Thursday, June 29, 2006

Memories, Memories and more Memories: How to mix N match

There is a recent article on Mixing Memories regarding the processes used in 'false memory' research.

Elizabeth Loftus has done a lot of research in false memory area and to summarise one of the methods refer to this article .You either invoke a true memory (in this case memories of say a visit to disneyland or similar place while reading an article or photo advertisement regarding the same and then while the memory is being reconstructed you insert a false element(in this case say showing a picture of Disneyland with Bugs bunny present) and later after the memory has been consolidated, the people may actually remember meeting bugs Bunny at Disneyland and when explanations are sought as to what they did while meeting bugs bunny, they would rationalize and come up with all sorts of activities. The explanations themselves may be very vivid and intriguing, but split-brain research by Micheal Gazzinga et al has shown that in split brain patients, when one eye(or brain region) is primed to one word (say scissors) and the other hand told to do something , it may pick up a scissor and start cutting things, the human as a whole though unaware of why he started doing the action may come up with all sorts of rationalizations. Thus all rational explanation by a person once the false memory got planted is of no relevance. We humans, as narrators of our life story, would come up with any explanation as to why that memory exists in our brain. This is called postevent misinformation and it is necessary that we keep this separate from the other type of false memory formation. In this typically Videos...or remembrance of visual stimuli...is needed as the contextual element and the false information has to be in the form of images or photos. Thus, this uses visual modality for it to be successful and relies on the visual perception, imagination, ideation etc. An interesting study could be to use lexical input, say asking someone to write about the trip to disneyland, show them or ask them to write bugs bunny, and then see if the memory can still be implanted. Imaginative activity during writing about one's trip to Disneyland may be a confound as it normally involves conjuring up past images. A better scenario would be writing a visit to Disneyland from a third person perspective.

The second mechanism of false memory introduction is less spectacular. It is known as Deese-Roediger-McDermott paradigm (DRM) and involves semantic priming. It involves presenting a semantically or conceptually related list of words and then when tested for a semantically or conceptually related word (that was not present in the initial list) as part of the test list and if the person remembers that critical word on test trial then it is assumed that a memory for that word has been formed in the subjects brain. Here it is not really possible to assert a-priori that the memory for the critical word was present prior to seeing the word on the word list or gets implanted the moment one sees the word on the word list. This test is mostly verbal based and uses the language or speech modality and thought/ memory processes that are mediated by language. Not sure whether this would work after memory consolidation too, as normally the 2 lists (original word list and trial list) are presented with little time for consolidation.

In my view it is important to keep the modalities different in the 2 cases.

The particular article referenced above refers to an article below
>>
From Sahlin, B.H., Harding, M.G., & Seamon, J.G. (2005). When do false memories cross language boundaries in English-Spanish bilinguals? Memory and Cognition, 33(8), 1414-1421.
>>
and here is some data from that study

This study uses DRM as the tool of study and Mixing Memory makes these inferences

>>
The first is that there is a big difference in accuracy between words that were presented in the same language in the learning and recall lists (English-English and Spanish-Spanish in the table) and words that were presented in different languages in the learning and recall lists (English-Spanish and Spanish-English). Participants were much more accurate when recognizing list words, and were much more likely to mistakenly recognize critical words, for words presented in the same language.
>>


The conclusions I draw are slightly different. For one it is clear that if we just look at the 'same language' studied-words data vis-a-vis the critical related-words data, then on trial 1 both show the same proportions of rememberence....i.e. after the learning phase the accuracy of list words is equivalent to the critical word due to semantic priming and as a matter of fact one can make a bold statement that the individual remembers none of the words, but only a semantic cloud...and when the trial list is presented to the individual it jutes picks those words from the list that belong to the semantic cloud and reconstructs his memory anew each time he sees the words.

This explanation works for the first trial. However when trials of similar nature keep on repeating, he sort of starts getting a hang of what the experiment is about and stops recognizing the critical words. This of course he unconsciously does and cannot verbalize that he has understood the experiment. If sufficient trials are run he would be able to understand consciously too the experiments. However, this unconscious understanding reflects on his performance in 2 ways. First, his recall of critical words that were not presented earlier drops. Secondly his recall of list words increases. How the individual does is questionable, but apparently instead of using semantic cloud as remembrance of lists studied, he may start using some other mechanism that allows accurate retention of actual words and also active inhibition of 'semantic cloud' way of memory. This is a classical case of mice getting insight in mouse trap and only utilizes data from same-language trials.


Lets look closely on the different-languages data. Here for the first trial itself, the proportion of right recognition of words-studied as well as critical related-words is lower than in the case of same-language scenario. This may suggest that the 'semantic cloud' metaphor may not work that great in different-languages situation or it works to the contrary by making the 'semantic cloud' or determining set too vast as to making recognition of a word as belonging to the earlier list become so problematic as to result in low proportions for both critical and regular words.

interestingly, over the 5 trials the memory for studied words decreases while that of critical words increases in this case of different-language condition. This suggests that while the 'semantic cloud' may be replaced by some other mechanism (say the classical right brain generalists to left brain particlarizer processing) , still the dual-language condition puts constraints on the use of particularistic method of perception, storage or retrieval. This can easily be tested in tri-or more linguals and see, if the same pattern emerges.

Mixing Memory concludes as
>>However, I think their data actually indicates something slightly different, and perhaps more interesting. While accuracy does increase over the five recall trials for critical words in the different language conditions, it dramatically decreases for list words over the five trials. I interpret this as an indication that as the availability of conceptual information decreased over the trials, participants had a difficult time recognizing list words, and an easier time rejecting critical words. This would imply that recognizing words learned in one language and then presented in another relies heavily on language-independent conceptual information.
>>
I draw a similar conclusion, that the encoding process that is necessitated in this case of different-languages, demands that 'semantic clouds' be used while encoding because the semantic cloud may be the most apt/only way of encoding information when 2 languages are involved. If particulars may be used, they might be limited to keywords or critical words (that may overlap lexically in the two languages) and hence these may be stored with an explicit rider that these were not part of the list. However, the subsequent task demands that particular words be recalled. So while reconstructing, one might recall the critical word that was not related to the cloud, but was stored and hence that word may be rejected at a higher frequency; while the semantic cloud being too broad may result in less proportions for the trail words leading to less 'truer' recalls.

Endgame
: would this suggest that if one wants not to fall in the false memory trap, one needs to study more and more languages? Also, would in that case, the rememberence of 'true' memories may also become less and less as one's 'semantic cloud' keeps expanding? Is it necessary to remember what you did not see in order to be a good eye witness who cannot be implanted with false memories by lexical suggestions?



Sphere: Related Content

No comments: