Perception TopDown and BottomUp Processing n TopDown Goaldriven

  • Slides: 12
Download presentation
Perception: Top-Down and Bottom-Up Processing n Top‑Down Goal‑driven n Processing dominated by context, expectations,

Perception: Top-Down and Bottom-Up Processing n Top‑Down Goal‑driven n Processing dominated by context, expectations, hypotheses, "the big picture" n n Bottom‑Up Data‑driven n Processing dominated by characteristics of stimulus, "details” n Feature analysis n n ISE 412 Both types of processing occur simultaneously and are highly interactive. 1

Some examples n Problem-solving ¨ Wheel of Fortune ¨ Crossword puzzles ¨ Design n

Some examples n Problem-solving ¨ Wheel of Fortune ¨ Crossword puzzles ¨ Design n ISE 412 Hearing: conversation, listening to music, etc. 2

An example … n Listen to the song and write down the words as

An example … n Listen to the song and write down the words as you hear them. n To discuss … ¨ What were doing during “bottom-up” processing (with and without visual cues? ¨ What ISE 412 were you doing during “top-down” processing? 3

Perception of print n Feature analysis: n n n Bottom-up, we perceive features of

Perception of print n Feature analysis: n n n Bottom-up, we perceive features of letters, words Involves pattern matching Top-down, we use context of surrounding letters and words to limit alternative interpretations Examples: Sawple ISE 412 Sample 4

Context-Data Tradeoff n Assuming limited display size for text, trade off context and data

Context-Data Tradeoff n Assuming limited display size for text, trade off context and data quality: n Many words: High contextual redundancy (easy to comprehend) ¨ Poor data quality (hard to read) ¨ n Few words: Good data quality (easy to read) ¨ Low contextual redundancy (hard to comprehend) ¨ ¨ Example: weather alerts on a heads-up display WEATHER ISE 412 Will encounter bad weather 5

Creating context: Redundancy gain n n ISE 412 The information is presented in more

Creating context: Redundancy gain n n ISE 412 The information is presented in more than one way. If the viewer is not able to process the information in one mode (e. g. , color blind) can rely on another. Reaction time is faster if information is presented redundantly. Promotes top-down processing. 6

Object perception n Objects are perceived holistically i h c hi er g on

Object perception n Objects are perceived holistically i h c hi er g on l s ? W n Empirical evidence on card‑sorting tasks: n objects categorized more quickly and accurately than "separate" displays such as numbers or words ♣ ♣ 4 of Clubs Carrot ♣ ♣ ISE 412 7

Advantages & disadvantages … n Advantages n Disadvantages ISE 412 8

Advantages & disadvantages … n Advantages n Disadvantages ISE 412 8

Object perception n Wickens et al. studies: ¨ object displays facilitate performance on information

Object perception n Wickens et al. studies: ¨ object displays facilitate performance on information integration tasks GPM, y 5. 5 5. 9 6. 5 3. 3 3. 6 4. 6 2. 9 3. 6 3. 1 4. 9 ¨ separable ISE 412 Wt. , tons 3. 4 3. 8 4. 1 2. 2 2. 6 2. 9 2 2. 7 1. 9 3. 4 displays support diagnosis tasks 9

Examples … Phone ISE 412 Easy Difficult A 0 2 B 1 2 C

Examples … Phone ISE 412 Easy Difficult A 0 2 B 1 2 C 0 3 10

Implications for display design n Optimize bottom-up processing ¨ size, contrast, font (for text),

Implications for display design n Optimize bottom-up processing ¨ size, contrast, font (for text), appropriate upper/lower case ¨ raw data on object displays n Optimize top-down processing ¨ avoid abbreviations & acronyms whenever possible ¨ provide context (more words in text, recognizable object displays) ¨ restrict “vocabulary” (text and picture) and optimize distinction between words/pictures n n ISE 412 Evaluate tradeoffs Usability testing in context 11

Some fun examples: http: //exp. kyb. tuebingen. mpg. de/web-experiment/list. html http: //www. grand-illusions. com/index.

Some fun examples: http: //exp. kyb. tuebingen. mpg. de/web-experiment/list. html http: //www. grand-illusions. com/index. htm ISE 412 12