Gestalt Perception and pattern recognition

?
  • Created by: shauna
  • Created on: 10-12-12 17:55

Pattern Recognition

Pattern

  • complex composition of snesory stimuli

Pattern Recognition

  • step between tranduction and perception of a stimuli and categorizing it meaningfully

Agnosia

  • Patients with lesions in area of Brain can see objects but CAN'T RECOGNISE THEM!
1 of 14

Challenges of object Perception

Inverse Projection problem

  • an image on retina could be caused by various objects

Viewpoint invariance

  • (objects look different from different viewpoints, yet )we can recognise an object regardless of viewpoint
2 of 14

Structuralist theory of object perception

  • (Wundt) claimed perceptions are created by combining elements (sensations) From this basic notion Gestalt Psychology was founded 

x Couldn't explain apparent movement

x Couldn't explain reversible images (old/young lady)

x Couldn't explain illusory contours

Movement Perception

  • (Wertheimer) 'Phi Phenomenom': illusion of motion between stationary points viewed in rapid succession. (e.g viewing a movie at the cinema, as its still frames flicked rapidly)
3 of 14

Gestalt Theory of pereption

  • Gestaltists were German Psychologists (Wertheimer, Koffka & Kohler) who proposed things are best described as a WHOLE, and 'more than a sum of its parts'.

"THE WHOLE IS GREATER THAN THE SUM OF ITS PARTS"

  • Gestaltists believed context played role in perception and were intersted in perceptual segregation: ability to decipher which parts of visual info form seperate objetcs.

Thus the theory is CONGRUENT WITH TOP DOWN PROCESSING!

Figure- Ground Principle

  • organise visual field into foreground and background. When figure-ground relationships are unstable, our perceptions become unstable

x Limited in only explaining artificial figures.

4 of 14

Principles of perceptual organisation

Law of Pragnanz (Law of simplicity)

  • we view and organise stimulus in the simplest possible manner. (Used as guiding principle for memory, learning personality and psycotherapy)
5 of 14

Gestalt Laws of perceptual grouping

  • Similarity ~ things with similar characteristics are grouped together (E.g dot painting)
  • Proximity ~ objects close together in space or time tend to be grouped together, or perceived as belonging together
  • Continuity ~ we tend to perceive things as having a continuos rather than distorted pattern
  • Common Fate ~ objects with similar motion/ destnation are grouped together
  • Closure ~ tendency to complete (i.e close) incomplete figures to form meaningful objects. Perceive things as whole figures regardless of gaps in sensory information.
6 of 14

Laws continued

  • Familiarity ~ familiar or meaningful things form groups (e.g flower face illusion)
  • Common Region ~ stimuli in same area are grouped togethe
  • Connectedness ~ things physically connected are perceived as a unit
  • Global superiority effect ~ Our tendency to notice wholes before parts. (Linked to similarity and proximity)
  • Texture Segmentation ~ segmenting image into regions according to texture

x Simplicity hard to define. Element of SUBJECTIVITY regarding how individuals simplify things!

x Only DESCRIBES perceptual phenomena but DOES NOT EXPLAIN THEM!

x Evidence based on 2-d line drawings

7 of 14

Types of Processing

Bottom-up Processing (Structuralism)

  • Perception comes from the environmental stimuli. Analyse smaller features to build WHOLE!

Top-dopwn Processing (Constructivism)

  • Use pre-existing knowledge/ expereince to organise individual featrues into a WHOLE UNIT! Recognition of whole pattern leads to recognition of its component parts.
8 of 14

Bottom- up Processing Theories

Template Matching

  • Match observed image to stored internal templates (external pattern matched to internal representation)

x Would need too many templates. (E.G A comes in various fonts, colours etc.)

x Cannot account for novel stimuli

Prototype Theory

  • Take various instances of object and take out  common characterisitcs. Recognise objects on the basis of their similarity to the prototype. No match is perfect!

Study: (Solso & Mc Carthy, 81) Pp's shown faces similar to prototype, given recognition test with old & new faces & prototype, pp's more confident they had seen prototype than actual faces they had seen. Prototpye face recognised despite not being studied.

 More flexible (manageable representation of faces in memory trace)

Accounts for intuition that some features matter more than others

9 of 14

BU Theories continued

Feature analysis Theory

  • Every stimuli has specific features. I pattern detection we match features of a pattern with features stored in memory (flexible)

1. Break stimulus into parts

2.Match features to stores in LTM

3. Decide which stored pattern is the best match

Study: (Gibson, 69) Found indeciding whether two letters differ takes longer with pairings E&F VS J&Q, as E&F share similar features.

Neurological evidence: (Hubel & Weisel) discovered Brain cells in a Cats visual cortex responding to bars set at specific orientations

10 of 14

Feature Detection Theories

Pandemonium Model (Selfridge, 59)

  • Hierachal Model, combination of serial and parallel processing. Pattern recognition occurs in STAGES. Image demon receives sensory input,feature demons detect features, cognitive demons SHOUT upon receival of certain combinations of features, Decision demon listens for loudest shout to identify input

Feature Intergration (Treisman, 87, 93)

  • single feature allows us to distinguish target from distractors
  • combination of two or more features necessary to distinguish target from the distractors

Feature search asymetries: The search rate for a target among distracters may vary dramatically depending on which stimulus plays the role of target and which that of distracters

11 of 14

Feature analysis Evaluation

Pro's 

  • Economical storing features in memory
  • Experimental evidence ( neurological) consistent with features

Con's

x Lacks applicability to range of stimuli

x Analysing stimuli doesn't always begin with features

x Doesn't explain recognition of complex objects

x Doesn't account for how features are put together

x Some features may be obscured form different viewpoints.

Structuraltheory accounts for last four

12 of 14

Recognition via components/ Structural Description

  • (Biederman,87) stated existance of "Geons" (simple 3-d forms) which can generate 3-d objects. Visual input is matched against structural representations of objects in the Brain.
  • Pattern recognition as well as Geons requires their RELATION to each other. If 65% of details are removed pattern recognition differs depending on parts removed.
  • Easier to recognise objects with vertices intact. WHAT is missing matters NOT HOW MUCH!

Evaluation

  • recognises importance of arrangement of parts
  • explains difficulty recognisingfamiliar objects from unusual perspectives

x Absense of physiological evidence

x Structure not always key to recognition (E.g Peach Vs. Nectarine)

x Which Geons? (simplicity Vs explanatory adequacy)

13 of 14

Context Effect/ Top down processing

  • Non of Prior theories can account for role of context. Setting is important! (E.g may not recognise lecturer outside lecture)
  • Top down processing strongest when stimuli is incomplete or ambiguos. Also strong when stimulus registered fro fraction of a second.

Word superiorty effect

  •  (Wheeler) pp's presented with word or letter briefly, asked to identify which of two items were presented, 10% better recognising words than letters, as words have contextual cues.
  • Cultural Context: Cultural context influences perception
  • Perceptual set: tendency to perceive things a certain way becuse previous expereince influences those perceptions

Study: (Palmer, 75) Kitchen quickly shown, observer asked to identify one of the objects from 3, targets congruent with the context identified 80% of the time, icongruenct 40% of time.

14 of 14

Comments

No comments have yet been made

Similar Psychology resources:

See all Psychology resources »