change blindness/deafness tutorial Flashcards
what is change blindness?
failure to detect changes to visual details of objects and scenes
what is the most common explanation of change-blindness?
fail to detect changes because the changed display masks or overwrites the initial display
how can you induce change blindness in the natural scene?
- simulate saccades by disrupting retinal transient normally accompanying a change
- e.g inserting a blank screen between original and changed image, separating images by a ‘mudsplash’/cut/pan
what is the word recognition view of change deafness?
- focus is on words and meaning, voice doesn’t add to the message
- so changing speaker has no effect on ability to encode and recall a message
- suggests semantic processing is independent of talker identity
what is the episodic view of change deafness?
- non-lexical info also automatically encoded and maintained in LTM
- listener has to be sensitive to change in speakers dialect and vocal characteristics to understand the message
- listeners should be sensitive to change in speakers voice to maintain perceptual phonological processing
what studies did Fenn et al. (2011) carry out on change deafness?
- phone call- person on other end changed during conversation
- aim was to investigate whether talker changes would be detected when listeners are actively engaged in normal conversation
what is the standard model of speech perception?
- speech signals are complex and vary in many dimensions (emotional info, referential info, source info)
- acoustic details are lost as a result of categorical perception
- word recognition proceeds from phonetic patterns/structure without consideration of source
- speaker-specific info stripped away during word recognition
how does conversation affect change deafness?
- when listening: are interpreting syntactic, lexical info and arrive at pragmatic interpretation and have to formulate reply
- assume messenger changes but talker does not -> change deafness higher
- constraints on timing of response, but flexible to allow listeners to delay response
what did Fenn’s experiment 1 find?
whether detected change in speaker AND memory for the two voices -> few noticed change, could distinguish between voices they’d heard/not but did not reliably recall the most recent voice
listeners may retain some sort of general memory for a voice (auditory gist) or only remember certain aspects - e.g reliably decide between young/old, male/female
what did Fenn conclude from the 5 experiments?
- when engaged in conversation, people prioritize message over speaker identity
- unless acoustic different exists or attention is directed toward voice features, speaker changes go unnoticed
- this challenges theories that voice identity is automatically encoded and supports that expectations shape auditory perception
- but p’s could be performing by chance
what did Fenn’s experiment 2 find?
- test memory for a voice that was encountered in the conversation compared to a voice that had not been encountered during the conversation
- if retains memory of both and memory is not sufficient for change detection → should be able to discriminate
- but if no memory (perform by chance)- should not discriminate
- many correctly identified heard voice, retained some voice info from both but still didn’t detect switch
- people encode some memory of a speaker’s voice but do not compare voices in real-time to detect changes
what did Fenn’s experiment 3 find?
- if voices too similar, detecting change while carrying out convo may be too demanding
- if participants assume talking to one person, may ignore voice properties that are inconsistent with this belief or interpret voice characteristics in a way that confirms their expectations
- so alerted participants to the possibility of change may facilitate detection
- increase in detection, conversation did not impose cognitive load
- detection failure in previous experiments was likely due to lack of expectation rather than a fundamental inability to perceive talker changes
what did Fenn’s experiment 4 find?
- all p’s told there would be a change in speaker, for half there was and for the other half there wasn’t
- all in actual change condition detected change, 20% p’s in same talker condition falsely believed a switch occurred
- people may interpret acoustic–phonetic differences according to their expectations
- small differences that usually would be considered within-talker when p’s assumed there was no change became indicating of a different talker
what did Fenn’s experiment 5 find?
- investigate what sort of changes people spontaneously notice (if not change in talker)
- changed gender of speaker
- many p’s detected switch (accepted as part of experiment)
- if encode gist of voice memory- gender changes would be large enough deviation (salient) from original gist to enable detection
- possible that the two voices does not require comparison as gender is very central feature
what are the 5 causes of change blindness?
- overwriting
- first impressions
- nothing is stored
- everything is stored but nothing is compared
- feature combination
which models are suited to different types of change blindness?
- overwriting: changes to simple visual stimuli
- first-impressions: complex, sematically-codable stimuli
- also depends on whether observers perform repeated intentional detection trials or single-trial incidental task
what is one of the main findings from the discovery of change blindness?
- memory for visual or verbal details is fallible
- we do not retain a representation of all the visual details of our world from one fixation onto the next
- shows that under precisely controlled timing and response conditions, observers sometimes fail to detect stimuli that are otherwise clearly visible
what are the two most used paradigms to study changed detection
FLICKER: original and modified image are presented in rapid alternation with a blank screen between them- respond as soon as they detect the changing object
- changes to objects in the “centre of interest” of a scene are detected more readily than peripheral or “marginal interest” changes
- suggested attention focused on central objects more rapidly OR more often
- therefore allowing faster change detection
FORCED CHOICE: observers only receive one view of each scene before responding, so the total duration of exposure to the initial scene can be controlled more precisely
- only a sub-set of the images have changes- accuracy and latency can be DVs
- both are intentional coding- observers know changes will occur and actively try to find them
- indicates change blind even when searching for change
what other type of coding approach can you use?
incidental coding- view display without knowing it might change
- also use motion-picture/real-world methodologies → insight into spontaneous representations formed under natural view conditions
- also blind to marginal interest changes
how do we detect change in visual scenes?
- change blindness for marginal interest + changes for central detected more readily = attention necessary for change detection
- details of object only retained if attention is focused on changing feature
- observers must scan and encode scene in pieces
- in order to retain info about an object from one view to next, must recode and compare abstracted representation of inital object to changed
- objects not recoded = not remembered in detail
- more important in centre = more likely to focus attention = more likely to notice changes
