments, we investigated the influence of segmentation cues on scene categorization. We presented participants with two scenes divided into four parts using different segmentation cues displayed for 300 ms prior to image onset. These cues established either a congruent (supporting the correct image segmentation into two scenes) or incongruent (prompting observers to incorrectly group scene segments) segmentation. Participants were less accurate in scene categorization when incongruent segmentation cues were presented, indicating the segmentation can influence categorization. Moreover, the effect remained robust even when the cues were presented concurrently with the images, suggesting that, whilst scene categorization might be rapid, it can also be influenced by segmentation mechanisms.