Aging, memory, and language in amnesic H.M.

For many years, amnesic H.M. has been ‘‘a touchstone for research on amnesia and memory systems’’ (Manns, 2004), and the usual assumption is that H.M. exhibits a pure memory deficit, reflecting lesion-induced impairment that affects memory-encoding but not language systems. Related to this assumption, Kensinger, et al. (2001) made three significant claims that are examined here: (1) in 1999–2000 H.M. was unimpaired in processing preoperatively-acquired word-knowledge; (2) H.M.’s lexical information remained constant with aging from 1953 to 2000; (3) H.M.’s language processing was unimpaired at grammatical levels in 1999–2000. These Kensinger et al. claims are important because they strongly support systems-theory over competing theories in widespread use in the cognitiveand neurosciences. Under systemstheory, independent systems process language vs. memory. A language system comprehends verbal inputs, and then transmits the products of comprehension to a completely separate memory system for long term storage; a retrieval system later recovers the stored memory for transmission to a language production system, enabling verbal expression of the recovered memory. If correct, the claims of Kensinger et al. support these fundamental processing relations in systems-theories by dissociating the retrieval, comprehension, and production systems (undamaged in H.M.) from the memory storage system (damaged in H.M.). However, all three claims of Kensinger et al. (2001) are in conflict with other evidence indicating that: (1) H.M. exhibits recent deficits in processing low-frequency (LF) words (James and MacKay, 2001; MacKay and James, 2001, 2002); (2) from 1983 to 1997, H.M. exhibited exaggerated age-linked declines in processing LF words that he knew and used correctly in 1970 (James and MacKay, 2001; MacKay and James, 2001, 2002); (3) H.M. exhibited syntax-level processing deficits in more than 30 sources of evidence from 1967–1973 (Lackner, 1974; MacKay, Burke, and Stewart, 1998a; MacKay, Stewart, and Burke, 1998b) to 1997–1999 (MacKay and James, 2001, 2002; MacKay, James, Taylor, and Marian, in press). It is therefore important to examine procedural differences between the studies of Kensinger et al. vs. the studies of MacKay et al. that might explain their seemingly contradictory results. This letter does this for each claim of Kensinger et al. in turn and concludes that the two sets of results are compatible under a new theoretical framework known as binding theory (see e.g., MacKay and James, 2002; MacKay et al., in press). Under binding theory, H.M. exhibits a binding deficit that predicts both sparing and impairment of a parallel nature in language and memory (see e.g., MacKay et al., in press, for detailed theoretical rationale). As applied to H.M.’s lexical knowledge in 2000, 47 years postlesion, H.M.’s binding deficit predicts selective impairment in processing LF but not familiar or highfrequency (HF) information. Second, for information encountered since his lesion, H.M.’s binding deficit predicts selective impairment for rapid (single trial) learning but not for slow learning, resulting from massive repetition over many trials or encounters with new lexical information. Third, binding theory predicts an age 3 lexical deficit interaction, with exaggerated age-linked declines from 1991 (age 65) to 2000 for processing LF but not HF lexical information. The data of MacKay–James supported all three of these binding-theory predictions. Consider now the first claim of Kensinger et al. (2001)—H.M.’s lexical processing is unimpaired. To support this claim, Kensinger et al. presented three sources of data. Source one was experimental. However, the relevant experiments of Kensinger et al. mainly involved information that most certainly has been highly familiar to H.M. since childhood, e.g., recognizing and producing names of familiar objects, spelling familiar (mostly HF) words, identifying place names associated with highly familiar landmarks or events, e.g., Alamo-Texas, multiple-choice category recognition for familiar birds, fruits, furniture, and insects, and production of highly familiar plural and past-tense forms, e.g., boy–boys and dig–dug. These results of Kensinger et al. simply reillustrate H.M.’s preserved function for highly familiar lexical information and do not discriminate between systems theory vs. binding theory, because binding theory predicts selective deficits that do not include HF information. By contrast, the research of MacKay–James satisfied this and other preconditions for discriminating between systems theory vs. binding theory. Consider the meaning–definition experiment of James and MacKay Psychology Department, University of California, Los Angeles, California Grant sponsor: Samuel A. MacKay Memorial Research Fund. *Correspondence to: Donald G. MacKay, Psychology Department, University of California, Los Angeles, CA 90095-1563. E-mail: [email protected] Accepted for publication 31 January 2006 DOI 10.1002/hipo.20176 Published online 29 March 2006 in Wiley InterScience (www.interscience. wiley.com). HIPPOCAMPUS 16:491–494 (2006)