lecture 4 - object and face recognition

What is object recognition?

The ability to identify objects despite variation in viewing conditions

What are intrinsic factors in object recognition?

Object properties such as shape, surfaces, and parts

What are extrinsic factors in object recognition?

Viewing conditions such as position, lighting, occlusion, and deformation

Why is object recognition considered a difficult problem?

Objects must be recognised despite large variations in visual input

What is constancy in object recognition?

The ability to recognise an object despite changes in viewpoint, illumination, or appearance

What are view-independent theories of object recognition?

Theories proposing that objects are recognised via abstract structural descriptions

Which model is a key example of a view-independent theory?

Biederman’s Recognition-by-Components (RBC) model

Who proposed the Recognition-by-Components model?

Irving Biederman (1987)

What are geons?

Basic volumetric components used to represent objects

How many geons did Biederman propose?

36

What is the first step in Recognition-by-Components?

Breaking the object into basic sub-objects

What is the second step in Recognition-by-Components?

Categorising each sub-object as a geon

What is the third step in Recognition-by-Components?

Recognising the object by combining geons and their relations

Why are geons useful for object recognition?

They are easily discriminable, view-invariant, and robust to noise

What does view-invariant mean in relation to geons?

They look the same from most viewpoints

What does robustness to noise mean for geons?

They can be identified even when parts of the image are missing

What is an advantage of view-independent theories?

Only one object representation needs to be stored in memory

What is a limitation of view-independent theories?

Some objects are difficult to describe using a single structural representation

Why does object recognition not appear to be completely view-independent?

Recognition performance often depends on viewpoint

What are view-dependent theories of object recognition?

Theories proposing that recognition relies on stored view-specific representations

Which study supports view-dependent theories?

Gauthier et al. (1998) – Greeble expertise

What are Greebles?

Novel objects made of geon-like components

What was the aim of Gauthier et al. (1998)?

To study expert object recognition

What happened during the training phase in the Greeble study?

Participants learned to identify and name Greebles

What happened during the test phase in the Greeble study?

Participants identified Greebles under different conditions

What were the three test conditions in the Greeble study?

Normal view, rotated view, and brightness change

What was the main finding of Gauthier et al. (1998)?

Faster recognition for Greebles shown from the trained viewpoint

What does the Greeble study suggest about object recognition?

It is at least partly view-dependent

What is the current consensus on object recognition theories?

Multiple mechanisms are used depending on task and expertise

What are the three levels of object categorisation?

Superordinate, entry-level, and subordinate

What is a superordinate category?

A broad category (e.g. animal)

What is an entry-level category?

The most natural/basic category (e.g. bird)

What is a subordinate category?

A very specific category (e.g. sparrow)

At which categorisation level is recognition typically fastest?

Entry-level

When is a geon-based account most effective?

At the entry-level of categorisation

Where in the brain does object recognition primarily occur?

The inferotemporal (IT) cortex

Which visual pathway is involved in object recognition?

The ventral (“what”) pathway

What did Gross et al. (1970s) discover about the IT cortex?

Neurons respond selectively to specific objects

What is visual agnosia?

The inability to recognise previously learned objects by sight

Is visual agnosia caused by basic sensory deficits?

No

Is visual agnosia a memory disorder?

No

What causes visual agnosia?

Lesions in the inferotemporal cortex

What are the three main types of visual agnosia?

Apperceptive, associative, and category-specific agnosia

What is apperceptive agnosia?

A failure to form a coherent percept

What is associative agnosia?

Inability to link visual input to meaning

What is category-specific agnosia?

Impaired recognition of a specific class of objects

Why is face recognition considered special?

Faces are processed holistically and by specialised brain regions

What does holistic processing of faces mean?

Faces are perceived as integrated wholes rather than individual features

Which brain area is strongly associated with face recognition?

The fusiform face area (FFA)

Where is the fusiform face area located?

In the ventral temporal cortex

What evidence supports specialised face processing?

Difficulty recognising inverted or configural changes in faces

What is prosopagnosia?

The inability to recognise faces by sight

Is prosopagnosia due to visual or intellectual impairment?

No

What causes acquired prosopagnosia?

Brain injury or stroke affecting the FFA

Can prosopagnosia be congenital?

Yes

What proportion of the population is affected by prosopagnosia?

Approximately 1–2.5%

What debate surrounds the fusiform face area?

Whether it is face-specific or expertise-specific

Which study investigated expertise and FFA activation?

Gauthier et al. (2000)

What did Gauthier et al. (2000) find about FFA activation?

FFA activation increased for objects of expertise (Greebles)

What does increased FFA activation for Greebles suggest?

The FFA may be involved in expertise rather than faces alone

What does object recognition integrate to identify objects?

Information about form, depth, and colour

What happens when ventral visual areas are damaged?

Object recognition impairments such as visual agnosia occur