Feature integration theory (FIT), proposed by Anne Treisman and Garry Gelade in 1980, is one of the most influential theories in the study of visual attention. FIT addresses a fundamental question: how does the visual system combine separately processed features (color, orientation, size, motion) into coherent object representations? The answer, Treisman proposed, is that focused spatial attention serves as the "glue" that binds features together.
Two Stages of Processing
FIT proposes two stages of visual processing. In the pre-attentive stage, basic features are registered automatically and in parallel across the visual field, coded on separate feature maps (one for color, one for orientation, etc.). Individual features can be detected rapidly and independently of the number of items in the display — producing the pop-out effect. In the focused attention stage, attention is directed to specific locations, binding the features present at each attended location into coherent object representations via a "master map" of locations.
Conjunction search: RT = a + b × set_size (serial, attention required)
Search slope ≈ 0 ms/item → pre-attentive feature detection
Search slope ≈ 20-30 ms/item → serial attentive binding
Illusory Conjunctions
A key prediction of FIT is that without focused attention, features from different objects may be incorrectly combined — producing "illusory conjunctions." Treisman and Schmidt (1982) demonstrated this: when attention was diverted, observers reported seeing illusory combinations of features (e.g., a red X when the display contained a red O and a blue X). These binding errors were not random — they preserved the features present in the display but miscombined them, exactly as FIT predicts for unattended locations.
Patients with bilateral parietal lobe damage (Balint's syndrome) show dramatically increased illusory conjunctions, consistent with FIT's claim that spatial attention — which depends on parietal cortex — is necessary for correct feature binding. These patients can perceive individual features but consistently fail to combine them correctly, providing neuropsychological support for the theory's central mechanism.
Visual Search Evidence
FIT's most testable predictions concern visual search. Searching for a target defined by a single feature (e.g., a red item among green items) should be fast and independent of the number of distractors (parallel search, slope near zero). Searching for a target defined by a conjunction of features (e.g., a red vertical bar among red horizontal and green vertical bars) should require serial attention to each item, producing response times that increase linearly with set size.
This basic prediction has been largely supported, though the clean dichotomy between parallel and serial search has been blurred by many intermediate cases and alternative models. Jeremy Wolfe's Guided Search model extended FIT by proposing that top-down knowledge and pre-attentive feature information can guide attention toward likely targets, producing efficient search even for some conjunctions.
Legacy and Influence
While FIT's specific mechanisms have been modified and debated over 40+ years, its core insights remain foundational: features are processed in parallel, binding requires attention, and the deployment of attention determines which object representations reach awareness. The theory launched the visual search paradigm as a major tool for studying attention, inspired computational models, and raised the binding problem to a central position in cognitive science.