Week 5-High-Level Perception

Question 1

Define Object Recognition

Answer 1

The ability to know what an object is (and process sensory information using our rod and cone cells) – it involves identifying the shape of
the object (despite changes in sensory input) and retrieving information from long-term memory about the object (e.g., its function, size, colour etc.). Object recognition takes around 200ms (very rapid).

Question 2

Why is Object Recognition Important?

Answer 2

• Neuropsychology - Brain injury / damage – affecting object and face recognition (hard to understand without research=hard to implement treatments and intervention techniques for specific brain regions and deficits).
• Computational modelling / machine learning / robotics (can embody into different technological systems which can benefit us).

Question 3

Define Object Constancy

Answer 3

The ability to recognise objects across variation in sensory input caused by changes in light (shadow), scale (size), viewpoint and occlusion (causing a different imprint on the retina). Essential for object / face recognition and high-level vision

Question 4

Object Recognition: What is the problem of shadow?

Answer 4

• When there is shadow it is
  challenging to identify the edges that define the object shape (using the primary visual cortex i.e., V1).
• The visual system must work to figure out which edges belong to the object and which belong to the shadow.
• Our rod and cone cells can distinguish information based on lighting allowing us to extract information about edges

Question 5

What other problems are there in Object Recognition?

Answer 5

• The problem of the variations in scale (size) – object size changes on
  the retina depending on how far away they are e.g., closer to retina =larger (is there a notion of size in our long-term memory and representations? must have some stored information on what the size of an object is typically)
• Object recognition is also difficult because of variations in spatial location– object position on the retina changes as objects move about.
• The problem of occlusion
  caused by scene clutter – foreground objects partially
  occlude background objects (note: we can still recognise objects despite incomplete sensory input! i.e., the brain has to reconstruct the missing information)

Question 6

What are the Gestalt Laws of Perceptual Organisation?

Answer 6

These are principles that describe how the visual system can use ‘bottom-up’ processes (driven by sensory-input) to group image features into shapes to form a whole (i.e., what the brain has to overcome for successful object recognition)

-Kurt Koffka (1886-1941)
-Wolfgang Kohler (1887-1967)
-Max Wertheimer (1880-1943)

Question 7

What are the 6 Gestalt Laws?

Answer 7

1. Law of Similarity (Items that are similar to each other are grouped together)
2. Law of Pragnanz (Reality should be transformed or reduced into the simplest form)
3. Law of Proximity (The objects that are close to each other are grouped together)
4. Law of Continuity (How our brain experiences visual line of elements that are grouped together)
5. Law of Closure (The tendency of our mind to perceive incomplete shapes as a whole figure)
6. The Law of Common Region (items within a boundary are perceived as a group and assumed to share some common characteristic or functionality).

Question 8

What does the perception of shape involve?

Answer 8

Both bottom-up (for our sensory information) and top-down (previous knowledge on things) knowledge.

-NOTE! The brain uses the principles of perceptual organisation to help us make sense of sensory input

Question 9

Where does Perceptual Organisation happen in the brain? Anatomical basis for high-level vision

Answer 9

-The VENTRAL pathway is associated with object recognition. Also associated with the parvocellular cells which rapidly respond to high spatial frequency (i.e., detailed images). Damage can vary e.g., face recognition problems but not object recognition problems

-Some of the main regions of the brain have been linked to the visual perception of different types of objects: LOC = Lateral Occipital Complex (objects).

-Other views of the brain regions associated with object recognition.
LO/pFs = Lateral Occipal Complex; FFA = Fusiform Face Area; PPA = Parahippocampal Place Area (around the hippocampus-activates when shown scenes)

-The best way to recognise things is through specialisation in brain regions according to the brain

Question 10

What shape information is used to represent objects?

Answer 10

-Focuses on low-level image features such as edges and vertices (intersections between edges) to provide important shape information.

-We need to be able to reconstruct the shape of things.

Question 11

What is Attneave’s (1954) ‘Sleeping Cat’?

Answer 11

• A simple demonstration that we can derive a lot of useful information about shape from the edge information alone.
• Also suggested we can recognise things from very limited information
• “Line drawing of a sleeping cat can still be identified when the smoothly curved contours are replaced by straight-line segments”

Question 12

But….What is this object? (Biederman, 1987)

Answer 12

-The image on the right is easier to identify despite containing less edge contour. This shows that VERTICES are important, not just edges (i.e., edges aren’t enough for recognising an object vertices gives us more information about shape).

Question 13

What evidence is there for edges and vertices being important sources of information for shape recognition? (Biederman, 1987)

Answer 13

-He presented objects with different amounts of edge contour (mid-segments) and vertices deleted, and measured recognition accuracy.

-The results showed that deletion of vertices affects recognition more than deletion of other edge contour (i.e., made more errors).

-Shows vertices are more important.

Question 14

What other kind of features can describe shape?

Answer 14

1. Surfaces (gives information about texture, patterns etc.,)
2. Volumetric parts (individual parts of an object rather than a whole)

-Study note: These shape features are sometimes called ‘primitives’ which means elementary ‘units’ of shape. A volumetric primitive is one that has 3D volume (e.g., a volumetric part). E.g., we could say a table has five volumetric parts (four legs and a top)

-Low-level image features such as edges and vertices (intersections
between edges) provide important shape information.

Question 15

What is one of the Structural Description Models of Object Recognition? (Biederman, 1987)

Answer 15

-In Biederman’s (1987) model 3D objects are represented and recognised using basic volumetric parts (primitives) known as ‘geons’ (like drawing art).

-We have 36 geons which we can mix and match to create a variety of abstract shapes

-Study note: You can think of geons as basic shapes like cylinders, bricks, pyramids etc., that you could use to ‘build’ more complex objects (like ‘Lego’ bricks!).

Goal of visual system - identify 3D parts that comprise the object

Question 16

What is one of the Structural Description Models of Object Recognition? (Marr, 1982)

Answer 16

Structural description theory of high-level vision:
-Marr’s model is another structural
description where complex 3D object shapes are represented by their parts.

-Argues we have 1 geon (‘a generalised cylinder’) and we can build our objects using this general shape e.g., human, then arm, then forearm, then hand (all using the generalised cylinder).

-Both models argue object recognition is achieved through having an abstract representation in memory.

Question 17

What does the question “are faces special?” mean?

Answer 17

-Are faces represented differently to common objects (and words)?

-Are faces processed differently to other categories such as common objects (and words)?

-Do different categories of objects have distinct neural correlates?

-NOTE! The general topic is whether objects, faces and words are processed by different systems in the brain (what we will call ‘domain-specific’ processing).

Question 18

What is the problem of Face Recognition?

Answer 18

Face recognition presents a difficult computational problem for the visual system:
* Faces are visually similar – requires subordinate level classification.

• Perceptual input highly variable across viewpoint.
• Faces are dynamic, moving stimuli (i.e., facial expressions).
• Faces are variable because of emotional expressions and speech.

-Issues with object constancy is also seen with faces

-NOTE! Remember: Subordinate-level classification occurs when we have to distinguish one category member from another (e.g., the faces of two different people, or two different kinds of car). Basic-level classification would be recognition of a particular category (e.g., distinguishing a table from a chair). Face recognition is a subordinate-level classification problem. They also move and express emotion (unlike other objects – so faces pose a challenge for recognition).

Question 19

What is an Information Processing Model of Face Recognition? (Bruce & Young, 1986)

Answer 19

-In this model face perception involves the abstraction of facial expression.

-There are separate representations of expression and facial identity (i.e., you can tell the identity of a person regardless if they are crying, laughing etc.,).

-NOTE! This should remind you of the general vision problem of abstraction – to recognise a face do we have to ‘remove’ emotional expression? This is like with objects when we asked whether we had to abstract (remove) shadow and orientation from shape. Same old
problem! Does your ‘mental representation’ of your best friend have an expression or is it ‘neutral’ (in some sense). What information about your friend’s face does it contain? Also note – you do NOT have to learn this model – we are only using it to illustrate how face
identity and expression might be separately stored in memory

Question 20

What was the importance of faces in evolution?

Answer 20

-Faces are important to survival and social interaction

-We are remarkably adept at seeing faces –have we evolved specialised brain mechanisms for face perception and recognition?

Question 21

Are our brains ‘tuned’ to organise perceptual input into meaningful objects?

Answer 21

If you think about how we impose order - the brain automatically uses the principles of perceptual organisation to impose order on sensory information (e.g., seeing faces in clouds).

-This is because we can extract information quickly to see faces (faces must be somewhat abstract if we can see it in random things).

Question 22

What are the effects of Face Inversion?

Answer 22

We find it harder to recognise inverted faces than inverted objects

Question 23

What is the Parts and configural processing of face stimuli?

Answer 23

-It takes us longer to process our own face when inverted

-With an object we care more about the parts whereas with faces we care more configurally (i.e., as a whole and the overall relationship between the parts).

Question 24

What does the ‘Thatcher’ Illusion tell us? (Thompson, 1980)

Answer 24

• Face perception is ‘tuned’ to the upright orientation. We do not
  notice feature inversion in the upside down image.
• This suggests that face processing relies on configural or ‘holistic’
  information about the whole stimulus, rather than an ‘analytical’
  processing of each face feature (part).

-Supports the idea that the face is processed configurally.

Question 25

What was found in this task? Passive viewing of faces and objects during fMRI scanning

Answer 25

The Fusiform face gyrus is a key area in the brain responding to faces.

Question 26

What was Bentin et al’s (1996) study?

Answer 26

-Task: Subjects presented with five categories of visual stimuli (faces,
scrambled faces, cars, scrambled cars and butterflies and asked to mentally count the number of occurrences of a specified target
category.

-The N170: A Face-Specific ERP Component - PO9 (one electrode) there is a positive deflection (rise) then negative deflection (decrease) and so on. N170=Negative 170ms is time we get the negative deflection (you only get this when you show a face stimulus).

-Study note: This slides shows the EEG effect known as the N170 which (it has been claimed – see Thierry et al. (2007) for an alternative view (this paper is on Blackboard) is specific to faces. It is called the N170 because it is a negative (downward) change in the EEG that is found 170ms post-stimulus.

Question 27

What is the Perceptual Expertise Hypothesis? (Gauthier et al., 1999)

Answer 27

-Based off Bentin et al’s (1996) results, the ‘Perceptual Expertise Hypothesis’ was proposed. It states
that there is NO domain-specific processing for faces. Rather, the face-specific effects previously found in other studies solely reflect subordinate-level processing of faces (since subordinate-level processing of novel objects like Greebles gives you the same results!

-N170 doesn’t respond to faces BUT instead things that are similar visually (e.g., distinguishing between houses, cars, faces etc.,)

-Study note: Here is a definition of the ‘Perceptual Expertise Hypothesis’. In brief, it says that it is not FACES per se that matter, but that faces are processed at a subordinate-level. So anything processed to a subordinate level should give face-like effects (at least, according to the Perceptual Expertise Hypothesis).