# Capacity vs. bottleneck theories

## Presentation on theme: "Capacity vs. bottleneck theories"— Presentation transcript:

Capacity vs. bottleneck theories
Capacity theory: minds have limited amount of mental fuel; different tasks share the amount of mental fuel available Can do two tasks in parallel, if enough mental fuel available Bottleneck theory: point in information processing where only one piece of information processed at a time Serial processing  only one thing done at a time

Have people do two very simple tasks at the same time (2 choice RT tasks) E.g., task 1 = hear a tone, press a key to signify HI or LOW pitch tone E.g., task 2 = see a letter, press a key (with other hand) to signify “A” or “B” Can you do them at the same time?

Can’t do them at the same time
Measure RTs: RT to one of the tasks gets slower and slower the more the two tasks overlap Psychological Refractory Period (Welford, 1967)  Doing one task after the other RT2 SOA

More on capacity theory
Sometimes, when you try to do more than one thing at a time, you exceed your mental fuel (capacity), and still do tasks just slower and less accurately The more capacity given to task, the faster and more accurate your performance will be

Capacity interpretation
Maybe people slow down on the second task (RT2) because they give less capacity to that task and more to the first task (Task 1)

Pashler (1996) [capacity theory developed Kahneman, 1973]
Have people do two simple tasks Two tasks always happen at exactly the same time (SOA = 0 ms) Measure – time between responding to one task and responding to the other (Inter-Response Interval or IRI)

predictions Bottleneck theory: IRI A TASK 1 Resp “Hi” TASK 2 Resp

Predictions (cont.) Capacity theory: are doing both tasks at the same time, just giving more energy to one or the other A TASK 1 Resp TASK 2 “Hi” Resp

More on capacity prediction
Will be variability in the IRIs because people will devote varying amounts of energy to task 1 and task 2 each time they do it.

Results Support bottleneck because there are no IRIs = 0 % of cases
% of cases IRI Support bottleneck because there are no IRIs = 0

Automaticity Task repeated enough times where it apparently no longer requires attention Driving is a good example Some tasks can become automatic and others can’t

Def. of automaticity Memory for task is not related to whether you’re trying to remember it More practice doesn’t help; hard to change how you do the task Can do this task and another task at the same time (no capacity nor bottleneck) Hasher & Zacks (1979)

example Reed text, pp. 70-1 LaBerge & Samuels (1974)
People did simple task with regular letters or strange new letters At first, people not very good working with new letters With enough practice, people are as good with new letters as with regular letters

Selective attention Def.: pay attention to one thing and ignore something else Cherry (1950s) created task to measure people’s ability to do selective attention “Shadowing task” = to repeat a message out loud as you hear the message (to shadow)

Selection appears fairly complete
Hear 2 messages simultaneously (one read to each ear); told to repeat one and ignore the other Results  people CAN do it; can shadow one message and ignore the other Surprise test of what is remembered from the other, ignored, message: none of the content, or what language, but could tell it is a language, and did know you heard something and gender of speaker

When does the selection happen?
Early processes are sensation, perception, etc. Late processes start with memory, thinking, problem-solving, etc. Question: Early or late selection?

Filter theory Broadbent (1959) A bottleneck theory
We “filter” out one message based on its early characteristics (sensation and perception) and let the other message through

Moray (1959) Cocktail party effect: trying to pay attention to your conversation while ignoring conversations around you Used shadowing technique: hearing one message and ignoring another Trick: secretly place the S’s name within the ignored message

predictions According to early selection, no one should notice their name in the ignored message (because you’re not processing the meanings of the words) According to late selection, people will hear their names (because you ARE processing the meanings of the words)

Results People DO notice their names, supporting late selection theory