Attention Flashcards
What is attention?
- The human information processing system is limited in terms of allocation of resources to sensory and perceptual information
- The term attention s sued to refer to this allocation of processing resources
- Attention acts as a filter (Broadbent) or an attenuator (Triesman), to prevent this limited set of resources from being overloaded
- It can act as a ‘bender’ of object features and as a ‘binder’ of related information from different sensory modalities.
What is the special problem of auditory attention?
Auditory system has a special problem: it requires processes that permit a listener to attend to the specific set of sounds without being confused by the overlap of other, irrelevant noises.
Auditory system is able to separate different, superimposed sounds on the basis of their different source directions.
Uncertainty remains over the fate of the unattended material.
Ears are positioned in a way that we cant direct them to avoid certain stimuli, unlike the eyes where you can just look away.
If the way we hear so much auditory stimuli would be like seeing loads of sentences superimposed on each other if you cold see it.
We have two ears which focus on long wavelengths, so we can focus on some stimuli over others.
What is the dichotic/shadowing procedure?
(Broadbent, 1952, 1954; Treisman, 1960)
- Subjects wear a set of headphones and made to hear two messages at the same time, one entering each ear, and asked to shadow (repeat back the words from one message only)
- A typical shadowing task:
· Left ear: and then John turned rapidly toward
· Right ear: a series of words - ran,house,ox,cat
Ask to repeat back the left ear material
Shadowing procedure was the first wave of research. Subjects wear headphones, hear one in each ear and told to only repeat the words from one ear only.
Typically one is asked to repeat the prose not the random works.
What was found from the dichotic listening/shadowing procedure?
- Unattended material appears not to be processed in a shadowing tasks; only most recent unattended material is available, while still preserved in the echoic memory
- The listener is normally unable to report significant details concerning the unattended information: can only tell whether the unattended message is a human voice or a noise, or if human, whether male or female and the language used by the voice
- These results suggest parallel acquisition of all available information, followed by serial processing to determine meaning for one attended message
Found people couldn’t really repeat the unattended words except possibly the last 3 words. Beyond that it would fade away incredibly quickly, only in very short term memory.
However, were able to register sensory aspects of it, such as pitch, language, male or female voice etc. Just cant get semantic value from it.
What is broad bent’s 1958 filter theory?
The basic claims of the model:
• Sensory channels have an unlimited capacity
• There is a bottleneck allowing only one piece of information into working memory at a time
• A selective filter allows in information from only one channel at a time based on the physical characteristics of the message (ear, pitch, etc)
• Information from unattended channel is completely blocked
Take information in via sensory organs and then there’s a gate at which point we decide which message we are going to receive. After that choice is made only one stream of information will go through. Then goes onto short term memory.Biggest mistake is that the filter could only be done on superficial sensory aspects (Eg pitch)
What are the problems with the bottleneck theory?
- Although there is little conscious awareness of unattended material, it may receive more processing than the above results imply
- Sometimes selected messages are processed on the basis of their semantic content rather than their physical characteristics (eg cocktail party phenomenon)
- Words presented to the unattended ear can produce priming and physiological effects (eg ‘ignored’ shock words give rise to galvanic skin responses; Corteen and Wood, 1972)
- Trying to ‘shadow’ one ear will follow the message to the other ear (Treisman, 1960)
Wasn’t true. Found that although there wasn’t much awareness of the unattenuated information, it was actually processed at a further level than initially thought.
Eg cocktail party phenomenon - shows it must be in some way
Can measure physiological aspects eg simple word association task in one ear and critical words every 5 or 6 words are followed by an electric shock and this is repeated a lot of times. After that dichotic listening task where have to shadow one ear, but in the other year every time the signal word is said their skin responds and this suggests there was some processing to the stimuli in the first part.
What is evidence for parallel processing (Treisman)?
• Treisman instructed subjects to shadow a particular ear into which was played a meaningful message
• The nonshadowed ear received a random string of words
• At some point in delivery, the meaningful message switched ears and the random words were switched into the nonshadowed ear
• left ear: In the picnic basket she had peanut butter book, leaf, roof, sample, always
• right ear: cat, large, day, apple, friend, every, select, sandwiches and chocolate brownies
• Although instructed to shadow a certain ear, many ignored this and followed the meaningful message instead (temporary sensitisation?)
Results imply that processing takes place in parallel, to the extent that meaning is extracted even from unattended material
Start sentence in one ear and then switch to the other and people will follow it seamlessly. Treisman said this is because you are primed to hear the next words in the sentence, as stimuli around us prime us. When you hear your name you have a very low threshold to recognise your name, very few processing resources needed for it as it is naturally primed. Similar with the sentences.
Some semantic value in the other information.
What is visual attention?
- Attention can be directed selectively towards different areas of the visual field, without the need to re-focus
- Visual attention linked to specific objects rather than to general regions of space
- Unlike hearing, seeing is typically extended over space (and not time), although seeing does require some finite time to capture and analyse information – this process has been the focus of much research
Similar message but a little more complex. With visual attention we are changing the focus. In auditory it’s the info available to us but we choose to ignore it, in visual its on information that’s there are we really want to process, but might not have time to do so because of its brief duration.
We an read a sentence, take visual imagery of a split second as opposed to auditory which takes longer to understand a sentence.
What are the ways of testing iconic memory (visual)?
• Presented an array of letters for 50 ms
X M R J
C N K P
V F L B
• Whole-report method: recall as much as possible from array
• People recall 3-6 letters; report that the letters “fade away” before they can report them all
• Part-report method: only certain elements from array
• Employs a tone (high, medium, or low) to cue subjects to report a particular row (top, middle, or bottom)
• Recall a higher proportion of letters: labelled the ‘partial report superiority effect’
All material captured in parallel, some selected for further, serial processing on the basis of position or colour
How long is long enough to process something?
1/20 of a second was shown letters and asked to repeat as many as possible. Found that generally get between 3-6 letters and say saw all the other letters but they faded away while repeating the ones did remember.
Part report method - straight after the letters shown a tone is sounded. Only have to repeat one row and found people pretty good at this suggesting that have taken on the whole thing but only need to say some so can manage it.
What was the backwards masking technique?
• Backwards masking procedure (Evett & Humphreys, 1981; Pecher et al., 2002)
• the mask is presented after the target, usually appearing in the order of 10-50ms after target first appeared
• time between onset of target display and onset of mask is called the stimulus onset asynchrony (SOA)
Experiments using this technique have provided persuasive evidence (e.g., significant priming effects) that meaning can be extracted from material of which participant is unaware
Seemed too simple. Became a more complex system.
Backwards masking - generally where a stimulus is shown and then something immediately replaces it. Can be done now digitally to the point where people don’t know what the initial stimulus was or can do it so fast they don’t even know there was an initial stimulus.
Normally around 40ms after.
There is persuasive evidence that even though people cant necessarily say the initial stimulus, it has subsequently affected their perception of something else.
What evidence was found from backwards masking studies?
- Participants say they cannot see masked words but often do better than chance when forced to guess whether or not one had actually been presented (Cheesman et al, 1984)
- Evett & Humphrey (1981) used a perceptual identification task - stimulus sequences containing 2 words, both of which masked (SOA = 40ms) – when second word related to first, it was more likely to be reported accurately
- Pecher et al. (2002)
- perceptual identification study using a potential prime (lion) followed by hard-to-see masked target which was related (tiger) or unrelated (dice)
- manipulated priming word duration (short vs long) and proportion of related target words (10% vs 90%)
If people say they didn’t see it but then get a 50/50 choice of words they do better than chance.
Experiments where they use words such as lion and the second will be tiger or dice, where it is related people are much better at identifying the words. Cant report the first word but much more able to get the second word if they’ve been primed for it, even though cant recall seeing lion. Both words shown very briefly.
They looked at the priming word duration. Where you have a 1 second prime you can prepare much easier for the second word and see much higher percentages well above chance (50%). People then are good at picking up on what’s going on and see how it works and get a massive priming advantage.
One is called a strategic prime, and one is called a hope for the best prime.
The 40ms is enough to get a small priming advantage.
Some of the time people are taking the semantic value of the words without being able to consciously report it, some of the time.
What results were found be Pecher et al?
2002
At 1 sec, benefit of related prime when targets more likely to be related to preceding prime (participants spot connection)
At short duration, priming advantages are far more modest – do not produce large increase in priming effect
Participants presumably unable to guess in brief condition but did produce a small priming effect – must have received sufficient analysis to activate their meaning
Automatic priming effects are caused by spreading activation between nodes at level of representation.
What is the test of of RSVP (rapid serial visual presentation) technique?
The sequence of stimuli, shown in the same location on a computer screen, in which the participant has to identify a white letter, then decide whether an X was also present
Number of stimuli occurring rapidly one after the other. Done so guessing wont help.
How spaced to they need to be before they are frequently seen
Typical results, showing the likelihood of detecting the X, when presented in the first and subsequent positions following the white target - the attentional blink
(a) Target 2 is seen more easily when Target 1 is made easier to see by removing the following item
(b) Target 2 is also seen easily when items following it are omitted
Found that if follows straight after people are pretty good, thought that they just put them together as they came so closely.
If have a lag of 2/3/4 then get pretty bad at seeing it. 6/7 later then people have recovered from first letter and start seeing it more frequently again.
Why is the lag of 2/3/4 so bad? Found that in effect after target one, have a bit of a gap then can remove the intentional blink then people are much better. If put at the very end irrespective if straight after or not then people are much better.
What was the Giesbrecht and Di Lollo model (1998)?
Two-stage model of visual processing to account for RSVP findings
• Stage 1: a range of info about target characteristics is captured in parallel (identity, size, colour, position)
• Stage 2: serial processes act upon information preparing it for awareness and report
• While Stage 2 is engaged, later info cannot be processed so has to remain at Stage 1
• Disruption to Stage 1 (masking) increases processing difficulty, so info from T2 is kept waiting longer
• If T2 masked by following stimulus, then run risk of overwriting it
• Damaging to episodic information; semantics info may be able to survive (revealed through priming effects, EEG)
Individual differences in RSVP: personality traits important to distinguish blinkers from non-blinkers (Morrison et al., 2016)?
First trying to get target one which you capture, taking in colour and identity of the letter which takes resources. People are good at doing this. However while doing this don’t have control over resource allocation to do something else to onto stage 2. serial processes are needed, preparing it for conscious awareness, and while still doing that you cant really process other information coming later. Disruption to this stage increases the difficulty.
If target 2 is then masked you have lost the change of carrying on doing it. But if its not masked, eg comes at the end, then you can do it. It is the damage to episodic information that is bad. Lose the ability to consciously report on the event, that is what attention is, the ability to take on the semantics of what’s in front of you and being able to report on it.
What is the visual awareness and attention model?
- Inability to report detail from brief, masked visual displays is linked to need to assemble various information components
- The visual information is captured in parallel, but assembly is a serial process
- Episodic detail (e.g. colour, position) is vulnerable to the passage of time, or to ‘overwriting’ by a mask
- Semantic information (i.e. identity/meaning) is relatively enduring, but does not reach conscious awareness unless bound to the episodic information(Coltheart, 1980)
- Attention, in this context, is the process of binding the information about an item’s identity to its particular episodic characteristics
- ‘Unbound’ semantic activation can be detected by priming and electrophysiological techniques
Many inputs are unconscious. This comes at a later stage, not the same as consciousness. There is a difference between consciousness and being consciously aware (being able to report).
Blinkers and non-blinkers in the RSVP task. People who are particularly prone to blinkers is more likely to be in people who: lack of sleep, depression and autism - but not anxiety (don’t really know why though).
What is subliminal messaging?
- Registration of sensory input without ‘conscious awareness’
- Limen is another word for Threshold - so subliminal means “below the threshold”
- Controversy began in 1957 when hidden messages such as “Eat Popcorn” were placed in films
- James Vicary claimed to have flashed the words “eat popcorn” and “Drink Coca-Cola” on a movie screen for 1/200th of a second, every 5 seconds during the movie Picnic
- He claimed popcorn sales increased 58% and Coke sales 18%
- Vicary’s experiment was never successfully replicated
- He later acknowledged the study was a fraud (Advertising Age ,1962)
Even though cant consciously report something can take some semantic value from the stimulus.
Hard to accept that there are things there that we aren’t aware of but are affecting our actions.
What are the types of subliminals?
- Embedded images: pictures or words that are hidden or flashed quickly (in 100ths of a second)
- Sub-audible messages: sounds or words that are too faint to be heard, or are played at extremely high frequencies
- Electronically altered signals: backward masking and other voice alterations
Can messages lead to priming, the activation of various mental constructs unbeknownst to individuals via perception of external stimuli, which not only alters beliefs or perception, but instead reaches the domain of action?
People are very interested in very quick images being flashed up.
Backmasking different to backwards masking. Beatles first to do it. Its where you play something backwards, record that, and then put it in the album.
Question is can this lead to priming?
What is public belief in the power of subliminals?
- 75% of Americans believe that subliminal messages are omnipresent in advertising, and that they work (Rogers & Seiler, 1994)
- Why?
- Vicary’s purported movie theater experiment in 1957
- Wilson Brian Keys claims of planted images in advertising
- Claims of subliminals in Disney movies and other media
- Media spoofs: In a Simpson’s episode, Homer receives a subliminal self-help tape which increases vocabulary instead of weight loss. He begins talking like Shakespeare
People do believe in this - they believe they occur and are very powerful despite advertising companies saying they don’t bother etc.
There is a belief that musicians etc are backmasking and putting in messages.
What is evidence for the power of subliminals?
• Meta-analysis assessed behavioural impact of and psychological processes associated with presenting words - revealed a small behavioural priming effect which was robust across methodological procedures (Weingarten et al., 2016)
Priming words, put a criteria for where awareness was there the study didn’t count. Controlled studies that may not extend outside the lab, but found some effects.
Betting where on fruit machines had the same 20 spins, for those who they displayed $$$ signs very briefly they tended to bet more.
What is the Stroop Effect?
• The effect was first demonstrated by J. R. Stroop (1935), who found that people required an average of 110 seconds to name the ink color of 100 words that were incongruent color names (eg, blue ink used in writing the word red). In contrast, people required an average of only 63 seconds to name the ink color of 100 solid color squares
• Since the original experiment, more than 400 additional studies have examined variations of the Stroop effect (e.g., MacLeod, 1991; Richards et al., 1992)
Older adult find the Stroop task to be even more difficult than do younger adults (Hartley, 1993)
People have to name the ink colour of words as quickly as possible.
• The effect was first demonstrated by J. R. Stroop (1935), who found that people required an average of 110 seconds to name the ink color of 100 words that were incongruent color names (eg, blue ink used in writing the word red). In contrast, people required an average of only 63 seconds to name the ink color of 100 solid color squares
• Since the original experiment, more than 400 additional studies have examined variations of the Stroop effect (e.g., MacLeod, 1991; Richards et al., 1992)
Older adult find the Stroop task to be even more difficult than do younger adults (Hartley, 1993)
How do we explain the Stroop Effect?
- Most promising account is provided by a parallel distributed processing approach (e.g., Cohen et al., 1990): Stroop task activates two pathways at the same time. One pathway is activated by the task naming the ink color, and the other pathway is activated by the task of reading the word. Interference occurs when two competing pathways are active at the same time. As a result, task performance suffers.
- Usually find significant effect in reverse Stroop (identify word; ignore colour) although weaker; account for that with parallel processing, but with faster accumulation of evidence for word recognition
Two pathways interfering. There seems to be interference. Slight slowing.
