Saturday 2 January 2016

4a. Cook, R. et al (2014). Mirror neurons: from origin to function

Cook, R., Bird, G., Catmur, C., Press, C., & Heyes, C. (2014). Mirror neurons: from origin to function. Behavioral and Brain Sciences, 37(02), 177-192.

This article argues that mirror neurons originate in sensorimotor associative learning and therefore a new approach is needed to investigate their functions. Mirror neurons were discovered about 20 years ago in the monkey brain, and there is now evidence that they are also present in the human brain. The intriguing feature of many mirror neurons is that they fire not only when the animal is performing an action, such as grasping an object using a power grip, but also when the animal passively observes a similar action performed by another agent. It is widely believed that mirror neurons are a genetic adaptation for action understanding; that they were designed by evolution to fulfill a specific socio-cognitive function. In contrast, we argue that mirror neurons are forged by domain-general processes of associative learning in the course of individual development, and, although they may have psychological functions, they do not necessarily have a specific evolutionary purpose or adaptive function. The evidence supporting this view shows that (1) mirror neurons do not consistently encode action “goals”; (2) the contingency- and context-sensitive nature of associative learning explains the full range of mirror neuron properties; (3) human infants receive enough sensorimotor experience to support associative learning of mirror neurons (“wealth of the stimulus”); and (4) mirror neurons can be changed in radical ways by sensorimotor training. The associative account implies that reliable information about the function of mirror neurons can be obtained only by research based on developmental history, system-level theory, and careful experimentation.




88 comments:

  1. Are we also suppose to read all the "Open Peer Commentary" (starting on p.192) (as well as the author's responses to it) at the end of the article?

    ReplyDelete
    Replies
    1. I think based on the heading we're only responsible for pages 177-192

      Delete
    2. oh right, didn't see that, thanks!

      Delete
    3. No need to bother with the BBS commentary (unless you want to see how the pro's do skywriting...)

      Delete
  2. I just have a question regarding some of the discussion from last class. We consider mental rotation of a shape a dynamic process. But isn't this just a simulation of the shape being rotated in the physical world? And if so couldn't a computer execute this simulation?

    ReplyDelete
    Replies
    1. I discussed this with prof.Harnad a couple weeks back i'll try and explain his response. He argued that because it takes more time the more an object gets mentally rotated (just as if we would physically rotate it) that it makes more sense for a dynamic system to be carrying out this function. I then asked your question saying that of course we could just as easily make a computer program that rotates objects in the same way as we would do in real life and thus we would get the similar time function for mental rotation. His response was essentially "why would we do that?". He argued that mentally rotating objects like that is very inefficient and to counter that computer program, we could make a much more efficient computer program to solve the mental rotation task. Instead our brain does this function as if it mirrored how we do it in the real world, in a dynamic analog fashion. He argued that for the sake of efficiency and practicality it makes more sense to imagine an internal analog system carrying out this function than an inefficient computational one. We finished the conversation by conceding that this may not actually be an analog task but may instead just be a good example of why not all of cognition needs to be computation.

      Prof. Harnad please correct me if I made any mistakes representing you or your argument here!

      Delete
    2. Hi Jordan, you're roughly right. The logic, written out longhand, is this:

      1. The Shepard task task of saying whether two shapes are different or the same (but one of them is a rotation of the other) can be solved many ways, analog rotation being one of them, but a particularly simple one.

      2. Because of the Strong Church-Turing Thesis (that just about anything can be simulated computationally), analog rotation can be simulated computationally.

      3. Because cognition is invisible and because we can't (yet) detect analog brain activity via neuro-imagery, we can't say for sure whether people do the Shepard task via analog rotation or computation.

      4. But what we can say is that all you need in order to perform the task is whatever it takes to detect whether the shapes are the same or different.

      5. Analog rotation does that by analog rotation, so it takes longer to say "same" the more the shape is rotated.

      6. There is no computational reason why computational shape-matching should take longer the bigger the rotation.

      7. By analogy, it takes no longer to multiply a number by 3, 30, or 300: there is no "growth" process going on there. Multiplication is not the simulation of growth.

      8. But computation can make multiplication simulate growth, so that it takes progressively longer to multiply something by 3, 30, or 300. (You can even make it simulate linear vs exponential growth, if you want to.)

      9. Ditto for the Shepard task: Instead of just computing the outcome from a simple symbolic coding of Cartesian (X/Y/Z) coordinates, you could simulate rotation so as to make it take longer to match 30', 45' or 60'.

      10. The question is: Why on earth would you want to do that? To find a mechanism that can do the job? Analog rotation already does that.

      11. Or to simulate the reaction time of what an analog mechanism would do, by simulating the timing of the rotation?

      12. If so, then you are no longer testing how the brain does the Shepard task (or anything): You are just applying the Strong Church-Turing thesis to any task where dynamics are not absolutely necessary for success.

      13. If you want a task for which dynamics are absolutely necessary for success, you need go no further than vision itself: The transduction of photons cannot be replaced by a simulation unless you simulate the whole world too, because transduction is the way the retina (a part of the brain) turns photon input into something internal (whether analog or digital). It is the interface with the outside (analog!) world.

      14. Simulating the reaction time (which would be naturally explained by internal analog rotation) by simulating analog rotation computationally is simply adding an arbitrary constraint in order to show that I can simulate that too, like everything else, because of the (Strong) Church-Turing Thesis.

      15. But as far as T3 is concerned, nothing is at stake here, because T3 just requires input/output equivalence; and in the Shepard task, you could do it either way. Only T4 would care about whether it was done dynamically or computationally by the brain.

      16. Unless it turned out that T3 could not be successfully passed by a simulated T4 component, but only by the dynamic T4 component.

      Delete
  3. “Automatic imitation is said to occur when observation of an action involuntarily facilitates performance of a topographically similar action (body parts make the same movements relative, not to external frames of reference, but to one another) and/or interferes with performance of a topographically dissimilar action (Brass et al. 2001; Stürmer et al. 2000).”

    Do mirror neurons fire when we imagine ourselves performing an action? Before practicing a new move, many gymnasts or athletes create a mental image of themselves performing that action. This has been said to facilitate the process of learning the new action. In unrelated studies, mental images have been shown to activate similar brain areas to their real-life counterparts. For example, imagining a face activates the fusiform face area, just like seeing a real face does. I am wondering if imagining an action activates these same mirror neurons, as seeing that same action would, to help us prepare to perform the action.

    ReplyDelete
    Replies
    1. From my understanding, the answer would be yes. Filimon et al 2007 used fMRI to study mirror neurons and map their cortical representations under three tasks: observed reaching, executed action of reaching, and imagined reaching in humans in which their arms were secured to their chest. Thus, as all three conditions yielded activation of the fronto-parietal mirror neuron network, they concluded that the system is specific to the type of hand action performed. It seems that mirror neurons do play a role in imagining performing an action.

      Delete
  4. 1.
    I’m not sure to understand what the concept of “field properties of MNs” mentioned in the article refers to. I tried looking it up and from what I gathered it refers to the property of some MNs to be assigned to a specific region in the body. So, for example, whenever a motion of the hands is observed or executed the MNs associated to the hands will fire. I don’t know if my understanding is correct.

    2.
    If I understand the Associative Theory correctly, the learning that causes the properties of MNs to occur is a result of seeing an action and executing a similar action at the same time. It is later said in the article that it is a result of social interactions and that there is a “strong experience-dependence”.

    “The kind of learning that produces MNs occurs when there is correlated (i.e., contiguous and con- tingent) excitation of sensory neurons and motor neurons that code similar actions. For example, when an adult imi- tates an infant’s facial movements, there might be corre- lated excitation of neurons that are responsive to the observation and execution of lip protrusion. […] Thereafter, the motor neuron fires, not only during execution of lip protrusion, but also, via its connection with the sensory neuron, during observation of lip protrusion; what was originally a motor neuron has become a lip protrusion MN.” (Cook et al., 2014, p. 181)

    The “wealth of the stimulus ” argument developed in this paper states that infants have access to a great number of stimuli and that each stimulus is rich enough to allow them to quickly accomplish associative learning.

    “The range of sources available to young human infants includes self-observation, being imitated by adults, being rewarded by adults for imitation, and the kind of experience in which, for example, lip movements make the same smacking or popping sound when observed and executed. Evidence of the richness of these sources comes from studies showing that infants spend a high pro- portion of their waking hours observing their own hands in motion (P. Rochat 1998; White et al. 1964); in face-to-face interaction with a caregiver, they are imitated on average once every minute (Jones 2009; Pawlby 1977; Uzgiris et al. 1989); and “noisy actions,” which provide an early source of acquired equivalence experience, are among the first that infants imitate (Jones 2009).” (Cook et al., 2014, p. 178).

    It is also said, in the introduction that MNs may be related to “empathy (Avenanti et al. 2005) [and] emotion recognition (Enticott et al. 2008)” (Cook et al., 2014, p. 178). What I don’t understand is how the “wealth of stimulus” argument could account for that. From my own intuitions, I don’t see how a child could have access to the sort of stimulus that could allow them to do associative learning between emotions and their physical manifestations (e.g., the facial expression associated to sadness). Would a baby/child really ever encounter enough situations where (a) they see their reflection at the moment where they are experiencing emotions and so get to see their own facial expressions or (b) have their facial expressions mimicked by adults?

    ReplyDelete
    Replies
    1. branching off of your mention of empathy...

      As mentioned in the paper, MNs may have a role in empathy, intention-reading and theory of mind. I’m very curious as to whether or not we can consider MNs being involved with emotion recognition. As of now, there are many studies that attempt to claim how mirror neuron systems could be “dysfunctional” in autism spectrum disorders which could be partly responsible for social deficits.

      Does anyone agree that investigating disorders in which this “dyfunction” could exist (ASD, Asperger’s, etc) is a good method for figuring out what MNs are doing? Of course, there are many other players (abnormalities in the frontal lobes and limbic system, affected genes, ‘hyperconnectivity’/more synapses) that are involved with these disorders but can we suppose that mirror neuron networks are a correlate for social cognition?

      Delete
    2. HC-G:

      1. By "field properties" of Mirror Neurons (MNs) I think Cook et al mean something similar to what "receptive field" meant with sensory neurons: It's the kinds of input for which they are selectively sensitive. The kind that makes them fire. In the case of MNs, it's whenever the organism does a certain action and also whenever another organism does that same action.

      2. According to the "associative theory," the MNs fire when the organism does X and when it sees another organism do X because it has learned the correlations between the two from experience, not because the MNs "know" it innately.

      (But the question you should be asking yourself is: "whether MNs "do" it innately or through learning, how do they do it? Even before we learned there were MNs, we knew people could do what MNs do. Now that we know we have MNs, do we now know how we do it? Is there something we could tell a roboticist who trying to give his robot the capacity to do what we can do -- recognize and imitate one another robot does the same movement as itself -- would it help to advise the roboticist: "Put in MNs"?

      We'll be talking about the "poverty of the stimulus" when we get to Chomsky. But before someone can claim that there is a "wealth of stimulus" on the basis of which a robot could learn "by association" to detect and generate the kinds of correlations that make MNs fire, it would be a good idea to provide a learning mechanism that can actually do it using the "wealth of the stimulus." Again, we already knew people could do it, even before we heard of MNs. The question remains: how?

      3. Not that knowing we have MNs helps, but it is not out of the question that an infant could learn the correlations between the facial expressions of others (angry, friendly) and what they are likely to do (hit, caress). They could also learn what it feels like for themselves to want do that (hit, caress). So they could learn the association without having to literally compare their own facial expressions with those of others. (Besides, the tendency to smile back when smiled on is probably innate, and probably so are a lot of other correlations between the facial expressions of others and what the baby feels and does.)

      LK: Can you think of one new thing we have learned about autism from the fact that there are MNs? Or anything we might have wanted to test in autists that we would not have thought of testing if we had not known there were MNs (other than, of course, recording their brain activity for MN activity!)?

      Ditto for what we have learned about empathy from finding out that there are MNs (other than the obvious: another T4 correlate to measure)...

      Delete
    3. Hi Prof Harnad

      "Now that we know we have MNs, do we now know how we do it?"
      No, but it's a good start!

      I've gotten a sneaking suspicion after reading the assigned article by Fodor, and after reading a number of your replies to commentaries on this article (and after you said it), that you favour achievement of Turing indistinguishability at the level of T3 rather than T4. You've (not-so-) subtly hinted that the data acquired from neuroimaging studies (particularly this one) is in "the land of non-causal correlations", and that we probably shouldn't obsess over the brain (or at least the popular practise of functional localization) and pursue a reverse engineered T3 machine as our causal explanation of cognition. This, I suppose, is my defense of T4 and of neuroscience in general.

      You ask, "Now that we know we have MNs, do we now know how we do it?" and I would say, "Of course not!" But I'd say we have a piece of the puzzle. Correlation is the first step to determining whether something plays a causal role. It is clear that MNs play some role in the recognition/imitation of actions, and only by manipulating variables (neuron excitability, input/output activity) can we begin to access the causal side of things, which I'll admit is also very challenging. But not impossible. We have techniques like optogenetics, which allow us to selectively activate particular brain circuitry, something that only works if we understand brain anatomy, and what functions we might be impinging upon. We also have techniques like TMS which are less precise, but are still a step in the direction of sussing out the causative roles of brain areas. I'll concede that imaging studies like the one described (and much of neuroscience at the moment) is correlation, but I think the question "do we know how we do it?" is a little premature since neuroscience has tools for finding out the causal role of MNs, it will just take some time.

      You go on to ask whether this information about MNs would have any bearing on our ability to reverse engineer a cognizing robot. You ask whether it would help to "advise the roboticist: 'Put in MNs,'" and I would say "Why not?" I don't mean actual physical MNs, but in principle, why shouldn't we use the apparent "programming" of MNs to model our own machines or their programs? Why doesn't it make sense to develop a system which links visual input and motor output on the basis of associative learning. Why can't we put this information about MNs to good use in our model of cognition?

      In class it was implied that we should eschew neuroscience and focus on computational modeling, but how much of computational modeling is actually based on our understanding of the brain? There is nothing in the world that comes close to the computational/dynamical capacities of the brain, and I think it would be crazy not to use brain as inspiration, or as the template on which to build our models. I would argue that computational neuroscience (a field still in its infancy) and 'biological' neuroscience cannot and should not exist independently. Models of cognition should not develop in a vacuum, and the biology should inform the computational side of neuroscience. You may point out the successes computational modeling has had in isolation, but I would say that we have a very long way to go, and that the headway we've made has been mostly in parsing apart the "how's" of sensory processing. Integrating and regulating sensory processes is something the brain does well, and something we have no idea how to model.

      Delete
    4. This is why I think it's important to strive for a T4 robot rather than a T3. You have claimed that it would be better to figure out just "one way to do cognition": to develop one particular hybrid of computation/dynamics that can do what we do, without bothering to make sure that it does so using the same causative mechanisms that we do. But how do we figure out "even one causal mechanism" without having even an inkling of how our own brains do it? Maybe I'm not imaginative enough, but I think we'd be hard-pressed to come up with our own independent computation/dynamic implementation without using our obvious cheat-sheet: the brain. I don't think it's important that our Turing indistinguishable robot be biological in nature; that's not what I'm saying. What I'm saying, is that I think we'd have an easier time reverse engineering a T4 robot than a T3 robot, and that it would be a mistake to specifically pursue a T3 robot to the point of excluding the obvious blueprint for cognition that we already have.

      Delete
  5. Cook et al. argue that although natural selection may have set the background for later development of mirror neurons, the matching properties of observation to execution are due to sensorimotor learning. To further the associative learning theory he argues that "MNs respond maximally to unnatural stimuli - that is, stimuli to which the evolutionary ancestors of contemporary monkeys could not possibly have been exposed - is hard to reconcile with the genetic hypothesis"

    This statement immediately had me questioning his claims; just because a brain property has been acquired ancestrally via natural selection does not mean that the practical function of the cell/region is 'time-locked' to the social, cultural, survival etc. needs of that time. For instance, cells in the Fusiform Face Area are specialized to recognize faces. Perhaps generations and generations ago the most important, highly similar items to distinguish amongst were human faces. However, today, the cells in the FFA of individuals who are experts in say bird watching, or cars respond preferentially to faces of birds or types cars (Gauthier, I., Williams, P., Tarr, M., & Tanaka, J. (1997). Training ‘greeble’ experts: A framework for studying expert object recognition processes.Vision Research, 38(15), 2401-2428). Although the FFA might be specified for face recognition (just like mirror neurons might have been evolutionarily selected), it can still alter the objects to which it responds. Thus, I don't buy Cook’s claim that the fact that mirror neurons respond to ‘contemporary, modern’ stimuli means that they could not have possibly been genetically specified.

    On another note, I found an interesting study regarding the mirror neuron system of individuals who have been visually deprived of observing other human behaviour, i.e. have not developed strong connections between their visual sensory neurons and motor neurons coding similar actions. Kupers et al. found an efficient premotor-temporoparietal network mirror neuron system in congenitally blind individuals in response to aurally presented actions (Kupers, R., Pietrini, P., Ricciardi, E., & Ptito, M. (2011). The Nature of Consciousness in the Visually Deprived Brain. Frontiers in Psychology Front. Psychology, 2.) I believe this fuels support for Cook’s belief that the matching properties of the mirror neurons are due to experiential sensorimotor associative learning, which may take the form of visual observation or auditory observation.

    ReplyDelete
    Replies
    1. "just because a brain property has been acquired ancestrally via natural selection does not mean that the practical function of the cell/region is 'time-locked' to the social, cultural, survival etc. needs of that time"

      I had the same feeling while reading this. The important aspect that both sides "seem" to agree on is that mirror neurons evolved with the capacity to perform the action they perform, but WHAT they perform their action to (modern stimuli, in the case of Cook) doesn't necessarily have to be concluded to be evolutionary in nature and I think if we do conclude that Cook is correct then we are jumping the gun quite a bit and perhaps seriously overlooking functions of neural plasticity. I'm glad someone else vocalized this!

      Delete
    2. Yes, inborn ancestral detectors could change their function through learning. But I think Cook et al agree: They think MNs learn the correlations they detect through "associative learning." And yes, the plasticity might even extend to another sensory modality.

      Delete
  6. While reading this paper, I found the authors advanced some good claims with substantial evidence to back them up. I agree that there is not enough evidence to conclusively say mirror neurons are involved in high-level “action understanding,” partly because this is such an ambiguous term and will be tough to dissociate from action perception and action execution. Also, it seems to be amalgamating cognition with the physical aspects of the mirror neuron system without adequately explaining the causal mechanism of how this would come about.

    “Using all of the measures of mirror mechanism activity commonly applied to humans, they have shown that relatively brief periods of sensorimotor experience can enhance, abolish, reverse, and induce mirror mechanism activity. Each of these findings confirms a novel prediction of the associative account: it reveals flexibility of exactly the kind one would expect if MNs/mechanisms are forged by sensorimotor associative learning. In contrast, this kind of flexibility does not provide any support for the genetic hypothesis” (187). Here the authors are arguing purely for a domain general view of associative learning. I think this is strong evidence that there has to be some learning from sensorimotor behaviour going on, and that a totally domain-specific view is hard to argue. However, it still may be possible that we have a genetic predisposition in our perceptual-motor system that has been shaped through evolution, such as babies focusing on faces or self-motion of hands/legs (this is a little in support of the canalization view).

    It is clear that through present-day functional localization and neuroimaging techniques, we are no closer to answering the hard questions such as how we do what we do. Rather, we are left with huge amounts of correlational information of physical processes and their relationship to observed behaviour. We still do not know the precise mechanisms behind our actions. At this point, we are only solving toy problems (in terms of Turing Hierarchy), instead of explaining the unified cognition that we experience. That being said, I do still see the value in the studies of mirror neurons. The knowledge behind the mirror neuron system has allowed new clinical techniques for those with autism or those undergoing motor rehabilitation after a stroke. Thus, as imitation deficits in children with autism is a robust finding, there are studies stating the beneficial role of treatment involving imitation. As well, the activation of premotor neurons while simply observing actions to aid in rehabilitation of motor functions is another appealing clinical finding.

    ReplyDelete
    Replies
    1. What is it about autism and imitation that we learned from the discovery of MNs?

      Delete
    2. Hopefully this is helps elaborate on autism (ASD)

      Deficits that appear in social cognition in schizophrenia and autism spectrum disorder (ASDs) may be explained by MNS dysfunction. Both studies conducted on people with ASD and people with schizophrenia have reported mixed findings. Abnormal mu rhythm suppression is reported in people with ASD while they observe actions. And both lowering and heighten of mu suppression has been reported with people who have schizophrenia. There is a relationship between mu suppression and psychotic symptoms.

      It is suggested that the mixed results may be happening because of deficits in reward processing. Reward processing has a significant influence on vicarious motor activity and motor learning. It is possible that these abnormalities in reward processing could have a detrimental impact on the development of the MNS. Which, as a result, leads to impairments in social cognition present in people with ASD and in people with schizophrenia. Social deficits in people with ASD could also be happing because of impaired responses to social rewards.

      Conclusion: Reward influences the motor-related ares of the MNS. This (reward influences) may help to explain the inconsistencies found in previous studies.

      Delete
    3. We knew that people had mirror capacity before discovering MNs. I have no idea what the mu-acivity differences between autists and schizophrenics mean, but what do MNs add?

      We are in non-causal correlation land in all this. Just weak correlations between various very remotely related variables. Are we really gaining causal explanations or understanding of anything from this, or just correlations to hang a horoscope on?

      (It's good to ask ourselves that question now and again!)

      BTW, Lucy, do you think kid-sib would have been informed (uncertainty reduced) by the "deficits in reward processing" explanation? Would he have learned something, given that he has no particular loyalty to the jargon that is momentarily generating the doctoral dissertations in the field?

      Delete
    4. This comment has been removed by the author.

      Delete
    5. Here is my attempt at the kid-sib definition for 'deficits in reward processing’:

      A reward is a thing (stimulus) given to someone. Rewards are usually pleasurable things, and are often used as reinforcers. Reinforcers are something used to alter a behaviour. For example, if a soccer team wins a soccer tournament. If the coach wanted the team to win the next tournament, the coach might give the team a pleasurable reward (ex: ice-cream). In other words, if the coach wanted to see the behaviour again, he might give them a reward to reinforce the behaviour that the team just displayed (winning the tournament). The players would hopefully be more driven to win again since they would receive the reward of ice-cream after the next tournament if they win.

      People who are diagnosed with ASD often have problems with reward processing. They have a difficult time displaying the miserable behaviour in order to receive the reward again.

      Delete
  7. [Action understanding]. “The term plays a key role in the genetic account; it describes the adaptive function of MNs, the effects that made them a target of positive selection pressure. However, there is still no consensus about exactly what is meant by “action understanding,” or how it differs from cognate functions such as “action perception,” “action recognition,” and “action selection””… “these descriptions do not provide an operational definition of action understanding, that is, a definition that would allow behavior based on (this kind of) action understanding to be distinguished empirically from behavior based on other processes.”

    The relevant part of this article to this class seems to me to be about how mirror neurons may allow for “action understanding”. The problem is I have no idea what “action understanding is”. Near the introduction they give the quote I have above, promising to clarify things in section 8. If they tried doing so then they failed because I found section 8 to be of no help here. The article seems to throw around a bunch of synonyms or weasel words to ascribe more words to the term “action understanding” without actually saying what any of it means.

    My immediate impression from reading through the article once was that one of the functions of mirror neurons was for “action understanding” and my intuition led me to believe that this was related to cognition. I imagined “action understanding” as a sort of underlying cognitive function that allowed us to make sense of how we are able to create an action and what to expect of that action in the real world. Similar to how we may see someone do a movement and we can then can fire the correct neurons so that we too may do a similar movement of our own.

    The article talked about action understanding in terms of goals or intentions for a few paragraphs and so I associated it with some sort of executive function ability that mirror neurons helped us to control the initiation of movements. Thus in this sense and the one described above mirror neurons may even help with what we know as consciousness because they could help give us the “control” that we associate with being conscious.

    This could all be very wrong but since I found the article failing to explain their points on the matter (at least in a coherent or clear way), I found my conjecture to be much more interesting.

    -Jordan

    ReplyDelete
    Replies
    1. Hi Jordan,

      I’m in the same boat as you. It seems as though one of the most pertinent concepts or ideas in this paper is the idea of the action understanding. If I understand correctly, action understanding seems to be the adaptation that has resulted natural selection according to the Genetic account for MNs. Although we don’t seem to be given a true definition of what this means in section 8, a couple interesting points have been brought up later in the article that may clarify (or confuse) us, and/or prompt us to think about revising the term “action understanding”.

      “Supporters of the genetic hypothesis argue that examination of the field properties of MNs shows that they encode “goals,” and this characteristic indicates that they were designed by genetic evolution to mediate action understanding (Bonini & Ferrari2011 ; Rizzolatti & Craigh-ero2004; Rizzolatti & Sinigaglia2010 ). We therefore begin our survey of the evidence by considering how well the neurophysiological data accord with this view.” (Section 4)

      This statement says that neurophysiological data should be examined to observe if MNs can encode goals and mediate action understanding. In short, it’s suggesting that electrical activity in the brain will be able to discern if neurons are encoding goals. This seems like a very quasi-objective way of proving a hypothesis. While we can record electrical activity in the brain (hence “objective”), it takes some interpretive stretch (“quasi”) to attribute this electrical activity to “goals” in the outside world. This got me thinking that maybe “action understanding” or “goal encoding” are two things for which we cannot provide evidence.

      “Is this hypothesis claiming that MN activity causes or constitutes “action understanding” ?” (Section 9.1.2)

      To me, this quotation parallels the discussion we have been having in class. The one in which we argue if neurons firing, or computers/machines/brains performing computations IS cognition, or if it causes cognition. In the Searle article the Brain Simulator reply was dismissed as insufficient because although you can create something that works like a brain, if you haven’t created something that simulates its “causal powers” which is claimed to be consciousness/feeling. So although we could record electrophysiologically to identify if the brain is firing we still have no way of knowing if this causally results in “understanding”.

      Delete
    2. "9.1.2. System-level theory. If MNs were a genetic adaptation, one could argue that new categories of psychological functioning – such as “action understanding” and “motor resonance” – are necessary to characterize what they do. It could be argued that, since they were “specially created” by evolution, MNs are likely to have a highly distinctive, largely independent, and previously unrecognized
      psychological function. In contrast, by showing that established psychological theory – associative learning theory – can cast light on the origin of MNs, the associative account underlines the value of embedding research on MN function within system-level psychological and computational theories of how the brain produces behavior (Giese
      & Poggio 2003; Kilner 2011; Kilner et al. 2007a). This implies that hypotheses about MN function should specify a part in a process – a process that goes all the way from peripheral sensory input to overt motor output – that MNs are thought to fulfill. The name assigned to this part is not important in itself. What is important is that the hypothetical function of MNs is distinguished clearly from other components of the same overall process. For example, in this kind of system-level, theory-guided approach, “action understanding” would be distinguished from components that are likely to be more purely perceptual (which might be called “action perception” or “action
      recognition”), more purely motoric (e.g., “action execution”), or to constitute a higher level of “understanding” (e.g., “mentalizing”). This approach would also make it clear whether the hypothetical function is thought to be optional or obligatory; whether it can be, or must be,
      done by MNs. The kind of system-level theoretical approach required in research on the functions of MNs is exemplified by studies of their role in speech perception (Lotto et al. 2009; Scott et al. 2009).
      A system-level theoretical approach would also overcome a problem that has haunted discussions of the “action understanding” hypothesis since MNs were discovered: Is this hypothesis claiming that MN activity causes or constitutes “action understanding”? The former is an
      empirically testable hypothesis suggesting that there is a distinctive behavioral competence (the nature of which has not yet been specified, see sect. 3.1), called “action understanding,” to which the activity of MNs contributes. The latter implies that the firing of MNs during action observation is, in itself, a form of “action understanding”;
      it does not need to have further consequences in order to qualify as “action understanding.” This claim is not subject to empirical evaluation; it is true, or otherwise, by virtue of the meanings of words."

      Delete
    3. Above I put the full section on System Level Theory because it seems to provide the best explanation for the authors' understanding (and confusion) about what is meant by 'action understanding' being 'encoded' in MNs according to the genetic account.

      I am still confused about what is meant by "system-level psychological and computational theories of how the brain produces behavior" though! The authors specify later in the excerpt that "in this kind of system-level, theory-guided approach" action understanding would be differentiated from action 'perception', and elevated to a "higher level".

      For me, this comes back to the issue raised in class about most of us (students) not understanding the actual power of computational theory. The authors seem to advocate the importance of this theory, while maintaining that the concept of a MN intrinsically having 'action understanding' needs to be revisited: "The latter implies that the firing of MNs during action observation is, in itself, a form of “action understanding”; it does not need to have further consequences in order to qualify as “action understanding.” This claim is not subject to empirical evaluation; it is true, or otherwise, by virtue of the meanings of words." "

      Delete
    4. Understanding Action Understanding

      The path from Mirror Neurons (MNs) to imitation is fairly straightforward (though MNs explain nothing).

      The path from MNs to empathy is a little more fanciful, but it still makes sense: Emotions feel like something. To the extent that we do something when we are feeling that emotion (e.g., smile), we could recognize the action in others and correlate it with what we are feeling when we do that.

      "Mind-reading" in general would go along much the same lines: We can detect the correlation between what is going on in our own minds when we do certain things, and from that we can infer that others have in mind. (This is especially true when their actions are verbal!)

      Let remind you again, though, that the existence of MNs does not add to or clarify any of this: It's just another brain correlate. We don't know how the brain does it even for just detecting that another is making the same movement as me. And we already know that we could all tell when someone else is making the same movement as me, when their voice of facial expression was like what I do when I feel like that, and, in general, that when from many of the things people do, we can guess what they are thinking and feelng -- especially when what they are doing is talking, hence telling us what they have in mind. Yes, we knew all that. The discovery of MNs did not tell us we could do all that. Nor did they explain how we can do all that.

      Now action "understanding." Well, where that came from was this: If we can take it that (1) to "understand" that when another moves his finger, he's doing this (I move my finger), and I know what it feels like for me to do this, and hence that that's what it means to "understand" that movement, and that (2) when someone else makes a grimace, it means he's feeling what I feel when I make that grimace, and that (3) when he does certain things that I do when I have certain things in mind, then I understand what he has in mind -- if all of that (1-3) is what we mean by "understanding" a movement, a facial expression, or actions (including especially verbal actions) -- then (somehow) having discovered that there are MNs, we now understand something about action understanding: the action of the other, in relation to my own action, and what it feels like to perform that action.

      In particular, if I stretch out to try to grasp something out of reach, I can also understand when someone else stretches out like that, that he wants to grasp something out of reach. I know what that feels like to want to reach, so I understand that he wants to reach, and hence I understand what it means to want to reach something. This is even more dramatic in the case of language, where -- if this kind of reasoning makes sense and is explaining something -- MNs have now even given me a clue as to what it is to understand not just a purposive action like trying to reach something, but even the action of verbalizing "I have a splitting headache" -- or "I want to reach that cup over there."

      If I don't sound very impressed by what we have learned from the discovery of MNs then I have managed to convey my understanding...

      Delete
  8. "After training, this pattern was reversed, for example, observation of index finger movement elicited more activity in the little finger muscle than observation of little finger movement” (Cooke et al. 2014).

    This experiment was perhaps the most fascinating presented in the article, presenting strong evidence for the associative account over the genetic account of mirror neurons. Because a genetic basis for the development of mirror neurons would involve inflexibility (through a stimulus-specific response), this experiment demonstrates that it is possible to form new associations. By showing the motor evoked potentials (the increase in activity recorded in a specific muscle) can be formed, through association, with a completely different muscle after a 90 minute training session, the genetic hypothesis seems to fall apart. Correspondingly, evidence for the associative account increases.

    What I find remarkable about the experiment is that it only took 90 minutes to discover a scientifically significant riding that the associations did reverse. On one hand, it shows how adaptive this response can be, if in the future it is shown that Mirror Neurons do help with behavioural discrimination. One might argue such a flexible process would have a hard time storing useful associations if they could be altered so quickly by different experiences. However, on the whole, this learning process would provide adaptive learning even if some instances resulted in a maladaptive response. In addition, in the real world people don’t repeat associations for 90 minutes straight and I think in the environment the formation of the associations responsible for MNs would occur over time, so the process is not as flexible as presented in the study. This author was able to expedite the process to show that MNs do appear to function according to the associative account.

    ReplyDelete
    Replies
    1. So the brain correlates between my actions and the actions of others can be modified by associative learning. MNs, whatever they do, and however they do it, have some plasticity. They are not rigid inborn correlations. (But didn't we know we could learn, and that sensorimotor correlations, in particular, are very modifiable, as the prism adaptation experiments have shown us?) Why shouldn't the same be possible with mirror action correlations between self and others?

      Delete
  9. 1
    “In its starkest form, the genetic hypothesis would suggest that gene-based natural selection has provided each individual – monkey and human – with MNs that code the mapping between a fixed set of observed and executed actions, and that experience plays a minimal role in the development of the observation-execution matching properties of these neurons.”

    “However, [the associative learning hypothesis] suggests that the cardinal matching properties of MNs are a product, not of a specific genetic predisposition, but of domain-general processes of associative learning – the same kind of learning that produces Pavlovian and instrumental conditioning phenomena.”

    After having finished reading the article, it struck me that the associative learning hypothesis fits better with the evolutionary psychology argument (despite the fact that the structure of that argument can be tenuous) than the genetic hypothesis does in the first place. This is especially true if MN function is built through social interaction, as the authors suggest. Being able to function efficiently in different environments would require MNs to be able to engage in different functions, and not be strictly limited to certain movements that were adaptive at one point in our evolutionary history.

    My point analogous to the argument the authors presented about the Visual Word Form Area (VWFA) being specialized for object recognition, though through this it has endowed us with the ability to read. In my opinion, having neurocognitive functions that have the general ability to mold to our specific needs is a more interesting and impressive evolutionary feature to inherit.

    2
    A second thing that struck me about this article is the implication that “action understanding” is a result of specific, discrete neural activity. Despite the fact that understanding wasn’t defined in a satisfactory manner (“meaning” of an action, a “richer understanding,” “real understanding,” or “understanding from within” are all awfully circular), it made me a little confused about something. Would the genetic hypothesis imply a computationalist account of cognition? Or would the mirror neurons have (somehow) encoded the goals of the actions which cause them to fire, and the meaning associated with that? If so, would that be related to the symbol grounding problem? Obviously, this all hinges on the genetic account being accurate, which seems unlikely given the solid arguments and evidence presented in the paper.

    ReplyDelete
    Replies
    1. "Would the genetic hypothesis imply a computationalist account of cognition? Or would the mirror neurons have (somehow) encoded the goals of the actions which cause them to fire, and the meaning associated with that? If so, would that be related to the symbol grounding problem? "
      I quite agree with your line of question! the genetic account, certainly seems to give a 'we just know' because MNs 'just know' type of account which would seem to be similarly dispelled by Searle's CRA.

      Delete
    2. Yes, the brain has a lot of plasticity -- in other words, we can learn and adapt. But this would be equally true whether the learning was computational, dynamic or hybrid. So Searle argument is not relevant here. But since MNs don't explain anything, whether they are rigid or plastic, there is not really anything to argue for or against!

      Delete
    3. "Yes, the brain has a lot of plasticity -- in other words, we can learn and adapt. But this would be equally true whether the learning was computational, dynamic or hybrid. So Searle argument is not relevant here. But since MNs don't explain anything, whether they are rigid or plastic, there is not really anything to argue for or against!"

      Hi Prof Harnad,

      This is somewhat off-topic, but in reflecting on the course I have somewhat of a basic question about what the tenets of a 'causal explanation' are.

      Indeed our discussion (and criticism) of neuroscience and imaging studies in particular hinge on the fact that they DON'T EXPLAIN ANYTHING. (Not angry just capped for emphasis.)

      But how do we ascertain the level of causal explanation that is sufficient to "explain" something. We consistently talk about causal mechanisms, but what constitutes a causal mechanism? (Note I'm playing a bit of devil's advocate here.)

      In one of my later skywritings, I bring up an analogy between the study of gravitation and the study of feeling (using heterophenomenology), and here I would invoke a similar argument. Why do we all agree that an equation that encapsulates the law of gravitation is sufficient to causally explain why planets behave the way they do, and yet a (hypothetical) algorithm that links neural correlates to inputs and behaviours is insufficient because it "explains nothing". What if it's predictive and accurate?

      I can feel the criticism already pointing out that our 'hypothetical algorithm' actually sidesteps the problem and doesn't explain how our brains do it... but then how does the law of gravitation explain how the universe does it? These laws which we tout as causal explanations seem to be merely descriptive.

      Furthermore, what level of causal explanation is sufficient for "explanation"? As I pointed out in another skywriting, there will always be an infinite list of why's associated with any explanation (see: annoying toddlers), so when do we decide to stop? Why is our predictive heterophenomenological algorithm not explanatory enough?

      Maybe I'm getting lost in my own head, but I feel like I'm touching on a kernel of truth somewhere in there...

      Delete
  10. If I understand this paper correctly, the distinction between the genetic and associative views is really a chicken and egg problem. Did motor neurons come first via evolutionary mechanisms and develop with associative learning? (genetic view) or did associative learning come first through evolution and serve to produce these mirror neurons? (associative view). I think the author is very throughout and has well-supported claims. Just to play devil’s advocate though, there is something for which clarification would be nice. As I understand, there are neurons in the brain which are involved in sensorimotor and associative learning but which are not motor neurons. Therefore, it might be reasonable to say that the capacity for a particular set of neurons to become “mirror” neurons as opposed to other varieties of neurons involved in associative learning was evolutionarily dictated. Of course, their functions as mirror neurons may have to be developed via associative learning but perhaps this capacity to become mirror neurons is genetic? This might represent a more nature and nature view that they talk about in the paper where nature provides the propensity and nurture allows it to blossom. Thus wouldn’t be almost all genetics or almost all learning but a nice mix of both where one lays the groundwork for the other.

    ReplyDelete
    Replies
    1. While reading the paper, I had the same feeling, that the distinction between the two views was actually a chicken and egg problem. Even if they defend the associative view with robust claims, I felt that the two view’s distinction was sometime aimless. Is it really important to know whether or not the MN became to be as a result of “genetic adaptation for action understanding” or as a result of associative learning need for such neurons? If MN evolve from genetic or to a specific need from social interaction is a question without answer, and to argue for one or the other theory isn’t really relevant to the real issue: to know if the activation of such neurons really are encoding the understanding of others actions.

      Delete
    2. Though I had the same intuition initially (chicken vs.egg) on re-reading the section '3.3 Not nature versus nurture', I think the point is a little more subtle. The authors specify "the two accounts differ in the specific roles they assign to genetic evolution and to learning, and
      in the types of experience they take to be important, in producing
      the characteristic matching properties of MNs. The
      genetic hypothesis says that genetic evolution has played a
      specific and decisive role, and learning – based on sensory
      and/or motor experience – plays a merely facilitative role,
      in the development of MNs. In contrast, the associative
      hypothesis says that genetic evolution has played a nonspecific
      background role, and that the characteristic matching
      properties of MNs are forged by sensorimotor learning."

      I think the differences in the two accounts can be (very) roughly chalked up to the importance they afford to questions of function and question of origin. As they mention themselves in the beginning, "much of the first 20 years ofMN research has been
      devoted to theorizing and speculation about their functions.
      In contrast, the primary focus of this article is the
      origin of MNs. Our principal questions are not “What do
      MNs do?” or “What are they for?”, but “What is the
      process that gives MNs their ‘mirrorness’; their fascinating,
      cardinal capacity to match observed with executed
      actions?”
      The standard view of MNs, which we will call the
      “genetic account,” alloys a claim about the origin of MNs
      with a claim about their function. It suggests that the mirrorness
      of MNs is due primarily to heritable genetic factors,
      and that the genetic predisposition to develop MNs evolved
      because MNs facilitate action understanding. In the sense
      of “an adaptation” developed by G. C. Williams, and used
      in Evolutionary Psychology, the genetic account casts
      MNs as an adaptation for action understanding. In contrast,
      we argue in this article that the balance of evidence currently
      favors an “associative account” of MNs, which separates
      questions about their origin and function."

      The genetic account assumes that 'action understanding' is inherently encoded in mirror neurons and that these neurons have precisely developed in order to afford us this 'understanding', thus that genetic evolution specifically selected them for this purpose (and associative learning plays a minor role similar to that assumed in the 'poverty of stimulus' argument). On the other hand the associative account disagrees that MNs have 'action understanding' already inherently encoded in them and argue instead that they have the inherent 'capacity' to encode these understandings and that the way they develop to 'understand' will be majorly forged by associative learning (observation, imitation, trial and error, etc...). I think this statement illustrates nicely the point of view of the associative account : "Rather, it is possible that MNs are constructed by
      domain-general processes of associative learning, and are
      recruited in the course of development to contribute
      to one or more psychological functions, without either
      the construction or the recruitment processes having
      become a specific target of gene-based selection"

      Delete
    3. As to whether or not it is relevant to know if its more genetically predetermined or social learning based, I think the explanations of 'action understanding' each account offers is useful to keep in mind. In the larger scope of how it is we are able to 'understand' I was definitely disappointed by the genetic account's account of action understanding mediated by MNs:
      "The term “action understanding” was introduced by Rizzolatti
      and colleagues to characterize the function of MNs
      (Rizzolatti & Fadiga 1998; Rizzolatti et al. 1996). As far as
      we are aware, it had not previously been used in research
      on animal or human cognition. The term plays a key role
      in the genetic account; it describes the adaptive function
      of MNs, the effects that made them a target of positive
      selection pressure. However, there is still no consensus
      about exactly what is meant by “action understanding,” or
      how it differs from cognate functions such as “action perception,”
      “action recognition,” and “action selection”
      (Gallese et al. 2011). Attempts to clarify have emphasized
      that, in comparison with purely visual processing of
      action, MN activity relates to the “meaning” of an action
      and yields a “richer understanding,” “real understanding,”
      or “understanding from within” (Gallese et al. 2011; Rizzolatti
      & Sinigaglia 2010)."

      Delete
    4. Yes, there is not just sensorimotor learning, but also sensory-sensory learning, and it too has evolutionary bases and plasticity. But passive sensory-sensory learning (Pavlovian) has much less power than sensorimotor learning (operant), if only because when you are just viewing things passively, all your brain can pick up is patterns of S-S correlation, without learning what's what and what's right or wrong. Once you have to do something, and that something can be right or wrong, you can learn from the feedback from the consequences of doing the right or wrong thing. (Think of information and uncertainty-reduction in having to choose among options that matter to you.)

      But this is neither here nor there for MNs, since we haven't the faintest idea what they do or how -- just when and where they are active. (See Fodor's critique.)

      On understanding "action understanding," see above...

      I would say the genetic and associative alternatives are too vague to even be called "chicken vs egg" (or, as I prefer it, "seed vs plant"!).

      Delete
  11. I found a lot of the debate and argument in this paper is grounded in the need to reduce brain function into one simple explanation. One thing I have learned from the many neuroscience courses that I have taken is that yes you can localize function to parts of the brain but these localizations and functions are never total. In Cook et al. (2014) evidence is continuously put forward, for and against, either the “genetic account” or the “associative learning” account of mirror neuron function. The authors then go on to present different evidence for the genetic account, such as the “poverty of the stimulus” argument, and then counter it with evidence from the, in the authors position, superior associative learning account. What I find hard to believe though is that the explanation for mirror neurons is simply only one of these arguments. I do agree evidence presented under the “Sensorimotor learning changes mirror neurons” heading enforces the associative learning argument, but this only tells a portion of the story and does not conclusively deny other arguments entirely. Instead of confining our scope of research to if something is either “this” or “that,” research should strive to develop multidimensional theories that do not fall into the trap of definitive assumption. In the case of mirror neurons, maybe associative learning explains part of the picture, leaving room for genetic or “hybrid” accounts discussed later on, but why assume that the answer has to be either or.

    ReplyDelete
    Replies
    1. Hey Cristian, I do agree with you, that the article does seem to simplify the brain structure and the tunnel focus on sensorimotor learning instead of a multidimensional scope of the mirror neurons. However, in simplifying their approach to brain system the authors was able to fully expand in their argument focus for, such as 'genetic account' vs 'associative learning' and the 'wealth and poverty of stimulus' with well supported examples. On the other hand, coming back to your point, I do also find that the article lacking the complete picture to fully explore the potential of mirror neuron, even with its attempt on the conclusion on p.191 where it does state that there exists "the possibility that MN's play one or more important roles in the control of social interactions." (Cook et al 2014). Which to me, may be merely an attempt to not neglect other functions. What do you think? Thanks

      Delete
    2. Maybe trying to figure out how MNs (or whatever bigger structures they are part of) do what they do -- thereby explaining how we do it would be more informative than just worrying about they do it innately or through learning?

      Delete
  12. Even before I read Fodor, I had the nagging question while I read Cook et al.: why should I care?

    Mirror neurons are kind of interesting to me because they potentially disclose part of our capacities for "mind reading," our ability to determine the goals/ambitions of our peers in their actions. Presumably the genetic account says more about this capacity than the associative learning model presented here, though neither is truly groundbreaking.

    Maybe I'm just being grumpy...

    ReplyDelete
    Replies
    1. Interested in why you think the genetic account potentially tells us more about our mind-reading capacity than the associative. I suppose that if you were to consider it just in terms of having an evolutionary benefit re: the ability to function well as a social animal, that makes sense. But I was also nagged by this passage:

      In its starkest form, the genetic hypothesis would suggest that gene-based natural selection has provided each individual … with MNs that code the mapping between a fixed set of observed and executed actions, and that experience plays a minimal role …

      (I know the authors also go on to say experience isn’t totally discounted by the genetic hypothesis, but all they credit it with is ‘maturation’ of the MNs/ 'sensorimotor links’, so I don’t think that bit is of much importance here.)

      What struck me from the sentence above is that it seems that if humans were genetically endowed with a fixed set of actions that MNs can map, that wouldn’t really tell us much about our ability to determine the goals/ambitions of our peers with such aptitude, since the evolutionary timespan required for that fixed set to be augmented or altered is far longer than the comparatively brief amount of time during which human life has gone from somewhat primitive (for lack of better term) to highly socially complex. It just seems to me that the fact that we are so skillful at 'mind-reading' very subtle and often complicated intentions/goals/ambitions etc. would favour the associative account, which allows for a lot more flexibility in the types of actions and observations that can be paired.

      P.S. I do share in your grumpiness though re: why should we really care

      Delete
    2. I share the grumpiness too, but that was intentional (dictated by my MNs!). Kid-sib is grumpy too. So far, he says, I haven't learned anything from the existence of MNs that I did not know already about what people (and hence their brains, somehow) could do. All I learned from MNs is that they seem to be (perhaps part of) what the brain is doing when people do those things.

      But I haven't learned how they (or their MNs) do it. Fussing over whether that "how" is inborn or learned does not help either, though it's obvious that it's at least partly learned.

      Delete
  13. First of all, does anyone have an idea why researchers would be satisfied to use the term 'mirror neurons' to signify neurons that fire "regardless of whether the observed and executed actions are even broadly similar to one another"(section 2.1) Does this not seem like it would defeat the entire purpose of singling out mirror neurons in the first place? I have a feeling that these neurons are mistakenly being regarded as mirror neurons when they really could be firing in response to other processes happening at the same time, but maybe this is just something related to ‘mirror systems’.

    Secondly, I’m wondering what would happen if someone were not exposed to facial expressions or normal visual stimuli at a young age and if this would affect the generation of MNs. It seems that if the hypothesis this paper puts forth is taken to be true, then MNs would not develop in these people.

    I’m also highly unsatisfied with the argument presented against the genetic hypothesis of the MNs that says “the fact that these MNs respond maximally to unnatural stimuli—that is, stimuli to which the evolutionary ancestors of contemporary monkeys could not possibly have been exposed—is hard to reconcile with the genetic hypothesis” Does anybody agree with it? I’m curious to know because to me it seems this argument is grasping at straws. In order for something to be evolutionary adaptive that doesn’t meant that it had to only be beneficial/useful at the time. That would be like saying that taste is not evolutionarily adaptive because we can now taste artificially created substances that weren’t available to our ancestors. Maybe I’m missing the point but this doesn’t seem like a sound argument.

    ReplyDelete
    Replies
    1. Yes, MN's could be part of a larger system, and it is not always clear what their activity is correlated with. And yes, there's plasticity in just about everything else -- why wouldn't there be for MNs?

      Delete
  14. This whole talk about the genetic versus the associate view of mirror neurons is basically a battle of nature versus nurture isn’t it? There are so many things in psychology and science and in different fields concerning humans (and animals) about whether things are nature versus nurture. Is language nature versus nurture? Is our personality nature versus nurture? Is sexual orientation nature versus nurture? I think for these three questions in particular it is a mix. I also think it is a bit of both for the mirror neurons. I don’t think one side could be right and one could be wrong. Anything a human adapts and is able to do: speak, have a certain personality, have a specific orientation; I believe they have a predisposition for it. So in this case, I think that humans (and animals) have a predisposition for the mirror neuron. I think it is there when we are born, from genetics and from over the years but that it isn’t used or becomes visible until it needs to be. Once it is used and exercised (the associative part) then it becomes visible at the microscopic level that it exists. Think about the personality traits of someone. Someone may be genetically predisposed to be super organized and a perfectionist. But that part of them may not become visible and come about till they need it to (at a job, in school), but the part of them was always there invisibly. At least that’s my opinion.

    ReplyDelete
    Replies
    1. Hi Jordana,
      I have a similar opinion on the matter, but I think that the associative view still recognizes the nature aspect. From what I understood, the associative hypothesis recognizes that genes create the domain-general neurons, but that these neural circuits require sensorimotor experience to develop into motor neurons. So, while the associative account is more supportive of nurture, it still recognizes a role of nature, just a more general role.

      When you say: "So in this case, I think that humans (and animals) have a predisposition for the mirror neuron. I think it is there when we are born, from genetics and from over the years but that it isn’t used or becomes visible until it needs to be. "
      It seems to me that you support the associative account, that the neurons are there at birth, but must be used to develop this specific functionality. One of the difficult parts about this debate is that animals use these neurons at a very young age/ infancy, so it is difficult to tell at which point the MNs developed their function.

      Delete
    2. Hey guys, this is an interesting topic. I found it humorous that the authors felt the need to include an entire subsection titled “Not nature versus nurture,” because like Jordana pointed out, the argument resembles the “battle of nature versus nurture” a lot of the time. However, I do agree with Amanda that “the associative hypothesis recognizes that genes create the domain-general neurons, but that these neural circuits require sensorimotor experience to develop into motor neurons.”

      Throughout the paper, I couldn't help but feel like the answer may be a hybrid of these two hypotheses. Rather than just having genes that encode domain-general neurons, we could have genes that encode the neural circuits, its just that sensorimotor experience is necessary to give these neural circuits MN function. This would fit with the finding that, as Amanda mentioned, “animals use these neurons at a very young age/ infancy, so it is difficult to tell at which point the MNs developed their function,” as well as the beak morphology and honeybee examples mentioned on page 182.

      From my point of view, this hybrid theory is just the exaptation argument presented on page 188, minus a few semantic differences about the origins of MN function. However, I found the evidence Cook et al. presented to refute the exaptation argument to be rather weak and inadequate, so I'm still not convinced that the associative hypothesis trumps the exaptation hypothesis. Either way, I think some hybrid theory of the associative and the genetic hypotheses will turn out to be correct, and the authors seem to concede this idea as well on page 189: “Hybrid modeling is a promising direction for future research.”

      Delete
    3. But notice that we can get so carried away with whether "it" is nature or nurture that we forget that we still don't know what "it" -- i.e., what MNs really do, and how -- is.

      Delete
  15. The associative hypothesis argues that MN originates in sensorimotor associative learning. So, MN would be the product of a strong association between the motor system and the perceptual system. Convergence of both those system onto another special unit (MN) would result in a more complex and comprehensive understanding of the world. Since MN are thought to encode other people behaviour, their presence must be necessary for the unique human sociability. “It has been suggested that MN dysfunction contributes to a number of disorders, including autism, schizophrenia, Down’s syndrome”. If this is the case, MN must have a strong implication in human social behavior. Does believing in the associative hypothesis give us a clue into how much those special neurons are implicated in human social behavior? When they say: “the associative hypothesis implies that the characteristic, matching properties of MNs result from a genetically evolved process, associative learning, but this process was not “designed” by genetic evolution to produce MNs. Rather, it just happens to produce MNs when the developing system receives correlated experience of observing and executing similar actions”, it seem like it implies the genetic didn’t know what it was doing, in contrast to the associative system. It seems to me that the article gives too much credit to the associative system, as if it was aware of what it was doing. I think that, whether the apparition of MN was due to genetic evolution or to motor and perceptual connexion, both were arbitrary. What really is discernible is the fact that MN emergence might have allowed for human, and its animal relative, to grow sociable and aware of other behavior.

    ReplyDelete
    Replies
    1. Well at least association has the virtue that it is a process rather than just genetic magic. But it is not an explanatory model that generates associative capacity. It just presumes it.

      Delete
  16. Re: 'action understanding' (genetic hypothesis)
    "...in comparison with purely visual processing of action, MN activity relates to the “meaning” of an action and yields a “richer understanding,” “real understanding,” or “understanding from within” (Gallese et al. 2011; Rizzo- latti & Sinigaglia 2010)" (Cook et al 180).

    Borrowing from computation discourse, "behavioural equivalence" seems to be a relevant point in this argument. The obvious connection is that the firing of MN is the behavioural 'output,' and action understanding is arguing that there is genetic software that predisposes the subject to 'compute' the input in ways that a different subject without the genetic software cannot.

    Interestingly enough, how this genetic predisposition functions is pitted against the "purely visual processing of action."

    Correct me if this is a stretch, but could parallels be made to Searle's CRA, such that the genetic hypothesis is saying that its account is comparable to a native Chinese speaker in the room who actually understands the characters and their meanings, whereas the associative account is comparable to a non Chinese speaker who must manipulate the data in order to produce the same output (behavioural equivalence)? If so, the obvious point of contention would be that the native Chinese speaker was not born a fluent Chinese speaker, but rather acquired the language through time and frequent usage.

    The genetic hypothesis is significantly lacking because the usage and definition of "understanding" or "meaning" is so ambiguous. This provokes the reader to think perhaps this simply does become a more basic debate of the nuances of nature vs nurture.

    ReplyDelete
    Replies
    1. Hi Annie,

      It seems to me that the genetic account is more like a computationalist one with “strong equivalency” (the idea that the exact system is present in all of us to have MNs because of the selective advantage given to them when we were evolving). We are all programmed the same way and have the exact same way of using our mirror neurons.

      The associative account seems to be more about trying to say that there is a “weak equivalency” in that we are able to all have the same behavioral output but we don't all have the exact same genetic make up because of evolution.

      So in terms of the behavioral output being the firing of the mirror neurons, I find it relevant to look at those two positions in terms of how they relate to the accounts given in the article for the development of mirror neurons.

      I feel like (as in most cases) the case for nature and nurture independent of one another isn’t strong enough to explain Mirror Neurons.

      Delete
    2. Neither the genetic interpretation of MNs nor the associative interpretation is an "account," in the sense that it explains how they do what they do. Searle's argument applies only to T2 (verbal capacity) and only if it is generated purely computationally.

      Delete

  17. “Correlated excitation of the sensory and motor neurons increases the strength of the connection between them, so that subsequent excitation of the sensory neuron propagates to the motor neuron. Thereafter, the motor neuron fires, not only during execution of lip protrusion, but also, via its connection with the sensory neuron, during observation of lip protrusion; what was originally a motor neuron has become a lip protrusion MN.”
    “When the system receives correlated experience of observing one action and executing a different action, the same associative process produces logically related MNs.”
    - These suggest the possibility of Grandma Cell, which arose from different networks connected through associative learning that brings a MN its function.


    "It is possible that MNs are constructed by domain-general processes of associative learning, and are recruited in the course of development to contribute to one or more psychological functions, without either the construction or the recruitment processes having become a specific target of gene-based selection”
    - This is suggesting that there is functional specialization that occurs in the brain, and through recruitment processes, the brain works as a whole; in this case, is the brain a network in itself, formed of subnetworks of networks?


    "Parieto-frontal mirror neurons are motor neurons. When MNs discharge, they “ignite” motor schemata similar to those endogenously activated during motor imagery and, within limits, during actual motor act execution. In other words, my motor schemata are activated during the observation of similar motor schemata of others."
    ^not from the main article but from the one in the link posted below.
    My main question here about action understanding; this is similar to the other minds problem in that when we view a behavior, how do we understand what the other person is doing without understanding what the action, in itself, is? This article below points out something similar to that specificity of a MN, in which even though MN may have specialized function, it itself still fires when we, ourselves, perform the action that was observed because it's still a motor or sensory neuron itself. This suggests that "action understanding" is an interpretation of a person's behavior that we can understand because we ourselves, must have an understanding of the behavior beforehand. With associative learning, it's simply a connection of what we interpret as associated, thus bringing about this "understanding" of observation with action. The following article, suggests that instead of associative learning, it's rather top-down processes that causes the origin of mirror neurons and it's worth checking out after reading the critics from the main article, such as tongue protrusion and newborns' early associative learning, to see where they differ.

    http://journals.cambridge.org/action/displayFulltext?type=6&fid=9248911&jid=BBS&volumeId=37&issueId=02&aid=9248910&bodyId=&membershipNumber=&societyETOCSession=&fulltextType=AC&fileId=S0140525X13002471title=

    ReplyDelete
    Replies
    1. See "Understanding Action Understanding" above.

      Delete
  18. I was reading through this article feeling pretty bad that I didn’t understand just exactly how Cook’s article connected to Searle’s of yesterweek until I got to page 4 and saw that Rizzolatti had dubbed “Action Understanding” as the functional property of mirror neurons. Later this is coupled with the encoding of goals (object directed actions and high level action-intentions).

    I’m not sure if this is where we are supposed to be going with this, but all of these higher order cognitive words being used to characterize a type of neuron is maybe affording them more capacity then they really have. Surely enough a quick google search and it seems to me that several people are referring to them as the bridge to understand consciousness.

    Using Cook’s evidence, I don’t think mirror neurons are getting us wildly closer to solving the mystery of consciousness, but I do think they’re a great example of an act of computation within the so-far unexplained phenomenon of cognition (cognition is not computation but computation can play a role in cognition; mirror neurons can play a role in cognition). To me, these neurons look like Searle in the Chinese room and are in no way understanding the actions they are firing for. If I put a mirror neuron network in a petri dish and send an electric signal to them identical to the one they receive in the brain, won’t they produce the same output? If so, there’s no understanding to be found, just a transduction of signals. The crucial part is still missing.

    This (to me) looks like one of those times when we ascribe our own abstract cognitive processes and anthropomorphize our brain hardware (such as “The amygdala sees and hears everything” and “how your brain tells you where you are”) because we are so very much uncomfortable with the fact that we do not understand our own consciousness. I think this should be avoided in the realm of cognitive science and consciousness studies.

    ReplyDelete
    Replies
    1. "Using Cook’s evidence, I don’t think mirror neurons are getting us wildly closer to solving the mystery of consciousness, but I do think they’re a great example of an act of computation within the so-far unexplained phenomenon of cognition (cognition is not computation but computation can play a role in cognition; mirror neurons can play a role in cognition)."

      I definitely agree with your comments! While attempting to ascribe a decisive cognitive function to a single neuron or set of neurons seems to be misguided (as you mentioned, a mirror neuron in a petri dish is most likely not providing the output of "understanding"), it doesn't undermine the discovery of said neurons. If we decide to accept that cognition is a hybrid dynamic/computational process, then acquiring knowledge of our own hardware and hypothesizing how it may contribute to the instantiation of certain cognitive functions is not all for naught. The issue is when we get to excited and attribute the fact of their existence as an explanation of "how". I think you put it nicely when you said:

      "This (to me) looks like one of those times when we ascribe our own abstract cognitive processes and anthropomorphize our brain hardware (such as “The amygdala sees and hears everything” and “how your brain tells you where you are”) because we are so very much uncomfortable with the fact that we do not understand our own consciousness. I think this should be avoided in the realm of cognitive science and consciousness studies."

      Delete
    2. The punchline of Searle's Chinese Room Argument was only partly right: He showed that cognition (understanding) could not be just computation. But he thought he had shown that cognition could not be computation at all, even partially, and that the only way to understand its causal mechanism was to study the brain, T4. Well, here we are with MNs, studying the brain. Now are we getting an inkling of how the brain generates cognition (i.e. the capacity to do all that we can do)? Even a little bit, for example, to explain how we can do what the MNs can do, namely, detect when someone else is doing the same thing we are doing?

      Delete
    3. “Using Cook’s evidence, I don’t think mirror neurons are getting us wildly closer to solving the mystery of consciousness, but I do think they’re a great example of an act of computation within the so-far unexplained phenomenon of cognition (cognition is not computation but computation can play a role in cognition; mirror neurons can play a role in cognition). To me, these neurons look like Searle in the Chinese room and are in no way understanding the actions they are firing for.”

      Riona’s post gave me a different perspective on MNs and I find the connection made between Searle and the MNs really interesting. I agree with the idea of computation within cognition. But then again it raises ‘homunculus’ related questions and arguments (or shall I say multiple homunculi inside the head in the form of MNs). Unfortunately the article states that only one study purports direct evidence in humans, so we don’t know to what extent the MNs are responsible for this.

      Additionally, the article states that MNs have been implicated in abilities such as empathy and emotion recognition. Clearly, we do not know how emotions/feelings work and saying that MNs play a role in their functioning is an example of “those times when we ascribe our own abstract cognitive processes … because we are so very much uncomfortable with the fact that we do not understand our own consciousness”. But I don’t see why this should be completely avoided. I believe that it would help us gain some perspective on why humans do what they do (and intentionality maybe?).

      Delete
  19. This meta-analysis of studies investigating mirror neurons leads to some feelings of uncertainty about how the data has been collected and interpreted. As the authors pointed out, there are many things going in the brain. It is difficult, if not impossible to separate what pattern of firing is responsible for which action the person or the animal is performing. With this in mind, it is dubious that mirror neurons even exist and that they are not simply perceptual or motor neurons. As it can be seen in the meta-analysis, the pattern of firing, the location and even the definition of what are these neurons changes from one study to another which makes it even harder to believe in their existence. Furthermore, many of the studies investigating mirror neurons seem to be using techniques that make the study not an experiment but an associative study. For example, fMRI scanning that highlights the relationship between actions being performed and activity in the brain seems to be more of an associative study rather than an experiment. In the studies being performed, it is difficult to be certain that the increased activity in the brain is due to the doing of a particular action and not of some other mechanism happening in the brain. The fact that most of the studies investigating mirror neuron activity relate more to associative studies that can only highlight associations between variables is addressed by the authors at the end of the article. They argue that identifying commonalities between different studies involving different techniques can allow causal inferences to be made from the data gathered in these associative studies, however, is it possible to infer causal relationships by simply analyzing the relationship between associative studies?

    ReplyDelete
    Replies
    1. I'm not sure what you mean by an "associative study," but my guess is you mean a correlational study. If so, you are right that MN data (like almost all brain data) only shows us the correlates of our cognitive capacities. It doesn't explain how they cause them.

      Delete
  20. Within the abstract article, it presented four views to demonstrate, one of them being that mirror neurons do not consistently encode action “goals”. Which raises my question in the discussion of part four where they talked about design for action understanding by dividing into two sections of: goal as object-directed actions, where it is noted that mirror neurons response maximally to unnatural stimuli so it is hard to reconcile with genetic hypothesis; and goal as high-level intentions, where the classes of mirror neurons is sensitive to features of actions that also fall below the generality of intentions.

    However, because the approach is seemingly highly specific to target one “goal” as an intention, so it begs the question in how to consistently track the intended goal for the neuron? As for the conclusion of the paragraph, the article mentioned “[t]hus, these MN [mirror neurons] respond to different object-directed actions, with different intentions, in observe and execute condition. Would this suggest that there still underlies encoding of action ‘goals’ but due to our experiment limitation that it may render impossible for us to predict consistently the nature of the neuron intention?

    ReplyDelete
    Replies
    1. Hi Grace,

      While the I found that Cook et al. put forward compelling evidence in support of the associative account of MNs, I was also left wondering how would we go about truly verifying if MNs are encoding “action goals”. It was in fact the genetic account that was arguing how MNs have this sort of action understanding due to the encoding of action goals but as you mentioned, the associative theory proved that MNs are active even when there is no intentionality/object-directed action.
      There could be such a loose interpretation when it comes to the definition of a “goal” (e.g. ‘reaching out for the cup vs ‘reaching out for the cup to drink’). As in mentioned in the paper, the idea that MNs can encode for “high-level action intentions” becomes quite fuzzy when you take into account their firing for irrelevant features such as handedness, direction sensitivity, view dependence, etc. I feel that there is a lack of convincing evidence to support this goal-coding hypothesis.

      I’m not sure what you mean by “…it may render impossible for us to predict consistently the nature of the neuron intention”? I don’t think the neurons themselves have any intentionality – everything is working together as a system (e.g. we receive sensorimotor input, process it, multiple brain regions work together to produce some output). Although Libet did have that whole argument about freewill… how we see this ‘readiness potential’ (measured by EEG) in neuronal activity before we ourselves become consciously aware of our urge to act (ie. to push a button in his experiment).

      I think the most important question to ask is whether MNs should be given this fancy categorization all on their own; are they really a special kind of neuron that “understands” or is this “understanding” contained in all kinds of neurons and networks?

      Delete
    2. Hello Linda,
      Thank you for pointing out my misphrasing on 'nature of neuron intention'. I do realize that it doesn't make much sense within context, sorry. I do agree with you that there may be some loss of interpretation when it comes to the many possibilities that can be defined as 'goal' or 'intention' But going back to trying to reiterate my question, hopefully more clearly this time, I was thinking along the lines that

      The associative goal-oriented approach is seemingly highly specific to focus on an intention, so how would the experimenter determine what is the initial stimulus that triggers the intention for the goal? That will the experimenter only knows the intention behind the firing of neurons through solely the behavioural output? I am not sure if I am fully expression my idea.

      Even so, with this in mind, we do know that the paper already mentioned (first point on the abstract) that mirror neurons do not consistently encode action ‘goals’. Does that mean that the mirror neurons encode more of a network that focuses more, as the latter part of the article describes, on associative learning of sensorimotor training? What do you think?

      Delete
    3. Hmm...

      "The term 'goal' affords numerous interpretations (Hickok 2009).
      We will consider two definitions commonly adopted, assuming that MNs encode “goals” if they encode object-directed actions (sect. 4.1) or high-level action intentions." (Cook et al. 2014)


      1) Object-directed actions
      The first definition suggests that MNs fire when there is an action involving an object (for example, reaching out to grab a cup). As the paper goes on to mention, there is a subset of MNs that will also fire when there is NO object as well (for example, the sound of tearing of paper or pantomimed actions). Further, it is argued that MNs cannot be genetically evolved because they are indeed responding to unnatural sounds (plastic crumpling sounds) and use of tools (ex. grasping with pliers) which are things that "the evolutionary
      ancestors of contemporary monkeys could not possibly
      have been exposed to" (Cook et al 2014).

      2) High-level action intentions
      This kind of definition is now referring to the end-goal of an action. For example, it's not that there is simply an action of reaching out for the cup anymore, but this action of reaching out for the cup has the purpose of feeding the animal because there's food inside. Once again, they go on to disprove that MNs are solely encoding for this kind of goal because they also fire for similar actions and have firing specificity for which hand is being used, direction sensitivity and other seemingly irrelevant characteristics.

      So basically, what I got out of it, is that MNs are encoding a lot of different things - "action-goals" could be one of them but certainly not all that they're active for.

      Delete
    4. "MNs are encoding a lot of different things" -- or, their activity is correlated with a lot of different things, and we have no idea what (or whether) they are "encoding" anything.

      Delete
  21. I’m not sure I fully understand the distinction that the author makes between associative versus genetic theories about mirror neurons. I understand the logical reasoning that is highlighted in the article: Cook et al. explain that the debate is not between nature versus nurture, but rather over whether the idea of mirror neurons in humans is primarily genetic or environmental (as both are a combination of the two to some ratio or another). The author suggests that the genetic theory misses the point that mirror neurons are not actually the adaptive mechanism, but rather, as associative theory explains, associative learning is the adaptive mechanism, and mirror neurons and whatever role they play in social interaction is a byproduct of that. The point that I am confused by is why mirror neurons couldn’t have first come about via associative learning, and then based on their adaptive role in social behaviour, were then selected for as a genetic trait. I think that the the hand/face hybrid model that suggests that hand and face MNs have different origins seems to be the most plausible, as it allows for both the genetic theory and the associative theory, depending on the how important the context of the mirror neurons is to survival. The author is making a case for associative learning, suggesting that all mirror neurons are a by-product of a genetic predisposition to learn by association over time. However, it seems more plausible that a hybrid of these two theories exist, depending on how adaptive an associative mechanism is. In terms of facial features, it is arguable highly adaptive, and fits in with the evidence that instinctive facial expressions in response to certain emotions are the same across all cultures. I had the impression throughout the article that Cook et al. were arguing against genetic theory altogether, but towards the end they seemed to open up the possibility of a hybrid model. Arguably, there are many brain areas that have been naturally selected for by long-term associative learning already, so there is no reason that mirror neurons can’t be genetically inherited in some systems, and associatively learned in others. I wasn’t sure if the authors were swaying back and forth between hybrid and associative theories, or if they argued for associative learning because it left room for the possibility of a hybrid system, whereas the genetic model does not. I would argue that perhaps the genetic model simply needs to be further refined (the authors touch on this slightly by pointing out several times the lack of definitive definition of the term “action understanding”) so as to categorize the various types of motor neurons in terms of those that are adaptive and those that are a byproduct of another adaptation.

    ReplyDelete
    Replies
    1. It's important to notice that associative learning is not a mechanism, it is a behavior and a behavioral capacity. There are models for simple associative learning -- making associations between unique individual items, like nonsense syllables -- but not for "associations" between my movements and your movements.

      There are sensory-sensory and sensorimotor correlations, so that, potentially, a robot could be designed with the capacity to learn to detect its own movements in a mirror, and eventually to detect movement patterns in other robots that are similar to its own. But that requires designing and texting a real model, to see whether it can do the job, which Cook et al do not do.

      But even given such a model with for the capacity to learn mirror correlations, it is extremely unlikely that single neurons are doing this. And it is unlikely that these mirror correlation detectors would scale up to the fancy capacities that MN researchers routinely claim to be MN functions, such as emotion detection, mind reading and "action understanding" (whatever that means) let alone the most powerful "mirror" capacity of all, which is to understand what you and I both mean by the action of saying "the cat is on the mat."

      Delete
  22. The way Cook et al. addressed one of their counterarguments, the possibility that action understanding could have emerged as an exaptation of MNs, left me both confused and unsatisfied. First, they directly raise the possibility that MN functions have co-opted existing genetically selected neural circuitry in section 3.3:

    Rather, it is possible that MNs are constructed by domain-general processes of associative learning, and are recruited in the course of development to contribute to one or more psychological functions, without either the construction or the recruitment processes having become a specific target of gene-based selection. (page 182)

    Putting aside the debate over how MN circuits were created, they concede that the functions produced by MNs could simply be making use of the pre-existing structures. Why create entirely new neural networks when one you already have will work fine? They even illustrate this idea with the honeybee example. Throughout the paper, this possibility kept popping back into my mind. Okay, MNs have definitely not emerged as a genetic adaptation for action understanding, Cook et al. have convinced me of that. But what about the possibility they raised earlier, that action understanding could have emerged due to associative learning that took advantage of pre-existing circuits that were a genetic adaptation for something else?

    Cook et al. go on to directly address this idea section 8.2. However, I found the first chunk of their counterarguments to the exaptation hypothesis to focus on trivial aspects of it. While I agree that MNs are probably not produced by “a special kind of sensorimotor learning which receives input from self-observation of hand motion” (page 189), that's not really the issue here. What matters is whether or not MN function could simply be an exaptation of pre-existing structures. The authors do eventually address this issue, but I found their evidence against this idea to be rather weak:

    However, as we have argued in section 4, the evidence from single-unit recording in monkeys suggests that MNs do not consistently encode goals. Therefore, the primary motivation for invoking exaptation is not compelling. (page 189)

    Okay, well let's flip back to section 4 and see what they had to say. Arguing against goals as object-directed actions:

    The fact that these MNs respond maximally to unnatural stimuli – that is, stimuli to which the evolutionary ancestors of contemporary monkeys could not possibly have been exposed – is hard to reconcile with the genetic hypothesis. (page 183)

    And arguing against goals as high-level intentions:

    However, the single-cell data again suggest that relatively few MNs have the field properties one would expect of a system designed by genetic evolution to represent high-level action intentions. (page 183)

    The evidence they point to in order to refute the exaptation hypothesis only serves to counter the genetic hypothesis. It simply proves that MNs were not a genetic adaptation for encoding goals. The possibility that they are an exaptation for encoding goals is still not adequately addressed. Because this idea bothered me throughout the paper, I was disappointed to see they only addressed it superficially and briefly at the end, kind of like an inconsequential after thought. But maybe I'm missing something that makes a drawn-out argument against this idea unnecessary. What did you all think?

    ReplyDelete
  23. "Many commentators have presented convergent evidence supporting the associative account. Oosterhof, Wiggett, & Cross (Oosterhof et al.) report multivariate pattern analyses (MVPA) of fMRI data that indicate mirror – sensorimotor matching – responses in brain areas outside the classical parietal-frontal mirror circuit. As Oosterhof et al. note, the finding that mirror responses are widespread across the brain is in line with the predictions of the associative account. It also therefore counters those commentators who argue that the associative account must be wrong because MNs are only found in restricted brain areas”

    I found this point to be particularly interesting, although I had some trouble understanding some aspects of it. The expeirmentors are saying that fMRI studies show that mirror responses are widespread across the brain, which contradicts other scientists who believe that MN’s are primarily located in restricted brain areas. Firstly, what is “restricted brain areas” - it seems like a very loose term, and without exact clarification, can be misrepresented quite easily. Secondly, I seem to agree with the point that mirror neutrons are widespread. It is closed-minded to say that MN are restricted. Different situations can trigger different emotions and responses which are widespread throughout the brain. Also, if you injure that “restricted brain area” - would one most likely be incapable of a mirror response.

    Another point I wanted to make was the consistent comparison between monkeys and humans. Although the article primarily uses its support from monkeys, is this a safe comparison? It seems like although we have a lot of similarities, there are also a lot of differences.

    ReplyDelete
    Replies
    1. "Another point I wanted to make was the consistent comparison between monkeys and humans. Although the article primarily uses its support from monkeys, is this a safe comparison? It seems like although we have a lot of similarities, there are also a lot of differences."

      I think this consistent comparison comes from the genetic account which talks about genetic predispositions in a common ancestors of monkeys and humans to develop MNs. So if we assume the genetic account, then I see the comparison making sense.

      But personally I agree with the associative learning account and find the human-monkey comparisons to be a bit weak. Also, we need to take into account that the article states only one study has 'direct evidence' for MNs in humans.

      Delete
  24. While I find this article touches upon some interesting research, I think we need to be careful in regards to the conclusions we can draw. I think we can firmly state that mirror neurons are involved in action understanding. They fire when someone is “performing an action, such as grasping an object using a power grip, but also when one passively observes a similar action performed by another” (p.177). They fire when particular shapes and movements are performed and they are involved in associative learning – but I think that (for the moment, anyways) we cannot conclude much more than that.

    More specifically, it is not exactly clear that mirror neurons explain our capacity to do something. Yes, both monkeys and humans have the capacity of doing many things and yes mirror neurons are activated during some of those capacities – but the activation of mirror neurons itself doesn’t explain how we go about having the capacity of doing those things. Mirror neurons seem to tell us where they are neutrally located when we do certain capacities, but they don’t explain how we can do those things. What we’re getting is correlational data – the activation of mirror neurons is correlated with those capacities – but can this correlational data help us generate causal explanation? Can they help us build a robot that can do these capacities?

    Something else that came to mind was whether we can be certain that mirror neurons “understand” actions? Yes, these neurons are firing under these circumstances (so we assume the brain is understanding “something”), but does that necessarily mean that mirror “understand” and recognize these actions? We’re assuming the two are connected but again, it is not clear what we get from neuroanatomatical data—“there’s no firm data for any but the grossest correspondence between types of psychological states & types of neurological states” (Fodor).

    Lastly, the authors talk about mirror neurons relation to our capacity of natural language. I am little unclear with this. It is clear that mirror neurons are involved in our capacity to mimic actions, but language isn’t simply the mimicry of vocalizations. So again, I don’t think mirror neurons explain our linguistic capacity.

    I know I seem like such a neuro-hater. I do acknowledge the power of neuroimaging techniques and that there is important information to be gained by process of activation patterns – but in reference to mirror neurons (for the time being), we cannot conclude that they are the causal mechanism for action understanding.

    ReplyDelete
    Replies
    1. "As discussed in section 3.1, consensus regarding the term “action understanding” has yet to be reached."

      So it really depends on what can be agreed upon as the term for understanding.

      "...a recent meta-analysis of fMRI data on action observation revealed that observation of hand movements produces responses in both pre- motor and parietal cortex, while face and body movements recruit premotor and parietal cortex, respectively (Grosbras et al. 2012)."

      So from this point of view, we assess "action understanding" as being the observation of certain movements producing MN responses in the appropriate brain areas, i.e. "understanding" as in areas of the brain respond to movements that would be implemented in those areas. Which makes sense and we have the fMRI data for, as stated above. But...

      "If, however, “action understanding” is operationalized in terms of understanding others’ intentions, then the role of mirror areas is less clear: A meta-analysis of studies using mentalizing tasks concluded that classical mirror areas are not recruited unless the tasks involve action stimuli (Van Overwalle & Baetens 2009)"

      "Intention understanding does not necessarily recruit mirror neurons.." So if we to understand 'action understanding' in this way, it becomes very unclear. In the article they also talk about the confusion between action "perceptual processes" and understanding.

      As for language, I assume you're talking about the passage: "Research investigating the effects of damage to parietal cortex supports a role for this area in imitation: for example, of mimed tool use (Halsband et al. 2001), of meaningless and object-related gestures (Buxbaum et al. 2005; Goldenberg & Karnath 2006; Tessari et al. 2007), and of phonetic detail in speech (Kappes et al. 2010)."
      (unless I missed some other part they talked about language, but this is the only part I remember it being mentioned, save for the brief mention in the beginning).

      I completely agree with you that clearly language isn't simply the mimicry of vocalizations. However I don't think that this is not what the author is suggesting. I interpret it as, perhaps the learning of certain more complex phonemes is facilitated by MNs. For example, a child may consistency mispronounce a word due to the constraints imposed by their vocal tract/mouth (for ex they have trouble with dentals because they are missing their two front teeth). So the mirror neurons are firing when adult are correctly pronouncing dental phonemes, so that when the child grows in their teeth, they too are able to correctly produce the dental phonemes. This is perhaps more of an interpretation based on a very short mention by the author, but that's my thinking.



      Delete
  25. I definitely agree more with the genetic hypothesis than with the associative hypothesis.

    "The genetic hypothesis says that genetic evolution has played a specific and decisive role, and learning – based on sensory and/or motor experience – plays a merely facilitative role, in the development of MNs. In contrast, the associative hypothesis says that genetic evolution has played a nonspecific background role, and that the characteristic matching properties of MNs are forged by sensorimotor learning."

    At first, I didn't really think that genetics played a large/important role in the development of MNs. I thought definitely that the development of MNs was solely adaptive. However, after reading further, I found the genetic hypothesis more inclusive and all-encompassing than the associative hypothesis.

    "Regarding the function of MNs, the genetic account assumes that they play a fundamental role in action understanding, and that this is why a specific genetic predisposition to develop MNs was favored by natural selection."

    I like this and, to me, it makes a lot of sense, as it allows me to believe still that MNs are developed for adaptive reason, but it does not negate the role of genetics (and it seems a little ignorant to completely ignore something as powerful as genetics when it comes to neuron development). The fact that natural selection favours a genetic predisposition to develop MNs, to me, means that the fittest organisms, I guess, get the "privilege" of developing mirror neurons. I call this a "privilege" because the development of mirror neurons is crucial for survival in the sense that you "learn from the mistakes of others" (in "kid sib" terms). The fact that the associative hypothesis says that "the characteristic matching properties of MNs were forged by sensorimotor learning" doesn't make much sense. I don't think that the "mirrorness" of mirror neurons can develop by individual learning. Meaning, I don't think that one can do something and then have the "mirrorness" develop after the fact. I think that the associative hypothesis is flawed in this respect. It seems only logical that the mirrorness of a mirror neuron, its characteristic matching properties, could only develop after one sees something and before one does what one saw. It's the mechanics that happen in that "in-between" moment that develop a mirror neuron... Or at least that's what makes the most sense to me.

    ReplyDelete
    Replies
    1. "I found the genetic hypothesis more inclusive and all-encompassing than the associative hypothesis"

      Personally, I actually believe the exact opposite: that the associative hypothesis is much more inclusive. The genetic hypothesis is pretty rigid and suggests that we would have to have genetically encoded mirror neurons (meaning they will appear whether or not we ever have the experience of imitation / learning motor movements visually) and that these neurons must play an evolutionarily decisive role.

      The associative hypothesis on the other hand acknowledges that, while the general topography of the brain is indeed genetically determined (the developing brain will differentiate into functionally separate areas such as the telencephalon, diencephalon, etc), the fine wiring of our brains is usually also heavily dependent upon experiential, epigenetic factors as well. Since suggestions for the functional roles of mirror neurons are still only suggestions, it is possible that they do not play a decisive role on their own. It is possible that they are simply a step in the process of the ill-defined "action understanding" process and simply integrate visual and motor information and then link up to higher order areas contributing to "understanding." This hypothesis also leaves room for the fact that their firing could be totally coincident and their roles as individual neurons be relatively trivial in the grand scheme of things. So the associative hypothesis actually is more inclusive (in my opinion).

      "[the genetic hypothesis] does not negate the role of genetics (and it seems a little ignorant to completely ignore something as powerful as genetics when it comes to neuron development)"

      As I mentioned in the paragraph above, the associative hypothesis does not negate the genetically encoded architecture of the brain.

      ("The associative hypothesis assumes that gene-based natural selection has played a significant background role with respect to the development of MNs; for example, in shaping the anatomy of visual and motor cortex for visual guidance of action, and in producing the capacity for associ- ative learning in neural tissue.")

      In fact, since mirror neurons usually exist in smaller populations, sometimes individually, it seems to me to be a bit too deterministic to claim that there are neurons that are genetically predisposed to fire when viewing or performing an action and that this firing necessarily has an evolutionarily positive function.

      I'm not entirely sure of what you mean by the "mirrorness" of mirror neurons, but if you mean the fact that they fire both when doing and when watching an action, then I think that this property could easily be acquired during development. As a child, learning how to grasp objects and perform movements you will imitate others around you. This would result in certain visual and motor neurons firing at relatively the same time and often (since you would have to perform the action repeatedly to fully learn it). Thus, if the two inputs happen to converge upon a certain neuron (which would end up becoming a mirror neuron), you would end up changing the properties of that neuron, and after a while the neuron will fire if it receives visual input alone or motor input alone.

      Delete
  26. "What is important is that the hypothetical function of MNs is distinguished clearly from other components of the same overall process. For example, in this kind of system-level, theory-guided approach, “action understanding” would be distinguished from components that are likely to be more purely perceptual (which might be called “action perception” or “action recognition”), more purely motoric (e.g., “action execution”), or to constitute a higher level of “understanding” (e.g., “mentalizing”)….Empirical (rather than constitutive) claims about the function of MNs need to be tested by experiments looking for, at minimum, covariation between MN activity and behavioral competence, and, ideally, testing for effects on behavioral competence of interventions that change MN activity”

    This essentially equates action understanding with behaviour, but if action understanding (which is not, as is admitted by the authors, by any means precisely defined in the paper) is in any way like any other type of understanding, then surely it can occur entirely independent of any behaviour. In a sense it must by default occur independent of any behaviour — the fact that it might then lead to behaviour or cause it somehow is a separate matter and not in any way obligatory, I would think.

    ReplyDelete
  27. The genetic hypothesis holds that mirror neurons (MN) developed precisely due to the functional benefits which they confer to organisms that posses them. Even under this hypothesis we can only speculate (given the evidence presented in the article) as to the actual function that they perform. The associative hypothesis goes on to divorce the origin of motor neurons from their function. Given the evidence that the authors present favouring the associative hypothesis, it seems we are at quite a loss when it comes to figuring out their role; we can only speculate about whether they fulfill any function at all in their own right, since the associative hypothesis leaves open the possibility that they merely develop as an incidental byproduct of underlying associative learning processes.

    ReplyDelete
  28. In comparing the genetic and associative hypotheses, I think in vitro experiments to supplement the fMRI data can shed further light into whether mirror neurons have the capacity to change its responses. A relatively simple study can include isolation of mirror neurons and measuring the responses to different types of stimuli. If there is "learning" on the molecular level, this may show evidence to the associative hypothesis.

    Overall, I agreed with Cook et al. that the genetic hypothesis seems less comprehensive because it does not take into account many forms of learning. It was curious to compare mirror neurons to domain-general processes of associative learning, such as Pavlovian responses and conditioning.

    ReplyDelete
  29. Glancing at the already published skywrittings, my concern seems to be one that has already been raised by a few other people. That concern is about action understanding. At the beginning of the text it says that the concept will be explained in section 8. But after reading up to and past that point, there is still no clear explanation as to what action understanding is. Synonyms are used but there is no (kidsib!!!) clarification for what it is. It says it results from natural selection (?), used for achieving goals, but this all seems like incidences leading to it or caused by it, not what it actually is. So I guess what I’m asking for is some sort of clarification?

    ReplyDelete
  30. I found it especially striking that in the midst of all the hypothesizing and speculating about the true mechanism of MNs there had only been one experimental study directly measuring MNs in the human brain, via single cell recording (Mukamel et al., 2010) at the time of publication in 2014. The authors are forthcoming about this caveat, and suggest that adopting the associative learning paradigm of MN function would serve as a much needed “new approach” to investigating MN function. I wholeheartedly agree with the idea that associative learning approaches could and should be adopted to facilitate the study of MN function and social behavior.

    The authors state “associative learning is found in a wide range of vertebrate and invertebrate species, indicating that it is an evolutionarily ancient and highly conserved adaptation for tracking predictive relationships between events”. The authors then explain that the associative learning account allows for the influence of environment on the subsequent mechanisms of MNs, while this flexibility is not afforded by the genetic account. “Indeed, if MNs were some kind of genetic adaptation, some evolutionary frameworks would predict that the development of MNs would be protected or ‘buffered’ against environmental perturbations that could interfere with their adaptive function”. I disagree with this claim. The authors do say “some evolutionary frameworks”, i.e. they do not necessarily support these evolutionary theories but if one was to assume them, this conclusion would follow. I still do not think this point is a valid or valuable piece of evidence in support of their overarching argument for sensorimotor associative learning. Associative learning is an adaptive trait that is flexible, but is there such thing as an adaptive trait that is inflexible? Would an inflexible trait ever constitute something biologically/evolutionarily “adaptive”? I don’t know enough about evolutionary biology to form a definite opinion, but I turned to some of the literature to help me out. The Princeton Guide to Evolution defines a ‘adaptive’ as “a feature of an organism that fits into its conditions of existence, giving rise to similarity among organisms leading the same or similar ways of life” (Losos, Baum & Futuyma, 2013). The authors write “Phenotypic plasticity – the ability of a single genotype to produce different phenotypes when exposed to different environments – may itself be adaptive. If individuals in a population are likely to experience different conditions as they develop, then the evolution of a genotype that could produce appropriate phenotypes depending on circumstances would be advantageous” (Losos, Baum & Futuyma, 2013). This clearly contradicts the authors’ suggestion that some kind of biological or evolutionary “buffering” mechanism would preserve an adaptive function.

    ReplyDelete
  31. I think the mirror neuron problematic is fascinating because it shows how sense activities and motor activities are inextricably intertwined as sensorimotor capabilities. Whether I am performing a motor activity (efference copy), or someone else is performing a motor activity (mirror neurons), the process of sensation is essentially involved.

    I do not think that this paper contradicts the idea that sensory and motor capabilities can’t be separated. We need the term sensorimotor. In fact, I think this paper can be used to support the sensorimotor claim (the two are inseparable) underlying T3. If it is true that mirror neurons are forged by domain-specific processes of associative learning in the course of individual learning, the implication is that sense and motor must co-function. We cannot have sense come before motor, and we cannot have motor come before sense. They are existentially equi-primordial. We need sensing to move, and we need moving to sense. This article argues against the idea that something like mirror neurons could spring up fully-developed and independent of the each individual’s personal understanding (associative learning). Mirror neurons, therefore, go through development, and are dependent upon personal understanding. I think this is an essential aspect of the existence of sensorimotor capabilities.

    ReplyDelete
  32. "Each time an individual sees an action done by another individual, neurons that represent that action are activated in the observer's premotor cortex. This automatically induced, motor representation of the observed action corresponds to that which is spontaneously generated during active action and whose outcome the acting individual knows. Thus, the mirror system transforms visual information into knowledge

    I am having some questions when it pertains to this statement. It seems to me that mirror neurons are responsible for the mirror reactions of learning and things. For example, If I see someone else grab a water bottle, it will be easier and faster for me to do the same thing. But what about all of the knowledge when it comes to the grip or hand placement or knowing that a hand has to wrap itself around the bottle. Just because our mirror neurons allow us to imitate the action, our level of understanding is probably different. Prior knowledge and experience may play a role; therefore how big of a role do mirror neurons play and is there a basis of understanding that must take place beforehand?

    ReplyDelete