Lec 5: Object Recognition

Question 1

object constancy

Answer 1

ability to recognize an object in countless situations

Question 2

3 factors in which the visual information from an object varies

Answer 2

1. Viewing position
2. Illumination conditions
3. Context

Question 3

Ventral stream

Answer 3

• “What”
• neurons of the temporal lobe can be activated by a stimulus that falls within either the left or right visual field
• essential for determining identity of an object

Question 4

Dorsal stream

Answer 4

• “Where”
• neurons in the parietal lobe may respond similarly to many different stimuli
• include cells representing both the fovea and the periphery
• critical for determining locations of objects and guiding interactions with these objects
• helps us construct 3D map of space, so you know where objects are

Question 5

visual agnosia

Answer 5

deficit in recognizing objects even when the processes for analyzing basic properties such as shape, colour, and motion are relatively intact

Question 6

lateral occipital cortex (LOC)

Answer 6

• part of ventral pathway
• processing in this area is essential for shape perception and recognition

Question 7

optic ataxia

Answer 7

• patient has difficulty using visual information to guide actions, even though they are unimpaired in ability to recognize objects
• associated with lesions of the parietal lobe

Question 8

Functional near-infraed spectroscopy (fNIRS)

Answer 8

• good for younger children
• non-invasive
• measures infrared light absorbance in oxygenated and de-oxygenated hemoglobin
• provides info about ongoing brain activity

Question 9

repetition suppression (RS) effect

Answer 9

• hypothesized to indicate increased neural efficiency
• seen in fMRI in which the BOLD response to a stimulus decreases with each subsequent stimulus repetition
• neural response to the stimulus is more efficient and perhaps faster when the pattern has been recently activated

Question 10

multistable perception

Answer 10

• phenomenon where an ambiguous or reversible visual stimulus can lead to multiple, stable interpretations that switch back and forth in perception
• stimulus information does not change at the points of transition from one percept to the other, but interpretation of the pictorial cues does

Question 11

gnostic unit

Answer 11

• type of neuron that can recognize a complex object
• tuned for a specific percept (ie. an apple)
• referring to the idea that the cell signals the presence of a known stimulus

Question 12

hierarchal theories of object perception

Answer 12

• suggest that our brains process visual information in layers or stages, moving from simple features to complex representations
• perception is organized so that lower-level areas of the visual cortex detect basic elements like edges, colors, and orientations, and higher-level areas progressively integrate these elements to recognize complex shapes and ultimately identify objects

Question 13

grandmother-cell hypothesis

Answer 13

• proposes the existence of highly specialized neurons that respond to specific, complex stimuli, such as the face of one’s grandmother
• a single neuron (or a very small population of neurons) in the brain becomes uniquely tuned to recognize a specific individual or concept based on repeated exposure and learning

Question 14

ensemble hypothesis

Answer 14

• suggests that object recognition is achieved through a distributed network of neurons, each of which responds to different features or aspects of the object
• network of neurons, collectively referred to as an “ensemble,” works together to represent the object as a whole
• account for why we can recognize similarities between objects and may confuse one visually similar object with another because both objects activate many of the same neurons
• also account for our ability to recognize novel objects, because they bear a similarity to familiar things and our percepts result from activating units that represent their features

Question 15

Template theory

Answer 15

• brain compares incoming sensory input to stored “templates” or “representations” of known patterns or objects
• consistency of the representation of the object is important
• recognizes an object as a whole, does not break it down
• would have to have templates of every object you have ever seen in every orientation you have ever seen it
• no biological plausibility: can’t be how it works from a neural standpoint because it would require too many neurons and representations

Question 16

Feature analysis theories

Answer 16

• may work for some things we see
• we recognize objects and patterns by detecting and analyzing their basic features—such as lines, angles, shapes, colors, and textures
• visually breaking down an object and then reassembling it back into what it is

Question 17

Recognition by components (RBC)

Answer 17

• “Lego Theory”
• objects consist of 3D pieces called geons
• have to think about how the geons come together to recognize an object
• insensitive to object orientation but sensitive to geon visibility (have to see all the geons to recognize an object, not always possible)
• arrangement of the geons matters
• deficit in recognizing particular objects because distinguishing factor between 2 objects may have nothing to do with shape

Question 18

view-dependent

Answer 18

suggests that object recognition depends on the specific view or perspective from which the object is seen

Question 19

view-invariant

Answer 19

suggests that object recognition is based on abstracted, stable features of the object that do not change regardless of the viewpoint or perspective

Question 20

Repetition suppression

Answer 20

• brain’s response to a stimulus decreases upon repeated presentation of the same stimulus
• first time seeing, lots of brain areas respond to properties of the object
• after seeing a few times, start to recognize that certain features have to be there for it to be a specific object and other features don’t matter ie. colour
• maybe over time brain is fine-tuning the features that it is paying attention to and filtering out the things that don’t really matter

Question 21

prosopagnosia

Answer 21

• inability to distinguish between faces
• can recognize individual features on the face but cannot put them together
• may be acquired or developmental

Question 22

Multi-voxel Pattern Analysis for “reading minds”

Answer 22

• looks at how different parts of the brain activate when you are thinking about something
• brain activates slightly differently when looking at one thing vs another

Question 23

Training Pattern Classification for “mind reading”

Answer 23

• can show people lots of different faces, then average brain activity across trials
• now have a template for brain activity when looking at a face
• do the same with houses
• feed these templates into the software and can now guess if you’re looking at a face or name
• not very accurate

Question 24

Object recognition in congenitally blind

Answer 24

• found activation of ventral stream is very similar whether you are born without vision and are touching an object or you are born with typical vision and seeing the same object
• organization of knowledge in ventral stream does not differ

Question 25

associative visual agnosia

Answer 25

• inability to recognize or identify objects, despite having intact visual perception and memory
• can see objects clearly but cannot associate the visual information with stored knowledge about the object’s meaning or use
• can recreate the drawing of an object but can’t identify what it is
• disconnection of the ventral stream

Question 26

Apperceptive visual agnosia

Answer 26

• deficit in perceiving or recognizing objects despite having intact basic sensory functions such as vision and memory
• unable to perceive objects as whole, unified entities, making it difficult for them to identify, copy, or match objects, even though they can see their individual features (like shape, color, or size)

Question 27

Fusiform expertise area and greebles

Answer 27

• wanted to show that face area did more than just faces
• created unique playdough faces (greebles)
• before having learned the specific greebles, FFA did not activate
• after becoming experts on greebles, FFA activated
• found that the area activates for areas of expertise

Question 28

Discovering form and object processing in ventral stream

Answer 28

• 1st showed people things that looked like they could be objects but weren’t (nonsense objects)
• compared to brain activity while looking at nonsense objects and visual noise
• this was looking at the part of the brain that analyzed form
• found activity along ventral stream in both hemispheres
• then compared brain activity while looking at objects and nonsense objects
• difference between conditions is object recognition
• found activity in ventral stream in both hemispheres
• conclude that ventral stream is involved in form processing and object recognition