How The Mind Works Pinker
<p class="default"><b>How the Mind Works: Steve Pinker; 1997</b> Norton and Co. actual text is 565 pages. total book is 660.</p>
<p>
Feb 23, 2009 6:48 PM</p>
<p>
</p>
<p>
</p>
<p>
</p>
<p>
03:13 PM, July 13, 2007
 
<b>How the Mind Works: Steve Pinker; 1997</b> Norton and Co. actual text is 565 pages. total book is 660.
 
put a summary of the vision chapter here.
 
Espec see the mental models pages:
 
we use both propositions and images: good but v. superficial discussion
 
 
page 290 thru 298 at end of The Mind's Eye
 
page 290
He compares an image of a teddy bear with a
an Is-A network and has-part and is-attached diagram of propositions.
Claim is that we use each for difft purposes. Both are present.
 
page 291: visualize a lemon AND a banana (one has to be the left).
Also note: you can use the transitive relationship to say
Gates is richer than Bob is richer than Ghandi. (you visualize all 3 in a row).
 
Mental images store shape, location, orientation, pattern of contours.
You can mentally rotate it (Bob: or in fact do anything you can do with an actual object in your hands
- that's how it works: see it once and experience it repeatedly).
 
Images capture the geometry.
We can cache assertions as propositions like cats have claws but if you ask
which is smaller, a cat's head or its stomach then the person must visualize them
and do mental manipulations to compare in mind's eye.
 
page 293: The DUCK-RABBIT ambiguous interpretation (but it is stored with its frame: therefore, hard to flip)
 
Steve and colleagues did experiments:
 
Imagine the letter B. Rotate it 90 degrees counterclockwise. Glue a triangle to it directly below it.
Now remove the horizontal line. What do you see?
 
also 2 more examples.
 
page 294:
 
Images are great but have big limitations:
 
page 295: Memory images must be labeled and organized within a propositional superstructure, perhaps like hypermedia
where graphics files are linked with a large database.
 
<b>CHESS:</b> masters are no better than beginners when it comes to remembering randomly arranged pieces.
Their memory captures meaningful relations among the pieces, such as threats and defenses, not just their distribution in space.
Bob: also see genius article in Sci American Secrets of the Expert Mind Aug 2006.
 
page 295: Psychologs. asked people to draw both sides of a penny. Only 5% drew all 8 features (without looking).
 
Also, which of these is true?
 
Madrid is farthernorth than Wash DC
Seattle is farther north than Montreal
Portland is farther north than Toronto
Reno is farther west than San Diego.
Atlantic entrace to Panama Canal is to the WEST of the Pacific entrance.
 
ALL are TRUE!!!
 
page 296: IMAGES cannot serve as CONCEPTS!!!
somehow combined with edges shaved and mooshed together.
eg, even the concept of triangle (must either be isosceles, or scalene, or equilate)
 
How do you represent FREEDOM (Statue of Liberty is already taken!)
 
Your common sense is NOT just a collection of pictures.
We need abstract symbols and propositions.
 
PINKER does NOT SAY what those abstractions are or those propositions.
How about a high level (a la Hawkins) common features from filtered images
(remove orientation, remove size etc.: normalize the resulting abstraction ... that's what they do at Numenta.)
 
WE can do these transformations... rotations, panning, zooming: why can't a computer do that - or can they?
 
 
 
 
 </p>