Perception and Perceptual Distortions

1
Perception and Perceptual Distortions
Vaughan Bell vaughan_at_backspace.org
2
Outline

• We are going to focus on Visual Perception
• Approaches
• David Marr - Bottom-Up Computational Account
• Richard Gregory Top-Down Influence
• JJ Gibson - An Ecological Approach
• Sources of perceptual distortion and their
  relation to theories of visual perception

3
What needs explaining

• The information we receive by our sense organs is
  relatively impoverished.
• For example, the retina receives a grainy 2D
  image of the visual scene
• that includes large gaps (blind spots)
• and an uneven representation of colour (cones)
  and luminance (rods).

4
What needs explaining

• This information is transformed into a rich
  visual experience.
• Theories of visual perception attempt to explain
  how this happens.
• And describe how perceptual distortions occur
  with the context of a model of normal
  psychological function.

5
Marr Computational Approach

• Marr wanted to understand mechanisms of vision
  rather than just behaviours associated with it.
• Particularly, he wanted to link neurophysiology
  with psychology.
• He took an information processing view of the
  mind
• and aimed to describe perception in terms of
  computations on sense data
• to extract high level visual experience.

6
Marrs Stages of Visual Processing

• Marr proposed there were distinct stages of
  processing in visual perception
• Raw Primal Sketch
• Complete Primal Sketch
• 2½D Sketch
• Full 3D Representation

7
Marr Early Primal Sketch
This involves the extraction of information
regarding edges and intensity changes.
8
Marr Complete Primal Sketch

• After the Raw Primal Sketch
• Marr argued we create a Complete Primal Sketch by
  grouping surfaces and common areas.
• The Gestalt Psychologists of the early 19th
  Century demonstrated many different ways in which
  we can group objects.

9
Proximity
Things that are close together seem part of the
same group.
10
Similarity
Things that are similar seem part of the same
group.
11
Common Fate
12
Continuation
We see this as two crossing lines, rather than
two V shapes.
13
Closure
Images are seen to form closed rather than open
patterns.
14
Marr 2½D Sketch

• After gaining information about groupings and
  surfaces the viewer needs some spatial
  information.
• Marr called this stage the 2½D Sketch to emphasis
  that this stage didnt give a full 3D
  representation.
• Just an estimate of the spatial locations of
  objects and materials in relation to the viewer.
• This is involved in some important processes

15
Perceptual Constancy

• We tend to experience objects as the same,
  despite the image they produce on the retina may
  vary greatly.
• Shape constancy objects seen from different
  angles do not appear different or to change shape
• Size constancy objects do not seem to change
  size when they move nearer or further away.
• Colour constancy differing illumination does
  not affect colour despite changes in the actual
  reflected light.

16
Figure / Ground Segregation
17
Figure / Ground Segregation

• Symmetry is one of the factors known to be
  important

18
2½D Sketch Depth Cues

• We perceive much information from which we can
  infer depth
• Binocular disparity
• Texture gradients
• Occlusion
• Convergence
• Relative Size

19
Depth Binocular Disparity
Each eye receives a slight different image of the
world from which 3D positions can be inferred.
20
Depth Textural Gradient
Surfaces appear to have a finer texture as they
recede into the distance.
21
Depth Convergence / Perspective
Lines will appear to draw closer together as they
go farther into the distance.
22
Depth Occlusion
Nearer objects appear to cover over more distant
objects.
23
Marr 3D Model Representation

• The final stage of Marrs theory.
• A full 3D description of our spatial environment
  involving the identification of the structure of
  objects and materials in our visual scene.
• It allows us to work out the 3D environment from
  a non-egocentric point-of-view.

24
Gregory Top-Down Perception

• Gregory believes that we use our experience of
  the world to shape how we perceive it.
• In this way Gregory theories are Top-Down.
• i.e. using high level concepts (knowledge of the
  world) to shape low level perceptions.
• He demonstrated many of his points by using
  visual illusions.

25
Moving Dots Illusion
26
Gregory Illusions

• Gregory would argue that this is an example of
  our conceptual knowledge affecting how we
  perceive simple visual phenomenon.
• The Necker Cube is another example.

27
Necker Cube
28
Gregory Illusions

• It would seem we can get two mutually exclusive
  percepts from a single visual input.
• Gregory argues that Bottom-Up theories, such as
  those of Marr and Gibson, would not explain this.
• Bottom-Up theories would seem to suggest that we
  should always perceive the same thing from the
  same input.

29
Rotating Mask
30
Rotating Mask

• Gregory argues this sort of illusion happens
  because we are not used to seeing hollow faces.
• Therefore our beliefs and expectations are
  applied to make best sense of the data.
• Leading to the illusion.

31
JJ Gibson Ecological Perception

• Gibson sought to explain perception in terms of
  how it relates to properties of the world around
  us.
• He saw movement and change within our visual
  field as providing crucial information to the
  viewer about the world and their place within it.
• He also saw the perception of surfaces as
  important, rather than the interpretation of
  spots or patches of light.

32
JJ Gibson Optic Flow

• Gibson saw the perception of movement as much
  more than simply the effect of changes in the
  retinal image.
• Gibson saw the perception of optic flow as one
  of our main sources of visual information.
• When we move our visual environment seems to flow
  past at different rates, closer things seem to
  move past faster than more distant things.

33
JJ Gibson Optic Flow

• This information can give us a great deal of
  information about our action within the world.
• For example, the point of minimum optic flow
  within our field of vision is often where we are
  heading.
• Optic flow can also give us information about the
  relative distances of objects and surfaces.

34
JJ Gibson Surfaces

• Gibson saw surfaces as one of the constants on
  which can base many perceptual judgements.
• When we grow up we are constantly surrounded by
  surfaces which communicate different properties.
• Differences in shading and texture of surfaces
  communicate a great deal about the shape and size
  of an object or area.

35
Gibsonian Optics Demonstration
36
Perceptual Distortions

• Perceptual distortions are usually classified
  into
• Hallucinations sensory perception without
  external stimulation of sense organ
• Illusions Misperceived or distorted perception
  of real physical stimuli.
• There are many theories of how such distortions
  occur so the following is a taste of how some
  theories relate to theories of normal visual
  perception.

37
Bottom-Up Form Constants

• Mescaline is a phenethylamine hallucinogen and is
  found naturally in certain cacti.

• Hallucinations associated with mescaline (et al)
  can be varied, although various commonalities
  have been observed.

Peyote cactus

• Named form constants by Klüver (1966)

38
Mescaline Form Constants

• These images appear in both eyes and move when
  the eyes move.
• Suggesting they are generated early in the visual
  pathway.

39
Bressloff et al (2002)

• Bressloff et al (2002) built a mathematical model
  of the primary visual cortex (V1).
• This is largely involved with detecting lines,
  edges, contours etc.
• Like Marr, Bressloff is taking a computational
  approach.
• And modeling the equivalent to Marrs primary
  sketch stage.

40
Bressloff et al (2002)

• Bressloff et al simulated destabilisation of V1
  neurons in their resting state.
• And produced results very similar to form
  constant images.

41
Integration

• This suggests that form constant hallucinations
  can be explained in term of Marrs theories.
• Particularly as a disruption to the primary
  sketch stage of visual processing.
• Which feeds up to the final experience of vision.

42
Top-Down White Christmas

• Mercklebach and van de Ven (2001) told
  participants they had hidden White Christmas
  amid white noise.
• Participants were asked to press a button if they
  heard any of the tune.
• Actually, they was no song hidden in the noise.
• 32 of the participants pressed the button at
  least once.

43
Top-Down Random Patterns

• Brugger et al (1993) ran a similar experiment
  with random dot patterns.

• And found that some people will report pictures
  when none are there.

• Particularly if they believe in ESP, or the
  patterns are presented to the left visual field.

44
Top-Down Distortions

• This suggests prior expectations can influence
  what we perceive.
• To the extent of experiencing meaningful
  information in a completely random background
  environment.

45
Ecological Approach

• Gibson placed the environment at the centre of
  visual perception.
• We know that perceptual distortions are not
  necessarily related to pathology and can be
  influenced by social and cultural factors.
• Ohayons (2000) survey of over 13,000 people
  found that 40 had experienced hallucinations,
  and less than half were related to pathology.

46
Ecological Approach

• Turner (1992) studied the Ndembu people of Zambia
  who regularly see spirit manifestations during
  rituals.
• When talking part in a ritual, Turner reports
  seeing one herself.
• Bartholomew (2001) reports on sightings of German
  bombers over Canada during times of national
  distress in World War I.

47
Ecological Approach

• Romme and Escher (1993) report on many people who
  hear voices but never become distressed or
  impaired, indeed voice hearing is considered
  normal in many cultures

48
Conclusions

• Major themes in approaching visual perception
  include
• Bottom-Up e.g. Marr
• Top-down e.g. Gregory
• Ecological e.g. Gibson
• Perceptual distortions may occur due to
  influences on any of these processes.