Andy Clark describes delusions as the dark side of the seamless story for predictive coding in which, “In place of any real distinction between perception and belief we now get variable differences in the mixture of top-down and bottom-up influence, and differences of temporal and spatial scale in the internal models that are making the predictions” (sect. 2.3, para. 8).
Theorists who endorse the predictive coding model have argued that in delusions of alien control, patients actually experience being controlled by an external agent. As Gallagher puts it, “the attribution of agency to another is a genuine result of what is truly experienced” (Gallagher Reference Gallagher2004, p. 17, my italics). Some experiments suggest that this experience is the result of a prior belief about the external origin of movement. This would be a nice vindication of the seamless story. I think, however, that the mind is not quite so seamless and that there is another explanation consistent with the predictive coding framework.
How could someone experience his or her own movements as alienated actions? The short answer is that right inferior parietal activation represents “surprisal” for intended movements. Surprisal is minimised for intended movements because the motor command from the supplementary motor area (SMA) attenuates activity in the right interior parietal cortex. On the seamless story, unpredicted/unattenuated parietal activation (surprisal) arising in the context of action observation is experienced as alienation: “The patients really had no cues (as inferred from the change in activity in the parietal lobe) about whether they saw their own movements or those of an alien agent” (Jeannerod Reference Jeannerod2006, my italics). Thus, they experience their own movements as alienated.
In an important experiment Daprati and collaborators had subjects trace a path from their body midline to a target directly in front of them. The subjects' view of their moving hands was occluded until the final 30% of the movement. For the first 70%, patients saw a computer-generated trace of the movement path. On some trials the experimenters introduced a deviation of 15% into the movement path so that if uncorrected the trace would veer off to the right. Both schizophrenic and neurotypical subjects were able to compensate for the perturbation, during the occluded section of the movement, with the result that when the hand came into view, the hand was to the left of the midline. Danckert et al. (Reference Danckert, Saoud and Maruff2004) express the consensus in a large literature when they say that such cases show that “on-line monitoring and adjustment of action is unaffected in patients with schizophrenia” (p. 253).
In Daprati's experiment, the last 30% of the movement is not occluded. When the subject sees the hand it is 15 degrees to the left of a straight line to the target. Neurotypical subjects attributed this discrepancy to the computer, indicating that they were able to become aware that they had intended a different movement than the one they actually made. Schizophrenics with positive symptoms did not, leading to the conclusion that “online control can coexist with a tendency to misattribute the source of error” (Daprati et al. Reference Daprati, Franck, Georgieff, Proust, Pacherie, Dalery and Jeanerod1997, p. 253, emphasis theirs).
This tendency arises for schizophrenics when they visually attend to the movement. In this case they seem lose access to information about self-initiation. (Note: this is a problem of degree not kind. The dominance of visual attention over proprioceptive/motor information generates similar misattributions in many conditions).
Blakemore et al. (Reference Blakemore, Oakley and Frith2003) hypnotized subjects whose arms were attached to a pulley apparatus and gave them two instructions. In the first they were told to raise their arms and in the second that the pulley would raise their arms. The pulley did not actually exert any force. Highly hypnotizable subjects moved their arms in response to both instructions but in the second case they reported no feeling of agency, attributing the movement to the pulley. In effect, hypnosis induced the experience of failed action monitoring characteristic of delusions of alien control. The authors explain: “The prediction made by the parietal cortex is concerned more with high level prediction such as strategic planning actions.” Furthermore, they suggest, “Perhaps the predictions made by the parietal cortex can be made available to consciousness” (Blakemore et al. Reference Blakemore, Oakley and Frith2003, p. 243, my italics). In other words we can experience ourselves as authors of our actions in virtue of attenuated parietal activity. Because schizophrenics cannot attenuate this activity, they cannot become aware of themselves as authors of their actions in some conditions.
Does it follow that unattenuated parietal activity represents that someone else is the author of the action? From what we have seen so far, the modulation of parietal activity only tells the subject whether a movement is produced by the SMA. That is a very low level of cognitive processing from which information about agency is absent.
Evolution has not posed us with the problem of determining which movements are ours rather than someone else's. It has posed us with the problem of determining which aspects of a movement are consequences of motor intentions in order to compute and resolve error. Therefore, there seems no reason to think that we would need to use predictive coding to disambiguate the agent of an action rather than to simply control our own action. This is true both at the level of automatic and of controlled processing.
In general, then, I conclude that parietal activation is not specialised for determining who intended the action. Rather, it determines for any movement whether it is a consequence of a motor instruction. It evolved to control movement, not to identify the agent. Because schizophrenics cannot attenuate this activity when visually monitoring actions, they cannot experience themselves as authors of those actions. In both experiments, however, the context provides a default interpretation of alienation.
If the fabric of the mind is stitched together seamlessly with predictive coding threads we should be able to unravel it entirely from the top down. But the fact that online control in schizophrenia is intact suggests that the seam linking automatic and visually guided motor control, while flexible, has been robustly tailored by evolution.
Andy Clark describes delusions as the dark side of the seamless story for predictive coding in which, “In place of any real distinction between perception and belief we now get variable differences in the mixture of top-down and bottom-up influence, and differences of temporal and spatial scale in the internal models that are making the predictions” (sect. 2.3, para. 8).
Theorists who endorse the predictive coding model have argued that in delusions of alien control, patients actually experience being controlled by an external agent. As Gallagher puts it, “the attribution of agency to another is a genuine result of what is truly experienced” (Gallagher Reference Gallagher2004, p. 17, my italics). Some experiments suggest that this experience is the result of a prior belief about the external origin of movement. This would be a nice vindication of the seamless story. I think, however, that the mind is not quite so seamless and that there is another explanation consistent with the predictive coding framework.
How could someone experience his or her own movements as alienated actions? The short answer is that right inferior parietal activation represents “surprisal” for intended movements. Surprisal is minimised for intended movements because the motor command from the supplementary motor area (SMA) attenuates activity in the right interior parietal cortex. On the seamless story, unpredicted/unattenuated parietal activation (surprisal) arising in the context of action observation is experienced as alienation: “The patients really had no cues (as inferred from the change in activity in the parietal lobe) about whether they saw their own movements or those of an alien agent” (Jeannerod Reference Jeannerod2006, my italics). Thus, they experience their own movements as alienated.
In an important experiment Daprati and collaborators had subjects trace a path from their body midline to a target directly in front of them. The subjects' view of their moving hands was occluded until the final 30% of the movement. For the first 70%, patients saw a computer-generated trace of the movement path. On some trials the experimenters introduced a deviation of 15% into the movement path so that if uncorrected the trace would veer off to the right. Both schizophrenic and neurotypical subjects were able to compensate for the perturbation, during the occluded section of the movement, with the result that when the hand came into view, the hand was to the left of the midline. Danckert et al. (Reference Danckert, Saoud and Maruff2004) express the consensus in a large literature when they say that such cases show that “on-line monitoring and adjustment of action is unaffected in patients with schizophrenia” (p. 253).
In Daprati's experiment, the last 30% of the movement is not occluded. When the subject sees the hand it is 15 degrees to the left of a straight line to the target. Neurotypical subjects attributed this discrepancy to the computer, indicating that they were able to become aware that they had intended a different movement than the one they actually made. Schizophrenics with positive symptoms did not, leading to the conclusion that “online control can coexist with a tendency to misattribute the source of error” (Daprati et al. Reference Daprati, Franck, Georgieff, Proust, Pacherie, Dalery and Jeanerod1997, p. 253, emphasis theirs).
This tendency arises for schizophrenics when they visually attend to the movement. In this case they seem lose access to information about self-initiation. (Note: this is a problem of degree not kind. The dominance of visual attention over proprioceptive/motor information generates similar misattributions in many conditions).
Blakemore et al. (Reference Blakemore, Oakley and Frith2003) hypnotized subjects whose arms were attached to a pulley apparatus and gave them two instructions. In the first they were told to raise their arms and in the second that the pulley would raise their arms. The pulley did not actually exert any force. Highly hypnotizable subjects moved their arms in response to both instructions but in the second case they reported no feeling of agency, attributing the movement to the pulley. In effect, hypnosis induced the experience of failed action monitoring characteristic of delusions of alien control. The authors explain: “The prediction made by the parietal cortex is concerned more with high level prediction such as strategic planning actions.” Furthermore, they suggest, “Perhaps the predictions made by the parietal cortex can be made available to consciousness” (Blakemore et al. Reference Blakemore, Oakley and Frith2003, p. 243, my italics). In other words we can experience ourselves as authors of our actions in virtue of attenuated parietal activity. Because schizophrenics cannot attenuate this activity, they cannot become aware of themselves as authors of their actions in some conditions.
Does it follow that unattenuated parietal activity represents that someone else is the author of the action? From what we have seen so far, the modulation of parietal activity only tells the subject whether a movement is produced by the SMA. That is a very low level of cognitive processing from which information about agency is absent.
Evolution has not posed us with the problem of determining which movements are ours rather than someone else's. It has posed us with the problem of determining which aspects of a movement are consequences of motor intentions in order to compute and resolve error. Therefore, there seems no reason to think that we would need to use predictive coding to disambiguate the agent of an action rather than to simply control our own action. This is true both at the level of automatic and of controlled processing.
In general, then, I conclude that parietal activation is not specialised for determining who intended the action. Rather, it determines for any movement whether it is a consequence of a motor instruction. It evolved to control movement, not to identify the agent. Because schizophrenics cannot attenuate this activity when visually monitoring actions, they cannot experience themselves as authors of those actions. In both experiments, however, the context provides a default interpretation of alienation.
If the fabric of the mind is stitched together seamlessly with predictive coding threads we should be able to unravel it entirely from the top down. But the fact that online control in schizophrenia is intact suggests that the seam linking automatic and visually guided motor control, while flexible, has been robustly tailored by evolution.