Scene consistency enhances state representations of real-world objects

Abstract

Previous research has shown that the context in which objects are located significantly influences how efficiently they are categorized. However, less is known about whether scene consistency can also affect the processing of finer object features, such as the state of an object (e.g., the angle of a Swiss army knife or the fill level of a bottle). Therefore, across three experiments, we presented a subset of the JURICS stimulus set, in which each object exists in 20 continuously varying states (e.g., from fully closed to fully open) in scenes that were either contextually consistent or inconsistent. Participants were asked to report the specific state of the object using a continuous report task. Our results showed that scene consistency enhanced the precision of state judgments; that is, participants made significantly larger errors in reporting object states when objects were presented in inconsistent compared to consistent scenes. These findings suggest that scene context exhibits its effect already at the level of fine-grained perceptual processing of objects, affecting not only object categorization but also the accuracy of its perceived features.

Keywords: Consistency effect; Object recognition; Objects states; Real-world objects; Scene context.

Fig. 1

(A) Example stimuli from the JURICS stimulus base; objects change state in 20 continuous steps. (B) The trial sequence for Experiments 1, 2, and 3. First, a scene alone was presented, followed by the brief presentation of the target object embedded within the scene. Thereafter, a dynamic white noise mask appeared. Participants then adjusted the object’s state using a continuous report task and rated object-scene consistency and confidence. After completing the continuous report task, participants were asked to rate the scene-object consistency and their confidence on a scale from 1 to 6, where 1 indicated low consistency/confidence and 6 indicated high consistency/confidence. (C) Three example object-scene combinations (experimental conditions): consistent, inconsistent, and control. The stimuli shown in the figure are not drawn to scale.

Fig. 2

(A) Mean error as a function of conditions; (B) Mean error as consistency reports ratings. Transparent colored circles represent individual participants’ data points. Error bars depict 95% CIs.

Fig. 3

(A) Mean error as a function of conditions; (B) Mean error as a function of consistency reports ratings; (C) Confidence reports as a function of conditions. Error bars depict 95% CIs.

Fig. 4

Mean error as a function of the block number and condition. Error bars depict 95% CIs.

Fig. 5

(A) Mean error as a function of conditions; (B) Mean error as a function of consistency reports ratings. Error bars depict 95% CIs.

Fig. 6

Combined data from Experiment 2 and 3: (A) (left) Mean error as a function of conditions; (B) Mean error as consistency reports ratings. Error bars depict 95% CIs.