Scene grammar facilitates object-location binding in realistic scenes
Every day we encounter various objects in various places: whether it be morning coffee on the table, fresh pizza in the microwave, or clothes scattered around the room. Studies demonstrated that swap errors could occur in VWM - incorrect binding of object and location information. However, in realistic scenes, attention is guided to specific locations under the specific scene laws. Thus, this study aimed to examine how scene grammar could influence object-location binding. We conducted an experiment where subjects memorized locations of objects presented in scenes. Real-world objects and realistic scenes were used as stimuli. Three objects in the scene were presented for 1 second, and after 1 second participants had to choose which object was presented at the marked location. There were two conditions: 1) target object had a position according to scene rules (e.g., pillow on the bed), and both distractors - against scene rules; 2) the target object had a position against scene rules (e.g., pillow on the floor), one distractor was in a consistent position (e.g., an alarm clock on the bed table) and another distractor was on the position which target object should have according to the scene grammar (e.g., slippers on the bed). The results showed that less errors were made in the condition where the target had position according to scene rules. Additional analysis of incorrect answers for the second condition demonstrated no difference between choosing the distractor which was located on the target’s location according to scene rules and choosing the second distractor. Errors did not follow location expectations, thus, errors probably occurred during the encoding or storing stage and not during the response. We suggest that scene grammar strengthens object-location binding via proper attention guidance.