Theories of Visual Perception Flashcards
Why is the perception of form and organisation important?
Because the environment contains hundreds of overlapping objects yet we have a perceptual experience of structured coherent objects that we can recognise and often name
Why does the visual system as a camera analogy easily break down?
- the left and the right visual fields go to the left and right visual cortex: there is no screen
- No 1:1 correspondence with the image at the back of the retina and visual processing
- The brain has to adjust for the retina being curved
- Receptors are unevenly distributed
- The image is inverted, tiny, curved and flat
What is perception?
The means by which information acquired from the environment via the sense organs is transformed into experiences of objects, events, sounds, tastes, ect
What is the whole of perception related to in terms of the brain working something out?
related to brain trying to work out what the distal stimulus is causing the proximal stimulus
What is the difference between sensation and perception?
Perception is the brain trying to work out what the distal stimulus is that causes the proximal stimulus. Sensation is the information that comes in
What is transduction?
Interpreting sense through electrical signals
What do different approaches to perception vary in?
- Bottom up vs top down processing
- Goal of perception
- Methods of study
What is Gestalt psychology
- the whole is greater than the sum of its parts
What are Gestalt psychologists interested in?
How we group parts of a stimulus together and the way we separate figure from ground: segregation and grouping
What is structuralism?
believes perception is a combination of individual sensations that can be reduced to simple individual elements such as lines
What are the 8 different rules of the Gestalt school of perception?
- Similarity
- Good continuation
- Proximity
- Connectedness
- Closure
- Common fate
- Familiarity
- Invariance
What is the similarity rule?
Similar objects are grouped together
What is the good continuation rule?
Points when connected result in straight or smoothly curving lines that follow each other
What is the proximity rule?
Things that are near to one another tend to be grouped together
What is the connectedness rule?
Things, physically connected together are perceived as a unit
What is the closure rule?
We tend to complete broken figures
What is the Common fate rule?
Things moving in the same direction are grouped together
What is the familiarity rule?
Things are more likely to form groups if groups appear familiar or meaningful
What is the invariance rule?
We have flexibility to recognise objects under different circumstances. E.g. we can see objects in different locations, orientations ect.
What are the problems with the Gestalt approach?
- Underplay the parallel processing and unconscious processing that the brain does
- Explanation of how some of their laws worked was wrong
- Their laws provide a description of how things work rather than an explanation
- Their laws are ill defined – Pragnanz – what is the simplest and most stable shape?
- Are they just stating the obvious?
What is Gibson’s approach for perception?
It’s a bottom up approach that states perception is direct. It thinks that the information we receive is sufficient for interaction with the environment, there is no difference between sensation and perception and complex cognitive processing is unnecessary
How was Gibson’s perception theory studied?
Ecologically: in natural environments, not the lab
what did the Gibson approach stress the importance of?
movement
What does Gibson mean when he says movement is active?
movement of the observer provides an additional stream of information. WE are active agents in the world.
What does ‘Ambient optic array’ mean?
- structure of light is reflected by textured surfaces
- changes are due to observer movement
What are invariants?
- unambiguous information about the environment
- can be directly perceived
- E.g. Horizon ratio relation: proportion of object above horizon is constant with changes in distance but not size
- Texture gradient gives cues about where the object is located
What is the empirical support for the Gibson theory?
- Gibson & Bridegeman (1987): participants could correctly identify objects, state their colour, identify the lighting conditions and the objects spatial orientation
- The average subject identified about 2/3 of photographs correctly
Why is motion important?
Because it is necessary in order to perceive invariant information in a static scene. We introduce motion to highlight the invariant properties (the unambiguous properties)
What is motion parallax?
Things far away move more slowly than things nearby - the speed of movement tells us about the distance to the object
What is motion parallax used by?
Animals that don’t have much binocular overlap e.g. head bob and orthogonal running
What is optical flow?
The combination of parallax and retinal size
What uses motion features such as motion parallax and optic flow?
Drones and autonomous cars
What are practical implications of Gibson’s optic flow ideas?
Horizontal lines are painted on the road, becoming closer together as driver approach junctions. These markings are often seen on exit roads from motorways and create the illusion of increasing speed which causes the driver to slow down
What is affordance?
What the surface/ object offers the animals. E.g. can it be grasped, eaten, sat upon.
Are memory and experience necessary for affordance?
No
What are criticisms of the Gibson perception approach?
- Vague: how is the information picked up?
- Ignores top down processing
- Ignores neuroscience
In Marr’s information processing approach to perception what are the four stages of image analysis?
- Grey level description (black and white image)
- Primal sketch (raw and full)
- 2 ½ D sketch (put together features according to certain rules
- 3D object centred description
What are the main features of Marr’s information processing approach?
It is a bottom up approach but emphasises the computation nature of perception. It takes the output from the previous stage to the input for the next stage
What is the grey level description?
- The intensity of light is measured at each point of the retina
- It’s produced by the activation of retinal photoreceptors
What is the raw primal sketch?
It aims to identify object edges. It does this by:
- Gaussian blurring: it blurs images to different degrees
- Identify intensity changes: those present at more than 2 levels of blurring
- Assign primitives: there are 4 types of intensity change: edge-segment, bar, termination, blob
What is the full primal sketch?
The goal is to identify the object outline. It does this by:
- grouping primitives together and assign place tokens
- place tokens can be grouped to form higher order place tokens
- grouping based on clustering (like Gestalt proximity) and curvilinear aggregation (like Gestalt good continuation
How does the primal sketch go the 2 1/2 D sketch?
The primal sketch is combined with depth cues, colour, motion.
How does the 2 1/2 D sketch go to the 3D sketch?
The 2 1/2 D sketch is analysed for 3D volume primitives such as cylinders, cones, cubes etc.
What is the importance of computational approaches?
- An algorithm/ rule/ system is more likely to be understood by understanding the problem that has to be solved rather than examining the mechanism (and hardware) in which it is embodied (AI arguments)
- TO understand perception (purely) by studying neurons is like trying to understand bird flight by studying only feathers
- All neurons are doing is computation: other systems could do computation
- The computation is the important part
What is the criticism of Marr’s approach?
The retinal image is not always sufficient to allow reconstruction. What about the role of memory and experience?
What is the constructivist approach of perception?
The retinal image does not provide sufficient information. Perception depends upon stored knowledge 9memory) and experience
What did Helmholtz say was to do with the perception in the constructivist approach?
- Unconscious interference
- likelihood principle - what is more likely? (based on stored knowledge)
What is unconscious interference?
Involuntary, pre-rational and reflex like mechanism which is part of the formation of visual impression. Fixed unconscious neural processing
What are illusions impervious to?
Experience - prior knowledge doesn’t help
What are many illusions explained by (the perceptual hypotheses)?
Many illusions are explained by stored knowledge leading to inaccurate perceptual hypotheses
What is the advantage of the constructivist approach?
Highlighted importance of both bottom up and top down processing
What is the criticism of the Constructivist approach
Vague
