Object Recognition

Question 1

What are the stages of recognition?

Answer 1

Perceptual recognition-allows us to recognise objects from different angles/lighting is and when parts are occluded.

Semantic recognition-allows us to know what object is and its function, and recall any associations we have with it.

Question 2

What is apperceptive agnosia?

Answer 2

Patients with this are unable to properly assemble the individual attributes of objects. They have an impairment in their perceptual recognition.
Lissauer 1890 - this was tested with figure copying tasks, which patients with apperceptive agnosia had an impaired performance.
Neuropsychological evidence supporting the different stages of recognition.

Question 3

Who is patient AS?

Answer 3

She could not recognise objects but using shape information could pick them up. Reaching versus orienting tests showed that she could easily reach for objects but when she was tasked with orientating the objects to a specific way she struggled. Grasping versus matching tests were also carried out and showed that she couldn’t tell if the two objects were the same or if there were two objects on the table but when told to pick them up she could with ease. This tells us that vision for action draws on different perceptual processes to vision for recognition. Plant et al 2015

Question 4

What is associative agnosia?

Answer 4

Patients can properly form object structure, but are unable to access stored knowledge about this. Can copy an object accurately but have no idea what it is. Semantic recognition is impaired. Person can perceive the object, redraw it and whatnot but can’t state what it is.
Lissauer 1890

Question 5

What is optic aphasia?

Answer 5

Lhermitte & Beauvois, 1973
In some instances, patients can perceive objects (perceptual recognition) and show semantic knowledge through mime and use (semantic recognition) but cannot name this object aloud. This type of disorder is called aphasia and can make it appear as if the patient has recognition failure when in fact it is a language difficulty.

Question 6

What is object constancy?

Answer 6

It is a concept that allows us to know that an object has a constant shape despite how it can look different from different view positions. People with occipital injury can struggle with this.

Question 7

How can object constancy be tested?

Answer 7

It can be tested by showing a picture of a drunk object to the subject, then showing two further pictures, where one of the pictures is the same as the first one but shown at a different angle. You would then ask which picture is the same as the first, if they get it right then they have no impairment in their object constancy.

Question 8

What are the routes to object constancy?

Answer 8

1) Template matching (Sutherland & Williams, 1969) - when you mentally rotate the object to see if it fits a template stored in the mind. Good for distinctive objects but not for objects that are simple and can looks similar to others.
2) Critical features (Selfridge, 1959) - when you recognise an object but something distinctive. Good for distinctive objects, but if that feature is obstructed then there is a problem. Plus any objects that are defined by their spatial arrangement e.g. faces.
3) Structural descriptions (Marr and Nishihara, 1979) - looking at how individual distinctive features are positioned to one another and the other aspects of the object.

Question 9

What is Marr and Nishihara (1978) theory of object recognition?

Answer 9

Suggests that when mapping the different aspects of the objects, you can link them to a common invariant property that does not tend to change with viewpoint. The main axis is normally the longest axis and tends to remain unobscured making it the best fit for the common invariant property.
Generalised cones - objects can be broken down into different axes and parts. Classes of objects can be distinguished by the various lengths and arrangements of these cones. Identifying the various axes of an object makes it possible to categorise the objects.

Question 10

What are the cons for the theory of object recognition?

Answer 10

The main axis can be obscured, either by an overlapping object or from the viewing angle.
Given that we can still recognise the stimuli under these conditions, there must be more than deriving axis-based descriptions.

Question 11

What is the Geon Theory (Biederman, 1987)?

Answer 11

All objects can be reduced to basic shapes, known as geons, and every object can be defined by the arrangement of these geons. This theory explains object recognition by suggesting that we simply see what shapes make up the object and how they are arranged together. Empirical support for this states that when geons intersect with one another they form a concavity. This is an area Where two edges meet to and they enable one Geon to be distinguished from another. When these intersections are removed, recognition becomes difficult.

Question 12

What is the problem with view-point independent theories?

Answer 12

Geon and Theory of object recognition are view-point independent. View-point independent theories try to explain how you can recognise an object from different viewpoints. These work for when talking about recognition that is independent of view point but there are some objects that are better recognised from certain viewpoints e.g. faces. This type of theory also makes it difficult when differentiating between different breeds of dog, for example dogs have the same parts and configurations but look different and this doesn’t explain how we can recognise different types. Another issue is objects that aren’t necessarily made up of obvious components or axes e.g. clouds.

Question 13

Considering most faces contain the same geons, how are we able to recognise them? How do we get the sensitivities for the subtle differences?

Answer 13

We do this through three different ways: facial consideration, within category discrimination and curvature.

Question 14

How do we recognise faces through facial configuration?

Answer 14

We can only recognise faces when they are configured to a specific way. The face inversion effect demonstrates how recognition is difficult when the facial features are inverted. This inversion disrupts the familiar configural cues. If the whole face is inverted with the inverted features, the face will look normal. This demonstrates how facial recognition is viewpoint dependent.

Question 15

What is within category discrimination?

Answer 15

With normal objects we don’t look to individuate but we do with faces. The brain processes for facial and object recognition are different, processes for faces are higher. It places much more rigourous demands on the visual system than the recognition of other objects which are recognised as subordinate levels. This is because we look to see the small individuations that differentiate faces from one another.

Question 16

How does curvature impact facial recognition?

Answer 16

It’s been seen that faces are harder to recognise when the curvature is affected but objects can still be recognised even with a significant curvature manipulation. Done with computer simulations.

Question 17

What is the reasoning behind why faces are processed differently to objects?

Answer 17

It is due to faces having social significance to us. This argument draws on three different lines of research: behavioural, brain imaging and neuropsychological.

Question 18

What is the behavioural evidence for why we process faces differently?

Answer 18

Support - inversion effect demonstrates that faces are harder to recognise than objects when inverted. May suggest that faces are processed in a more configurable manner than objects
Against - inversion test has the same results with animals as it does with faces if the participant is an expert in that animal I.e dog breeders (Diamond & Carey, 1986). This suggests that we do use the same processes and the differences in recognition are reflected in our familiarity with the stimuli.

Question 19

What is the brain-imaging evidence for why we process faces differently?

Answer 19

Support - fMRI scans show different brain areas activate for different objects and faces.
Against - Gauthier et al., 1999 created objects that he named ‘greeble’. These objects had different families where they shared a common attribute but each still remained individual. When the participants looked at faces the fusiform activated and did not for greebles but once they became experts in greebles the fusiform activated. Suggests that maybe the fusiform isn’t just for facial processing.

Question 20

What is the neuropsychological evidence for why we process faces differently?

Answer 20

Support - pure propagnosia makes patients poor at recognising faces but they can still recognise objects. Three patients found with this - Whiteley & Warrington 1977
Patient R.C. (Wilkinson et al., 2009) - suffered a stroke with unilateral right hemisphere damage. Still able to recognise objects but couldn’t recognise his closest relatives by facial appearance. MRI scan showed that the fusiform was not damaged but the higher processing areas had been damaged e.g. The parietal lobe. This demonstrates that a whole system is needed and not just one brain area.
Support 2 - pure object agnosia makes patients poor at recognising objects but normal with faces. Patient C.K (Moscovitch et al., 1999) bilateral brain damage sustained in z road accident. Visual acuity was intact, able to recognise real faces, caricatures, cartoons and faces composed of objects. Suffered with associated agnosia.
Against - demonstrated that pure propagnosics struggled with object recognition when the process required them to encode configuration, curvature and within-category detail. This was tested by using 2 very similar objects. This demonstrated that propagnosics is in fact a more general processing impairment that under the right condition is accompanied by object agnosia.