Brain Plasticity and Recovery from Stroke

Nina P Azari & Rudiger J Seitz. American Scientist. Volume 88, Issue 5. Sep/Oct 2000.

As you read this page, your brain is engaged in several tasks. It observes the words and decodes their meaning, it senses the weight and texture of the magazine and controls the muscles in your hands to hold the page at the proper angle with the appropriate pressure, and it coordinates your head and your eyes so that the whole process of reading doesn’t make you queasy. The brain is able to perform all of these tasks (and much more) at the same time because it is not merely a homogeneous blob of cells. It is made of separate parts-neural networks-that are specifically dedicated to doing each task independently Such a division of labor has obvious benefits, and it is performed so seamlessly that we never need to think about it-until something goes wrong.

When a particular neural network is damaged, as often happens in a stroke, the system fails and function is lost because no other neurons in the brain “know” how to do the task formerly performed by the damaged network. Thus, the result may be paralysis or the loss of speech or the inability to comprehend speech or any one of a number of actions we take for granted until we can’t perform them anymore. Curiously, however, many people who have suffered a stroke regain some or most of the lost function after a brief recovery period, sometimes in a matter of weeks.

The story of a 25-year-old woman who was an accomplished pianist offers a dramatic example. The young woman had a stroke that damaged the left hemisphere of her brain. Because the left hemisphere controls the right side of the body, she lost complete use of her right hand and was unable to speak (because the left hemisphere tends to dominate language functions). It was a devastating loss at a young age, and her inability to play the piano only added to the tragedy. She was placed in a therapeutic program that called for repeated attempts to engage the right side of her body, including speech therapy and piano playing. After several months in therapy her diligent attempts were rewarded: She regained nearly full use of her right hand, and she was again able to speak. Remarkably she also demonstrated exceptionally rapid finger movements in both hands, displaying speed and coordination beyond those of the average (non-stroke-affected) person. Today she has resumed her piano playing and has fully recovered her abilities to the virtuoso levels attained before the stroke.

Scientists and physicians have long noticed such astonishing recoveries. The brain’s capacity to respond to injury in these instances has always posed a mystery: Since the pertinent neural systems are permanently damaged, it is not clear how the patients’ brains are able to compensate for the loss. The question also seems impossible to answer. How does one look inside the skull of a living person to see how the brain might be changing in response to an injury?

The answer appeared in the past few years with the development of new technologies that allow scientists to observe neural activity in the living brain. Two of these methods-positron emission tomography (PET) and functional magnetic resonance imaging (fMRI)have been especially valuable in the clinical neurosciences. We have employed these “functional imaging” techniques in our own studies of people who have suffered a stroke. By looking inside the brain of a recovered stroke patient while he or she performs a particular task, we have been able to see how the brain compensates for its injury.

Modular … Yet Plastic

The capacity of the brain to reorganize itself-its “plasticity”-in the process of learning a task is perhaps the most interesting phenomenon that distinguishes the nervous system from all other tissues in the body The brain’s plasticity appears to be greatest when we are young (from infancy through early adolescence), a time when many of the neural pathways that will be used for the acquisition of language and motor skills are formed. Our ability to learn new languages and new skills as adults indicates that the brain retains a certain level of plasticity throughout our lives (although our potential may be less).

The neural changes associated with learning in the uninjured brain generally involve neural networks-specialized modules-that are designed for a given task. Neural networks involved in specific functions are almost always found in the same part of the human brain. For example, a region involved in the production of speech (Broca’s area) is found in the same part of the cerebral cortex in the left hemisphere of the brain of most people. The small percentage of people (mostly left-handers) who don’t have Broca’s area in the left hemisphere have it in the same (homologous) part of the cerebral cortex in the right hemisphere. Similarly, the regions of the cerebral cortex concerned with control of the muscles (the motor cortex), the tactile sense (the somatosensory cortex) and the visual system (the visual cortex) are in pretty much the same place from one individual to the next. These regions may develop to differing degrees through our interactions with the environment, but their locations in the brain have a strong genetic component.

In spite of this tendency to partition tasks into specialized modules, it now appears that the brain can also allocate functions outside the traditional boundaries-like a creative young child, it can “color outside the lines”when it needs to. Such instances of “adaptive plasticity” have been found to occur when the brain must compensate for neuronal damage produced by chronic degeneration (as in aging or dementia). PET studies by scientists in Canada and the United States show that elderly people process less visual information in the primary visual cortex than younger people do, but more in the pre-frontal cortex (a nonvisual site). The recruitment of nonprimary cortex is even more pronounced in people with Alzheimer’s type dementia. It has been suggested that these compensatory changes may underly the changes seen in task-solving strategies among the elderly.

Other PET studies reveal adaptive plasticity in response to acute events (such as a stroke or head trauma) and congenital conditions. Wolf-Dieter Heiss and co-workers at the MaxPlanck Institute of Cologne found that adults who had language difficulties (aphasias) caused by a stroke in the language-dominant left hemisphere were able to recruit homologous areas in the right hemisphere to perform language tasks. Leonard Cohen and his colleagues at the National Institutes of Health showed that people who are blind from birth engage parts of the visual cortex when they read Braille or perform other tactile discrimination tasks. Since the visual cortex is not engaged in normal-sighted people when they perform the same tasks, the authors suggest that the recruitment of “extra” cortex may partially explain the superior tactile sense of blind people.

Taking a Detour

These compelling studies of functional recovery and neural plasticity raised some questions in our minds. Are there specific parts of the brain that compensate for neuronal losses when a particular region is damaged? What sorts of neural networks take the place of the damaged networks? We explored these questions in a series of studies on stroke patients who had lost and then regained motor control to one side of the body.

In one study we looked at seven patients who had sustained damage to the primary motor cortex after a first-ever stroke. The study took place after the patients demonstrated significant recovery in the stroke-affected hand, about six months after the stroke occurred. The patients were scanned while they performed (blindfolded) a simple sequential finger-movement task. When they performed the task with the unaffected hand they demonstrated a normal pattern of neural activation, engaging several areas in the unlesioned hemisphere of the brain: motor cortex, premotor cortex, somatosensory cortex, parietal cortex and the supplementary motor area. In contrast, when they performed the task with the affected hand they showed an abnormal pattern of activity, recruiting premotor cortex and the supplementary motor area in both hemispheres of the brain as well as prefrontal cortex in the lesioned hemisphere.

The abnormal pattern of neural activity involved cortical regions that were not normally recruited for this kind of finger-movement task. Nevertheless, these areas were active during complicated motor tasks in the normal brain. In that sense, these patients recovered the use of their affected hand by recruiting cortex that is within the system that was damaged-that is, the motor system.

The abnormal recruitment of cerebral cortex by the stroke patients raised the question of whether the recovered brain also used different pathways to control the affected hand. In the normal brain, neurons in the motor cortex send long projections to the spinal cord (the corticospinal fibers or pyramidal tract), which excite spinal motor neurons that effect muscle contractions in the hand. Since this direct route from the cortex to the spine is compromised in the stroke patients, they must be using some other pathway.

In an atttempt to answer this question we analyzed neural activity (using a PET scan technique) in 21 stroke patients who had lesions involving the motor cortex. All of the patients had suffered severe paralysis to one hand, but 12 of the patients recovered function within about four weeks. Our PET study suggested that, instead of the pyramidal tract, the recovered patients were using a compensatory path from the supplementary motor area to the spinal cord. This compensatory path was also accompanied by abnormally enhanced connections between the thalamus and the cerebellum (both parts of the brain’s motor system). The cerebellum provides critical information about movement control to the supplementary motor area by way of the thalamus. The unusually strong “cerbello-thalamo-cortical” pathway we observed in the recovered patients may serve as a “resetting” mechanism for the compensating neural pathway. That is, enhanced signals from the cerebellum about demands for movement may reset the normal operating mode of the supplementary motor area, making it available for compensatory plasticity. So the brain appears to have the ability to seek out a detour if the “main road” is blocked.

Cross-Modal Plasticity

As we continued to study people who had recovered from stroke, we observed that some patients also recruited networks in areas of the brain that were not normally involved in the motor system. That is, the patients engaged parts of the visual cortex (as well as unusual parts of the motor cortex) when they moved the fingers of the stroke-affected hand. We should emphasize that the patients had been blindfolded, so they were not receiving any visual stimulation. Normal blindfolded subjects do not engage the visual cortex when performing the task. Thus these patients had recruited networks outside the damaged system, an alternative network, as well as cortex within the system that was damaged.

The patients who recruited an alternative network had been recovering for at least six months, whereas patients who recruited within-system networks had been recovering for merely four weeks. This distinction suggested a time course to recovery: Within-system networks are recruited fairly early (about four weeks) after a stroke, whereas alternative-system networks need a longer time to be effective.

It is important to point out that the method we used in these studies (a network analysis, see box, page 428) indicates that the active areas are functionally associated. They are not simply isolated areas of activity. Thus the recruitment of visual cortex in late-stage recovery from stroke involves a “crossmodal” adaptive plasticity: Visual cortex appeared to be subserving a motor function. This cross-modal plasticity may be akin to that seen in blind patients who engage the visual cortex for the tactile sense.

These imaging studies suggest that the visual cortex is critical to crossmodal plasticity. This part of the human brain has now been shown to function in motor and sensory roles, as well as in the visual mode for which it was designed. But the visual cortex alone may not be able to reorganize its function without the demand from another system that has become incapable. Future imaging studies may be able to address this issue.

A Model of Recovery

Many studies have shown that stroke patients require time to regain function. During this time the brain is evidently sorting out how it might compensate for the damaged neurons, perhaps “seeking out” alternative routes or detours. So although no neurons “know” immediately how to compensate for the lost network, apparently some of them can learn to do so.

The process of neural recovery can be divided into several stages. As other studies have shown, initially there is a passive tissue response in the first few hours and days following brain-tissue injury This passive response involves the reperfusion of ischemic tissue and cessation of inflammatory processes secondary to brain damage. This leads to a regression of dysfunction associated with a temporary “shock” to the neurons near (and away from) the lesion. Medical intervention that facilitates these processes can determine the extent to which recovery will proceed in the subsequent stages.

In the days and weeks following a stroke the brain begins active processes of recovery involving adaptive plasticity. In the early stages this may include within-system pathways if any have survived undamaged. Such pathways may normally play a mere supporting role in the undamaged brain. For instance, the within-system pathways may only be engaged by normal adults when they are first learning the task. These same pathways come to play central roles when needed in adaptive plasticity. Since there is already a general familiarity with the task, the within-system pathway must only relearn how to perform it. This may explain why recovery is often seen within a few weeks following a stroke.

If, however, there is complete damage to a neural system, then the brain may still have the capacity to recruit an alternative system, one not generally activated for the task by normal subjects. In such instances the alternative system is naive to the task so that the patient must relearn not only how to do the task but also what to do in the first place. This takes more time, and may explain why alternative-system pathways have not been heretofore evident in the functional imaging studies.


The existence of distinct stages in the recovery process has only become evident through the use of functional imaging techniques. As the technology develops we have little doubt that we will come to appreciate progressively finer aspects of adaptive plasticity and its role in a patient’s recovery from brain lesions such as stroke.