I read two books last week. The first was Mnemenology: Mnemonics for the 21st Century by Worthen and Hunt, a fantastic exploration of psychological research concerning what we know for sure about how the mind works with mnemonic systems. I’ll quote some of the best passages below.
The other book was The Memory Palace of Matteo Ricci by Jonathan Spence. it was a fascinating history of the Jesuit attempt of evangelizing 16th century China through the efforts of gifting Western gadgets and treasures as well as showing off mnemonic feats of Ricci’s Chinese language learning. But the book and history fails on telling us much about his techniques other than he used memory cities. @metivier even wrote about it on his blog just a few days ago.
So, here’s the quotes from Mnemonology with bolding that I provided to highlight the important part of each section.
Selective Encoding. Hunt (2008) suggested that encoding processes may influence memory in three ways. First, encoding may function to direct attention such that only a select portion of the to-be-remembered information is represented in memory. A good deal of research from the levels-of-processing approach to memory (Craik & Lockhart, 1972) suggests that successful memory depends on orientation toward specific features of to-be-remembered information. For example, Craik and Tulving (1975) presented participants with a list of words with instructions to determine whether each was typed in uppercase or lowercase, whether each word rhymed with a presented word, or whether each word fit meaningfully into a provided sentence frame. The results indicated that the latter condition led to better memory for target words than either of the former conditions. This outcome indicates that orientation toward semantic features leads to better memory than orientation toward superficial features of to-be-remembered information. Extended to mnemonics, these results suggest that to achieve maximal effectiveness, a mnemonic technique must encourage orientation toward meaning rather than the perceptual features of to-be-remembered information.
However, existing empirical research on the effectiveness of bizarre imagery as a mnemonic suggests that there are limits to its effectiveness. First, bizarre imagery leads to better memory than does common imagery only when both types of imagery are used within the same list. Thus, the bizarreness advantage occurs when some list items are learned using bizarre imagery and other items are learned with common imagery, but not when bizarre imagery is applied to the entire list (McDaniel et al., 2000; McDaniel & Einstein, DeLosh, May, & Brady, 1995). Perhaps even more revealing is the finding that when both imagery types are used in the same list, memory for information encoded with bizarre imagery is remembered at the expense of items encoded with common imagery (Kroll & Tu, 1988; Lang, 1995). This suggests that bizarre imagery does not increase the total number of items remembered but simply allows the learner to select which list items will have the highest probability of being recalled.
… research (De Beni & Cornoldi, 1988; Massen & Vaterrodt-Plunnecke, 2006) has found that repeated use of the same loci for different lists does not diminish the effectiveness of the method. However, research with normal populations has shown that self-generated loci are more effective than loci that are supplied by others (see Moe & De Beni, 2004).
… Cornoldi and De Beni (1991) and Moe and De Beni (2005) demonstrated the method of loci to be more effective than cumulative verbal rehearsal for learning complicated verbal material (e.g., extensive discourse) with oral but not written presentation.
Importantly, research (Massen & Vaterrodt-Plunnecke, 2006; Morris & Reid, 1970) also indicates that repeated use of a single list of peg words does not result in interference effects. However, category-relatedness among to-be-remembered items (Reddy & Bellezza, 1986) and rapid stimulus presentation (Bugelski et al., 1968) may reduce the effectiveness of the method.
The keyword method was designed by Raugh and Atkinson (1975) specifically to enhance second-language acquisition. Unlike the method of loci and the peg-word method, the keyword method was designed not to enhance serial recall but to enhance cued recall. Specifically, the method was designed to aid memory for definitions when presented with foreign vocabulary. However, the application of the keyword method is not limited to second-language learning, as it can be useful for any vocabulary-learning situation.
The results of initial research investigating the efficacy of the phonetic system were mixed. Bruce and Clemons (1982) found the method to be ineffective in enhancing memory for metric conversions and standard measurements. Similarly, Patton (1986) found that the phonetic system did not facilitate and sometimes hindered recall of dates, addresses, and phone numbers. Contrary to these results, Morris and Greer (1983) found the system to be effective for remembering lists of two-digit numbers. Subsequent research (Patton, DÁgaro, & Gaudette, 1991; Patton & Lanzy, 1987) provided clues to understanding the discrepancy in previous results by demonstrating that the effectiveness of the phonetic system depends on whether participants were supplied with conversions or generated conversions themselves. Specifically, these studies indicated that the system is effective when conversions are provided but not when they are self-generated. These finding are not surprising given the overall difficulty of use of the method.
Empirical research supports the notion that constructing a story out of to-be-remembered items can be a highly effective mnemonic. The story mnemonic has been demonstrated to be effective in facilitation memory in both college-age (e.g., Herrmann, 1987; Herrmann, Geisler, & Atkinson, 1973) and older (e.g., Drevenstedt & Bellezza, 1993; Hill, Allen, & McWhorter, 1991) adults. In addition, the story mnemonic has been shown to be effective in improving memory in memory-impaired (Wilson, 1995) and mildly retarded (Glidden, 1983) participants. It should also be noted that story mnemonics can be effective both with long lists (Bellezza, Six, & Phillips, 1992) and over long retention intervals (Drevenstedt & Bellezza, 1993). Moreover, the effectiveness of the story mnemonic does not differ when self-generated and other-generated stories are compared (Buonassissi, Blick, & Kibler, 1972). However, as we alluded to earlier, the effectiveness of the story mnemonic may be reduced when the to-be-remembered information is particularly abstract (Manning & Bruning, 1975)
… we do believe that the use of mnemonics can significantly improve one’s memory ability across a wide domain of subject areas. Moreover, we believe that highly specialized feats of memory prowess could be achieved through extensive practice with appropriate techniques. However, truly exceptional broad-based memory is likely a product of multiple factors including genetic predispositions, environmental factors, and general aptitude.
…the loci mnemonic advantage applies only to spoken discourse. Written passages actually were better remembered following rehearsal. De Beni et al. (1997) offered an interesting explanation for this modality effect in the context of Brooks’s (1967) classic study of selective interference. When two similar tasks are performed in the same modality, performance is not as efficient as when the tasks are performed in separate modalities. Brooks showed that performing a response that required locating an object in space interfered with the use of a visual image. In an analogous fashion, imagery-based mnemonics would interfere with visual processing required by reading, just as verbally based rehearsal would interfere with processing of spoken discourse.
Final test memory is optimal when study episodes are separated by 10 to 20% of the amount of time from initial study to final test.
Pyc and Rawson (2009) demonstrated that the long-term benefits of testing are a direct function of the difficulty in retrieving the answer. In their research, Pyc and Rawson set up conditions in which study of the to-be-learned material involved retrieving correct answers, ensuring that the correct answer was always retrieved. They varied the intervals between retrieval attempts of particular answers and found that the longer the interval between retrievals, the better the ultimate learning, even though the longer intervals made for more difficult retrieval at study.
…basic research has established that the joint operation of organization and elaboration, herein referred to as distinctive processing, facilitates memory beyond the operation of either alone. Research even has shown that it is this distinctive processing that seems to characterize the domain-specific memory advantage of experts (Rawson & van Overschelde, 2008; van Overschelde, Rawson, Dunlosky, & Hunt, 2005). No mnemonic has been intentionally developed to capitalize on this discovery, although some applications of existing devices are probably doing just that. Successful creation of a technique based on these principles would be a powerful tool.