Pattern recognition (psychology)

From Wikipedia, the free encyclopedia
Jump to: navigation, search

Pattern recognition involves identification of faces, objects, words, melodies, etc. The visual system does more than just interpret forms, contours and colors. Pattern recognition refers to the process of recognizing a set of stimuli arranged in a certain pattern that is characteristic of that set of stimuli. Pattern recognition does not occur instantly, although it does happen automatically and spontaneously. Pattern recognition is an innate ability of animals.

Contents

[edit] Theories of pattern recognition

[edit] Template matching

The incoming sensory information is compared directly to copies (templates) stored in the long term memory. These copies are stored in the process of our past experiences and learning.

E.g. A A A are all recognized as the letter A but not B.

Note: This does not allow for variation in letters unless there are templates for each variation.

[edit] Prototype matching

Prototype means a concept of average characteristics of a particular subject. It can be found throughout the world. For instance a concept of small animal with feathers, beak, two wings that can fly is a prototype concept of a crow, sparrow, hen, eagle, etc. Prototype matching, unlike template matching, does not emphasize a perfect match between the incoming stimuli and the stored concept in the brain.

[edit] Feature analysis

According to this theory, the sensory system breaks down the incoming stimuli into its features and processes the information. Some features may be more important for recognition than others. All stimuli have a set of distinctive features. Feature analysis proceeds through 4 stages.[citation needed]

  1. Detection
  2. Pattern dissection
  3. Feature comparison in memory
  4. Recognition

[edit] Recognition of components

Irving Biederman theorizes that every object is made up of geons—the building blocks of all objects (cylinders, cones, are combined in many ways: on top of, to the side, etc.).[citation needed] (Citation - E. Bruce Goldstein; Cognitive Psychology; p. 50)

[edit] Top down and bottom up processing

  1. Top down processing can be seen as processing what one is perceiving using past information.It occurs when someone infers from a generalization,law etc. to conclude something about a particular example, instance, case etc.
  2. Bottom up processing can be seen as starting with no knowledge on a subject.It is said to occur when one draws generalizations from particular examples, instances, cases etc. to capture commonalities between them.

Hierarchy of detectors:

  1. Feature detectors — lowest and highest; respond to curves, edges, etc.
  2. Geon detectors — activated by feature detectors
  3. Higher level detectors — recognize combinations of features and geons

According to Biederman an individual on average is familiar with about 30,000 objects and recognizing them requires no more than 36 geons.[citation needed]

[edit] Multiple discrimination scaling

Template and feature analysis approaches to recognition of objects (and situations) have been merged / reconciled / overtaken by multiple discrimination theory. This states that the amounts in a test stimulus of each salient feature of a template are recognized in any perceptual judgment as being at a distance in the universal unit of 50% discrimination (the objective performance 'JND': Torgerson, 1958) from the amount of that feature in the template (Booth & Freeman, 1993, Acta Psychologica).

[edit] False pattern recognition

The human tendency to see patterns that do not actually exist is called apophenia. Examples of apophenia include the Man in the Moon, faces or figures in shadows, clouds and in patterns with no deliberate design, such as the swirls on a baked confection, and the perception of causal relationships between events which are, in fact, unrelated. Apophenia figures prominently in conspiracy theories, gambling, misinterpretation of statistics and scientific data, and some kinds of religious and paranormal experiences. Misperception of patterns in random data is called pareidolia.

[edit] See also

Personal tools
Namespaces
Variants
Actions
Navigation
Interaction
Toolbox
Print/export