Presentation on theme: "Is there evidence for the processing of different memory systems by different areas of the brain? Jeff Karpicke, Jessica Roland, Johnie Sanders. Department."— Presentation transcript:
Is there evidence for the processing of different memory systems by different areas of the brain? Jeff Karpicke, Jessica Roland, Johnie Sanders. Department of Cognitive Science, Indiana University.
Background There is evidence that humans have two separate, distinct memory stores: one that is “explicit” and one that is “implicit.” Explicit memory is distinguished as memory that requires conscious processing. Memorizing an item explicitly often requires either repetition or visualization of the item (i.e. Baddeley’s model of working memory). Implicit memory, however, does not require such conscious processing. Instead, an item can be memorized implicitly by virtue of being part of a larger pattern. Rather than the individual item itself, the repeated pattern is learned without conscious inference on the part of the learner (i.e. “The Nissen Task”). Familiarity with a repeated pattern leads to implicit memorization.
Hypothesis We used a word recognition task to attempt to distinguish between explicit and implicit memory stores and to determine whether these different stores are processed in different areas of the brain. Ordinary, everyday, “real” words that have semantic distinctiveness ought to be processed by an implicit memory system, because they are frequently encountered and are part of a large, repeated linguistic pattern. Pronounceable “nonsense” words – words without semantic distinctiveness – ought to be processed by an explicit memory system, because these novel items are encountered infrequently and are not part of a large pattern. Pronounceable nonsense words should require more conscious effort (such as repetition) in order to be stored in memory. It follows that there should be different patterns of brain activity corresponding with these different memory stores. The activity in the explicit store ought to be temporally delayed and larger (i.e. more active) compared with activity in the implicit store – since in theory, it should take more “effort” to encode novel, meaningless words than it would to encode ordinary, meaningful words.
Method One subject (female, age 20) was presented with a series of 400 words (200 real words and 200 non-words, intermixed) during the study phase of the task. Each word was presented for 500 ms on a computer monitor. In the testing phase of the task, the subject was presented with a series of 600 words (the 400 target words plus an additional 200 distracter words, intermixed). The subject was asked to decide whether they “remembered” the word (meaning that they were absolutely sure the word was on the list during the study phase), whether they “knew” the word (meaning that they felt the word was familiar, but were not sure that the word was in the study phase), or whether the word was a “new” word (meaning that they were sure the word had not been in the study phase). An electroencephalograph (EEG) reading of the brain activity of the subject was taken during the testing phase of the recognition memory task. Recordings were made for 1200 ms at 1000 Hz.
Data RemCond refers to the “Remember” condition; KnowCond refers to the “Know” condition; NewCond refers to the “New” condition. Condition “=1” refers to the Real Word condition; Condition “=2” refers to the Pronouncable Nonsense Word condition. The EEG data indicate a great difference in activity between the “new real word” response (in yellow) and the “know nonsense word” response (in purple) in several channels.
“I’m personally not convinced that the implicit system is active in word learning.”
Results Throughout the occipital, temporal, and parietal lobes, there appears to be a difference in the activity of the “new real word” condition around 200 ms. In general, the condition evokes a larger negative pattern of activity. The “know nonsense word” condition also exhibits activity different than the other conditions. The condition evokes a smaller and more positive pattern of activity around 200 ms – a pattern which is also quicker than others in the temporal lobe at T5 and T6.
Conclusions The large negative spike associated with the “new real word” condition probably is a reflection of the “extra effort” put forth to determine whether or not the word was a member of the target list based on its semantic characteristics. The smaller, more positive spike associated with the “know nonsense word” condition is most likely a reflection of the inability to make a semantic distinction between pronounceable nonsense words. The meaninglessness of the nonsense words makes them indiscriminable, and thus a response that the subject “knows” or is familiar with the word is more likely and effortless in the nonsense word condition (since they’re all semantically the same!).
More Conclusions These results don’t indicate that two different memory stores are at work in this word recognition task. Instead, they indicate that a conscious, explicit method was used to discriminate a new real word from the target real words. And rather than making an explicit decision based on phonological evidence alone to discriminate new nonwords from target nonwords, all nonwords appear to have been lumped together, becoming indiscriminable due to a lack of semantic information. Perhaps the real word, nonsense word conditions aren’t the best way to distinguish between implicit and explicit memory stores – other conditions should be explored in future research.