3. Sensation, Perception - Gestalt Principles Flashcards
(34 cards)
sensation
- neural activation, signal transduction, and basic sensory coding of info acquired from the world
- e.g. psychophysics: mathematical models use to understand how sensory phenomena (not love, rating music)
perception
use of sensory info to produce an understanding of the stimuli (e.g. identification, recognition); how do we make sense of the world?
- perceptual processing: typically requires matching an external stimulus to an internal (mental) representation
- interpretation of the sensory info is required
bottom-up processing
the perceptual system may rely on the physical input
- this is our default when we have no prior knowledge about the stimuli
- i.e. I have no experience or theory with a given stimulus
bottom-up or data-driven
- principal source of info arises from accumulated evidence that is sensory in nature
- this info is often defined as being external to the organisms and without attention
perceptual segmentation
- what is required to differentiate one object from another or a part from a whole
- perceptually continuous
- our perceptual system tend to switch back and forth - which one is more probable?
- e.g. Pop out effect - slide 8 of powerpoint
issues: identifying a stimulus
- as stimulus = exposed to an individual, they begin to accumulate information
- the more info that is accumulated, the more detailed the identification process can be
- the detail requires info in LTM
- e.g. when I see a cat, I’ve already started to categorize
1. cat
2. tabby cat
3. tabby cat named Tommy
perceptual constancy
- how does the perceptual system accomodate variation in physical appearance of an object allowing us to see it as the same thing
- tendency for objects within the env. to be perceived of as having a set of constant attributes (brightness, size, shapes)
- humans have binocular vision - pick up more depth cues
- we assume things have same set of attributes
object recognition (F example)
- how do people recognize objects in spite of all the variability?
- refer to page 14 of notebook for F example
perceptual representations and pattern recognition types
- templates: fixed pattern
- protoypes: mental average
- features: individual components
each type has its issues
template representation
- input is compared with memorized exempllars until a good fit is found
- search your memory until you get the appropriate match
- we would have a template for every colour shape font etc
-what happens when we encounter a new font?
template representation: issues
- parsimony
- we would need an infinite amount of templates to accommodate easy variant - flexibility
- has no means to account for the effect of context - similarity
- does not account for similarity between items
prototype representation
- stimulus info is compared to an “average” or idealized representation of stimulus info
- similar to a heuristic
- arbitrary labels: if we say “pizza”, it’s easier to visualise than “love” (abstract category)
- the triangle: refer to page 14 of notebook
- central tendency
prototype representation: central tendency
- avergae position of dots in all of the distorted figures is “roughly” equal to the prototype
- if you do not have sufficient nb of exemplars and have high distortion, the central tendency may be quite different from the prototype
prototype: the Classification Task (Posner et al., 1967)
- some studies found that participants could successfully categorize prototypes but not new distortions
prototype: recognition memory
- presented old and new patterns and participants had to indicate whether stimuli were old or new
- prototypes recognized as old
protoype: converging evidence from classification task and recognition memory
- participants categorized protoypes even though they had never been presented to them before and belive they were old
- why would you mistake a novel pattern for a new pattern
- representations that you refer to in our head CANNOT be an “old pattern” (you never saw the prototype), but MUST be something similar to the stimuli
prototypes: schematic faces
- refer to page 15 of notebook for example and more information
prototype benefits
- retaining a prototype allows us to store fewer representations (e.g. cognitive economy)
- provides a reasonable account of how people classify similar objects, events, etc, into a common category
- note: it does not suggest how prototypes are formed or how they are used to compare with a stimulus pattern
- we keep summary representations to make out the world
feature-based representation
- although we ofetn perceive objects in our env. as a whole, these objects can be decomposed into elementary components
- features in object have their own internal representation (e.g. letter T consists of two features: I and __
- we’ve created a mental template, yet we don’t realize we’ve done it due to mental processing
- for methods and results, refer to slides 33 and 34 of powerpoint
bottom-up: pandemonium
- data-driven reocgnition model based on feature analysis
- objects are recognized from an analysis of their components
- composed of 4 types of recognition units (demons)
pandemonium: level 1
IMAGE DEMONS
- registers the initial image of the external signal
- encodes objective stimulus and activates feature demons
FEATURE DEMONS
- look for particular features in the pattern
- presence of certain line, presence of an angle
COGNITIVE DEMONS
- they watch the response of the feature demons
- seek out one particular pattern
- respond when multiple features are represented at the cognitive level
DECISION DEMON
- listens to the pandemonium created by the demons
- answer = determined by the demon that is yelling the loudest
- the one with the most activation = the one being listened to
pandemonium evidence
- can account for a nb of common features of our perceptual processes
- with a finite set of feature detectors, pandemonium can recognize potentially infinite nb of objects
- will recognize letters in spite of changes in size, orientation and other distortions
- people’s performance in identification studies show a pattern of errors consistent with confusion based on shared features (e.g. O vs C, G vs Q)
pandemonium issues
- we also use contextual cues to determine what we are seeing
- top down processes
- those which are influenced by the context and higher level knowledge
- info in LTM helps us organize new info - how did certain elements get there in the first place
feature-based representations: support
- economical to store features in memory
- experimental and neurobiological evidence = consistent
- allows to decode
feature-based representations: criticisms
- does not apply to a wide range of stimuli
- does not account for specific feature arrangements
