Issue 3, September 2004

REVIEW: Psychological & Social Sciences

A Review of Parietal Lobe Functioning in Planning and Updating Motor Movements

Andreas Rauschecker
Georgetown University

According to Blakemore et al. (2002), most disorders of movement due to parietal damage may be explained by an internal model. For example, optic ataxia is believed to occur as a result of improper fine-tuning of the inverse model (another specific type of internal model). The inverse model normally predicts the proper motor movements to be made, including those needed to grasp an object. The authors do not describe why the inverse model should be affected while the forward model is not affected, or whether these models should be located in separate areas.

Perhaps a more satisfying explanation of phantom limbs, rather than of parietal damage, is given by the internal model. Immediately after amputation, and sometimes up to years later, patients can feel their missing limb as if it was still present. Many of these people experience being able to move their phantom limb. It is surmised that a forward model would allow movement predictions, and that the efference copies of these predictions would cause the feeling of the absent limb’s movement (Blakemore et al. 2002). After some time (on the order of weeks or months), as no limb actually moves in response to motor commands (i.e., there is no visual or other sensory feedback of the moving limb), neural plasticity will cause the forward model to reflect the body’s new state. In other words, the forward model will no longer predict movement of the limb in response to a motor command, and amputees no longer will have the experience of being able to move their phantom limb. This gradual loss of feeling of phantom limb movement is, indeed, experienced by many people who have had a limb amputation.

Evidence for a forward model in the parietal cortex comes from other types of studies as well. Desmurget et al. (1999) showed that a short neural disruption via transcranial magnetic stimulation (TMS) to the posterior parietal cortex (PPC, including intraparietal sulcus and superior parietal areas) does not inhibit pointing by a subject to a target light. In fact, subjects easily made a correct pointing motion as they normally would have without TMS stimulation. However, when the target was slightly moved at the moment a subject started to point, and without the participant’s awareness (by moving the target during a saccadic eye movement, when vision is suppressed), the participant did not make corrections for the target’s new location. Instead, the subject pointed at the original site of the target. It was also shown that without the TMS pulse, subjects automatically corrected for target movement. Therefore, it was shown that the posterior parietal cortex’s comparison between current body state and expected body state is disrupted by a TMS pulse. Furthermore, the study emphasized that the disruption caused by TMS is not a problem involving target localization, because the effect is not seen with the hand ipsilateral to the stimulation site.

The study by Desmurget et al. (1999) is a good example of how visual and tactile feedbacks optimize movement. Proprioceptive signals from the arm and shoulder determine the location of the arm (or other body part), while trajectory of the arm is adjusted with visual feedback. The study’s authors emphasize the multi-modal nature of the parietal lobes that make them suitable for converting between coordinates of one’s body and the outside world (see also Farrer and Frith 2002). One criticism of the Desmurget et al. study is that this multi-modal nature was only present in four out of five of the subjects. One subject showed preserved ability to correctly point to the target location when it was moved and the subject was given a TMS pulse. The authors explained this result as a possible consequence of the anatomical variability between people. Perhaps functional imaging studies of the parietal lobes will allow us to determine whether it is anatomical variability or altered mechanisms that cause people to respond differently to moving targets. For example, a subject unaffected by a TMS pulse may have a more conscious mechanism for correcting arm movement, or perhaps the saccadic suppression followed a slightly different time course in this individual. Another possible explanation for this discrepancy could be that some individual’s may exhibit unique responses to TMS pulses. Combined TMS and fMRI studies would need to be conducted to elucidate a clear explanation.

Lastly, it is unclear from the Desmurget et al. study (1999) why the feedback loop of the internal model was affected, while that of the forward model was not affected (since subjects still correctly pointed to the location of the original target.) The authors do not speculate about this point, except to say that a forward model, created by extra-foveal information (since subjects do not look directly at the target at first), must have been present for the action to occur in the absence of feedback. The most parsimonious explanation seems to be that different areas of the brain serve as neural correlates of the original movement and comparison process (i.e., the feedback loop). If this is the case, this difference should be emphasized in the future, and studies should be conducted to elucidate these neural correlates.

One major differentiation that has been made in studies such as those described is between movements of the body that are self-initiated and those that are externally produced (whether by an actual physical stimulus or a psychological one, such as when one directly copies another person’s movement). It should be noted that most voluntary movements fall within the first category, but that some can also be in the latter group. Naturally, then, attempts have been made to use the forward model to explain delusions over control of action (Blakemore et al. 2002). Schizophrenic patients with these delusions describe experiences in which they feel that someone is controlling their movements or that they are passively experiencing their own actions (i.e., that they are not in control). Blakemore et al. (2002) described how such feelings could be evoked by problems in the forward model. If the forward model does not correctly predict sensory consequences of an action, then the actual consequences of the action are not cancelled out, as they should be, by the prediction. This, in turn, would cause the subject to feel like the action was externally-produced. An externally generated action should lead to a greater differential activation in the comparator of the forward model (hypothesized to have its neural correlate in parietal cortex), since predicted sensory consequences do not match real consequences. Interestingly, Simeon et al. (2000) showed greater activation of the PPC, specifically the angular gyrus, in a positron emission tomography (PET) study of people with depersonalization disorder. Therefore, it seems clear that people with depersonalization disorder often experience their actions as being controlled by an outside source or as outside of voluntary control. However, the verbal learning task used by Simeon et al. (2000) may not be universally accepted as it was done without any motor actions during the PET scan. This is because the forward model is generally applied only to motor movements.

Therefore, it is necessary to look to other studies, which have made the distinction between self-generated and externally produced movements, to see whether the forward model applies to this distinction. This is important because, as Farrer and Frith (2002) have pointed out, most of our actions are carried out in social environments where interpersonal relations take place. Actions need to be appropriately referred to the correct agent so that actors knows whether an action was self-initiated or caused by an external agent. As seen by disorders of depersonalization or some symptoms of schizophrenia, the distinction can be very pertinent.

One study attempted to replicate delusions of control through hypnosis in participants without neurological disorders (Blakemore et al. 2003). Functional MRI was performed on these participants under three conditions: 1) subject lifting an arm on own accord (active condition), 2) subject having an arm lifted by a pulley system (passive condition), and 3) subject lifting arm on own accord but believing (through hypnosis) that the pulley system was lifting the arm (deluded passive condition). The tasks were all executed while participants were under hypnosis, which was performed by a psychologist on the hypnosis unit of University College, London. The authors asserted that since the predictive system (i.e., the forward model) distinguishes between sensory experiences that result from self-initiated movement and those that result from outside sources, more sensory activity should be cancelled in the active condition than in the deluded passive condition. This, in turn, would lead to more activity in those areas which correlate to the forward model in the deluded passive condition. Not surprisingly, the parietal cortex (including the parietal operculum, or secondary somatosensory cortex, and the left inferior parietal cortex) was more active in the deluded passive movement than in the active movement condition. The inferior parietal cortex is the traditional hypothesized locus of the neural correlate of the forward model comparator, so the higher activity in this region would suggest an experience of not being in control of the movement. Indeed, subjects’ ratings of the degree of “voluntarity” of actions under the third case was lower than under the other two cases. It must be emphasized, however, that since neither the mechanisms nor psychological experiences of hypnosis are poorly understood, subjects may have based their ratings on their view of the expected responses.

Parietal operculum activity is seen as a sign of decreased sensation during deluded passive movement. The forward model explains that sensory stimuli resulting from movement are not cancelled out as strongly in this condition as in the active movement condition (in which efference copies of the motor command are used to predict sensory consequences of the action). In the active movement condition, efference copies of the motor command cancel out sensory stimuli to a large extent. The forward model is also used by Blakemore et al. (2003) to similarly explain why people cannot tickle themselves. In this study, the experimenters used a robotic device connected to a participant’s finger to tickle the palm of the participant’s other hand. If the participant initiated tickling by himself or herself and the tickle occurred immediately (i.e., a self-initiated stimulus), then the stimulus was perceived as less intense and less ticklish than if the time duration between the command to start movement of the robot tickler and the actual movement of the tickler was lengthened (i.e., an externally-produced stimulus). Moreover, there was less activity in the primary somatosensory cortex in the latter case, indicating a neural correlate of the subjects’ reports. Although no differential activation was found in other parts of the parietal cortex, activation did change in the cerebellum, which may have a similar but simpler mechanism of comparison than the posterior parietal cortex.

It seems that the forward model may, after all, be used to explain the neural mechanisms underlying perceptions of self-initiated versus externally-produced movements and sensations. In schizophrenic patients experiencing passivity symptoms as those described above, both the parietal cortex and cerebellum are overly active, and this activity subsides when those symptoms decrease (Spence et al. 1997). Moreover, evidence from the animal literature corroborates these findings. It was shown that single neurons in the parietal cortices of cats were inhibited more strongly and often during self-initiated movements (i.e., in the absence of a conditioned stimulus) than during the same movements made in response to a conditioned stimulus. A large part of this inhibition was observed before onset of the movement, suggesting that these areas are at least partly involved in preparation of the movement (Khitrove-Orlova et al. 1997).

The story seems clear at this point, but some problems remain. For example, Farrer and Frith (2002) have recorded the neural activation of subjects when they used a joystick to move a dot along a T-shaped tube. In some conditions, the dot was controlled by the participant, while at other times it was controlled by a computer (with or without the participant knowing.) The neural activations arising from the self- and other-attributed conditions were compared, and each were also compared to a control condition, which just required the participant to watch the dot move either left or right. The authors did not point out, however, that some of the activations in relevant areas (i.e., the angular gyrus and right anterior insula) only appear when comparing other-attribution to self-attribution conditions, and not when comparing either of these conditions to the control condition. Therefore, it is possible that the difference between neural activation levels of these two conditions and the control condition was statistical insignificant. Yet, this explanation does not seem sufficient. One can only postulate about why one must compare the two conditions; a primary reason may be the involvement of the subject in the control task, which may easily have been nonexistent (i.e., no attention paid by the subject to the control stimulus). Moreover, the study can also be criticized for the dot moving away from a cue in the “other-attribution” condition while moving toward the cue in the “self-attribution” condition. This movement made subjects aware of which condition was being performed. However, these are two very different stimuli, and this difference in the stimuli may have been enough to account for some of the differences in activation between the two conditions.

If the forward internal model does hold up against criticism and can be applied to sensory stimuli, then it may be useful to explain why some people, after brain damage, attribute ownership of their limbs to other people. The proposed explanation would be that a limited cancellation of sensory stimuli occurs in these people because of damage to parietal areas, leading them to judge sensory events as external. Yet it remains to be seen why in these symptoms develop in these patients and optic ataxia or delusions of alien control appear in other patients. Whether such variety in symptoms can be the result of different specific lesion sites, or whether the symptoms that develop given some lesion site depend on other factors in the person’s life, remains to be seen. It is also peculiar that when subjects view one of their limbs as not being their own, it is usually only the case with one of their limbs, as opposed to their whole body or at least a large region of it. Perhaps this confusion must be coupled with a moving part of the body, but this reasoning still does not explain why only one limb would be affected, unless the lesions in these cases are always very confined. Case studies should be examined systematically to determine what other factors in a person’s life might be affecting the course of their symptoms, and whether the precise lesions cited make a difference.

Imagery Explained

Subjects with parietal lobe damage have been shown to be impaired at using mental imagery to predict the time necessary for completing a motor movement (Sirigu et al. 1996). This result fits nicely with the forward model, since a mental image could be seen as the conscious version of the efference copy of motor commands. In other words, a mental image of an action is the prediction of that action, and can be compared (mostly unconsciously) to the outcome of the action as it is occurring. Danckert et al. (2002) also presented a case study of a person with right parietal damage to the inferior parietal cortex who showed similar impairment. This person exhibited non-lateralized impairment of motor imagery when asked to judge the time necessary to point to targets of various sizes with a pen. The participant, LR, showed the normal trade-off between target size and movement duration (i.e., longer duration for small targets), but did not predict the length of his movements based on this law. This is in contrast to people without parietal lobe injury who correctly use imagery to predict movement duration. It is strange that no studies have asked subjects (or at least reported the answer) if they can imagine their movements or how they are estimating the time necessary to make a movement. The answer to such questions may be useful in elucidating whether the parietal cortex is actually necessary for the conscious perception of a motor image. It may be true that only the impairment of an unconscious mechanism is causing a person to lose conscious perception of motor images. It is interesting to note that LR performed as well as the control subjects visual imagery tasks; it was only when the task involved imagery of motor movements did LR show any impairment.

Another case study of a man with bilateral parietal lobe damage (Schwoebel et al. 2002), CW, showed a very different manifestation of a possibly similar phenomenon to that of LR. CW could imagine movements, but simultaneously executed those motor movements without being aware of it. Moreover, these movements were actually more accurate than a “volitional movement” (i.e., one that he intended to perform). The authors speculated that the forward model in CW may be faulty so that the mechanism used to update ongoing movements was not working properly. However, why CW showed this manifestation of a faulty forward model, while LR was not able to imagine motor actions, remains unclear.

Such different symptoms may be a result of injury affecting slightly different parts of the parietal lobes. Participant PJ, from a study by Wolpert et al. (1998), may shed some light on this issue. PJ sustained damage to her superior parietal lobe. Her proprioception has faded with time and she must be able to see her limb in order to know its location. The authors assert that PJ has an impaired representation of her body state so that she is aware of the location of each of her limbs for only a few seconds at a time. She is, therefore, also impaired at slow pointing movements (when without visual sight to guide her arm). Although the explanation of having no internal representation of body state neatly clarifies PJ’s impairments, it does not explain why a movement of the limb does fails to restore the body image. If the parietal areas are multisensory, then a change in position of the limb (which provides proprioceptive impulses) should be as effective as visual sensations in restoring knowledge of the location of the limb. Clearly, an additional deficit not explained by the authors must have been present in PJ. This deficit did not allow her to properly use sensory input to update her internal body representation already in existence.

It should be pointed out that some modifications to the forward model may explain the above phenomena. The “intentional binding” theory put forth by Tsakiris and Haggard (2003) is as an example of a theory which goes beyond the forward model. This theory states that the intention to make a motor movement acts as a key to associate the intention with the action and its consequences. Intended actions bind to their effects and are literally experienced as occurring closer together in time than involuntary actions. It is this time spacing, moreover, that determines whether or not actions are connected to the intention and sensory consequence. In other words, the intention itself binds the intention to the action by having these two components be experienced closer in time. This mechanism, though it should be explored more deeply, seems to be at least as plausible. Schizophrenics, who do have intentions to act but still perceive their actions as being controlled from the outside, would seem to have a faulty intentional binding mechanism. Moreover, the hypnosis study by Blakemore et al. (2003) shows hypnosis disrupting the link between intention and action due to interference with the intentional binding mechanism.

Both the intentional binding and the forward model explanations have one deeply rooted problem in common. This problem is one of correlation and causation. Activity in the parietal cortex may be correlated with the comparison mechanism of the forward model or the conscious sensation of intention may be correlated with this change in the perceived timing of action and effect of the intentional binding explanation. The problem, in both cases, is that the effect is not all-or-none. Some activity in the parietal cortex is still present even when the movement is purely self-generated (otherwise we wouldn’t feel anything when we tickled ourselves!) In addition, the brain would not be able to determine if its activation was reduced without the original sensation for comparison. Therefore, the brain may be using a signal indicating the difference between the unprocessed sensation and the reduced sensation to determine whether or not movements are self-initiated. If this is true, then the forward model would need to learn to predict the feelings of certain sensations. For example, it may expect babies to be more ticklish than adults, since adults can better predict, from experience, the feeling of being touched by another person. Whether or not babies are really more ticklish than adults remains to be seen; such a study would be quite difficult to carry out.

The same deeply rooted problem exists for the intentional binding mechanism: If a person has an intention to move, then the time difference between action and sensory response is changed. The change in timing then binds the intention to the action. But, it is not clear how the actor would know if the difference in timing was a real difference (i.e., that the action and effect were actually that far apart) or one due to just the intention. There would have to be two copies of the timing information, one with the actual time difference between action and effect, and one with the modified time difference. In this case (as opposed to in the forward model case), the modulation of the existing information seems completely unnecessary, since it does not actually allow for any reduction in information. There seems to be no reason, then, for timing differences to be used to bind an intention to an action. However, further studies are needed to shed more light on this issue.

Conscious Intention and Awareness of Action

It has been established by the literature reviewed above that the parietal cortex is involved, at least in some respects, in the prediction of motor actions and refinement of movements as they occur. The burning question, however, is whether or not a neural correlate of the conscious mechanisms behind these predictions can be found. Studies looking at conscious imagery of motor actions have investigated this question from one angle; however, further work in this direction should be completed to discern the parietal cortex’s necessity in forming conscious internal images of motor actions. Only recent studies, such as those discussed, have tried to tackle the question of neural correlates from different perspectives.

It may be useful to look at some of the studies cited above from the standpoint of someone interested in conscious awareness and intention of motor actions. Schwoebel et al. (2002) give the case of CW, a person who performed movements when he imagined them even though he was unaware of these actions. It may be interesting to know whether or not CW would be able to show some vague memory of performing the action if pressed on the issue or given a recognition test. However, if the observations of Schwoebel et al. are correct, then it may be presumed that the parietal cortex really does have a highly strong role in conscious awareness of movements and the intention to move. This is because CW did not actually intend to move when he imagined the movements. On the other hand, CW was still able to make internal images of his motor movements in the first place; it was the application or inhibition of these imagined movements that was disrupted along with the conscious awareness of actual movement.

Blakemore et al. (2003) clarified the role of intentions in their hypnosis study. They described intentions as a necessary part of the forward model because intentions act as an impetus for sending an efference copy of the motor movement. Thereby, intentions predict the sensory consequences of motor movements and allow the sensory consequences of intended movements to be cancelled out (at least to some extent.) The question is then whether or not the intention itself comes from the parietal cortex. Lau et al. (2004) explored this issue by having subjects perform a task developed by Libet et al. (1983) to measure the timing of readiness potentials. In this task, participants judged the time on an artificial clock just as they had the intention to press a button and just as they started to press a button. Lau et al. measured the fMRI activation in these two conditions, figuring that attention to the intention to move would make certain brain areas involved in the intention activated more strongly. They made this assumption because attention usually increases activity in any modality to which one pays attention. This design by Lau et al. was very elegant, and it found that certain brain areas were indeed activated more strongly when paying attention to intention. The main finding was the increased activation in the pre-SMA (Supplementary Motor Cortex) area. However, increased activity was also shown in the dorsal prefrontal cortex (DPFC) and left intraparietal sulcus (IPS). Therefore, the parietal cortex (specifically the left IPS) must be involved to some extent in intention. Based on the information described above, intention may be “created” in the frontal lobe areas and then used in the parietal area during a comparison process of the forward model.

Some of the biggest clues of the parietal cortex’s involvement in movement intention, however, come from a study of five people with stroke lesions in the parietal cortex. Subjects were asked to perform the Libet task described above. Interestingly, the study sample had trouble judging the onset of the intention to press the button. In addition, no overlap within one standard deviation existed between test and control subjects. In general, however, the brain-damaged subjects showed greater variability in their judgments. This may indicate that people deal with parietal cortex damage in different ways. The behavioral data from the study were matched to EEG data, and it was shown that a readiness potential, which usually marks the onset of an intention to move as well as the onset of movement, was barely (or not at all) detectable in subjects with parietal cortex damage (Sirigu et al. 2004). Thus, it seems that these patients cannot use their intentions in the parietal system; they do not have access to conscious awareness of intention timings. Therefore, some aspects of conscious awareness of intentions for motor movements seem to be dependent on areas in parietal cortex (note: the specific area of damage to subjects with parietal cortex lesions was the inferior parietal lobule, or more precisely the angular gyrus, BA 39, which was shown in fMRI studies to be activated more strongly to externally-caused actions; e.g., Farrer and Frith 2002; Blakemore et al. 2003.)

Blakemore SJ et al. (1998). Central cancellation of self-produced tickle sensation. Nature Neuroscience. 1:635-640.

Blakemore SJ et al. (2002). Abnormalities in the awareness of action. Trends in Cog Sci. 6:237-242.

Bálint R. (1909). Seelenlämung des ‘Schauens’, optisches Ataxie, räumliche Störung der Aufmerksamkeit. Monatsschrift für Psychiatrie und Neurologie. 25:51-81.

Critchley M. (1953). The Parietal Lobes. Hafner Press, New York.

Danckert J et al. (2002). Selective, non-lateralized impairment of motor imagery following right parietal damage. Neurocase. 8:194-204.

Desmurget M et al. (1999). Role of the posterior parietal cortex in updating reaching movements to a visual target. Nature Neuroscience. 2:563-567.

Farrer C and CD Frith. (2002). Experiencing oneself vs another person as being the cause of an action: The neural correlates of the experience of agency. NeuroImage. 15:596-603.

Khitrove-Orlova TV et al. (1997). Pre-movement activity of neurons in the parietal associative cortex of the cat during different types of voluntary movement. Neurosci Behav Physiol. 27:82-86.

Lau HC et al. (2004). Attention to intention. Science. 303:1208-1210.

Libet B et al. (1983). Time of conscious intention to act in relation to onset of cerebral activity (readiness potential). The unconscious initiation of a freely voluntary act. Brain. 106:623-642.

Liepmann H. (1905). Die Linke Hemisphäre und das Handeln. Münch Med Wochenschr, 49:2322-2326.

Schwoebel J et al. (2002). The man who executed “imagined” movements: evidence for dissociable components of the body schema. Brain Cogn. 50:1-16.

Sherrington CS. (1941). Man on his Nature. Cambridge: Cambridge University Press.

Simeon D et al. (2000). Feeling unreal: a PET study of depersonalization disorder. Am J Psychiatry. 157:1782-1788.

Sirigu A et al. (2004). Altered awareness of voluntary action after damage to the parietal cortex. Nature Neuroscience. 7:80-84.

Sirigu A et al. (1999). Perception of self-generated movement following left parietal lesion. Brain. 122:1867-1874.

Sirigu A et al. (1996). The mental representation of hand movements after parietal cortex damage. Science. 273:1564-1568.

Spence SA et al. (1997). A PET study of voluntary movement in schizophrenic patients experiencing passivity phenomena (delusions of alien control). Brain. 120:1997-2011.

Tsakiris M and P Haggard. (2003). Awareness of somatic events associated with a voluntary action. Exp Brain Res. 149:439-446.

Wolpert DM et al. (1995). An internal model for sensorimotor integration. Science. 269:1880-1882.

Wolpert DM et al. (1998). Maintaining internal representations: the role of the human superior parietal lobe. Nature Neuroscience. 1:529-533.