Pattern Recognition

Question 1

What is agnosia?

Answer 1

A condition where people have trouble recognising objects due to brain deficits.

Question 2

What do people with visual agnosia have trouble with?

Answer 2

The brain processes that understand details (colour, shape, motion) of an image are intact, however, they cannot put all the parts of an image together as a pattern of a coherent object.

Question 3

What do visual agnosia patients tell us about perception?

Answer 3

Recognising parts of an object is different to the recognition of an object as a whole. Therefore, there is a step where the parts need to be integrated into the perception of a whole object.

Question 4

What is the ventral stream in cortical pathways for vision responsible for and what happens when it is damaged?

Answer 4

This is the ‘what’ pathway that moves from V1 in the occipital lobe to the temporal lobe. Processes information about object appearance, identity and object perception. When damaged you lose your ability to recognise an object whilst keeping other parts of perception.

Question 5

What is the dorsal stream in the cortical pathways for vision responsible for?

Answer 5

This is the ‘where’ pathway, which moves from V1 to the parietal lobe. Processes spatial information about objects, is important for guiding action and object location.

Question 6

What is it called when pathways from the occipital lobe split into the temporal and parietal lobes?

Answer 6

The division of labour in information processing. If there is damage in either pathway, the ability to process information in that way is lost.

Question 7

Patient DF had severe agnosia. Her visual acuity and nonvisual object recognition were intact. She was able to correctly place a card into a slot by moving the card towards it. When the slot was rotated 90 degrees and she was asked to place it in the slot, she was not able to do this, why?

Answer 7

Because her dorsal “where” pathway was intact she was able to place the card in the slot. This means her ventral or “what” pathway was damaged, meaning she could not visually identify the slot and its orientation.

Question 8

Optic ataxia is the opposite of agnosia. Where is the damage located in patients with optic ataxia and what is the outcome of patients with these lesions?

Answer 8

This is where patients have damage in the dorsal pathway or the “where” stream. Meaning, they are able to recognise an object but unable to use visual information to guide their action.

Question 9

What are the four gestalt principles of object recognition?

Answer 9

Similarity (similar things tend to be grouped together)

Closure (We tend to find closure when we see an open object)

Good continuation (figures with edges that are smooth are more likely seen as continuous than edges that have abrupt or sharp angles)

Proximity (Things that are closer together have a relationship to one another)

Question 10

What is template-matching?

Answer 10

A theory of pattern recognition modelled on computers. Pattern recognition proceeds by comparing an incoming sensory stimulation pattern to mental images or representations of patterns until a match is found.

Question 11

How did gestalt theory of pattern recognition generate new theories?

Answer 11

Gestalt theory showed us that there is an intrinsic tendency to segment things together in a certain way, but it did not tell us why.

Question 12

Based on Selfridge’s pandemonium model, what is feature analysis theory?

Answer 12

When an object is received from the environment it is broken down into its smallest componential parts. There is a hierarchy of feature detectors that work to identify each individual element of the object. The pattern of that object is recognised and then put back together again in the brain. Based on this information the brain is able to identify the object holistically. This is the theoretical explanation behind gestalt theory.

Question 13

In feature analysis theory, the HB paradigm is used to test how our brain tries to comprehend the stimulus it is presented with. What happens when we stare at this image for a while?

Answer 13

Eventually, our rods and cones become tired and elements in the image become lost simultaneously. The way we lose our visual perception is not random, meaning the specific way of losing visual perception suggests that we are using each element as a component of a pattern.

Question 14

What is recognition-by-component theory?

Answer 14

This is an extension of feature analysis theory, whereby 3-D object recognition is segmented into a set of basic sub-objects (geons) and then recognised as a pattern composed of these geons. Essentially, decomposition of elements and putting them back together again in the brain.

Question 15

How do neural mechanisms within the brain, from the retina to the visual cortex, recognise patterns within the environment?

Answer 15

The processes become increasingly more complex as information moves through the brain. In the retina in the LGN it recognises a dot, V1 then recognises stipes, once you reach V4 they respond to something a little more complex than a dot and a line. Once in the ventral pathway, you reach the TE area where neurons respond to human faces.

As you move higher in the hierarchy, neurons are responding to increasingly complex patterns.

Question 16

As neurons become more complex in the chain of pattern recognition, it creates an interesting hypothesis called “grandmother cell”, where there is one particular neuron that is responsible for firing whenever it sees your grandmother. What are the criticisms of this hypothesis?

Answer 16

Neurons are unreliable, they are firing constantly, if there was a “grandmother cell” there’s a chance you may accidentally see your grandmother when she is not there. Or if that neuron is lost you will not be able to realise who your grandmother is.

Question 16

As neurons become more complex in the chain of pattern recognition, it creates an interesting hypothesis called “grandmother cell”, where there is one particular neuron that is responsible for firing whenever it sees your grandmother. What are the criticisms of this hypothesis?

Answer 16

Neurons are unreliable, they are firing constantly, if there was a “grandmother cell” there’s a chance you may accidentally see your grandmother when she is not there. Or if that neuron is lost you will not be able to realise who your grandmother is.

Therefore, the perception of novel objects cannot be explained well.
&
The flexibility of object recognition cannot be explained well.

Question 17

In the literature, what is the explanation that is most accepted by researchers as the reason to why the grandmother cell may not exist?

Answer 17

The view is, object recognition is the result of the firing of an ensemble of cells called ensemble coding. When all groups are firing, the combination is taken, and then you can perceive your grandmother.

Question 18

What needs to be considered when recognising patterns in the environment?

Answer 18

The context. We are constantly using top-down and bottom-up processing in order to completely understand a stimulus and recognise its pattern.

Question 19

What is a contextual effect?

Answer 19

When regarding object-recognition, there is usually a context it exists in. We bring this context into understanding a stimulus holistically.

Question 20

What is top-down processing, and why do we use it?

Answer 20

Your brain applying knowledge to situations in order to anticipate what is next. When perceiving something we begin with the most general and move towards the more specific. These perceptions are heavily influenced by our expectations and prior knowledge.

It helps us simply our understanding of the world and quickly make sense of all the information our senses bring in.

Question 21

What is bottom-up processing, why do we use it?

Answer 21

Bottom-up processing is an explanation for perceptions that start with an incoming stimulus and working upwards until a representation of the object is formed in our minds. This process suggests that our perceptual experience is based entirely on the sensory stimuli that we piece together using only data that is available from our senses. Bottom-up processing breaks the process down into its most basic elements.

Question 22

What is the difference between bottom-up and top-down processing?

Answer 22

Bottom-up:
Data driven
Focuses on incoming sensory data
Takes place in real time

Top-down:
Information is interpreted using contextual clues
Uses previous experience and expectations