Abstract: One important task for the visual system is to group image elements that belong to an object and to segregate them from other objects and the background. We here present an incremental grouping theory (IGT) that addresses the role of object-based attention in perceptual grouping at a psychological level and, at the same time, outlines the mechanisms for grouping at the neurophysiological level. The IGT proposes that there are two processes for perceptual grouping. The first process is base grouping and relies on neurons that are tuned to feature conjunctions. Base grouping is fast and occurs in parallel across the visual scene, but not all possible feature conjunctions can be coded as base groupings. If there are no neurons tuned to the relevant feature conjunctions, a second process called incremental grouping comes into play. Incremental grouping is a time-consuming and capacity-limited process that requires the gradual spread of enhanced neuronal activity across the representation of an object in the visual cortex. The spread of enhanced neuronal activity corresponds to the labeling of image elements with object-based attention.
Introduction: The Binding Problem and Gestalt Laws
Our visual system begins by decomposing the world into fragments—small edges, colors, and textures detected by individual neurons. The "binding problem" asks how the brain stitches these fragments back together into coherent objects while keeping them separate from the background. Psychologists have long identified "Gestalt laws" (such as proximity, similarity, and connectedness) that guide this organization.
A central paradox remains: why are some grouping tasks instantaneous, while others are slow and effortful? For example, detecting a red dot among green ones is fast and parallel, but tracing a single tangled fishing line in a drawing is a serial, time-consuming process. The Incremental Grouping Theory (IGT) seeks to reconcile these observations with the underlying neurophysiology of the visual cortex.
The Incremental Grouping Theory (IGT)
The IGT proposes a dual-process framework for perceptual grouping based on hierarchical recurrent processing in the visual cortex (V1, V2, V4, and IT).
1. Base Grouping
Base grouping relies on hard-wired neurons tuned to specific feature conjunctions (e.g., a "red horizontal edge"). This process is fast, occurs in parallel across the entire visual scene, and is mediated by feedforward connections.
2. Incremental Grouping
When no single neuron exists to represent a complex shape, the brain must "label" the object. This is achieved by increasing the firing rate of all neurons representing that object. This enhanced activity spreads through the network via lateral and feedback connections, effectively tagging image elements with object-based attention.
Key Insights
- Labeling via Attention: Object-based attention is the actual process of enhanced activity spreading through the neuronal network.
- N-neurons vs. A-neurons: The theory suggests a division of labor where N-neurons (non-attentional) provide a stable map of features, while A-neurons (attentional) carry the grouping label through enhanced firing. This explains why we can attend to an object without its perceived features changing.
- Transitivity: Grouping is transitive; if element A is linked to B, and B to C, the attentional label will eventually flow from A to C, even across large distances.
Key Results
- Linear Processing Time: In curve-tracing tasks, the time required to group elements increases linearly with the length of the curve. Neuronal data shows the attentional label spreads at a rate of roughly 50ms per receptive field in the primary visual cortex (V1).
- Latency of Grouping: While initial feature detection happens within ~40ms, the attentional labeling for complex grouping typically begins around 140ms after stimulus onset.
- Scale Invariance: In higher visual areas with larger receptive fields, the label can jump across larger portions of the image in fewer synaptic steps, allowing us to trace lines at similar speeds regardless of their distance from the eye.
Limitations and Future Work
While the IGT provides a robust conceptual framework, it requires more detailed implementation as a formal neural network model to handle natural scenes. Future research must also clarify the "read-out" mechanisms—how higher brain areas interpret these enhanced firing rates to trigger motor actions—and seek direct biophysical evidence for the multiplicative interactions required to enable specific lateral connections.
Summary updated