Permanent Present Tense

by Suzanne Corkin

  Although Henry lacked this declarative knowledge (awareness), he eventually demonstrated trace conditioning, requiring ninety-one trials. Thus, the mechanism that supported Henry’s learning must have been different. The California memory researchers suggested that the cerebellum, while necessary for delay and trace conditioning, could not sustain a representation of the tone across the half-second trace interval. In Henry’s case, then, that contribution must have come from a representation of the tone in his intact auditory cortex, which allowed him to acquire the conditioned responses—his nondeclarative knowledge.9

  Our eyeblink classical-conditioning experiments demonstrated the plasticity in Henry’s brain. Through these procedures, he could achieve associative learning, connecting a tone with an air puff to his eye. This nondeclarative learning was involuntary, confined to circuits that operated outside the realm of his conscious awareness. In contrast, if he explicitly tried to associate the tone and the air puff, he would fail, just as he would be unable to associate his doctor’s name with his face. He did not have relational, declarative memory circuits to rely on for that task, but he still maintained networks that allowed him to acquire, without conscious recollection, two different kinds of conditioned responses—delay and trace.

  Like classical conditioning, perceptual learning is also expressed through performance of a task. Perception in the visual system is the mind’s ability to detect movement and to identify by sight objects, faces, shapes, textures, orientation of lines, and colors. In the same way, the sense of touch allows the mind to appreciate roughness, temperature, shape, texture, and elasticity.

  Perceptual learning is different from perception. It occurs on top of the basic processing of stimuli. Perceptual learning is the ability to identify something more accurately and effortlessly after training, and it occurs incidentally and without conscious awareness of learning. The fine-tuning of perception through experience is apparent in just about any area of life, from the hobbyist who knows every make and model of antique cars, to the quality-control manager of an assembly line who can spot defects in an instant, to the radiologist who can identify a cancerous tumor from the shadings in an MRI image.10

  My colleagues and I were curious to see whether Henry’s medial temporal-lobe lesions would allow him to acquire new perceptual information without being aware that he was doing so. We addressed this question in 1968 when Milner gave Henry a test of perceptual learning, the Gollin Incomplete Pictures Test. This test was not directed at Henry’s visual perception, but rather at his ability to identify a less complete version of a picture the second time he saw it compared with the first. The task entailed viewing simple line drawings of twenty common objects and animals, such as an airplane and a duck. Henry viewed each object in five degrees of fragmentation. The test began with a very sketchy representation containing a few parts of each object, which was almost impossible to interpret, and ended with a complete, recognizable picture. Henry first saw the most fragmented set, one drawing at a time, each for about a second, and said what he thought the drawing might represent. He then saw progressively complete sets of pictures until he could name all twenty objects (see Fig. 17).11

  Milner administered the Gollin test to Henry on two consecutive days, with these instructions: “I’m going to show you some pictures that are incomplete. I want you to tell me what the figure would be if it were completed. Guess if you are not sure.” After a short practice test, she showed him the first twenty cards, the most difficult, and noted his errors. She then presented a less-fragmented version of the pictures in a different order so that Henry could not anticipate which drawing would appear next, and told him that this time they would be a little easier to identify. This procedure continued, with the drawings more complete on each trial, until Henry identified the twenty pictures. He completed the test without errors after four trials, and remarkably, his accuracy was a bit better than that of the ten control participants: he misnamed twenty-one pictures, whereas the controls, on average, misnamed twenty-six.12

  We knew from other testing that Henry’s visual perception was excellent, but would this first encounter with the pictures benefit his performance the next time he saw them? Would he show perceptual learning? An hour later and without warning, Milner showed Henry the same set of pictures. Henry did not remember having taken the test before; nonetheless, he identified the fragments in fewer trials.13

  Still, Henry did not show as much improvement as the controls did. Why not? The control participants had an advantage over Henry: they retained the names of the pictures in their long-term memory, so they had a menu of the correct names to choose from as they viewed the fragmented pictures a second time. They knew, for instance, that one of the pictures was of a duck, so when they saw a few fragments that suggested a beak and tail, they guessed duck. But Henry did improve from trial to trial, and amazingly, when I gave him the same test thirteen years later, his identification was even more accurate. Although he still had no conscious memory of ever having seen the pictures, he had learned a perceptual skill without explicit knowledge, and it stuck—solidly stored in the preserved cortical areas of his brain.14

  We now understand much more about how certain parts of the brain detect and classify information. In the early 1990s, studies revealed, for instance, that one brain area is dedicated to face processing and recognition. Using a functional imaging technique called positron emission tomography (PET), a cognitive neuroscientist at the Montreal Neurological Institute asked participants to identify faces, and found regional increases in cerebral blood flow—indicating heightened neural activity—in areas within the temporal cortex dedicated to the processing of visual information. Five years later, a cognitive neuroscientist at MIT turned to a brain-imaging method that gives more precise pictures of the brain than PET. She developed functional MRI protocols to define the limits of this face-selective area in the temporal lobe and establish its function, christening it the fusiform face area. This area was not damaged in Henry’s brain, so he could still recognize his parents, relatives, friends, and celebrities after his operation; he had stored those images in his long-term memory preoperatively. If we had shown Henry a series of unfamiliar faces inside an MRI scanner, his fusiform face area would have been active while he was looking at them. But after he left the scanner, he would not remember them, because he lacked the necessary areas in his medial temporal lobe to form these new memories.15

  The MIT researcher’s pivotal discovery inspired a team of scientists at Vanderbilt University to conduct further functional MRI studies, showing how the brain maps other kinds of expertise. They found that extensive knowledge of birds or cars also recruited the face-selective area in the brain. In the MRI scanner, all participants saw pairs of cars and pairs of birds, and judged whether the cars were the same model but different years, and whether the birds belonged to the same species. When the researchers compared brain activity associated with cars and birds in the two groups, she found greater activation for cars than birds among the car aficionados, and greater activation for birds than cars among the bird experts. The car- and bird-expertise effect occurred in the same cortical area as face expertise, suggesting that activity within this small area is differentially focused to support several specializations—face recognition and expert recognition of objects.16

  These experiments illustrated plasticity within individual human brains—the dedication of neurons in a precisely defined area resulting from long-term perceptual learning with specific objects, such as faces, cars, and birds. This ability is fundamental to successful interactions with other people and our environment. After his operation, Henry could still perceive faces, cars, and birds, and could show normal perceptual learning with the Gollin pictures—capacities that relied on his intact visual cortex. But those processes, by themselves, were insufficient for him to remember, in the everyday sense of the word, new faces and objects.

  We have continued to learn more about how our brains learn and classify inf
ormation. In 2009, neuroscientists identified white-matter pathways connecting the visual areas that support face and object processing with the amygdala and hippocampus. We will examine Henry’s autopsied brain to confirm the integrity of these connections. We assume, however, that these pathways to the medial temporal-lobe structures were intact, so information about faces and objects would have reached his medial temporal-lobe structures if they had not been removed. Henry lacked the machinery to receive, encode, and consolidate face information as a memory.17

  Not every kind of nondeclarative learning requires repeated exposures to a stimulus or procedure. Repetition priming can occur after one learning trial. In the lab, when Henry looked at a series of words, pictures, or patterns and encountered them a second time in a subsequent test, his perception of them or his response was often facilitated due to his earlier exposure. This enhanced processing is referred to as repetition priming—his brain was “primed” to respond in a certain way to a stimulus because he had seen it previously. Even when he was not intentionally trying to recall the past, his experience unconsciously influenced his memory.18

  Repetition priming occurs frequently in our daily lives but usually goes unnoticed. We may hear a song on the radio first thing in the morning and then find ourselves humming it during the day without knowing why. Priming is a favorite tool of the advertising industry. Frequent exposure to particular brand names on TV or in magazines may prime us to process them more and thus select them above other brands even if we do not consciously remember seeing them in advertisements. Political campaigns also take advantage of priming: little-known candidates can become popular overnight if constituents see and hear their names repeatedly. When we read these names on our voting ballot, we may mistakenly think that the candidates are seasoned politicians with formidable track records, simply because we process their names more easily.

  In the mid 1980s, we became interested in examining in detail Henry’s ability to prime because we wanted to know whether this form of memory was resistant to amnesia and whether different kinds of priming were equally robust in amnesia. Another focus of our research was demonstrating whether the priming effect was comparable when the test items were familiar to Henry, compared to those that were novel.

  We explored these topics in a series of experiments carried out in the late 1980s and throughout the 1990s, using a variety of priming tasks. In every case, the test had two parts: a study phase, in which Henry was exposed to words or pictures, followed by a test phase, in which he performed a task with studied and unstudied words or pictures. For example, in one study phase, we showed him a list of words on a computer screen, one at a time, and asked him to say “Yes” if the word had the letter A in it and “No” if it did not. This instruction led Henry to believe that we were simply testing his ability to detect As, so he did not think that this was a memory test.

  EPISODE

  FACULTY

  RADIUS

  STOVE

  CALCIUM

  ROUGH

  CLAY

  STAMP

  FROST

  Next, in the test phase, we showed Henry the first three letters of these words interspersed with the first three letters of comparable words that were not in the study list.

  CLA

  SER

  CAL

  ROU

  MED

  TRO

  EPI

  FAC

  SWI

  RAD

  BRE

  REC

  We told Henry that each three-letter stem was the beginning of an English word, and asked him to make each stem into a word. We encouraged him to write the very first word that popped into his mind, and did not mention the study list. Henry remained unaware that his memory was being tested.

  The studied words were not the most common completions of the stems in that they were not among the three most popular responses given in a pilot study in which we asked healthy participants simply to complete stems with the first word that came to mind. Common completions of the stems CLA, CAL, and ROU included clap, calendar, and round. CLAY, CALCIUM, and ROUGH were more unusual choices. Remarkably, after a single exposure to the study list, Henry gave the less-common completions, indicating a priming effect. His priming score took into account the number of items he would get correct just by chance. It was the number of times he completed the stems to the studied words minus the number of times he completed other stems to comparable unstudied words—words similar to the studied words in number of letters and frequency of occurrence in the English language. During the test, priming occurred as the result of activation in Henry’s brain of the representation of the word he had just encountered in the study list.19

  We compared Henry’s performance on this nondeclarative memory task with his scores on two measures of declarative memory, in which his task was to consciously recollect studied words in a similar format. He viewed the study list on a computer screen as before, and after a short delay, we asked him to orally recall the words he had just seen. Next, we gave him a recognition-memory test, in which three words appeared on the computer screen, all of which began with the same three-letter stem, for example, CLAY—CLAM—CLAP. Henry’s task was to select the studied word from among the three choices. He was impaired on both measures—recall and recognition.20

  The critical difference between these declarative tasks and the nondeclarative priming task was in the instructions. For the recall and recognition tests, we asked Henry to intentionally retrieve words from the study list—a memory test in the traditional sense. These results showed that he activated separate neural networks for declarative and nondeclarative learning. He failed the recall and recognition tests, showing that his declarative circuits were faulty, but performed normally on the word-completion priming test, proving that his nondeclarative circuits were preserved.21

  What brain mechanism allows people with amnesia to show normal priming? The first hints at an explanation came in 1984, when psychologists at the University of Pennsylvania made an astute observation during casual conversations with amnesic patients. The researchers noticed that densely amnesic patients, after lengthy exposure to a particular word or concept—for example, dogs or types of dogs—followed by fifteen seconds performing another task, would claim to have no memory for any particular conversation, and no idea what the topic of conversation might have been. But if the researchers then asked the patients to initiate a conversation on any topic they wished, they were likely to choose a topic or mention a word from the previous discussion—dogs or terrier, for example—even though they did not recognize the link between the new discourse and the prior conversation.22

  The researchers speculated that amnesic patients’ normal performance on priming tests resulted from trace activation, the excitation of intact mental representations—symbolic codes for information. They proposed that when participants read words out loud, such as CANDLE, PLEASANT, and BUTTON, they activate a mental depiction of that word. This activation lasts for minutes or hours—just as your hair dryer stays hot for a short time after you have switched it off—and it occurs similarly in normal and amnesic participants. On a subsequent test, when participants have to complete CAN, PLE, and BUT to the first word that comes to mind, the words CANDLE, PLEASANT, and BUTTON are highly activated and, therefore, more likely to be selected than are other possible completions.23

  In the mid 1980s, when my colleagues and I began to study the repetition priming-effect, we had several goals. One was to examine nonverbal priming by using unfamiliar patterns as test stimuli. Most demonstrations of intact priming in amnesia had used verbal tasks, such as reading, spelling, or word completion, but a larger theory of the nature of priming in amnesia would have to encompass information other than words. When amnesic patients see words, they can draw on the knowledge about those words that they acquired before they became amnesic: the stimuli are already stored in their mental dictionary, and can be activated and thus primed. But what about information t
hey are seeing for the first time? It was possible that amnesic patients would show intact priming only when they had knowledge of the primed response—when they already possessed a normal representation of the stimulus. Researchers could easily identify knowledge of words as the basis of verbal priming, but it was less clear what constituted a knowledge base for nonverbal (pattern) priming.

  In 1990, members of my lab set out to discover whether Henry would show normal priming when the stimuli were patterns drawn on paper. We created six target figures by connecting five dots from the nine possible dots in a three-by-three square matrix. We then asked Henry, along with a group of control participants, to draw any figure they wished, using straight lines to connect the five dots in each of the six dot patterns. These figures constituted the participants’ baseline figures, indicating what figures they would choose to draw spontaneously on their own. The priming test came six hours later. In the study phase, participants received a sheet of paper with the six target figures and were told to copy these figures onto corresponding dot patterns on the same page. We then removed this sheet of paper, and participants performed a distractor task for three minutes—writing down the names of as many famous entertainers from the twentieth century as possible (see Fig. 18).24