ACT-R Tutorial Unit3

Attention

1. Visual Locations

Visual-location requests
- :attended 该元素是否已经被看过
  - possible value : new , nil , t
  - 当多个符合条件，选择最近被加入 visual icon的 (chose the newest one )
  - 当没有符合条件的，then a failure will be signaled for the visual-location buffer.
  - Finsts
    - 数の制限：A limit of elements can be attented. “Attended”とタグされるアイコンの数の上限はFinst数　
    - 時間の制限：there is also a time limit on how long an item will remain marked as attended t
    - parameter in the system :
      - :visual-num-finsts = 4
      - :visual-finst- span = 3
Visual-location slots
- (chunk-type visual-location screen-x screen-y distance height width size color kind value)
- (the screen resolution of the experiment windows is assumed to be 72 pixels per inch
- unit : pixels
- screen-x : left to right
- screen-y : top to bottom
- distance = 1080 (15 inches)
- kind : text, line , …

Visual-location request specification

Exact values

   +visual-location>
      isa visual-location
      screen-x 50
      screen-y 124
      color black
      - kind text

General values

   +visual-location>
      > screen-x 50
      <= screen-y 124

Production variables

 (p find-by-color
  		=goal>
  	target =color
  ==>
	+visual-location>
  		color =color)

Relative values

 ;// code1 : 从左到右依次看过去的代码：
+visual-location>    
      :attended nil
      screen-x lowest

  ;// code2 : 先考虑绝对条件，再考虑相对条件的代码
  +visual-location>
      width highest
      screen-x lowest
      color red

lowest, highest, nearest
当查找符合条件的item，ACT-R会优先看有绝对数值的条件，然后再看相对条件，比如code2中，会先考虑红色，再考虑最宽的，再考虑位置最靠左边的
The current value
```
 ;// 查找位于当前元素的正上方，且拥有与当前元素不同颜色的元素
   +visual-location>
     screen-x current
     < screen-y current
     - color current
```
- It is also possible to use the special value :current in a visual-location request.
- If the model does not have a currently attended object (it has not yet attended to anything) then the tests for current are ignored. 如果当前没有正在看的元素（current），则包含current的条件将会被忽视

Request variables

 ;// 查找一个元素，与当前元素位于统一垂直线上，且位置低于100，且x不等于y,  返回符合条件中位置最高的元素
   +visual-location>
         screen-x current
         screen-x &x
         > screen-y 100
         screen-y lowest
         - screen-y &x

The :nearest request parameter

 +visual-location>
  	 :nearest current

用到了nearest的条件将会被最后考虑

Ordered Search
- 从左到右依次看过去，不会因为finsts的限制而陷入loop
- That will always be able to find the next word to the right of the currently attended one regardless of how many finsts are available and in use.
```
 +visual-location>
   	> screen-x current
  	 screen-x lowest
```
- By using the relative constraints along with the :nearest request parameter and the current indicator a variety of ordered search strategies can be implemented in a model.

2. The Sperling Task

for study iconic visual memory
- https://study.com/academy/lesson/iconic-memory-sperlings-partial-report-experiment.html
In the Sperling experiment subjects are briefly presented with a set of letters and must try to report them. Subjects see displays of 12 letters.
闪过一个画面，该画面仅维持50ms，实验者被要求回忆某一行的字母, 在原始的任务中，需要被回忆的那一行被用不同频率的声音读出来)
Original one: used a tone with a different frequency for each row and the model will hear simulated tones while it is doing the task. the display was only presented for 50 ms.
ACT-R で実行：

3. Visual Attention

Buffer Stuffing
- 一开始，模型会自动填充 visual-location buffer ，default是画面最左且最新的元素 :attended new and screen-x lowest. leftmost new item on the screen.
- 作用： Using buffer stuffing allows the model to detect changes to the screen automatically. 自动监测画面更新
- buffer stuffing will only occur if the buffer is empty.
- it occurs for both visual and aural percepts
- a simple approximation of a bottom-up mechanism of attention.
- “NIL” : this setting of the chunk in the buffer was not the result of a production’s request.
  
  0.000 VISION SET-BUFFER-CHUNK VISUAL-LOCATION VISUAL-LOCATION1 NIL
- Testing and Requesting Locations with Slot Modifiers
- strict harvesting : automatically clear buffers, otherwise it is modified on RHS

4. Auditory Attention

2 buffer
- aural-location : holds the location of an aural message
- aural : holds the representation of a sound attented
Only 2 steps to encode a sound ( in visual there is 3 )
- There is a delay between the initial onset of the sound and the audio-event for that sound (in visual it is immediately)
- depend on the type of sound : tone, digit, word, …
- tone : 0.5s

the following production responds to the appearance of an audio-event in the aural-location buffer:

   (p detect-sound
   	=aural-location>    ;// test if there is a chunck in the aural-location buffer
   	?aural>
   		state free
   ==>
   	+aural>        ;// request to shift attention and encode the event provided
   		event  =aural-location  
   )

it taks 285ms from when the request was made until the sound was attended and encoded
- chunk type of the sound has slots : kind, content, event
  - kind : indicate type of sound (digit, word, customized type)
  - content : representation of the sound heard - for tone : frequency - for words : strings - for digits : number
  - event : contains a chunk that relates to the event that was used to attend the sound
    (p sound-respond-low =goal> isa read-letters tone nil =aural> isa sound content 500 ;// a 500 Hz sound is considered low ==> =goal> tone low upper-y 500 lower-y 520 )

5. Typing and Control

goal module creates a new chunk immediately in response to a request (imaginal module takes time to create new chunk)

(p start-report
   	=goal>
   		isa read-letters
   		tone =tone
   	>visual>
   		state free
   ==>
   	+goal>    ;// create a new chunk to be placed into the goal buffer
   		isa report-row
   		row =tone
   	+retrieval>
   		status =tone
      )

only want this production to apply when there is nothing else to do !
make the production with a low utility
set production parameter (spp)
- (spp start-report :u -2)
- the default utility is 0
Let the sound to be processed as soon as possible
- (spp detected-sound :u 10)
- (spp sound-respond-low :u 10)
- (spp sound-respond-high :u 10)
- (spp sound-respond-middium :u 10)
```
...
==>
 	+mannual>
 		cmd  press-key
 		key  =val
```

6. Declarative Finsts

For a model to exhaustively search DM for items without repeating
- use finst :recently-retrived
- default value
  - number of finst : 4
  - finst span : 3s
- change the default value:
  - sgp command = set general paramter
  - (sgp :v t :declearative-finst-span10)
  - (sgp :v nil :declearative-num-finsts 5) not to print out traces

7. Data Fitting

取消随机种子，运行多次得到模拟数据
(sgp :seed (100,0))
improve speed of runing the model
turn off the trace :v nil
run in a virtual window instead of having a real window to watch
compare with human data
metric : number of items correctly recalled
conditions : 0s, 0.15s, 0.3s , 1.0s (delay of sound stimuli)
result: