I enjoy neuroscience articles! But I know from my own blog that philosophy ones tend to be more popular. Anyway, I didn't realize anesthetizing the amputation site was effective. Although the fact that that didn't end the investigation seems to show that no one answer will probably resolve it.
I think predictive coding has a lot of promise, so it's kind of exciting to see it possibly having some use in clinical scenarios. I don't see it as a complete alternative to other scientific theories like global workspace, attention schema, or higher order theories. I think they're all modeling different aspects of the overall reality.
But I like predictive coding because it seems to provide an explanation for the first glimmers of cognition in those early Cambrian or pre-Cambrian animals. We get stimulus-response from diffuse nerve nets and spinal cords, but prediction explains what the earliest additions to those automatic responses would have added, and how cognition overall began to evolve.
I completely agree -- I don't think any of the current theories are the universal theory we are looking for. Many of them have important things to say.
The thing I like about the predictive coding model is that it's able to step away from the classic stimulus-response framework that has dominated the field for so long. There's enough evidence to suggest that the brain doesn't work in this purely bottom-up way.
I think you're right. The idea that prediction might explain the first steps beyond simple stimulus-response mechanisms could be how we start to explain more complex forms of cognition.
Great work as usual. I think I've heard that some of the prosthetic devices were actually being attached to the damaged nerves, or that they were trying to do that, so I did a quick search for that (inconclusive). I found this interesting article, though, which might add grist to your mill: https://pubmed.ncbi.nlm.nih.gov/10507559/ in the abstract of which is said: "The amount of phantom limb pain has been found to be highly correlated with the amount of injury-related, afferent-decrease cortical reorganization. It is possible that the increased use of the amputation stump induced by wearing a Sauerbruch prosthesis produced a countervailing use-dependent, afferent-increase type of cortical reorganization that reversed the phantom limb pain."
Then I was curious as to whether EEGs (electroencephalography) might offer some help. I was wondering whether some kind of micro electric shock treatment might have been tried but didn't find anything there, but what I did find was this: https://pubmed.ncbi.nlm.nih.gov/38220575/ which I think suggests that people can be trained away from phantom limb pain. I'd guess all of this would be of great interest to cybernetics.
The nerve-connected prosthetics you mentioned are fascinating! Several labs are working on this sort of thing with BCI tech that not only lets amputees control movement but actually feel sensations through their prosthetic limbs. Pretty amazing work. Some folks can already sense pressure with their prosthetics, letting them do delicate things like picking up eggs or holding hands. Just imagine what that means for everyday life and human connection!
The use of prosthetics and pain is an interesting one. I believe there's a frustrating catch-22: phantom pain often gets worse when people remove their prosthetics, but when the pain hits, putting the prosthetic back on can be too painful.
Love your suggestion about stimulation treatments! TMS or TDCS might be worth exploring. I'm sure there's lots of work on this (adding that to my ever-growing 'must research' list).
Your responses are always such a gold mine. A million thanks.
TMS (Transcranial magnetic stimulation) at first struck me as perhaps too blunt an instrument, so to speak, since it would seem to set up magnetic fields. And I would have guessed TDCS (Transcranial direct current stimulation) would have the same problem, but then… given neuroplasticity perhaps I was very wrong about that, and that pinpoint accuracy would be counterproductive. Interesting that both of these things are used to treat depression, and generally fascinating to think of the ways electrical and magnetic fields interact with brain function. Is this another conspiracy theory come true, though, that telephone towers can affect behavior?? (now I’m just playing with you)
To me it makes sense that our brains would often cause us to feel like a missing arm is still with us. This is because we learn to operate them over the years, and indeed, they become an integral part of us. I wouldn’t expect a missing ear to stay with us quite so integrally. Though we often take credit for arm function, consciousness effectively teaches an arm to operate automatically. This is to say, algorithmic brain control on the basis of our desires.
Regarding pain from a missing arm (and beyond the effect of severed nerve endings themselves), consider the thought that this is a result of the severed arm not functioning as the brain demands it to. The non reception of healthy afference signals from the arm might be good reason for the brain to cause that non existent limb to hurt. For example, if you were an army general and some of your units refused to take your orders (somewhat like a severed arm won’t take your orders), you may have reason to punish those units. Of course here the brain effectively punishes a person given a severed arm, but under adverse body conditions like injury, that’s what brains do. This perspective lies in opposition with the maladaptive plasticity theory. Brain plasticity is considered beneficial in all cases that I currently know of, so that explanation might be reaching too far. What I’m proposing is instead quite consistent with the persistent representation theory.
Regarding the thought that pain exists as a body prediction error, in a natural world I don’t believe it could be this simple. If prediction error happens to be involved, then I think we ought to consider what that error information might inform to actually exist as the experiencer of that pain? Information seems only to exist as such to the extent that it informs something appropriate, so I don’t think we should expect an error in itself to do the trick.
Though I do consider our brains to effectively predict, it seems to me that pain should instead exist as punishment for body damage regardless of what the brain does or does not predict. It’s something to potentially learn from that encourages us to not let ourselves become damaged. And what might body damage signals inform to exist as an experiencer of pain? I believe that such signals cause neurons to fire with a synchrony that sets up an electromagnetic field which itself exists as the experiencer of the pain, and in combination with all other EMF elements of consciousness. All elements of consciousness seem bound together from moment to moment no less than a highly complex electromagnetic field happens to be.
"The non reception of healthy afference signals from the arm might be good reason for the brain to cause that non existent limb to hurt. "
Or, the limb keeps sending "pain, pain, pain" messages until it is amputated. At that point, the part of the brain that processes the signals only has the last messages to know anything about the state of the limb. Any activation of neurons in the brain near to the pain circuits, such as the parts that process the signals for the remaining part of the limb, could bleed over into the original pain circuits and activate them because of connectivity established during the pain episode prior to amputation.
Interestingly, a different part of the brain - the insula - is involved with "limb ownership." That syncs with my view that consciousness at any one time is simply a composite of fragments generated in many different parts of the brain but near in time (with no requirement for synchronicity). So, the consciousness of "owning" a limb and feeling pain in the limb are distinct qualia, although the first may be a requirement for the second.
I’m certainly on board with your fragmented consciousness position in that sense James. There are fragmented informational components associated with consciousness that happen all over the brain. So how do they unify into the bound experiences that we actually have? Furthermore “representational drift” is observed where different neurons seem responsible for a given element of consciousness in subsequent examples. Why? Perhaps because it’s not the specific neurons themselves which fire that reside as consciousness, but rather the specific electromagnetic field that’s created by which ever neurons fire to create such a field?
I'm not sure our experiences are as unified as they appear. I gave six reasons why consciousness might appear integrated even though it isn't in a post on my site.
I also think recent research with more precise timings of neuron firings are actually finding that that there is a not as much simultaneity happening as previously believed, especially across different regions. What was believed to be simultaneous is happening with milliseconds offset, possibly as part of traveling waves.
What's more, I think, it appears that information from one region is more widely disseminated than I previously believed possible. Retinotopic maps in the visual cortex, for example, are replicated in other regions of the brain. This dissemination of information from one region to another across the brain may be what consciousness does. If information is copied to many parts of the brain, any one part might have access to a large portion of the big picture. This brings to mind a holographic type of representation where even small parts of the hologram contain the entire image.
Since people can experience phantom limb sensations without pain, I am not sure how much predictive coding and error adds to the explanation of the pain. The people who are experiencing the limb without pain must also be having predictive errors but without the pain.
It seems to me more likely that phantom limb pain is more like a PTSD-type of learning originating from the pain experienced prior to amputation that gets reactivated either by stimulating the nerve endings at the amputation site or the efference copy from the the motor cortex.
"Persistent pre-operative pain, residual limb pain, and non-painful phantom limb sensations were identified as risk factors for PLP."
Apparently people born without limbs can also experience phantom limb. Would that suggest that circuits that produce the sensation of the limb, absent the pain, is to some degree hard-coded into the cortex from birth?
Well done James! If prediction error theory is correct then its proponents have some explaining to do. Let’s say you throw a ball to someone with a recently amputated arm, and they automatically try to catch it with both arms. Thus an obvious example of brain prediction error. Does the person then report that when their brain made that prediction error, pain also resulted? If so then fine, maybe it’s right. I suspect however that this isn’t what’s found, and specifically because brain prediction error does not creat pain.
I'm with you and Eric on the point that a prediction error might not be enough to explain pain. But the argument against your claim might be that the mismatch is different for those who experience pain and those who don't. For people with pain, what the former sensory hand area receives during passive stimulation and what it receives from active movement would be very different. This is because, as that paper from the 90s showed (and others like it) the amount of pain the amputee experiences depends on the amount of remapping in the former sensory hand area -- more remapping means more pain. For people with less pain or no pain the mismatch would be less -- less remapping means less mismatch between what is received during passive stimulation and what is received during movement, and, as the argument goes, less mismatch means less pain.
No doubt mechanisms involved in PTSD would be involved in phantom pain. That would make a lot of sense.
On being hard-coded, that's a difficult one to claim. There are obviously genetic influences (possibly very strong ones), but the environmental impacts during pregnancy must also be considered. Like most things, it's probably a complex interaction between both genes and the environment.
I enjoy neuroscience articles! But I know from my own blog that philosophy ones tend to be more popular. Anyway, I didn't realize anesthetizing the amputation site was effective. Although the fact that that didn't end the investigation seems to show that no one answer will probably resolve it.
I think predictive coding has a lot of promise, so it's kind of exciting to see it possibly having some use in clinical scenarios. I don't see it as a complete alternative to other scientific theories like global workspace, attention schema, or higher order theories. I think they're all modeling different aspects of the overall reality.
But I like predictive coding because it seems to provide an explanation for the first glimmers of cognition in those early Cambrian or pre-Cambrian animals. We get stimulus-response from diffuse nerve nets and spinal cords, but prediction explains what the earliest additions to those automatic responses would have added, and how cognition overall began to evolve.
Interesting article, as always Suzi!
Thanks Mike!
I completely agree -- I don't think any of the current theories are the universal theory we are looking for. Many of them have important things to say.
The thing I like about the predictive coding model is that it's able to step away from the classic stimulus-response framework that has dominated the field for so long. There's enough evidence to suggest that the brain doesn't work in this purely bottom-up way.
I think you're right. The idea that prediction might explain the first steps beyond simple stimulus-response mechanisms could be how we start to explain more complex forms of cognition.
Thanks again for a great comment!
Great work as usual. I think I've heard that some of the prosthetic devices were actually being attached to the damaged nerves, or that they were trying to do that, so I did a quick search for that (inconclusive). I found this interesting article, though, which might add grist to your mill: https://pubmed.ncbi.nlm.nih.gov/10507559/ in the abstract of which is said: "The amount of phantom limb pain has been found to be highly correlated with the amount of injury-related, afferent-decrease cortical reorganization. It is possible that the increased use of the amputation stump induced by wearing a Sauerbruch prosthesis produced a countervailing use-dependent, afferent-increase type of cortical reorganization that reversed the phantom limb pain."
Then I was curious as to whether EEGs (electroencephalography) might offer some help. I was wondering whether some kind of micro electric shock treatment might have been tried but didn't find anything there, but what I did find was this: https://pubmed.ncbi.nlm.nih.gov/38220575/ which I think suggests that people can be trained away from phantom limb pain. I'd guess all of this would be of great interest to cybernetics.
Thanks Jack!
The nerve-connected prosthetics you mentioned are fascinating! Several labs are working on this sort of thing with BCI tech that not only lets amputees control movement but actually feel sensations through their prosthetic limbs. Pretty amazing work. Some folks can already sense pressure with their prosthetics, letting them do delicate things like picking up eggs or holding hands. Just imagine what that means for everyday life and human connection!
The use of prosthetics and pain is an interesting one. I believe there's a frustrating catch-22: phantom pain often gets worse when people remove their prosthetics, but when the pain hits, putting the prosthetic back on can be too painful.
Love your suggestion about stimulation treatments! TMS or TDCS might be worth exploring. I'm sure there's lots of work on this (adding that to my ever-growing 'must research' list).
Your responses are always such a gold mine. A million thanks.
TMS (Transcranial magnetic stimulation) at first struck me as perhaps too blunt an instrument, so to speak, since it would seem to set up magnetic fields. And I would have guessed TDCS (Transcranial direct current stimulation) would have the same problem, but then… given neuroplasticity perhaps I was very wrong about that, and that pinpoint accuracy would be counterproductive. Interesting that both of these things are used to treat depression, and generally fascinating to think of the ways electrical and magnetic fields interact with brain function. Is this another conspiracy theory come true, though, that telephone towers can affect behavior?? (now I’m just playing with you)
haha! You almost got me there!
Great article on phantom pain. Always enjoy your musings and discussions.
Thx.
Respectfully.
Thanks David!
Excellent piece. Thanks again.
Thanks, John!
To me it makes sense that our brains would often cause us to feel like a missing arm is still with us. This is because we learn to operate them over the years, and indeed, they become an integral part of us. I wouldn’t expect a missing ear to stay with us quite so integrally. Though we often take credit for arm function, consciousness effectively teaches an arm to operate automatically. This is to say, algorithmic brain control on the basis of our desires.
Regarding pain from a missing arm (and beyond the effect of severed nerve endings themselves), consider the thought that this is a result of the severed arm not functioning as the brain demands it to. The non reception of healthy afference signals from the arm might be good reason for the brain to cause that non existent limb to hurt. For example, if you were an army general and some of your units refused to take your orders (somewhat like a severed arm won’t take your orders), you may have reason to punish those units. Of course here the brain effectively punishes a person given a severed arm, but under adverse body conditions like injury, that’s what brains do. This perspective lies in opposition with the maladaptive plasticity theory. Brain plasticity is considered beneficial in all cases that I currently know of, so that explanation might be reaching too far. What I’m proposing is instead quite consistent with the persistent representation theory.
Regarding the thought that pain exists as a body prediction error, in a natural world I don’t believe it could be this simple. If prediction error happens to be involved, then I think we ought to consider what that error information might inform to actually exist as the experiencer of that pain? Information seems only to exist as such to the extent that it informs something appropriate, so I don’t think we should expect an error in itself to do the trick.
Though I do consider our brains to effectively predict, it seems to me that pain should instead exist as punishment for body damage regardless of what the brain does or does not predict. It’s something to potentially learn from that encourages us to not let ourselves become damaged. And what might body damage signals inform to exist as an experiencer of pain? I believe that such signals cause neurons to fire with a synchrony that sets up an electromagnetic field which itself exists as the experiencer of the pain, and in combination with all other EMF elements of consciousness. All elements of consciousness seem bound together from moment to moment no less than a highly complex electromagnetic field happens to be.
"The non reception of healthy afference signals from the arm might be good reason for the brain to cause that non existent limb to hurt. "
Or, the limb keeps sending "pain, pain, pain" messages until it is amputated. At that point, the part of the brain that processes the signals only has the last messages to know anything about the state of the limb. Any activation of neurons in the brain near to the pain circuits, such as the parts that process the signals for the remaining part of the limb, could bleed over into the original pain circuits and activate them because of connectivity established during the pain episode prior to amputation.
Interestingly, a different part of the brain - the insula - is involved with "limb ownership." That syncs with my view that consciousness at any one time is simply a composite of fragments generated in many different parts of the brain but near in time (with no requirement for synchronicity). So, the consciousness of "owning" a limb and feeling pain in the limb are distinct qualia, although the first may be a requirement for the second.
I’m certainly on board with your fragmented consciousness position in that sense James. There are fragmented informational components associated with consciousness that happen all over the brain. So how do they unify into the bound experiences that we actually have? Furthermore “representational drift” is observed where different neurons seem responsible for a given element of consciousness in subsequent examples. Why? Perhaps because it’s not the specific neurons themselves which fire that reside as consciousness, but rather the specific electromagnetic field that’s created by which ever neurons fire to create such a field?
I'm not sure our experiences are as unified as they appear. I gave six reasons why consciousness might appear integrated even though it isn't in a post on my site.
I also think recent research with more precise timings of neuron firings are actually finding that that there is a not as much simultaneity happening as previously believed, especially across different regions. What was believed to be simultaneous is happening with milliseconds offset, possibly as part of traveling waves.
What's more, I think, it appears that information from one region is more widely disseminated than I previously believed possible. Retinotopic maps in the visual cortex, for example, are replicated in other regions of the brain. This dissemination of information from one region to another across the brain may be what consciousness does. If information is copied to many parts of the brain, any one part might have access to a large portion of the big picture. This brings to mind a holographic type of representation where even small parts of the hologram contain the entire image.
Since people can experience phantom limb sensations without pain, I am not sure how much predictive coding and error adds to the explanation of the pain. The people who are experiencing the limb without pain must also be having predictive errors but without the pain.
It seems to me more likely that phantom limb pain is more like a PTSD-type of learning originating from the pain experienced prior to amputation that gets reactivated either by stimulating the nerve endings at the amputation site or the efference copy from the the motor cortex.
"Persistent pre-operative pain, residual limb pain, and non-painful phantom limb sensations were identified as risk factors for PLP."
https://pmc.ncbi.nlm.nih.gov/articles/PMC10845739/
Apparently people born without limbs can also experience phantom limb. Would that suggest that circuits that produce the sensation of the limb, absent the pain, is to some degree hard-coded into the cortex from birth?
Well done James! If prediction error theory is correct then its proponents have some explaining to do. Let’s say you throw a ball to someone with a recently amputated arm, and they automatically try to catch it with both arms. Thus an obvious example of brain prediction error. Does the person then report that when their brain made that prediction error, pain also resulted? If so then fine, maybe it’s right. I suspect however that this isn’t what’s found, and specifically because brain prediction error does not creat pain.
I'm with you and Eric on the point that a prediction error might not be enough to explain pain. But the argument against your claim might be that the mismatch is different for those who experience pain and those who don't. For people with pain, what the former sensory hand area receives during passive stimulation and what it receives from active movement would be very different. This is because, as that paper from the 90s showed (and others like it) the amount of pain the amputee experiences depends on the amount of remapping in the former sensory hand area -- more remapping means more pain. For people with less pain or no pain the mismatch would be less -- less remapping means less mismatch between what is received during passive stimulation and what is received during movement, and, as the argument goes, less mismatch means less pain.
No doubt mechanisms involved in PTSD would be involved in phantom pain. That would make a lot of sense.
On being hard-coded, that's a difficult one to claim. There are obviously genetic influences (possibly very strong ones), but the environmental impacts during pregnancy must also be considered. Like most things, it's probably a complex interaction between both genes and the environment.