3 - Recognition Flashcards
Define bottom-up processing
Processes that are directly shaped by the stimulus
(Not expectations)
Is bottom up only processing realistic?
No we do most likely have some way to understand the world beyond just visual input
Describe apperceptive agnosia
Impaired early vision (can’t organize basic features)
- results in difficulties of processing stimuli at different stages
- could not copy what they were seeing, somewhere in visual system they could not put the features together to see a complete picture
- doesn’t mean they don’t know what the object is because they are able to draw from memory
**Problem is translation of visual input
Describe associative agnosia
Impaired late vision (can see object but not recognize/name it)
- no issues perceiving the visual input
- applying a label to object they are seeing is where the issue is
- ex. If you give someone a glove they would name the features but couldn’t see it as a glove (man seeing wife as a hat)
Perception is constructive. What does this mean
Perception is affected by our assumptions about the context
- visual system breaks down our world into features + perceptive fields and puts it all together
- the difference in perception is the reconstruction of this image in your brain
(Ex. Assumption of lighting conditions in controversial dress, top down influence)
Define top down influences
knowledge and expectations that influence and enhance our interpretation of sensory input
What two things play an important role in shaping perception?
Context and experience
(Lifetime experience can even influence colour perception! Lighting assumptions making illusions, interpreting blurry object as hair dryer based on bathroom context)
When do top down influences have more of an influence on perception?
When the brain doesn’t really know what’s going on (ex. When stuff is blurry)
What are visual search tasks?
Tasks in which participants examine a display and judge whether a particular target is present
* Efficient in examples where you spot lines with only 1 difference vs lines with 2 features
Type of processing guides visual search tasks?
Top down
- looking for a target that has different feature(s)
What are tachistoscopic presentation experiments?
Present an image on the wall super quickly and ask participants about the stimulus
- replace visual input with random letters after showing word so it limits processing time on the word
- later ask questions about the words they saw & if they recognize them
What categories do we sort words in for word recognition?
High frequency/low frequency words
- unprimed/primed within high/low frequency
How is recognition different between high frequency words and low frequency words
High frequency words come up a lot in articles and are better recognized and low frequency words
- suggests there is a role of experience
What effect did primed versus unprimed words have on recognition? What is this called?
A little presentation gave you an advantage for recognizing words
- get an advantage of recognizing words after little exposure even if they are low frequency words
= Repetition priming!
What is the word superiority effect?
It’s easier to perceive and recognize words in the context of a word rather than isolation
- people are better at recognizing letters within words
What task is used to look at the word superiority effect?
Two alternative forced choice (2AFC) where participants are shown a word (EG. dark) or a single letter (eg. E) and asked if it contained a certain letter (e or k?)
- even then we are able to finish the word correctly doesn’t even have to be an actual word, just words that sound like words
Define well-formedness (word recognition)
How closely a letter sequence conforms to the typical patterns of spelling in the language
- ex. HZYW vs FIKE vs HIKE
- CAN ALSO INFLUENCE ERRORS (dpum read as drum bc brain wants it to be real) *generally this helps us though
How can feature nets be used to recognize words?
- detect features of letters, combined to form the features themselves
- each detector has starting activation level (it’s just affected by recency and frequency of its firing)
- input increases activation level
- detector fires when response threshold reached
Describe the order of components in a feature net
Feature detectors>letter detectors>bigram detectors>word detector
How can feature nets explain well-formedness?
May have a representation of pairs of letters that typically occurring succession
What can feature detectors detect
Very basic features of letters (ex. curved edges that make up the letter C)
If several letters get activated we get _
A bigram
Great combination of biograms get activated, it will activate _
The word that we read/word brain thinks you’ve read
Each detector has its own baseline activation level. The activation level of each of these detectors is determined by which two things?
If the detector was recently used or activated (recency/warm up effect)
OR
It was used a lot in the past (in lifetime, exercise effect)
*Both of these lead to detectors being more readily activated, reach some threshold, and send signal to next stage
How to feature nets help us?
- allows us to understand our perception of the world
- allows us to interpret that or create a model of how that happened in terms of the stimulus itself (bottom up process, things progressively getting activated and driven by stimulus)
- don’t need to create special rules about language or objects to make this make sense
- helps think about top-down effects
How do future nets explain top-down effects?
- recency/warm up effect or exercise effect
- you experience have heightened activation level
- can have readiness of a certain word to get activated if you have something on your mind (ex. Run in late and word time is primed)
Are feature detectors neurons?
No
- a word is more likely to be able to network of neurons as opposed to individual neuron
- have some similarities (baseline activity level, fires and get action potential at threshold) but don’t necessarily exist in the brain
What are two important aspects of feature nets that we need to consider?
- knowledge is distributed: it is not locally represented in one detector (the detector doesn’t know it’s firing)
- perfect accuracy is sacrificed for efficiency (as processes like reading get more efficient it comes with the cost of accuracy because we trust by grams and receptors to recognize what we know)
More advanced and capable feature net models have what kind of connections?
Bottom up, top down, same level/lateral connections
*Both have excitatory and inhibitory connections
List the categories in feature nets for objects
Feature detectors>geon assemblies>geon detectors>object model
What are object feature detectors for?
Feature detectors (basic visual stimuli, edges, corners, textures, lines)
Who came up with the idea of geons? What was the idea behind them?
Irving Biederman
- much like we have basic building blocks for language we can have building blocks for the entire visual world
- came up with ~30
- can use it to come up with basic configuration of every object you could think of
What do geon detectors do?
Identify geons in the objects we see
The optic model should be viewpoint invariant, what does this mean?
- should be able to recognize object from any angle
- mental representation that involves these genons is complete (can fill in the blank if some parts of the geon are obstructed)
What is the thatcherization effect
Individual features of mouth and eyes and flip them, always tell what’s wrong when it’s upside down
What is pareidolia
- tendency to see faces everywhere in visual world
- also describe emotional/humanistic characteristics of the face
What happens when participants are shown composite faces
- if the top and bottom of face image belong to different people, and you tell people to ignore the bottom, they can’t
- BUT separating them/misaligning them breaks the illusion
What are the two extreme ends of face perception abilities?
Prosopagnosia and super-recognizers
Describe prosopagnosia
Have a good object recognition and good memory but they are really bad at recognizing faces
- will recognize that it is a face, but thinks the face looks generic and can’t identify whose it is
Who are super recognizers
Have really good face recognition, could see you once and then recognize you later
How do we know there is something special about faces + face perception?
- if we measure people’s recognition ability with upright faces we make one error on average versus five times as many errors with upside down faces
- only ever see houses upright so we don’t make any errors, how is this represent all non-face objects and they have a much smaller inversion effect
Special processing of faces may be due to _
Having a huge experience with faces from the moment you’re born
- quickly become the most important stimulus
**Expert bird watchers have the same inversion effect!
What is still a limit of object recognition?
Can’t explain super broad top down influencing well
- if you’re primed about the space time continuum and shown the word clock, you’ll be able to identify it quickly
If you saw places instead of faces you’d have which part of your brain activated
Parahippocampal place area (very important to navigate space, creating maps
What do I know about imagery and object perception
- areas used for visual processing are also used during a visualization imagery
- imagery and perception can rely on some of the same neural architecture, if you disrupt brain activity (tms or trauma) this disrupts perception and imagery
What relationship do we see between vividness of imagery and brain signaling?
- the vividness of your imagery is highly correlated with the change in your brain signal
- the more vivid your imagery is the bigger the change in the brain signal (recorded with fMRI)
What are the two types of imagery?
Visual and spatial
Describe visual imagery
Actually seeing the thing in your mind’s eye, ex. Being able to picture your favorite car and interacting with it
Describe spatial imagery
Involves moving through space (is it the same object if you rotate it)
What traits are correlated with visual or spatial imagery?
Stronger visual imagery: more successful in art fields, reliving memories, emotion processing
Stronger spatial imagery = more successful in science field
