Abstract
What is selected when attention is directed to a specific location of the visual field? Theories of object-based attention have suggested that when spatial attention is directed to part of an object, attention does not simply enhance the attended location but automatically spreads to enhance all locations that comprise the object. Here, we tested this hypothesis by reconstructing the distribution of attention from primary visual cortex (V1) population neuronal activity patterns in 24 human adults using functional magnetic resonance imaging (fMRI) and population-based receptive field (pRF) mapping. We find that attention spreads from a spatially cued location to the underlying object, and enhances all spatial locations that comprise the object. Importantly, this spreading was also evident when the object was not task relevant. These data suggest that attentional selection automatically operates at an object level, facilitating the reconstruction of coherent objects from fragmented representations in early visual cortex.
Introduction
Our visual system receives fragmented information—neurons in the primary visual cortex (V1) have very small receptive fields, meaning each one only "sees" a tiny speck of the world. A major mystery in neuroscience is how the brain glues these fragments into coherent objects. Theories of object-based attention suggest that when we focus on one part of an object, attention doesn't stay localized; it spreads to encompass the whole object. However, it has been unclear whether this happens as early as V1, whether it follows the specific "Gestalt" shape of the object, and whether this process is voluntary or automatic. This paper aims to visualize this attentional spread in the human brain with high spatial and temporal precision.
Experimental Design and Methodology
The researchers combined functional magnetic resonance imaging (fMRI) with population-based receptive field (pRF) mapping in 24 healthy human adults. Participants viewed two overlapping horseshoe-shaped objects. A spatial cue directed their attention to one end of one horseshoe.
To test the automaticity of this process, the study employed two distinct tasks:
- Object Task: Participants judged if two dots appeared on the same horseshoe, making the object shape relevant.
- Color Task: Participants judged if two dots were the same color, making the object shape irrelevant.
The core innovation was a multivariate reconstruction method. By knowing exactly which part of the visual field each V1 voxel represents, the authors could "project" V1 activity back into visual space to create a picture of the attentional field. In a second experiment (N=7), they used a high-speed fMRI sequence (TR = 262 ms) to measure the timing of the BOLD signal at different points along the object's curve.
Results: The 'Paintbrush' Effect in V1
The neural reconstructions showed that attention doesn't just enhance a circular "spotlight" around the cue. Instead, it acts like a "paintbrush," automatically spreading to enhance the entire cued object while ignoring the overlapping distractor.
Key findings include:
- Automaticity: This spreading occurred even during the Color Task. Even when the shape of the horseshoe didn't matter for the goal, the brain still automatically grouped the features of the object together ($p = 5.4 \times 10^{-3}$).
- Serial Spreading: The process is not instantaneous. Experiment 2 revealed a temporal delay: BOLD activity peaked later at locations further away from the initial cue along the object's contour, with a delay of approximately 0.43 seconds.
- Background Suppression: While the cued object was enhanced, the "background" (the empty space in the center of the horseshoe) was actively suppressed below baseline.
- Behavioral Advantage: Participants were significantly faster at responding to dots on the same object (728 ms) compared to different objects (791 ms), confirming a "same-object advantage."
Discussion and Future Directions
These results demonstrate that object selection occurs automatically in the early visual cortex, following Gestalt principles without requiring voluntary control. While the initial spread is automatic, the authors noted that the enhancement was more transient in the color task than in the object task, suggesting that maintaining object-level enhancement likely requires top-down voluntary control. Future research may investigate how these V1 representations interact with higher-order shape-selective regions like the Lateral Occipital (LO) complex to complete the feedback loop of object recognition.
Summary updated