Week 6: Object Recognition Flashcards
allows consistent identification despite changes
in viewpoint, size, lighting, and occlusion
Invariant object recognition
ensure recognition from various angles and sizes
Viewpoint and size invariance
Occlusion tolerance
allows recognition
even when objects are partially hidden
How does Context, memory, and attention further
enhance recognition?
by integrating
familiar environments and past experiences.
provide insights into the processes and mechanisms
of object recognition
Behavioral and psychophysics studies
reveals the specificity of visual deficits from brain damage, showing how
different areas contribute to various aspects of vision.
Lesion work
provides a macroscale view, identifying regions involved in visual processing
and their functions
Neuroimaging
examines individual neuron responses to objects and stimuli, offering a
microscale perspective.
Electrophysiology
helps uncover the underlying algorithms that simulate how the brain processes visual information for object recognition.
Computational modeling
Why is foveating crucial for object recognition?
it allows the visual system
to focus on specific regions of interest
with high resolution, while peripheral
vision captures less detailed information
As it highlights importance of foveating by
demonstrating how eye movements and
fixation patterns change based on the
task at hand; What does Yarbus experiments suggest?
that the brain
directs the fovea to areas most relevant for processing and interpreting visual information
extends beyond the primary visual cortex to multiple areas in the visual processing hierarchy
large cortical allocation to
foveal vision
extensive allocation
highlights the fovea’s key role in what?
high-resolution vision and
complex visual tasks,
underscoring its importance
throughout visual processing
stages.
Characteristics of p-cells
smaller cell bodies, smaller receptive fields, its color sensitive, sustained response, ventral pathway
Characteristics of m-cells
larger cell bodies, larger receptive fields, its NOT color sensitive, transient response, dorsal pathway
In the early stages, the representation of visual information does what?
Significantly diverges from the sensory input.
What does the combinatorial and hierarchical transformation of visual info enable?
the recognition of
increasingly complex patterns such as oriented lines and
gratings through the integration of simpler receptive field (RF)
inputs.
As visual information progresses from retina to the primary
visual cortex (V1), it undergoes a __________________________.
combinatorial and hierarchical transformation
motion detection
middle temporal area (MT)
color processing
V4
where information
from multiple neurons converges to shape more complex and specialized visual processing capabilities.
convergent inputs
What is the key to emerging complex receptive fields ?
The principle of multiple neurons from an earlier
area (closer to retina input) converging onto a
single neuron in a subsequent area
What does the Hierarchical Organization Across Visual Cortex allow for?
the integration of simple visual inputs, and further synthesis of sophisticated representations in extrastriate areas
The principle of multiple neurons from an earlier
area (closer to retina input) converging onto a
single neuron in a subsequent area is essential for?
the brain’s ability to
interpret complex visual scenes, gradually
constructing detailed and nuanced representations from basic visual elements.
In the primate inferotemporal cortex, what are the 3 key properties that emerge?
- Increasing spatial receptive field
- Longer responses latencies
- The gradual emergence of
selectivity for specific object
shapes
In the primate inferotemporal cortex, what allows for the processing of
more complex visual information?
Increasing spatial receptive field sizes
In the primate inferotemporal cortex, what causes Longer response latencies?
additional synapses involved in processing.
Describe the double dissociation of object and landmark recognition.
Object recognition: Monkeys were first shown an object and then, during a test phase, were given the choice to select an object they had not seen before (novel object) to receive a reward.
Landmark test: Monkeys were trained to recognize landmarks and subsequently required to choose the food well located nearest to a tall cylinder, serving as the landmark, to obtain a reward
What did Ungerleider and Mishkin (1982) identify?
two distinct
pathways in the visual system based on animal lesion
studies, later supported by human neuropsychological
evidence
Extends from the primary
visual cortex to the inferotemporal cortex, supporting object perception and recognition by processing
shape, color, and identity.
The ventral “what” pathway
Extends from the primary
visual cortex to the parietal lobe, specializing in spatial
perception, including object location, movement, and
visually guided action.
The dorsal “where” pathway
What does the distinction between the ventral and dorsal pathway highlight?
the functional specialization of
visual processing, demonstrating how different brain regions contribute to perception and interaction with the environment.
Where is the ventral visual pathway located?
medial, ventral
How is info processed?
serial and parallel processing
a condition that affects how your brain processes what you see. Your vision works correctly, but your brain doesn’t. This affects how you recognize objects, people, places and more. This issue usually happens because of injuries or diseases of the brain. It’s often permanent, but therapy can help people adapt to this condition. It is a condition that makes it difficult to recognize objects or faces, or to navigate
Visual agnosia
Describe Patient DF
Patient DF could detect light and colors, with
normal acuity and visual memory.
* She had no difficulty naming objects (not anomia)
but showed severely impaired picture
recognition.
* She produced crude descriptions of displayed
objects and struggled with simple geometric
forms.
what did findings suggest from Patient DF?
Findings suggest a deficit in form perception,
linked to ventral stream damage.
What was patients DF’s dissociation between Perception and action?
she could not visually copy objects when drawing them. But she could draw from memory.
In the card posting task, she could use vision for action, she failed to correctly orient the card for posting. but she was able to put card in slot if was just told to do that regularly.
What is the card posting task?
patients have to post a card (or their hand) through a slot of varying orientations while the matching task requires them to indicate the slot’s orientation as accurately as possible
How did Miner and Goodale redefine the dorsal stream?
They redefined it as the “How” pathway focused
on action guidance.
processes real-time visual information to coordinate movements like reaching and grasping
Dorsal “how” stream
What does the redefining of the dorsal stream highlight?
its role in visually guided behavior, directly linking perception to action. “How”
What did patient DF have?
Apperceptive Agnosia
Specific deficits in color
and motion processing
Achromotopsia,
akinestopia
Deficit in coordinating visual
input and hand movements
Optic Ataxia
is a rare condition that makes it difficult to recognize objects. People with this condition have intact knowledge of objects, but they can’t perceive their correct form
Apperceptive Agnosia
Core issue of Apperceptive Agnosia
A disruption in the earliest stages of perception,
making it difficult to form a coherent visual representation
of objects
Typical presentation of Apperceptive Agnosia
- Individuals can see basic features (e.g., lines, edges) but
cannot combine these features into a meaningful whole
shape. - They often cannot match or copy simple figures
accurately because the overall shape is not perceived in
an organized way.
What did Patient F.R.A. have?
associative visual agnosia
Describe patient F.R.A.
- Normal visual acuity, speech, spatial sensitivity, and memory, alongside an ability to detect basic shapes.
- Has challenges in naming objects presented visually and does not recognize ‘a’ and ‘A’ as representing the same letter.
- Cannot name objects he sees, but he
can color in parts of complex
drawings, demonstrating his ability to dissect visual stimuli without recognizing them. This indicates a nuanced understanding of the object’s structure despite his identification difficulties.
core issue of Associative Agnosia
An inability to link a visually formed percept
(which may be intact) to the stored knowledge of that
object’s meaning or identity
Typical Presentation: Associative Agnosia
- Individuals can typically copy or match objects quite
accurately, suggesting that “lower-level” perception is
intact. - However, they cannot recognize the object or link it to semantic knowledge (e.g., naming or stating its function)
What did patient CK have?
Integrative Visual Agnosia
Describe patient CK
Struggles to perceive objects as a unified whole
and tends to dissect objects into their
components.
* Capable of drawing objects even when he cannot
name them. e.g., when tasked with replicating a
figure, his drawing was accurate, with the two
diamonds and the circle being easily distinguishable.
However, the sequence in which he drew the
segments was unconventional.
* With appropriate contextual clues, C.K. could deduce
the identities of many objects by analyzing their
distinct parts. e.g., he identified an object on the table
as a “trophy” by recognizing a stand with an item
attached to it.
Core issue of integrative agnosia
A disruption in synthesizing multiple parts or
features into a single, integrated whole.
Typical Presentation: Integrative Agnosia
- Individuals can identify separate components (e.g., color,
lines, corners) but cannot effectively “group” these
features when the presentation is complex. - For simpler shapes, they might do better, but once the
object’s parts overlap or are jumbled, recognition falls
apart. - They may produce highly fragmented copies or drawings,
indicating they struggle to see how pieces fit together
Individuals can see elements
of objects but struggle to synthesize them into
recognizable shapes, affecting their ability to
recognize or copy designs.
Apperceptive Agnosia
Individuals can identify simple
shapes but cannot recognize the object as a whole,
especially in complex or jumbled presentations.
integrative Agnosia
Individuals can perceive and
detail objects but fail to connect these perceptions
to known objects, leading to an inability to recognize
objects despite clear visual representation.
Associative Agnosia
Types of visual agnosias
Apperceptive Agnosia
Integrative Agnosia
Associative Agnosia
Challenges to categorical distinctions of visual agnosias?
Overlap and Variability: Critics argue that the distinct categories often fail to capture the overlapping presentations seen in real patients, suggesting that these subtypes aren’t always clearly distinguishable in practice.
* Neural Underpinnings: The precise neural correlates of each agnosia subtype can vary or overlap significantly, challenging the idea of
distinct and separable pathways.
in the ventral temporal cortex is crucial for object
recognition, with particular sensitivity to object shapes.
- The ability to process shape information is essential for identifying and differentiating
objects, reflecting the brain’s specialization in detailed visual processing.
Lateral Occipital Complex (LOC)
responds consistently
to shapes, whether defined by light-dark contrast or by motion contrast between coherent and random dot movement.
LOC
Since the LOC responds consistently to shapes, not matter the contrast, what does this suggest?
The LOC processes shape
independently of specific
visual cues, highlighting its role in robust object recognition.
Will the LOC respond similarly to both familiar and novel objects?
yes, as long as they have
coherent shapes, but shows
reduced activity for
scrambled shapes.
The LOC responds similarly
to both familiar and novel
objects, as long as they have
coherent shapes, but shows
reduced activity for
scrambled shapes. What does this suggest?
This suggests that the LOC
prioritizes structural and
geometric configurations
rather than the semantic
meaning of visual stimuli.
True or false: Extrastriate (intermediate) visual areas along the
ventral visual pathway are sensitive to motion.
FALSE, they’re sensitive to object form
A selective response where a small subset of neurons is activated by a specific stimulus, creating an efficient and specialized representation of visual information
Localist (Clustered) Coding
A broader array of neurons responds to various stimuli, with each stimulus represented by a unique pattern of neural activity, allowing for a rich
and nuanced perception of the visual environment
Distributed Coding
How many neurons in a single fMRI voxel?
hundreds of thousands
What are all the face-selective regions identified with fMRI?
1, occipital face area
2. posterior FFA
3. anterior FFA
4. posterior STS
5. middle STS
6. anterior STS
7. anterior temporal
a region of the brain located in the medial temporal lobe, near the hippocampus. It is involved in processing and recognizing scenes, such as landscapes, cityscapes, and indoor environments
PPA ( parahippocampal place area )
is a part of the brain that’s involved in memory, decision-making, and spatial cognition. It’s located in the parietal lobe and is densely connected to the hippocampus.
RSC (retrosplenial cortex)
a brain region that processes visual information about scenes and is involved in navigation. It’s located near the transverse occipital sulcus.
OPA (occipital place area)
refers to the idea that different regions within the ventral visual pathway of the brain are specialized for processing specific types of visual information, such as recognizing faces, objects, scenes, or body parts, with each area contributing unique features to the overall perception of a visual stimulus
Functional Specificity Across Ventral Visual Cortex
located in the left ventral occipitotemporal cortex (vOTC) of the brain; essentially, it’s the part of the brain specifically designed for processing written words.
visual word form area (VWFA)
TRUE OR FALSE: multiple areas in dorsal cortex
selectively respond to specific
visual categories in primates
FALSE; ventral cortex
Why is it insufficient to infer certain neurons code for faces?
because we haven’t tested EVERYTHING
do the neurons in the Unit Ri-10 experient code for faces?
NO
A neurological disorder characterized by an
inability to recognize faces, including
familiar ones, while object recognition and
other visual processing typically remain
intact.
prosopagnosia
What damage does prosopagnosia result from?
damage to ventral stream regions, particularly the fusiform face area (FFA) in the temporal lobe, which is specialized for facial recognition
Despite face recognition deficits, what can individuals with prosopagnosia usually identify?
people using non-facial cues, such as voice,
clothing, or hairstyle.
Explain the case of WJ
- A 51-year-old professional developed
prosopagnosia following a series of strokes. - Afterward, he took up sheep farming.
- Although human face recognition is typically
easier than recognizing sheep, WJ found it
easier to recognize sheep than human faces.
What does the case of WJ suggest?
that face recognition
relies on distinct neural mechanisms
from general object recognition.
Explain the case of Patient KC
Able to perceive individual parts
but not the whole (integrative
agnosia).
* Experiences alexia, an inability to
recognize written words,
* Retains the ability to recognize
faces.
what does WJ have?
prosopagnosia
what does KC have?
agnosia without prosopagnosia (object agnosia)
how does patient KC compensate for their deficits?
Compensates for these deficits by
relying heavily on color
processing to aid in identification
tasks.
What does a single dissociation do between face and object agnosia?
A single dissociation (e.g., comparing a
face agnosia patient to a control) provides
weaker evidence because it does not confirm
that object recognition is independent from
face recognition
What does a double dissociation do between face and object agnosia?
A double dissociation (e.g., contrasting
face agnosia with object agnosia) shows
that face and object recognition rely on
distinct neural mechanisms.
what does a double dissociation (.g., contrasting
face agnosia with object agnosia) demonstrate?
that damage can selectively impair face recognition while sparing object recognition, and vice versa, highlighting the specificity and
independence of these cognitive processes.
explain Temporary Disruption Through TMS: Triple Dissociation
transcranial magnetic stimulation (TMS) over three adjacent functionally localized areas in extrastriate cortex. In three experiments, participants performed discrimination tasks involving faces, bodies, and objects while TMS was delivered over the right occipital face area (rOFA) [12], the right extrastriate body area (rEBA) [13], or the right lateral occipital area (rLO) [14]. All three experiments showed a task selective dissociation with performance impaired only by stimulation at the site selective for that category: TMS over rOFA impaired discrimination of faces but not objects or bodies; TMS over rEBA impaired discrimination of bodies but not faces or objects; TMS over rLO impaired discrimination of objects but not faces or bodies. The results support a modular account in which category-selective areas contribute solely to discrimination of their preferred categories.
an individual’s ability to recognize faces is impaired due to brain injury,
such as a stroke, affecting areas of the brain specialized in facial recognition.
Acquired Prosopagnosia
Face recognition impairment present from birth.
* Affecting approximately 2% of the population.
* Lacks a clear link to any specific neurological condition.
* This contrasts with “Super-Recognizers,” individuals who possess an
extraordinary ability to recall and recognize faces - showcasing the wide
spectrum of facial recognition capabilities.
Developmental (Congenital) Prosopagnosia
Normal activation in the right fusiform face area (FFA) during fMRI scans when viewing both
faces and objects, indicating typical activation patterns. However, multivariate pattern analysis
(MVPA) revealed what?
deficit in integrating facial features into a coherent whole
what does damage to the several areas within the ventral visual cortex that are selective
for specific visual shape categories lead to?
behavioral deficits in recognition of the corresponding category
Spatially Distributed Information for Visual Categories by Haxby
The functional architecture of the object vision pathway in the human brain was investigated using functional magnetic resonance imaging to measure patterns of response in ventral temporal cortex while subjects viewed faces, cats, five categories of man-made objects, and nonsense pictures. A distinct pattern of response was found for each stimulus category. The distinctiveness of the response to a given category was not due simply to the regions that responded maximally to that category, because the category being viewed also could be identified on the basis of the pattern of response when those regions were excluded from the analysis. Patterns of response that discriminated among all categories were found even within cortical regions that responded maximally to only one category. These results indicate that the representations of faces and objects in ventral temporal cortex are widely distributed and overlapping.
Suggests the brain stores multiple views of
an object, enabling recognition by
matching the current view with these
stored exemplars
View-Dependent, Exemplar-Specific Model of Object Recognition
View-Dependent, Exemplar-Specific Model of Object Recognition explain?
why we recognize
objects more quickly and accurately when
viewed from familiar or typical
viewpoints, based on prior experience.
what does the View-Dependent, Exemplar-Specific Model of Object Recognition account for?
for the difficulty in recognizing objects from uncommon or unusual perspectives, as these may not align with stored exemplars, leading to
slower or less accurate identification.
Proposes that the brain forms a viewpoint-
independent representation of objects by
experiencing them from different
perspectives.
View-Invariant, Categorical Model of Object Recognition
what does the View-Invariant, Categorical Model of Object Recognition excel in explaining?
how we can recognize objects from novel viewpoints by abstracting the essential features that define the object’s category, regardless of its orientation.
what idea does the View-Invariant, Categorical Model of Object Recognition support?
the idea that our ability to
generalize from known to unknown views is
key to our adaptability in dynamic
environments, improving our efficiency in
object recognition.
Objects are recognized by decomposing
them into basic geometric components
called geons
View-Invariant Object Recognition
How does the visual system segments objects? How?
into geons and their spatial relationships,
using nonaccidental properties (e.g.,
curvature, parallelism, symmetry) to ensure
stable recognition across viewpoints.
What is the use of geons?
They have distinct, viewpoint-invariant
properties that allow recognition even
under changes in perspective, occlusion,
or orientation, ensuring robust object
identification.
What does behavioral evidence suggest?
suggests viewpoint-dependent processing in
object recognition, even with
viewpoint-invariant geons
What do studies show in regards of reaction times for matching geons?
reaction times for
matching geons increase with angle of
rotation (remember back to week 3),
indicating the visual system does not
treat all orientations equally
What does the increase in reaction time for matching geons imply?
a blend of viewpoint-dependent and
viewpoint-invariant processes,
challenging the idea that geon-based
recognition is completely unaffected by
changes in perspective.
Some neurons in ventral visual
pathway are tuned for….?
for particular
views of faces or hands
neurons in ventral visual
pathway parallels to orientation tuning in _________.
V1
Neurons along the ventral visual pathway appear to
increase in ___________
tolerance
In the gnostic cell or grandmother cell
theory, what do neurons activate strongly for?
the stimulus they are tuned to and weakly, if at all, for others
what does the gnostic cell or grandmother cell
theory suggest?
a highly localized representation of complex external realities within the brain.
What is evidence for the gnostic cell or grandmother cell
theory?
the halle berry neuron; These neurons are found in the hippocampus, where damage leads to memory encoding deficits,
not object or visual agnosia
what does the halle berry neuron evidence support?
sparse coding
what is a challenge to the gnostic cell or grandmother cell
theory?
theory may be oversimplified, as complex stimuli are likely represented by distributed activity patterns across neural networks, rather than by single, specialized cells.
when the brain actively constructs interpretations of sensory input based on prior experiences, What does the brain make inferences about?
about incoming information to create the most probable understanding of the environment
what influences our perception?
our prior knowledge
Examples of Contextual Processing in Brain
- partially occluded images: response pattern to cat (intact vs. scrambled) is similar even when occluded.
- context of scene informs where missing image ought to be (Regions that respond to visual presentation of face
respond to images that indicate a face ought to be
there (even when absent))
acts as an early integrator of visual information, particularly focusing on the “gist” or overall shape of an object, allowing for rapid initial recognition based on low spatial frequencies
Orbitofrontal Cortex (OFC)
Lesion studies show that _________ visual pathway is critical for object recognition
ventral
Neuroimaging (and electrophysiology) has revealed an area in the __________ visual cortex that responds to object shape
ventral
