1 Cognitive Models CS 160, Spring 2004 February 18

2 Why Model Human Performance? 4 To test understanding 4 To predict influence of new technology

3 Model Human Processor

4 What is missing from MHP? 4 Haptic memory *For touch 4 Moving from sensory memory to WM *Attention filters stimuli & passes to WM 4 Moving from WM to LTM *Rehearsal

5 MHP Basics 4 Based on empirical data *Years of basic psychology experiments in the literature 4 Three interacting subsystems *Perceptual, motor, cognitive

6 MHP Basics 4 Sometimes serial, sometimes parallel *Serial in action & parallel in recognition +Pressing key in response to light +Driving, reading signs, & hearing at once 4 Parameters *Processors have cycle time (T) ~ ms *Memories have capacity, decay time, & type

7 Memory 4 Working memory (short term) *Small capacity (7 ± 2 “chunks”) vs. (617) DECIBMGMC vs. DEC IBM GMC *Rapid access (~ 70ms) & decay (~200 ms) +pass to LTM after a few seconds 4 Long-term memory *Huge (if not “unlimited”) *Slower access time (~100 ms) w/ little decay

8 MHP Principles of Operation 4 Recognize-Act Cycle of the CP *On each cycle contents in WM initiate actions associatively linked to them in LTM *Actions modify the contents of WM 4 Discrimination Principle *Retrieval is determined by candidates that exist in memory relative to retrieval cues *Interference by strongly activated chunks

9 The Model Human Processor Long-term Memory Working Memory Visual Image Store Auditory Image Store Perceptual Processor Cognitive Processor Motor Processor Eyes Ears Fingers, etc. sensory buffers

10 Principles of Operation (cont.) 4 Variable Cog. Processor Rate Principle *CP cycle time T c is shorter when greater effort *Induced by increased task demands/information *Decreases with practice

11 Principles of Operation (cont.) 4 Fitts’ Law *Moving hand is a series of microcorrections, each correction takes T p + T c + T m = 240 msec *Time T pos to move the hand to target size S which is distance D away is given by: T pos = a + b log 2 (D/S + 1) 4 Summary *Time to move the hand depends only on the relative precision required

12 Fitts’ Law Example 4 Which will be faster on average? Today Sunday Monday Tuesday Wednesday Thursday Friday Saturday Pop-up Linear Menu Pop-up Pie Menu

13 Fitts’ Law Example 4 Pie menu: bigger targets for a given distance; 6.2 / k vs. 2 / k Today Sunday Monday Tuesday Wednesday Thursday Friday Saturday Pop-up Linear Menu Pop-up Pie Menu

14 Pie Menus 4 Pie menus have proven advantages, but you rarely see them (QWERTY phenomenon?). 4 Examples: Maya (animation tool), and many research systems like DENIM. 4 Still, open-source code for them exists.

15 Principles of Operation (cont.) 4 Power Law of Practice *Task time on the nth trial follows a power law T n = T 1 n -a + c, where a =.4, c = limiting constant *i.e., you get faster the more times you do it! *Applies to skilled behavior (sensory & motor) *Does not apply to knowledge acquisition or quality

16 Power Law of Practice

17 Perception 4 Stimuli that occur within one PP cycle fuse into a single concept *Frame rate necessary for movies to look real? +time for 1 frame 10 frame/sec. *Max. morse code rate can be similarly calculated 4 Perceptual causality *Two distinct stimuli can fuse if the first event appears to cause the other *Events must occur in the same cycle

18 Perceptual Causality 4 How soon must red ball move after cue ball collides with it?

19 Perceptual Causality 4 Must move in < Tp (100 msec)

20 Perceptual Causality 4 Must move in < Tp (100 msec)

21 Break 4 Administrivia: Midterm is a week from Friday. *Material covered up to this Friday. *Closed book 4 In-class quizzes will start this week.

22 Simple Experiment 4 Volunteer 4 Start saying colors you see in list of words *When slide comes up *As fast as you can 4 Say “done” when finished 4 Everyone else time it…

Paper Home Back Schedule Page Change

24 Simple Experiment 4 Do it again 4 Say “done” when finished

Blue Red Black White Green Yellow

26 Memory 4 Interference *Two strong cues in working memory *Link to different chunks in long term memory 4 Why learn about memory? *Know what’s behind many HCI techniques *Helps you understand what users will “get” *Lots of people have memory difficulties

27 Stage Theory 4 Working memory is small *Temporary storage +decay +displacement 4 Maintenance rehearsal *Rote repetition *Not enough to learn information well 4 Answer to problem is organization *Faith Age Cold Idea Value Past Large *In a show of faith, the cold boy ran past the church

28 Stage Theory Working Memory Sensory Image Store Long Term Memory decaydecay, displacement chunking / elaboration decay? interference? maintenance rehearsal

29 Elaboration 4 Relate new material to already learned material 4 Recodes information 4 Attach meaning (make a story) *e.g., sentences 4 Visual imagery 4 Organize (chunking) 4 Link to existing knowledge, categories

30 LTM Forgetting 4 Causes for not remembering an item? *1) Never stored: encoding failure *2) Gone from storage: storage failure *3) Can’t get out of storage: retrieval failure

31 LTM Forgetting 4 Interference model of forgetting *One item reduces ability to retrieve another *Proactive interference +earlier learning reduces ability to retrieve later info. *Retroactive interference +later learning reduces the ability to retrieve earlier info.

32 Recognition over Recall 4 Recall *Info reproduced from memory 4 Recognition *Presentation of info provides knowledge that info has been seen before *Easier because of cues to retrieval 4 We want to design UIs that rely on recognition!

33 Facilitating Retrieval: Cues 4 Any stimulus that improves retrieval *Example: giving hints *Other examples in software? +icons, labels, menu names, etc. 4 Anything related to *Item or situation where it was learned 4 Can facilitate memory in any system 4 What are we taking advantage of? *Recognition over recall!

34 Summary 4 MHP: three interacting subsystems *Perceptual, motor, cognitive *Sometimes serial, sometimes parallel *We gave several principles of operation 4 Memory principles: *Several types: WM -> LTM progression *Interference causes recognition problems *Recognition over Recall