Feature Integration Theory Notes

Feature Integration Theory and Guided Search

Observation #1:
- Visual search is easy if the target differs from non-targets in a simple feature.
- Visual search is difficult if the target differs from non-targets in a conjunction of features.
Observation #2:
- At attended locations, features are correctly bound together.
- At unattended locations, features are often incorrectly bound together.
Observation #3:
- Arrangements of colored shapes into textures are much easier if different parts of a scene differ in simple features compared to a conjunction of features.

A cognitive model explaining that we perceive coherent objects by binding simple features together at attended locations.

Pre-attentive Stage:
- Simple features (e.g., color, orientation, edges, movement) are processed independently and in parallel across the visual field.
- Feature maps are created for each dimension, coding which features are at which locations.
- Examples include pre-attentive representations of:
  - Blue
  - Red
  - Vertical orientations
  - Horizontal orientations
  - Edges
Master Map of Locations:
- Codes where visual information is located but not what it is.
Attentional Stage:
- Attention is directed, like a spotlight, randomly around the master map of locations.
- When a location is attended, features from different dimensions are bound together into an object representation (called an object file).
Post-attentive Stage:
- Object representations (object files) are created one-by-one as attention is directed to locations in the master map.

Feature Search:
- Without attention, simple features are extracted simultaneously across the entire visual field.
- By monitoring for the presence of activity on the known target feature, the target can be found instantly (or rejected as being absent instantly).
Conjunction Search:
- Attention is required to bind features together.
- Each location needs to be selected (in random order) to bind features together from different feature dimensions until the proper binding is found.

Binding features into an object representation requires attention.
Without directed attention toward individual items, a texture display just looks like a bunch of simple features (e.g., circles, diamonds, reds, greens).
Only after directing attention toward several individual items can one notice differences between different parts of the scene.

The correct answer is:
- C. Attention binds features together into coherent object representations by selecting a location on the master map.