Abstraction vs. Hyperspecifity
Abstraction explains how people remember conglomerations (generalisation, the gist) NOT verbatim information. Literal meaning and surface/perceptual information appears to be lost. Research constructed sentences that contained four ideas, but presented participants with sentences containing only 3, 2 or 1 of those ideas. Some of them were old (the same sentences) and some were new. They then performed an old/new recognition test, and found that the more ideas in the sentence, the more confident they were that the sentence was old, and they'd seen it before. Even if they hadn't. This demonstrates how people remember the general gist of information, and the abstractions/inferences/generalisations they have made, but not word for word exactly what was presented. Research on Nixon's tapes revealed that Dean was entirely truthful in substance, but inaccurate in detail. Therefore, abstraction implies that LTM maintains the gist of information and the generalisations we have made, but not detail (surface/perceptual information).
Hyperspecificity is the reduction of the priming effect if the surface characteristics of an item are changed, demonstrating that surface/perceptual information is maintained in LTM. Here, priming refers to the difference in response time to an item you have already seen before, and a new item. Research has found that you respond faster to the word 'tiger' that has been presented before, than to the term 'car' that has not, by 100ms. However, if you change the font and colour of the word 'tiger' on the second presentation, there is only a 50ms priming effect, demonstrating that the perceptual characteristics are maintained in LTM and facilitate retrieval. Similar research initially presented people with an elephant facing left, then when presented with the same elephant facing left there was a 100ms priming effect, but when presented with a different elephant facing right there was only a 55ms priming effect. This demonstrates that the first 55ms speed up on RT is due to the conceptual ('gist') match; the fact that they're both elephants. There's an extra 45ms speed up on RT if the perceptual ('surface') features match, too. Research has shown that you're better at reading the SAME inverted text (therefore, having the same surface features) a year later, than new inverted text. Abstraction is generally found on unspecific (implicit memory) tests, such as free recall. Hyperspecificity is generally found on specific (direct memory) tests, such as priming and cued recall. It can be found on unspecific tests, too, depending on its novelty/distinctiveness. For example, you remember and are faster to respond to inverted text better when it's novel/new to you.