Humans

In discussing long-term memory, scientists have found it useful to distinguish between several kinds of memory that rely on different brain systems. The major distinction is between declarative memory, which refers to the conscious memory of facts and events, and nondeclarative memory, which refers to nonconscious memory of skills, habits, or other modes of learning that proceed beneath the surface of conscious awareness. Declarative memory is what most people call memory. It depends on the integrity of the hippocampus and related structures of the medial temporal lobe. Nondeclarative memory affects our behavior without our explicit knowledge. It is a heterogeneous collection of nonconscious memory abilities that depend on various other structures within the brain (Squire et al., 1993; see Figure 1).

One well-studied component of nondeclarative memory is procedural memory. The difference between declarative memory and procedural memory is the difference between "knowing that" and "knowing how." Procedural learning describes the formation of skills and habits. It is the most primitive form of learning, the first to develop in infancy (Tulving and Schacter, 1990). Because it requires extensive practice, it is a slow and inflexible learning system that eventually takes on an automatic or reflexive quality. It is, however, long-lasting and reliable, as any bike rider knows—even after years of absence from a bicycle, one never loses the skill. While most declarative learning is not impaired by rapid-eye-movement (REM) sleep deprivation, procedural learning is (Stickgold et al., 2001).

Skills, the procedures that allow us to function in the world, include motor, perceptual, and cognitive processes. Examples of learned skills are driving a car with a manual transmission (motor), a parent's attentiveness to his or her baby's cry in a distant room (perceptual), and increasing alacrity in solving a Rubik's Cube with practice (cognitive). Habits are a form of gradual, incremental learning, a settled pattern of responses toward repeated stimuli. An example of a habit is a person regularly opening the refrigerator door when he or she walks into the kitchen. Like skills, habits allow us to function efficiently in the world by responding to stimuli with minimal cognitive effort. While declarative memory can, in some cases, enhance or hasten the acquisition of skills and habits, usually conscious awareness of learning is not necessary; once the information is acquired, it often becomes difficult to verbalize it. This is why procedural learning is called "nondeclarative."

The Case of Patient H.M.

By far the most famous example of a patient who has retained procedural learning in the absence of declarative memory is the case of H.M., a patient with severe intractable epilepsy who underwent surgery as a last attempt at correcting his condition. Surgeons removed most of his medial temporal lobe structures bilaterally and left him devoid of declarative memory, and therefore suffering from amnesia. Brenda Milner (1962) later showed that he was capable of improving his performance on a mirror-drawing task in which participants trace the outline of a figure with a stylus (e.g., a star) while watching the reflection of their efforts in a mirror. While initially challenging, the task becomes easier with practice. Despite not having any recollection of ever performing the task, H.M. became more accurate across sessions, demonstrating that he had acquired this skill. This was the earliest evidence that skill-learning can occur without declarative memory. There is evidence that not only motor-skill learning (as tested by the mirror drawing task) but also perceptual and cognitive-skill learning can also proceed in the absence of declarative memory.

Motor-Skill Learning

Motor-skill learning is tested primarily by three tasks: the mirror drawing described above, rotary pursuit, and serial reaction time (SRT). In rotary pursuit, participants try to keep a hand-held stylus in contact with a nickel-sized metal disk that rotates on a table. With practice, participants are able to maintain this contact for longer periods of time. Suzanne Corkin (1968) has shown that patients with declarative memory deficits (i.e., amnesics) are able to improve their performance on this task. In the SRT task, participants press the corresponding button when a target item appears in one of four locations on a screen. The appearance of the target follows a ten-to-twelve trial sequence that the participants eventually learn implicitly. SRT learning remains intact in amnesics (Nissen and Bullemer, 1987).

Furthermore, both the basal ganglia and the cerebellum play a role in motor-skill learning (Gabrieli, 1998). The basal ganglia are a group of subcortical structures that surround the thalamus and include the caudate, putamen, and globus pallidus. Neuroimaging techniques have shown that SRT skill learning activates the basal ganglia (Karni et al., 1995). Lesions to the cerebellum, a structure at the base of the brain, can impair mirror-tracing skills (Sanes et al., 1990). Many scientists believe that the acquisition of motor skills depends on the basal ganglia, while the association of visual cues with motor actions depends on the cerebellum (Willingham et al., 1996).

Perceptual-Skill Learning

In addition to having intact motor-skill learning, patients with amnesia also have intact perceptual-skill learning. One of the most popular tests of perceptual-skill learning is reading mirror-reversed text. While people with declarative memory problems show improved mirror-reading with practice, patients with basal ganglia damage (e.g., people with Huntington's disease) evince mildly impaired learning (Martone et al., 1984).

Learning Cognitive Skills and Habits

Cognitive-skill learning has also been tested using several tasks, the most common of which are the tower tasks and the artificial grammar learning task. Habit learning in humans has been measured using a probabilistic classification task. The tower tasks require planning and problem solving: participants are asked to change the location of objects according to certain rules. For example, in the tower of Hanoi task, the subject is presented with three pegs with a set of disks on the leftmost peg. The goal is to move all the disks to the rightmost peg, with the disks piled in order from largest to smallest. The subject may move a disk to an adjacent peg on each move and may not put a larger disk on top of a smaller disk. The tower tasks have yielded mixed results: some researchers have found normal learning in amnesic patients under some circumstances (Cohen et al., 1985) but not others (Butters et al., 1985). It is likely that the tower tasks draw on both procedural memory and declarative memory for the consequences of particular moves that one has already tried.

The artificial grammar task requires the abstraction of rules and regularities underlying seemingly random strings of letters. The rules allow only certain letter strings to follow other letter strings. After viewing a series of these "grammatical" letter strings without being told about the rules, people are able to classify new strings as either "grammatical" or "nongrammatical" fairly accurately. However, they are not able to report much about the rules and generally feel that they had simply been guessing. Again, amnesic patients cannot remember individual training letter strings very well but perform as well as normal people when classifying new strings (Knowlton, Ramus, and Squire, 1992; Knowlton and Squire, 1996).

The probabilistic classification task that has been used to test habit learning is the "weather prediction" task, which involves a series of cues that are probabilistically associated with either a "sunny" or "rainy" outcome. On each trial participants try to guess which outcome will occur for the cues presented. Because the cues are associated with a particular outcome on only 60 to 90 percent of the trials, memorizing individual trials is not as helpful as accruing knowledge across many trials. Patients with amnesia perform normally on the probabilistic classification task, even in the absence of conscious recollection of the training episode (Knowlton et al, 1994; 1996).

Neural Substrates of Habit Learning in Humans

In neuropsychology, the best evidence for a brain region's involvement in a cognitive process is a double dissociation, in which patients with a particular lesion are impaired at task A but intact at task B, while patients with a different lesion show the opposite pattern. In humans, such a double dissociation was found by Knowlton and colleagues (1996) among amnesic patients and those with Parkinson's disease (PD). Both groups of patients were asked to perform the weather prediction task described above. Both groups were also asked some multiple-choice questions designed to investigate whether they remembered the learning situation. Patients with amnesia were able to learn the classification task but showed almost no declarative memory for the learning episode. In contrast, PD patients were able to remember details of the learning episode but showed no learning in the probabilistic classification task. Since the brain regions damaged by PD include the neostriatum (caudate and putamen structures that are part of the basal ganglia) but not the medial temporal lobes, the experimenters concluded that the neostriatum is responsible for habit learning.

The ultimate output of the basal ganglia is the frontal cortex. In turn, the frontal cortex sends a major projection to the neostriatum. This corticostriatal loop appears to be important for procedural learning. This finding is supported by reports that patients with striatal damage exhibit deficits in skill learning as well as habit learning. Furthermore, a recent neuroimaging study by Poldrack and colleagues (2001) has shown the medial temporal lobe-based declarative memory system and the cortico-striatal based procedural memory system may have a reciprocal relationship during learning. They administered the probabilistic classification (weather prediction) task to healthy young people and designed two conditions, one emphasizing the declarative aspects of the task and the other the nondeclarative aspects. The investigators found that the declarative version elicited medial temporal lobe activity, while the nondeclarative version elicited activity in the basal ganglia and other subcortical structures. Also, activity in the basal ganglia was negatively correlated with activity in the medial temporal lobes.

Conclusion

Procedural learning involves skill and habit learning, both of which are spared in the abolition of declarative memory. While declarative memory depends on medial temporal lobe structures (e.g, the hippocampus), skill and habit learning depend on the basal ganglia. Other structures, such as the cerebellum, may also play a role in some forms of procedural learning. The development of new procedural learning tasks that can be used with brain-injured patients or in neuroimaging studies will help elucidate the neural substrates of this form of learning.

See also:DECLARATIVE MEMORY

Bibliography

Butters, N., Wolfe, J., Martone, M., Granholm, E., and Cermak, L. S. (1985). Memory disorders associated with Huntington's disease: verbal recall, verbal recognition, and procedural memory. Neuropsychologia 6, 729-744.

Cohen, N. J., Eichembaum, H., Deacedo, B. S., and Corkin, S. (1985). Different memory systems underlying acquisition of procedural and declarative knowledge. Annals of the New York Academy of Sciences 444, 54-71.

Corkin, S. (1968). Acquisition of a motor skill after bilateral medial temporal-lobe excision. Neuropsychologia 6, 255-265.

Gabrieli, J. D. E. (1998). Cognitive Neuroscience of human memory. Annual Review of Psychology 49, 87-115.

Karni, A., Meyer, G., Jezzard, P., Adams, M. M., Turner, R., and Ungerleider, L. G. (1995). Functional MRI evidence for adult motor cortex plasticity during motor skill learning. Nature 377, 155-158.

Knowlton, B. J., Ramus, S. J., and Squire, L. R. (1992). Intact artificial grammar learning in amnesia: Dissociation of classification learning and explicit memory for specific instances. Psychological Science 3, 177-179.

Knowlton, B. J., and Squire, L. R. (1996). Artificial grammar learning depends on implicit acquisition of both abstract and exemplar-specific information. Journal of Experimental Psychology: Learning, Memory, and Cognition 22, 169-181.

Knowlton, B. J., Squire, L. R., and Gluck, M. A. (1994). Probabilistic classification in amnesia. Learning, Memory, and Cognition 21, 699-710.

Martone, M., Butters, N., Payne, M., Becker, J. T., and Sax, D. S. (1984). Dissociations between skill learning and verbal recognition in amnesia and dementia. Archives of Neurology 41, 965-970.

Milner, B. (1962). Les troubles de la mémoire accompagnant des lesions hippocampiques bilaterales. In P. Passouant, ed., Physiologie de l'hippocampe. Paris: Centre de la Recherche Scientifique.

Nissen, M. J., and Bullemer, P. (1987). Attentional requirements of learning: Evidence from performance measures. Cognitive Psychology 19, 1-32.

Poldrack, R. A., Clark. J, Pare-Blagoev, E. J., Shohamy, D., Creso Moyano, J., Myers, C., and Gluck, M. A. (2001). Interactive memory systems in the human brain. Nature 414, 546-550.

Sanes, J., Dimitrov, B., and Hallett, M. (1990). Motor learning in patients with cerebellar dysfunction. Brain 113, 103-120.

Squire, L. R., and Knowlton, B. J. (1995). Memory, hippocampus and brain systems. In M. Gazzaniga, ed., The cognitive neurosciences. Cambridge, MA: MIT Press.

Squire, L. R., Knowlton, B., and Musen, G. (1993). The structure and organization of memory. Annual Review of Psychology 44, 453-495.

Stickgold, R., Hobson, J. A., Fosse, R., and Fosse M. (2001). Sleep, learning, and dreams: Off-line memory reprocessing. Science 294, 1,052-1,057.

Tulving, E., and Schacter, D. L. (1990). Priming and human memory systems. Science 247, 301-306.

Willingham, D. B., Koroshetz, W. J., and Peterson, E. W. (1996). Motor skills have diverse neural bases: Spared and impaired skill acquisition in Huntington's disease. Neuropsychology 10, 315-321.

Indre V.Viskontas

Barbara J.Knowlton