3 – Attention 3 – Automatic Processes

Question 1

Describe the experiment in Shiffrin and Schneider’s classic 1977 study of automatic processing.

Answer 1

Subjects memorise 1-4 target letters/numbers (memory set), search a display containing 1-4 letters/numbers (search set).
Two conditions
1.consistent mapping – consonants always targets; numbers always distractors. This condition likely to lead to automaticity.
2. Variable mapping – both consonants and numbers can be targets or distractors. This condition not likely to lead to automaticity.

Question 2

What’s the difference between pre-attentive processes and automatic processes?

Answer 2

Automatic processes have to be learnt (e.g. decide letter A is important and train ourselves to see it quickly; walking). Pre-attentive processes are hardwired (e.g. pop-out of blue circle among red circles).

Question 3

What are 7 proposed features of automatic processes?

Answer 3

1. fast
2. parallel –can do more than one without overloading
3. do not draw on central capacity
4. unavailable to consciousness
5. unavoidable
6. effortless
7. ‘insulated’

Question 4

What were the results of Shiffrin and Schneider’s 1977 study?

Answer 4

1. Flat(ish) search slopes for consistent mapping, irrespective of number of distractors or memory set size. Suggests automaticity.
2. Even after 2100 trials, variable mapping was still slower when more things to search through or remember.

Question 5

What was the immediate criticism of Shiffrin and Schneider (to do with distractor shape)?

Answer 5

All distractors are numbers when consonants are targets. Most consonants are pointy, most numbers are round. The stimuli break down into different perceptual categories, creating a confound.

Question 6

When set size increases, the RT slopes in parallel search studies are never COMPLETELY flat. What does this imply?

Answer 6

It suggests that the process might not be categorically different from serial search, just much faster.

Question 7

How does a search switch from being serial to automatic during training, according to Shiffrin and Schneider? And what’s wrong with this explanation?

Answer 7

Sometime during practice, a task that was previously effortful and serial can suddenly be performed in parallel, with no draw on attentional resources. But when does this happen? If it’s gradual, then there’s no categorical distinction. Not a satisfying explanation theoretically.

Question 8

What explanation does Cheng (1985) propose for how people acquire skills?

Answer 8

Processes are restructured with practise. Practise doesn’t just make people faster – or make them do things in parallel –they also do things differently. E.g. chessmasters don’t have spectacularly fast brains that think through every combination – they retrieve games from memory. The skill has changed. A person develops heuristics, and they know what to look for. As skill progresses, you perform task differently.

Cognitive arithmetic –5 + 7 counting on fingers as a kid. As adult you know = 12. Are you counting 6789101112 really quickly? No you’re doing it differently.

Shifrin and Schneider’s simple tasks may not be able to model how experts perform skills.

Question 9

What does Logan (1988) identify with automaticity?

Answer 9

Automaticity is just memory retrieval.

Question 10

What is Logan’s ‘instance theory’ of automaticity?

Answer 10

Every time you respond to a stimulus, a memory trace is laid down. Repeated exposure means more memory traces (not one that is stronger). More memory traces means faster retrieval, because, statistically, one of these is going to be retrieved faster. This explains novice AND expert behaviour.

Question 11

In what way does Logan’s ‘instance theory’ of automaticity predict not loss of voluntary control, but just faster and smoother behaviour?

Answer 11

Memory can be chunked, and when you’re retrieving a sequence of responses, each of those memory cues can be strung together. E.g. learning how to turn corner on a car –skill begins as memory retrieval of several discrete hand movements, then when skilled it becomes one instance of memory retrieval.

Question 12

If automaticity is just memory retrieval, in what way are automatic processes then NOT unavoidable?

Answer 12

Because you can retrieve a memory based on a certain cue (e.g. seeing word ‘blue’ in red type in Stroop Test and wanting to say red), but not act on it.

Question 13

What evidence is there that ‘automatic’ processes are less controlled?

Answer 13

Not much. On the contrary, people who are skilled are more in control.
E.g.
A novice typist will finish the word when you say STOP. Skilled typist will stop on a letter.

Bilingual people are less likely to suffer from stroop effect in language they are dominant in.

Question 14

To what does Logan attribute the lack of awareness of ‘automatic tasks’?

Answer 14

Lack of awareness of a task may stem from an inability to verbalise it. These processes –such as driving –aren’t coded linguistically. And they may not be encoded in memory in a way that can be accessed verbally, assuming they are encoded at all.

Question 15

How does a memory-retrieval approach, such as that of Logan (1988) jibe with the idea of parallel processes?

Answer 15

Parallel processes don’t make sense with a memory retrieval-type theory.

If turning corner is just ONE instruction, you can move onto other tasks easily, because process of turning corner has been chunked. Then you turn on radio. Are you doing these things in parallel? No, just minute, really quick instructions.

This explains why search slopes are not completely flat in ‘parallel search’ studies. They draw on central capacity to a small extent.

Question 16

Are ‘automatic processes’, such as washing your hair, insulated?

Answer 16

No, many expert behaviours can be broken into and stopped. Hair-washing will be interrupted if there is an earthquake.

Question 17

Do ‘automatic processes’ draw on central capacity?

Answer 17

They must draw on central capacity to a small extent – if search slopes are not completely flat.

Question 18

What do Moors and De Houwers suggest is the difference between automatic and non-automatic process?

Answer 18

They suggest there is a continuum – that processes can be more or less automatic.

Question 19

How might Logan’s instance theory of automaticity repudiate claims that processes could lie on a spectrum of automaticity?

Answer 19

Processes are not more or less automatic, they are just faster or slower depending on how well the memory retrieval is chunked. If there’s lack of consciousness it’s because you only had to spend a MINIMAL amount of time retrieving a memory to perform that task – and that’s why it slips your memory. E.g. hair-washing.

Question 20

Has a clear dichotomy between automatic and conscious processes ever been established?

Answer 20

Nope.

Question 21

What is Logan and Crump’s 2009 double loop theory?

Answer 21

Within a specific task, there are parts of the task (outer loop) that are available to consciousness –that we are aware of. There are parts of the task (inner loop) where we are not aware what’s going on – because it’s a physical skill. There are no codes at that level (e.g. swallowing, breathing etc.).

Question 22

In what way does typing exemplify Logan and Crump’s 2009 double loop theory?

Answer 22

– An ‘outer loop’ selects words (from thoughts or vision)

– An ‘inner loop’ turns those words into finger movements

Question 23

What happened when Logan and Crump (2009) asked skilled typists to only type the ‘right hand letters’ from each word? Why was this disruptive?

Answer 23

This required attention to be shifted from the outer to the inner loop. And the outer loop clearly doesn’t know what the inner loop is doing.

Question 24

In what way is the effect of attention different on inner and outer loops?

Answer 24

Attention to the outer loop is helpful. Attention to the inner loop disrupts transfer of information.

Question 25

What happened when Logan and Crump (2009) inserted errors into the output?

Answer 25

If errors inserted into the output, even though words were typed perfectly, people reported they had made the errors themselves. But feedback from the hands –RT, flow of typing –suggested that the inner loop was not bothered. The inner loop knew it didn’t make a mistake.

And vice versa –when people made a mistake, Logan had software CORRECT the mistake. In that case, the outer loop didn’t hesitate, but the hands hesitated on the next keystroke.

Question 26

What generally happens when attention is divided between two tasks (dual task performance)? And what does this suggest?

Answer 26

Poorer performance on dual tasks suggests both tasks draw from a common pool of resources.

Question 27

When is there LESS interference in dual task performance?

Answer 27

When tasks use different stimulus and response modalities (e.g. driving and talking vs talking and listening; Wickens’ multiple resources, 1984)

Question 28

As processes get more ______, they draw on ______ resources and interfere with each other ______.

Answer 28

As processes get more automatic, they draw on fewer resources and interfere with each other less.

Question 29

Describe 4 types of theory of dual task performance?

Answer 29

1. Capacity theories: Graded sharing of a single pool (Kahneman, 1973) or multiple resources –if you’re doing tasks that draw on different resources, then no interference (Wickens, 1984). These assume processing is parallel, and that allocation of attention can be varied at will.
2. Bottleneck theories (aka single channel theory): queuing, and rapid shifting between tasks because of serial processing (Craik, 1947)
3. Crosstalk and task similarity: Limitations depend on the similarity of tasks and the information being processed. Focus on inputs and outputs. e.g. Can’t listen to two radio shows, but can listen to radio and watch TV.
   (e. g. “Outcome conflict” – Navon & Miller, 1987)
4. Neural theories: tasks can be performed together if the structures performing them are neurally distant (Kinsbourne, 1981)

Question 30

What is Pashler’s Psychological Refractory Period (PRP) paradigm of task performance (1998)?

Answer 30

It’s a bottleneck theory that states that two tasks cannot draw on central processing simultaneously.
Central processing has to finish its central processes for task 1 and then start task 2. That means the time it takes you to do task 2 is slowed by 1ms by every extra 1ms the central processes take for task 1.

Theory explains how we seem to be doing things at the same time, and yet everything seems to require our attention.

Question 31

How can tasks overlap in the PRP Paradigm (1998)? Two ways.

Answer 31

1. When you have a long stimulus onset asynchrony (SOA) –when two tasks begin a long way from each other. Stimulus 1 is presented and Stimulus 2 is presented while you are responding to Stimulus 1. But because there’s no overlap in CENTRAL processes, there’s no interference that’s observed – almost as though data for two DIFFERENT experiments.
2. When you have short stimulus onset asynchrony (SOA), Stimulus 1 is presented, and then central processes begin for Task 1. Stimulus 2 is presented, but central processes for Task 2 have to wait until central processes for task 1 are complete. WhenResponse for Task 1 is initiated, central processes for Task 2 can begin.

Question 32

How does the PRP paradigm explain the fact that we appear to be doing everything in parallel?

Answer 32

Some things can be done in parallel, such as producing a response. But central processes take less time as we become more skilled at tasks, and when task becomes less about algorithmic calculation and more about direct retrieval of memory/solutions. So we can really string central processes together at great speed.

Question 33

How does the PRP paradigm match data from empirical studies?

Answer 33

The theory concurs with the data The increased RT for multiple processes flattens out with practice but doesn’t disappear.

Question 34

How does the PRP paradigm contradict Wickens’ idea of multiple resources?

Answer 34

A multiple resource model suggests there would be no cost to doing activities in parallel drawing on different resources. But you can still find PRP when you’re responding to things in different modalities. Task 1 might be looking at something and responding with hand. Task 2 might be listening to something and responding with foot. With practice it becomes fast and smooth. But data suggests there’s still a cost to task 2.

Question 35

In what way does the PRP paradigm tie in with with Logan’s inner/outer loop idea?

Answer 35

In both theories, some processes require central processing, others don’t.

Question 36

How does the PRP paradigm explain the data from resource theory (capacity sharing) that people can divide attention 50/50 between two tasks?

Answer 36

It’s possible to switch rapidly and equally between tasks. We can share resources quite neatly by switching.

Question 37

In what way is Pashler’s PRP Paradigm like the computer game Tiny Tower?

Answer 37

In Tiny Tower you move between different floors of a building setting things in motion. Each floor looks after itself in general, but needs attention occasionally. Seems as though you are managing each floor in parallel, but you are not –the operations on each floor are happening without your attention. Control is needed – these process are not fully automatic, yet they overlap because processing before and afterwards CAN overlap.