Henry did, however, have difficulty on tests of language fluency. In one such test, we gave him a category, such as “fruits,” and asked him to name as many examples of the category as he could think of in one minute. In another, we asked him to name in one minute as many words as possible beginning with the letter F then A then S. These letters sampled a range of difficulty. Based on the number of words available to choose from, F words are the hardest, S words are the easiest, and A words fall in between. On both of these fluency tests, Henry’s scores were inferior to those of the nineteen control participants. Still, the results for all other language functions showed preserved lexical-memory performance (word knowledge).6
Henry’s scores on measures of fluency were impaired—of that, there was no doubt. The most straightforward explanation for this limited performance was his lower socioeconomic status. Prior to his operation, he was by no means a highly verbal person, and his poor naming ability after his operation probably reflected a general lack of verbal skills. His working-class upbringing may have limited the development of his language processes. He never went to college, and his skills and interests as a young man tended toward technology and science. Language was not his strong suit. The letters Henry received from friends stationed overseas during World War II, with their frequent misspellings and grammatical mistakes, underscored our impression that language skills were not a priority in his social group. The bulk of our research ultimately indicated that overall, Henry’s language abilities were consistent with his socioeconomic status and likely the same as before his operation.
Lab members and I found that during our informal interactions he could appreciate puns and linguistic ambiguities, such as words with double meanings. Henry rarely initiated conversations, but when we engaged him, he was always a willing, communicative, and entertaining participant. Once, when I said to him, “You’re the puzzle king of the world,” he replied, “I’m puzzling!”
Henry’s operation did not affect most of his linguistic abilities because the many brain areas that support the production and understanding of language are outside the medial temporal-lobe region. Beginning in the late 1980s, functional brain-imaging experiments added a different dimension of information to the understanding of language. Two new scanning tools, positron emission tomography (PET) and functional MRI, allowed researchers to observe brain activity while a healthy person in the scanner was performing various kinds of tasks that used words. Different technologies underlie PET and funtional MRI. For PET studies, the participant is injected with a radioactive tracer that is taken up by neurons—particularly those that are most active—and is detected by a complex X-ray machine. The analysis tools allow researchers to link discrete areas of activation to the specific cognitive processes engaged by the person in the scanner. Functional MRI relies on a different technology to link brain and behavior, one that uncovers areas of task-related activation using measures of blood flow. Functional MRI has largely replaced PET in cognitive neuroscience studies because it does not expose participants to radiation, and it gives a more precise picture of brain activity.
A 2012 review of 586 functional-imaging experiments synthesized results concerning the localization of activity related to heard speech, spoken language, and reading. This overview showed thirty-one areas of language-related activation in the cortex on the left side of the brain, as well as in structures under the cortex—the caudate nucleus, globus pallidus, and thalamus—and in two places on the right side of the cerebellum. Each area supports one or more aspects of language function, such as processing speech sounds, speech comprehension, speech production, processing written words, and converting spelling to sound. These cortical and subcortical areas are highly interconnected by white-matter fiber tracts, allowing them to communicate with one another efficiently. The right hemisphere also participates in language functions. A network in the right frontal and temporal lobes processes information about speech rhythms, accents, and pitch.7
Henry’s unique situation enabled our team to again make an impact on cognitive neuroscience—this time in the area of language. Our study of his lexical and grammatical processing was the first extensive analysis of these capacities in amnesia. His data revealed a striking distinction within lexical memory between retrieval of previously acquired information (which was preserved) and new learning (which was not). Henry’s test results showed clearly that medial temporal-lobe structures are not critical for the retention and use of preoperatively learned lexical information and grammar. He could spell familiar words, identify objects by name, match pictures with the corresponding words, and say where famous landmarks were located. This capacity to retrieve lexical information and use that information efficiently was rooted in the intact cortical networks that support language. By contrast, we learned from Henry’s case that medial temporal-lobe structures are required for learning new lexical information, as we shall see in his inability to learn words that were not in his preoperative vocabulary.
How did Henry’s preoperative semantic knowledge survive over time? Was he able to retain these memories as well as people without brain damage? In healthy people, semantic memory is less vulnerable to the ravages of time than is episodic memory. In fact, experiments conducted in the 1960s indicated that older adults often scored higher than young adults on tests of their general knowledge of the world. Of course, as you age, you have more opportunities to build up storehouses of words, concepts, and historical facts—as well as to reconsolidate information you have already learned. For instance, each time you hear or read the word espionage, you automatically access its meaning in your semantic storehouse and process it. In this way, you enrich your memory trace for espionage over time. This continual rehashing of information may explain why some semantic memories are indelible.8
We wanted to understand whether Henry had kept his preoperative semantic store intact in the same way that healthy older adults had, and whether he had shown consistent performance from year to year in retrieving word information. One of the advantages of studying him for decades was that we could compare his performance on the same IQ tests given over and over again. No studies had examined the stability of memory for words over time in amnesia, so we broke new ground when in 2001 we reviewed forty-eight years of Henry’s test results.
Researchers in my lab analyzed Henry’s test scores from twenty testing sessions carried out between August 24, 1953 (the day before his operation when his brain was still intact) and 2000. This analysis evaluated his performance on four subtests from a standardized IQ test: General Information (Who wrote Hamlet? On what continent is Brazil? How many weeks are there in a year?); Similarities (In what way are an eye and an ear the same?); General Comprehension (Why is it better to build a house of brick than of wood?); and Vocabulary (What does espionage mean?). We found that Henry’s performance on these four tests was consistent across forty-eight years. His memory for facts, concepts, and words remained constant from the day before his operation through 2000, indicating that medial temporal-lobe structures were not critical for retaining and using the word knowledge and concepts he had consolidated before his operation. Importantly, the results showed that Henry’s brain could maintain information it had already learned, without explicit practice. Because of his amnesia, he did not have the benefit of episodic learning, but he could still preserve word knowledge by engaging brain circuits outside the hippocampus, in his frontal, temporal, and parietal lobes. He believed that doing crossword puzzles helped his memory, and perhaps he was correct.9
This kind of retrospective study was possible only because we had been collecting detailed information about Henry’s semantic knowledge for decades. A related goal of our research was to further examine his semantic memory as thoroughly as we had scrutinized his episodic memory. We wanted to go beyond the routine assessment provided by standardized tests, and make sure we did not leave any hidden corners of his memory unexamined.
In 1970, memory researchers in England proposed th
e idea that the memory difficulty experienced by amnesic patients boiled down to an abnormality in retrieval—they could store new memories normally but were simply unable to consciously bring them back. The researchers further argued that ostensibly forgotten information could be lured out of amnesic patients by giving them prompts. If that were true, then giving Henry prompts related to postoperative material should have catapulted his performance into the normal range. A 1975 experiment in our lab tested this proposal. A graduate student working with Hans-Lukas Teuber designed a famous-faces test using news photographs of public figures who had been eminent at various points in time from the 1920s to the 1960s. He first asked Henry to identify these people without any hints. If unsuccessful, Henry received two kinds of prompts to help him—circumstantial and letter. For instance, the circumstantial prompt for Alfred Landon was, “He was the Republican presidential nominee in 1936; he ran against Roosevelt and lost; he was also Governor of Kansas.” If Henry still could not identify the public figure, he received phonemic prompts that provided an increasing number of letters in the first and then the last names of the public figures, moving from the initials and ending with almost the whole name. For Alfred Landon, the phonemic cues were “A.L., Alf. L., Alfred L., Alfred Lan., Alfred Land.” When we compared Henry’s semantic memory for the time periods before and after the onset of his amnesia, we found that he retained his memory for public figures from the years before his operation, but his memory for public figures from the period after was markedly inferior to that of the control participants. For the post-1950s public figures, the prompts were of little help. Henry had not successfully encoded, consolidated, and stored this semantic information; clearly his amnesia could not be dismissed as an instance of faulty retrieval.10
During the following decades, lab members updated this test and gave it to Henry in nine test sessions from 1974 through 2000. We then harnessed this huge dataset to determine whether he performed consistently across years of testing. When all the data were combined, we found that he was as good as, or better than, healthy control participants when asked about people who were famous in the 1920s through the 1940s but was woefully inferior to control participants for the 1950s through the 1980s. For example, without being prompted, he correctly identified Charles Lindbergh and Warren G. Harding from the 1920s, Joe Lewis and J. Edgar Hoover from the 1930s, and John L. Lewis and Jackie Robinson from the 1940s. After the 1940s, he was at a loss. He could not name Stan Musial and Joseph McCarthy from the 1950s, John Glenn or Joe Namath from the 1960s, Jimmy Carter or Princess Anne from the 1970s, or Oliver North or George H.W. Bush from the 1980s.
Still, it was apparent that Henry was storing some new information. With prompting, he was able to recognize a few public figures whose fame began after 1953, but the number of prompts he required was fifty percent greater than the average for the control participants. So, even with generous cueing, Henry’s memories were difficult to tease out. Clearly, more postoperative material is stored than can be retrieved without extensive prompting, but if a general failure of retrieval were the basis of Henry’s amnesia, then this deficit should have compromised the retrieval of preoperative material as well. The fact that it did not, reinforces the view that amnesia is rooted in an inability to continuously consolidate, store, and retrieve life’s experiences.
Even though Henry’s case provided proof that the medial temporal lobes were necessary to form both kinds of declarative memory—episodic and semantic—the idea was not accepted without controversy. Some researchers doubted that new semantic learning depended on medial temporal-lobe structures the way episodic memory did. In 1975, two clinicians in Toronto proposed that amnesic patients have damage to brain structures that support the acquisition and retrieval of episodic memories, but not to structures that mediate acquisition and retrieval of semantic memories. In 1987, one of them argued that amnesic patients should be able to learn new facts whenever retrieval of the facts did not depend on having explicit recollection of the specific event during which the patient learned the fact. This scientist predicted that patients like Henry could acquire general knowledge outside of conscious awareness, through nondeclarative memory circuits.11
In 1988, my colleagues and I put this view to the test by trying to teach Henry new vocabulary words. We wanted to see if he could learn the definitions of eight words that were in the English dictionary but were not in common use: quotidian, manumit, hegira, anchorite, minatory, egress, welkin, and tyro. These were all words, we suspected, that Henry would not have encountered preoperatively. Henry viewed these words on a computer monitor, one at a time, with a single definition. He read each word and its definition aloud. He then saw all eight definitions with one of the eight words below and had to select the correct definition for that word. If he answered correctly, the definition was removed from the choice list, and a new word appeared at the bottom of the screen. If he answered incorrectly, he was asked to select another definition. This procedure continued, word by word, until Henry selected the correct definition for each of the eight words. Control participants were, on average, able to match all of the words with the definitions in fewer than six trials. Henry, however, failed to learn these new vocabulary words in twenty trials.12
We persisted and tried to teach Henry the same words using two additional methods—giving a common single-word synonym for each of the eight words and filling in the blank in sentences in which one of the eight words was missing. Henry apparently knew the meaning of tyro (presumably he had learned it prior to his operation) because he selected the correct definition and synonym every time, and filled in the blank correctly on ninety percent of the trials. But he never mastered the meaning of any other word, providing unequivocal evidence that he could not learn, in a controlled laboratory setting, the meaning of any new words. In contrast, the control participants continued to absorb the novel words in fewer than six trials.13
Yet one could argue that these experiments did not mimic the natural way that people learn the meaning of new words in their everyday lives. Perhaps our laboratory setup was too artificial and did not sufficiently tap into Henry’s true ability to acquire semantic information. On a daily basis, we are exposed to new words in a variety of meaningful and relevant contexts in numerous situations. We often see or hear these words in pursuit of a goal, so we are motivated learners. Picking up on this idea, another team proposed in 1982 that the decisive test of whether an amnesic patient could acquire new vocabulary words would be to take the person to a country where he did not speak or understand the language. The researchers predicted that the patient would learn the new tongue slowly, as a child would, while afterward forgetting that he had even been there. This would be a more natural learning environment than a laboratory because it would combine hearing, speaking, reading, and writing the language. Rather than trying to learn words as Henry did in the lab, the amnesic person in a foreign country might learn phrases and sentences in meaningful contexts, such as the bakery, pharmacy, coffee shop, or park. According to the argument, he might, with repeated exposure, be able to build up a rich mental representation of the language, consisting of its speech sounds, vocabulary, concepts, and grammar.14
We sensed that this theory was wrong, but in order to prove it we needed to see if Henry had incidentally acquired some knowledge of words that were new to the English language since his operation in 1953—words that he may have encountered in daily life. Even if he could not remember the definition, it was possible, for example, that he knew a new word was a real word without knowing its meaning. This intuition is common in healthy people.
We set out to assess Henry’s knowledge of new words that had been added to the Merriam-Webster Dictionary after 1954, words he probably encountered after the onset of his amnesia. The test stimuli were words like charisma, psychedelic, granola, Jacuzzi, and palimony. They were intermixed with old words (butcher, gesture, shepherd) and pronounceable nonwords (phleague, thweige, phlawse). We wanted to know whether Hen
ry considered the post-1954 words and the pronounceable nonwords to be legitimate words. Each trial began with the following question on the computer screen, “Is the following a real word?” Henry read the word and responded “yes” or “no.” He was right if he answered “yes” for the legitimate words and “no” for the nonwords. He correctly said “yes” to ninety-three percent of the pre-1950s words (which was normal compared to the controls’ ninety-two percent) and fifty percent of the post-1950s words (which was impaired relative to the controls’ seventy-seven percent). His ability to categorize nonwords as nonwords was borderline normal—eighty-eight percent versus the control group’s score of ninety-four percent. This relatively simple experiment strengthened the distinction between Henry’s preoperative semantic knowledge, which was intact, and his postoperative semantic knowledge, which was severely lacking.15
To explore another dimension of Henry’s semantic knowledge, we designed a test to measure how much he knew about public figures. A minimal expression of knowledge for the names of famous people would be the ability to recognize a famous name as such. That is what we asked Henry to do in a name-categorization task—“Is or was the following a famous person?” Henry answered “yes” or “no.” The celebrities included movie stars, athletes, American politicians, foreign leaders, and writers. On this task, the names of people who became famous either before or after Henry’s operation were intermingled with similar names chosen from the Boston area telephone directory. Henry matched the control participants in rejecting the non-famous names as famous, and he was slightly better than the controls in identifying famous personalities in the 1930s and 1940s (eighty-eight percent for Henry and eighty-four percent for the healthy controls). For the people who became famous in the 1960s, 1970s, and 1980s, and who were almost certainly unknown to Henry before his operation, his score of fifty-three percent was well below the controls’ eighty percent correct. This pattern of results—intact memory of public figures for the preoperative period and impaired memory for such people from the postoperative period—reinforced the conclusion that Henry could not successfully store and retrieve factual information that had entered the world after the onset of his amnesia.16
Permanent Present Tense Page 29