How Configural Is the Configural Superiority Effect? A Neuroimaging Investigation of Emergent Features in Visual Cortex

Abstract

The perception of a visual stimulus is dependent not only upon local features, but also on the arrangement of those features. When stimulus features are perceptually well organized (e.g., symmetric or parallel), a global configuration with a high degree of salience emerges from the interactions between these features, often referred to as emergent features. Emergent features can be demonstrated in the Configural Superiority Effect (CSE): presenting a stimulus within an organized context relative to its presentation in a disarranged one results in better performance. Prior neuroimaging work on the perception of emergent features regards the CSE as an "all or none" phenomenon, focusing on the contrast between configural and non-configural stimuli. However, it is still not clear how emergent features are processed between these two endpoints. The current study examined the extent to which behavioral and neuroimaging markers of emergent features are responsive to the degree of configurality in visual displays. Subjects were tasked with reporting the anomalous quadrant in a visual search task while being scanned. Degree of configurality was manipulated by incrementally varying the rotational angle of low-level features within the stimulus arrays. Behaviorally, we observed faster response times with increasing levels of configurality. These behavioral changes were accompanied by increases in response magnitude across multiple visual areas in occipito-temporal cortex, primarily early visual cortex and object-selective cortex. Our findings suggest that the neural correlates of emergent features can be observed even in response to stimuli that are not fully configural, and demonstrate that configural information is already present at early stages of the visual hierarchy.

Keywords: configural; emergent features; fMRI; perception; perceptual organization; ventral visual pathway; vision; visual cortex.

FIGURE 1

A representative set of the full range of configural stimuli separated by Configurality Range, with a selected set of Scrambled stimuli.

FIGURE 2

A subset of the stimuli used in the category localizer experiment. (A) An example from the faces category. (B) An example from the tools category. (C) An example from the textures category. (D) An example from the houses category.

FIGURE 3

Mean response times for each Configurality Range with error bars denoting standard error.

FIGURE 4

Mean percent signal change of all Configurality Ranges compared to the percent signal change of the Scrambled condition within each ROI. Error bars denote standard error.

FIGURE 5

Percent signal change at each Configurality Range within EVC and LOC. Error bars denote standard error.

FIGURE 6

Percent signal change at each Configurality Range within FFA and PPA. Error bars denote standard error.