Last Monday (01/26/09), ETC-SV students, faculty and some other people were invited to a test with respect to the storage capacity of visual long-term memory. We were asked to view sequences of images and took 5-minute breaks in between. After several iterations, testers were handed a set of test papers with paired images on it. The images were paired in this way:
1) In the novel condition: the presented image was paired with a different categorized image to test the category memorization of the long-term memory.
2) In the exemplar condition: the presented image was paired with a phycally similar image to test the appearance memorization of the long-term memory.
3) In the state condition: the presented image was paired with a exactly the same but in different state image to test the state memorization of the long-term memory.
Testers needed to circle the image which appeared in the previous slices.
These images were our test results. From the tests we could know that: 93% of novel questions were correctly answered, 88% of exemplar questions were correctly answered, and 87% of state questions were correctly answered. The result indicates that our memory is as good at detail(shape and state) recognition as at category(nobel). This leads to a conclusion that our long-term memory is capable of maintaining detailed representations of thousands of images.
From the game developer’s perspect of view, players’ memories should not be under-estimated. If making a role playing game, the hints for a puzzle should not be too easy or too obvious to get. If making an action or a shooting game, the tempo should not be too slow so that players have enough time to observe the environment and react. Also, for both games or movies or any other works, the character , the NPC, and the world information need to be well-designed so that nothing will be illogical or mismatched when the story goes.