Do Mirror Neurons Support a Simulation Theory of Mind-Reading? Anna Christina Ribeiro Unpublished, 20031
Abstract: Both macaque monkeys and humans have been shown to have what are called „mirror neurons‟, a class of neurons that respond to goal-related motor-actions, both when these actions are performed by the subject and when they are performed by another individual observed by the subject. Gallese and Goldman (1998) contend that mirror neurons may be seen as „a part of, or a precursor to, a more general mind-reading ability‟, and that of the two competing theories of mindreading, mirror neurons lend support to simulation theory. I here offer four reasons why I think mirror neurons do not provide support for simulation theory over its contender, theory theory.
According to current research on primates, macaque monkeys have what are called „mirror neurons‟; research on human beings has subsequently attempted to show that we do, too. Mirror neurons respond to goal-related motor-actions, both when these actions are performed by the subject and when they are performed by another individual observed by the subject (the „target‟). They will fire in my brain when I grasp my cup of coffee and they will also fire in my brain when I see you grasp your cup of coffee. The fact that they fire when I perform an activity is not so surprising; something has to fire when I perform an action. But the fact that the same class of neurons fires when I see another perform the same kind of activity is at least interesting, if not surprising. Some theorists have suggested that mirror neurons are “a part of, or a precursor to, a more general mind-reading ability” (Gallese and Goldman 1998). Various theories have emerged in the past couple of decades that attempt to explain our ability to „read‟ the minds of others, i.e. to be able to predict what they might do and to explain what they have done. Most of these theories are subsumable under the umbrellas of “simulation theory”
Although I never published this paper, it has been cited in at least one published paper (Carcione et al., „States of mind and metacognitive dysfunctions are different in the various personality disorders: a reply to Ryle‟, in Clinical Psychology and Psychotheraphy 12:5, 2005, pp. 367-373: http://www3.interscience.wiley.com/journal/112094493/abstract?CRETRY=1&SRETRY=0) and it is listed online on PhilPapers (http://philpapers.org/rec/RIBDMN) and on David Chalmers‟ „Online Papers on Consciousness‟ web page (http://consc.net/online/7.3b.3), so I thought I should make it available. I presented an earlier version of this paper at the U.C. San Diego Graduate Student Conference on Philosophy of Science in April 2003. [1 of 19] 1
and “theory theory.” Briefly, simulationists claim that we read other people‟s minds by simulating their mental states on the basis of the information we have available (perceptual input, e.g.) and then, rather than acting on the results of the simulation (making the decision they would, say), we attribute that final mental state to our target. Theory theorists, alternatively, contend that we arrive at the decision our target is likely to make on the basis of information we have about the target‟s situation plus inferences made on the basis of our folk psychological knowledge. Because mirror neurons appear to simulate the brain state of another, several theorists have identified their discovery with the discovery of biological evidence for the simulation theory. That identification, I shall argue, is hasty and unwarranted. In what follows, I offer four reasons why mirror neurons do not provide support for simulation theory over theory theory. I first sketch the differences between simulation theory and theory theory. I next outline the account relating mirror neurons and simulation theory given in Gallese and Goldman 1998, according to which mirrorneuron activity is predicted by simulation theory and not by theory theory, which should give us reason (they say) to favor simulation theory over theory theory. Finally, I develop my four reasons for not seeing mirror-neuron activity as evidence in support of simulation theory.
I. Simulation Theories and Theory Theories Simulation theorists claim that we explain and predict behavior by using our own mental resources as a model of the target‟s mind rather than by third-person inferential reasoning ultimately based on a theory of the mind (a folk psychology). We take what we think is our target‟s „input‟ information (construed broadly) as pretend input and run our mental processes „off-line2‟, so that the end result is not an action on our part but a prediction (or an explanation) of our target‟s action.
A metaphor from computer technology that caught on and that, with respect to human „networks‟, means that these mental processes are disconnected from the mechanisms that occasion behavior. [2 of 19]
There are two main kinds of simulation theory: the Gordon-variety (similar in kind to the „replicative‟ or „co-cognition‟ model developed by Jane Heal [1986, 1998]) and the Goldman-variety (endorsed by Paul Harris among others).3 There are at least two ways in which Gordon-variety simulation is different from Goldman-variety simulation. First, according to Goldman, one must recognize kinds of mental states in oneself (and therefore have the relevant mental state concepts) prior to being able to attribute such states to others via simulation. Gordon, on the other hand, claims that “only those who can simulate can understand an ascription of, e.g., belief—that to [subject] S it is the case that p.”4 So for Goldman self-ascription comes first; for Gordon, otherascription does. Another important way in which these kinds of simulation theory differ is in how they understand the process of simulation itself. Whereas for Gordon and Heal it is the environment of the target that matters in simulation, for Goldman internal similarity between simulator and target counts at least as much. Gordon makes this explicit when he says the „one system modeling another‟ account may suggest that simulation is a device for discerning what goes on „inside‟ another, based on an assumption of internal similarity of the simulating system and the target system. However, where one explains or predicts another‟s behavior in terms of a shared, jointly known world, there is no question of internal resemblance between simulator and target, only one of what it is about the world that moves the other to action. What is presumed is not similarity but access, not that the other believes as one does but that the other has access to (what one presumes to be) the world.5
For simulationists like Gordon the important feature in simulation is not that I think as my target does, but that I think what my target does. By „internal similarity‟ Gordon seems to mean similar „cognitive systems‟.6 So it seems that for Gordon what matters is not so much that subject and 3 4 5 6
See Goldman (1989, 2000), Gordon (1986, 1997) and Harris (2000). See Goldman (1989) and Gordon (1995). This is an area of investigation in developmental psychology which Gordon thinks may provide support for one or the other kind of simulation theory. 2001, 2 (his emphasis). The term „off-line simulation‟, Gordon writes, „packs into the very name of the theory what I regard as an ancillary hypothesis: that when we simulate—that is, use our imagination to identify with others in order to explain or predict their behavior—many of the same cognitive systems that normally control our own behavior continue to run as if they were controlling our behavior, only they run off-line… This hypothesis is very plausible… But I think the imaginative identification theory remains attractive even without the off-line hypothesis‟ (1992, 174, my emphasis). [3 of 19]
target have neuronal networks of the same kind, but that they process beliefs in more or less the same manner: that they both decide on R if they desire Q and believe P. Gordon is not saying that internal similarity is not important—indeed, it is hard to see how simulation could take place at all if simulator and target were entirely dissimilar. What he is saying is that internal similarity, while presupposed in the process of simulation, is irrelevant to that process itself: the simulator is not interested in whether her target has a brain that functions like hers, but rather in what about the world around him can explain his behavior. From this it seems fair to conclude that the possible discovery of mirror-neurons in humans would be irrelevant to a Gordon-variety simulationist account. For what is relevant for such accounts is not whether the same set of neurons are at work in both action and simulation, but whether the simulator has access to the target‟s „world‟—to his „environment‟, broadly construed. I will call Gordon-variety simulation „externally-driven simulation‟. 7 Goldman, on the other hand, thinks that what goes on inside us during the process of „reading the minds‟ of others can not only show that simulating rather than theorizing is indeed the nature of that process, but also help solve what kind of simulation theory correctly describes it. Hence his claiming, in Gallese and Goldman 1998, that mirror-neuron activity may lend support to simulation theory. Elsewhere (1993, 2000) he also adverts to various experiments that might lend support to his view. Other defenders of Goldman‟s version of simulation theory have adopted a similar approach.8 I will call Goldman-variety simulation „internally-driven simulation‟. 9
Davies and Stone (Forthcoming, §2, 10f.) characterize the Gordon and Heal type of simulation theory as being pitched at the „personal‟ („external‟) level, while Goldman-style simulation theory is pitched at the „sub-personal‟ („internal‟) level. But Heal and Gordon differ on how they construe the epistemological character of the theory: Heal thinks the dispute is to be settled a priori, whereas Gordon thinks that which theory wins the day is an empirical matter. E.g., Currie (1995) and Currie and Ravenscroft (1997, 2002) For a critique of both versions of simulation that charges the former with Cartesianism and the latter with behaviorism, see Carruthers (1996). [4 of 19]
Theory theories also come in varieties. Theory theorists claim that our ability to understand and predict another‟s behavior is the result of our drawing upon a theory of mind—a folk psychology. Depending on how „folk psychology‟ is construed, theory theories may also be seen as externally or internally grounded.10
Internalist theory theories may be further distinguished
according to how they explain the development of the theory of mind. Some say it is mostly learned;11 other theorists claim that there is a theory of mind module, which develops according to a process of maturation.12 The reading of folk psychology as externally grounded derives largely from David Lewis‟s early work (Lewis 1972). According to this reading, folk psychology is the theory implicit in our everyday talk about mental states. The job of the theory theorist is to try to unravel this implicit theory by collecting all the platitudes causally relating mental states, sensory stimuli, and motor responses in a conjunctive sentence, and appropriately „Ramsify‟ it. From the Ramsey sentence one can then obtain a definition of any mental state term. Mental states are thus defined in terms of their causal-functional relations. Criticisms of the Lewisian functionalist approach include skepticism about our everyday talk about mental states being as systematic as Ramsey sentences make it, and about the theory implicit in that talk being exclusively functionalist.13 Lewis himself later opted for a reading of folk psychology as internally represented.14
11 12 13 14
Ravenscroft (1997) uses the terms externalist and internalist instead: “On the externalist reading of „theory theory‟, our everyday talk about mental states implicitly constitutes a theory of mind: folk psychology (external). On the internalist reading of „theory theory‟, our everyday capacity to predict and explain behavior is underpinned by an internally represented theory of mind: folk psychology (internal). Unfortunately, theory theorists are not always clear as one might hope about which sense of „theory theory‟ they are endorsing” (2f.). Although I prefer to avoid such loaded terms, I might resort to them on occasion for ease of reading. See Gopnik and Wellman (1992), Gopnik (1996), Wellman (1990). See Leslie (1994), Baron-Cohen and Swetenham (1996), Carruthers (1996). See Chalmers (1996). Lewis sums it up in his 1994, 416: “We have a very extensive shared understanding of how we work mentally. Think of it as a theory: folk psychology. It is common knowledge among us; but it is tacit, as our grammatical knowledge is. We can tell which particular predictions and explanations conform to its principles, but we cannot expound those principles systematically. (Pace Lewis, 1972, p.256, eliciting the general principles of folk psychology is no mere matter of gathering platitudes.)” [5 of 19]
On the opposing and now dominant camp are theory theorists who contend that our mindreading capacities are subserved by an internally represented theory of the mind—as Stich and Nichols put it, “a body of information (or perhaps misinformation) about psychological processes and the ways in which they give rise to behavior‟ (1995b, 88). How that body of information is represented is, however, a matter of dispute: in the language of thought (whatever that is), as a sentence-like or rule-based system, as a connectionist network, or as mental models? There is disagreement also with respect to how law-like the folk psychology is and how accessible it is to consciousness, and it is far from clear how the internally represented body of information relates to folk psychological platitudes we might recognize as such. There are also, of course, those who think there is no such thing as folk psychology at all (at least in the sense that whatever it is, folk psychology is a false theory).15 But among the believers in folk psychology perhaps the biggest bone of contention concerns the mode of acquisition of the theory of mind: here the camps split between those who think it is acquired and those who think it is innate. Gopnik and Wellman (1992) argue that folk psychology is largely learned, by inference and induction, more or less on the model of scientific theory construction and revision. Folk-psychology nativists challenge that claim on the basis of „poverty of stimulus‟ arguments similar to those made by Chomsky with respect to language. It is implausible, they claim, to think that a five-year old child has been exposed to enough „behaviors‟ on whose basis it could construct the highly sophisticated theory of mind it typically has by that age. An obvious difference between simulation theory and the theory theory approaches is that simulation is a first-person process (I think as if I were you), whereas theorizing occurs in the third person (he picked up the cup, he will take a sip of coffee). Another difference is that theory theory involves a theory (however „theory‟ is ultimately construed) while simulation theory does not. 15
See, e.g., Churchland (1981, 1988). [6 of 19]
Simulation theorists claim that we need no theory in order to understand one another‟s mental states and behavior: we use our own mental resources in order to simulate another‟s mental states, and thus to explain and predict behavior. By our „mental resources‟ simulationists seem to mean our own reasoning processes, but no theory—at least no theory of the mind. It is true that I can find out how many days John thinks there are till Christmas by counting the days myself and then attributing that number to him.16 Minimally, however, I need a theory that John thinks like me (I will also need a mechanism that allows me to attribute the end result of my reasoning processes to others17). Of course, this is far from being a full-fledged folk psychology, but it is information: information that is crucial to the simulation process and that could not be obtained by means of it.18 I have sketched the main varieties of both simulation theory and theory theory. Simulation can be interpreted as being internally or externally driven, while folk psychology may be interpreted as an internally represented body of information that may or may not embody the platitudes implicit in our everyday talk about mental states, or (though no theory theorist currently holds this view) as externally grounded in our everyday talk. This body of information may in turn be represented in various ways, and be either acquired or innately constrained. From these considerations, it seems clear that mirror-neuron activity, if it is relevant at all, would be relevant to the internally-driven kind of simulation theory and to the internally-grounded kind of theory theory (in any of its varieties),
16 17 18
Thanks to Peter Carruthers for this example. Gordon (1995) suggests a mechanism of pronoun substitution. Daniel Dennett already voiced skepticism on this issue in The Intentional Stance (1987, 100f.): „If I make believe that I am a suspension bridge and wonder what I will do when the wind blows, what „comes to me‟ in my make-believe depends on how sophisticated my knowledge is of the physics and engineering of suspension bridges. Why should my making believe I have your beliefs be any different? In both cases, knowledge of the imitated object is needed to drive the make-believe „simulation‟, and the knowledge must be organized into something rather like a theory.‟ But in cases where the imitated object is relevantly like me, perhaps no such theory is needed, except for the minimal theory, as I mentioned, that the object is indeed like me in the relevant respects. Nevertheless, any likeness beyond that of the cognitive systems will have greater claims to being called a theory. If instead of John I have Juruna, a native Brazilian of the Yanomami tribe, relying on my math skills may have little to do with how I find out how many days he thinks there are till Christmas. Which perhaps shows that the „minimal theory‟ that my target is like me is actually not so minimal after all. [7 of 19]
inasmuch as these theories posit internal systems of mechanisms underlying our mind-reading capacities. I will now proceed to the claims made in support of simulation theory on the basis of mirror-neuron activity.
II. Mirror Neurons and Simulation Theory The version of simulation theory as internally driven is responsive to findings in the cognitive sciences, and the discovery of mirror neurons in macaque monkeys has provided what supporters of that version of simulation theory take to be evidence in their favor. Gallese and Goldman (1998) defend the hypothesis that the existence of mirror neurons is evidence in favor of simulation theory and not of theory theory. In this section I summarize their claims concerning mirror-neuron activity and simulation theory. Mirror neurons are a class of visuomotor neurons found in area F5 of the monkey‟s ventral pre-motor cortex. They are responsive to goal-related motor actions (e.g., reaching out a hand in order to grasp a ball). Interestingly, they are responsive not only when the monkey performs the motor action himself, but also when he observes another agent performing the same kind of action. According to Gallese and Goldman, “grasping, manipulating, and holding objects are by far the most effective actions triggering their motor response,” while “grasping-with-the-hand neurons form the largest class of F5 neurons.”19 A notable feature of mirror neurons is that they are situationally constrained in their activity: Neither the sight of the object alone nor of the agent alone is effective in evoking the neuronal response. Mimicking the action without a target object, or performing the action by using tools is similarly ineffective.20
Gallese and Goldman (1998), 495a. Id., 495a. [8 of 19]
As Gallese and Goldman conjecture, mirror neurons “appear to form a cortical system that matches observation and execution of motor actions.”21 These neurons fire at the monkey‟s picking up a cup, and they fire at the monkey‟s seeing a target pick up a cup. They do not fire, however, if the monkey sees a target sitting in front of the cup, looking at it wide-eyed, and running his tongue around his mouth, for example. A similar picture emerges with respect to human beings. Gallese and Goldman cite a 1995 study (Fadiga et al.), which first provided evidence of the existence in humans of a mirror system similar to that of macaque monkeys: “Every time we are looking at someone performing an action, the same motor circuits that are recruited when we ourselves perform that action are concurrently activated.”22 Two other experiments, using Positron Emission Tomography, had similar findings, now with respect to the brain (rather than muscles): Rizzolatti et al. (1996) and Grafton et al. (1996). According to these studies, a cortical sector of the human brain is also activated during goal-related motor action observation.23 Gallese and Goldman then speculate that the functional role of this matching system is “to underlie the process of „mind-reading‟, or serve as a precursor to such a process.”24 This is not to say that monkeys have the capacity to attribute mental states such as beliefs and desires to their conspecifics the way we do. What Gallese and Goldman suggest is that there is a „cognitive continuity‟ between non-human primates and humans “within the domain of intentional state attribution.”25 Mirror neurons are supposed to be the neural correlate of the capacity for such attribution.
21 22 23 24 25
Ibid., 495a. Ibid., 495b. Id., 495b. Ibid., 495b. Ibid., 500a. [9 of 19]
As we have seen, the capacity to attribute mental states to others has been explained variously as the capacity to simulate their mental states and as the capacity to infer what those mental states are on the basis of folk psychological knowledge. Gallese and Goldman think that the existence of mirror neurons lends support to simulation theory rather than theory theory insofar as “observers undergo motor facilitation in the same muscular groups as those utilized by target agents,” and simulation theorists contend that we represent the mental states of others “by adopting their perspective: by tracking or matching their states with resonant states of our own.”26 Moreover, they claim, a matching system such as the one exemplified by mirror neurons would not be predicted by theory theory, though it would be predicted by simulation theory: The point is that MN [mirror neuron] activity is not mere theoretical inference. It creates in the observer a state that matches that of the target. This is how it resembles the simulation heuristic. Nothing about TT [theory theory] leads us to expect this kind of matching… If TT were correct, and an observer represents a target‟s behavior in purely theoretical fashion, it would not be predicted that the same muscle groups would be facilitated in the observer as in the target.27
This picture suggests a rather dualistic view of mind and body, according to which folkpsychological theorizing occurs at an impersonal level that does not engage one‟s physicality. We need not go to the lab to find out that merely thinking about something can make us excited, sad, fearful anxious, and so on (all states that engage muscle groups and more). And „something‟ may include trying to infer what someone will do—say, one‟s boss at a meeting. This being true, theorizing about whether my target will pick up a cup may likewise engage my grasping neurons. Granted, Gallese and Goldman do not go so far as to say that mirror-neuron activity would be inconsistent with theory theory; what they claim is that simulation theory, unlike theory theory, predicts such a matching system—i.e., simulation theory would lead us to expect something like it. But perhaps what mirror-neuron activity shows is not that simulation theory leads us to expect
Ibid., 493 (abstract). Ibid., 498b. [10 of 19]
matching systems, but that simulation theorists, unlike theory theorists, are conflating two levels of explanation. Gallese and Goldman provide a sketch of the neural correlate of the simulation process: Let us interpret internally generated activation in MNs as constituting a plan to execute a certain action, for example, the action of holding a certain object, grasping it, or manipulating it. When the same MNs are externally activated—by observing a target agent execute the same action—MN activation still constitutes a plan to execute this action. But in the latter case the subject of the MN activity knows (visually) that the observed target is concurrently performing this very action. So we assume that he „tags‟ the plan in question as belonging to that target. In fact, externally generated MN activity does not normally produce motor execution of the plan in question. Externally generated plans are largely inhibited, or taken „off-line‟, precisely as simulation theory postulates. Thus MN activity seems to be nature‟s way of getting the observer into the same „mental shoes‟ as the target—exactly what the conjectured simulation heuristic aims to do.28
Many questions naturally emerge from consideration of this picture. For example, why should the sight of someone grasping an object lead me to attribute to him the plan of grasping that object, when he already did so? Why should externally motivated neuronal activity constitute a plan—of my own—to do X, that is then transferred to another agent? Why not have it directly (and more parsimoniously) represent the target‟s plan? A plausible reason should be given to require that our neurons do double the necessary work. It seems unreasonable that every representation of another‟s action should first be represented as my own before I can predict or understand anything about his or her behavior. Finally, why should neuronal activity, just because it is analogous to a target‟s, not involve theorizing? I will consider these questions in the next section. To sum up the results of this one: we have seen that mirror neurons are a. visuomotor neurons, b. responsive to goal-related motor actions, and c. responsive in the agent‟s action and in his observation of that action by another.
And, according to Gallese and Goldman, mirror neurons are a. possibly part of, or precursors to, general mind-reading abilities, and b. more supportive of simulation theory than theory theory, because their activity can be interpreted as analogous to the simulation process. 28
Ibid., 497b-498a. [11 of 19]
I will now develop my reasons not to follow the second of Gallese and Goldman‟s hypotheses, even if the first turns out to be true.
III. Four Reasons Not to See Mirror Neurons as Evidence in Favor of Simulation Theory. Contra Gallese and Goldman, I think there are at least four reasons why we should not take mirrorneuron activity as evidence in support of simulation theory, namely: 1. 2. 3. 4.
Internal similarity is avowedly not relevant to externally-driven simulation. Internal similarity is not relevant to internally-driven simulation. Mirror-neuron activity is compatible with both simulation theories and theory theories. Mirror-neuron activity concerns a narrowly specific kind of motor action, whereas mindreading abilities are vastly broad. More importantly, it only occurs in „real time‟—i.e. only at the sight of the action actually being performed.
I will go through these reasons in turn. 1. „Externalist‟ simulation theorists take the internal similarity of simulator and target for granted, focusing instead on the information socially available. Gordon explicitly points out that simulation is not a matter of simulating what goes on “inside another,” but of finding out “what it is about the world that moves the other to action.”29 Under this construal of simulation theory, mirror neurons may be an interesting finding, but not one that sheds any light on what it means to simulate another in order to explain his behavior. 2,3. It is a different story with simulation theory of the internalist kind. Here internal similarity is all-important, inasmuch as simulation is supposed to occur at a level that is not necessarily available to consciousness. Even so, mirror neurons provide no support for the internalist simulation theory. My arguments against the relevance of mirror-neuron activity for internalist simulation theory are aimed also at explaining why such neuronal activity is compatible
2001, 2, his emphasis. [12 of 19]
with any of the theories of mind reading (i.e. they cover reasons 2 and 3). The first argument focuses on what goes on when we are not simulating; the second, on what goes on when we are. That mirror neurons are active does not mean that inferential reasoning is not taking place; it certainly does not mean that a theoretical background is not present, or that, if present, it is not being used. Nothing shows that the activity is exclusive of either reasoning or theory. It is a fact that mirror neurons are active not only when we see another perform a motor action, but also when we perform such an action ourselves. In the case where the neurons fire because the agent is performing an action rather than observing one, the agent is clearly not simulating himself. Imagine a man reaching out and grabbing a cup of coffee, and taking a sip from it. If called upon to explain his behavior, he might say “I wanted to have some coffee, so I reached out for the cup.” If we further asked him, “How did you know that you wanted to have coffee?,” a plausible answer would be “I just knew it.” This is (pace Sellars) the kind of non-inferential, intuitive knowledge we seem to have of (perhaps most of) our own mental states. But that this knowledge is non-inferential does not entail the absence of a background theory. For me to characterize my state as a desire for coffee, I must know what desires are, and what it is like to have them.30 So, on pain of repetition, the activity of a class of neurons during action performance does not exclude the possibility of a background theory of the mind—one that may guide, and one that would explain, that action. Mirror neurons fire when we observe a target performing a goal-related motor action. Simulation theorists seem to assume that because the mirror neuron system „mirrors‟ the neuronal activity of an observed target during the target‟s action, observers are not engaged in reasoning but exclusively in a simulation process that is subserved by that neuronal activity. But however the simulation is realized at the physical level, it seems clear that all simulation must occur within a
This is not to say that all of our folk psychological knowledge is consciously known to us. It seems quite plausible that the more general and abstracted from content folk-psychological propositions and/or rules become, the less available to consciousness they are. [13 of 19]
theoretical context. In Gordon-style externalist simulation, we must know what to take as pretend input—hence we must be able to differentiate among mental states. In the Goldman-style, internalist kind, we must be able to differentiate among mental states in order to know what to attribute to our target. So it seems that the activity of mirror neurons not only does not exclude the possibility of an internally represented folk psychology, it requires one. To say that theory theory subserves our mind reading abilities is not to say that simulation never takes place. It is to say that, if and when it does, simulation occurs within the context of a body of information—information that is more or less coherently related, information that is more or less correct: a sufficiently reliable folk psychology. When Gallese and Goldman claim that a matching system such as is exemplified by mirror neurons would not be predicted by theory theory, and that the existence of it comports better with simulation theory, they are using as their target a radical, exclusivist version of the theory theory that is no longer, if ever it was, defended by any theory theorist. While simulationists of any stripe seem to make little, if any, room for theory, 31 theory theorists are happy to include simulation as a heuristic device of considerable usefulness, albeit one that is ultimately grounded in a body of psychological knowledge. 4. A fourth reason not to settle on simulation theory on the basis of the activity of mirror neurons is that mirror-neuron activity concerns a narrowly specific kind of motor action, namely the action of grasping objects. Our mind-reading abilities are vastly broad. We can tell that someone is about to write something on a piece of paper, that someone was upset at a comment made in conversation, that someone is anxiously awaiting a piece of correspondence, that someone is saying one thing and meaning another, that someone is about to whack us over the head with a stick, that someone is filled with joy at the sight of his newborn.
A notable exception is found in Currie and Ravenscroft (2002). See §3.1. [14 of 19]
Perhaps more importantly, mirror-neuron activity only occurs in „real time‟—i.e. only at the sight of the action actually being performed. Gallese and Goldman themselves point out that mirror neurons “discharge when the monkey observes another individual performing an action” and that they discharge “during specific goal-related motor acts.”32 If this is the case, simulation promoted by mirror-neuron firings is hardly useful for the purposes of behavior prediction (i.e. before the action is taking place). What predictive use can it be to me if I can only infer (or simulate) your mental state when performing an action after you have already performed it? Not much. Of course, I could perhaps explain why you did what you did—because you had a plan to do so, because you wanted to do so, because somebody was threatening to whack you over the head with a stick if you didn‟t, because you had been meaning to do so for days and were finally able to—but it is hard to see how the firing of a class of my neurons can possibly be such as to facilitate my job of choosing one from even these few options. Unless Gallese and Goldman can show that mirror neuron activity occurs spontaneously, triggered by an imaginative project, or prior to the performance of an action (in sight of activities that may or may not explicitly foreshadow it), and unless they can show that mirror-neuron activity of this sort occurs in a broader spectrum of phenomena (not just the grasping of objects), they are unwarranted in claiming that such mirroring activity is “part of” the process of making sense of the behavior of others, and that that process is done exclusively via simulation. In particular, they cannot claim that it is part of the process of behavior prediction. At best, mirror neurons can aid in establishing (part of) what someone‟s mental state is at the time he performs a goal-related motoraction: i.e. that the agent has (had) a goal; in particular, the goal of grasping an object. We would be in bad shape if that were all we could tell about our conspecifics. We would probably not be here at
(1998) 495a, my emphasis. [15 of 19]
all if that were all we could tell about the many other kinds of living creatures with whom, once upon a time, we had no choice but to interact.33
I am especially thankful to Peter Carruthers for several stimulating conversations on the topic of this paper, and to Gregory Currie and Georges Rey for their comments on an earlier draft. I am also thankful to Jukka Appelqvist, Bryan Baltzly, Andrew Kania, and Bradley Rives for commenting on an earlier draft. [16 of 19]
Adams, Frederick.  Empathy, Neural Imaging and the Theory versus Simulation Debate. Mind and Language 16:4, 368-392. Baron-Cohen, Simon and Swetenham, John.  The Relationship between SAM and TOMM: Two Hypotheses. In Carruthers and Smith, editors, 1996. Botterill, George, and Carruthers, Peter.  The Philosophy of Psychology. Cambridge: Cambridge University Press. Carruthers, Peter.  Simulation and Self-Knowledge: A Defence of Theory-Theory. In Carruthers and Smith, editors, 1996. . [2004, Forthcoming] Distinctively Human Thinking: Modular Precursors and Components. The Structure of the Innate Mind. P. Carruthers, S. Laurence and S. Stich, editors. Carruthers, Peter, and Peter K. Smith, editors.  Theories of Theories of Mind. Cambridge: Cambridge University Press. Chalmers, David.  The Conscious Mind. New York: Oxford University Press. Churchland, Paul M.  Eliminative Materialism and the Propositional Attitudes. Journal of Philosophy 78: 2. .  Folk Psychology and the Explanation of Human Behaviour. Proceedings of the Aristotelian Society, suppl. vol. 62: 209-21. Currie, Gregory. [1995a] Visual Imagery as the Simulation of Vision. Mind and Language 10: 25-44. . [1995b] Imagination and Simulation: Aesthetics Meets Cognitive Science. In Davies and Stone, editors, 1995b. .  Simulation-theory, Theory-theory and the evidence from autism. Theories of Theories of Mind. Peter Carruthers and Peter Smith, editors. Cambridge: Cambridge University Press. Currie, G. and Ravenscroft, Ian.  Mental Simulation and Motor Imagery. Philosophy of Science 64: 161180. .  Recreative Minds. Oxford University Press. Davies, Martin, and Stone, David, editors. [1995a] Folk Psychology: The Theory of Mind Debate. Oxford: Blackwell Publishers, Ltd. . [1995b] Mental Simulation: Evaluations and Applications. Oxford: Blackwell Publishers, Ltd. .  Mental Simulation, Tacit Theory, and the Threat of Collapse. Philosophical Topics 29: 1 & 2, 127-173. Dennett, Daniel C.  The Intentional Stance. Cambridge, Mass.: MIT Press. [17 of 19]
Fadiga et al.  Motor Facilitation During Action Observation: A Magnetic Stimulation Study. Journal of Neurophysiology 73: 2608-11. Fodor, Jerry.  A Theory of the Child‟s Theory of Mind. In Davies and Stone, editors, 1995b. Oxford: Blackwell Publishers, Ltd. .  The Mind Doesn’t Work That Way. Cambridge, Mass.: MIT Press. Gallese, Vittorio, and Goldman, Alvin.  Mirror Neurons and the Simulation Theory of Mind-Reading. Trends in Cognitive Science 3: 493-501. Goldman, Alvin.  Interpretation Psychologized. Mind and Language 4: 161-85. Reprinted in Davies and Stone, editors, 1995a. .  Empathy, Mind and Morals. Proceedings and Addresses of the American Philosophical Association 67. Reprinted in Davies and Stone, editors, 1995b. .  Folk Psychology and Mental Concepts. Protosociology 14: 4-25. Series edited by Gerhard Preyer, J.W. Goethe University, Frankfurt am Main, Germany. Gopnik, Alison.  Theories and Modules: Creation Myths, Developmental Realities, and Neurath‟s Boat. In Carruthers and Smith, editors, 1996. Gopnik, A. and Wellman, H.  Why the Child‟s Theory of Mind Really is a Theory. Mind and Language 7: 145-171. Reprinted in Davies and Stone, editors, 1995a. Gordon, Robert M.  Folk Psychology as Simulation. Mind and Language 1: 158-171. Reprinted in Davies and Stone, editors, 1995a. .  Reply to Stich and Nichols. Mind and Language 7: 174-184. Reprinted in Davies and Stone, editors, 1995a. .  Simulation Without Introspection or Inference from Me to You. In Davies and Stone, editors, 1995b .  Folk Psychology as Mental Simulation. Stanford Encyclopedia of Philosophy (Winter 1997 edition). Edward N. Zalta, editor. http://plato.stanford.edu/archives/spr2001/entries/folkpsychsimulation Grafton, S.T. et al.  Localizations of grasp representations in humans by PET: 2. Observation Compared with Imagination. Experimental Brain Research 112: 103-111. Harris, Paul.  The Work of the Imagination. Oxford: Blackwell Publishers, Ltd. Heal, Jane.  Replication and Functionalism. Language, Mind and Logic. Jeremy Butterfield, editor. Cambridge: Cambridge University Press. Reprinted in Davies and Stone, editors, 1995a. .  Co-cognition and Off-line Simulation: Two Ways of Understanding the Simulation Approach. Mind and Language 14: 477-498. Leslie, Alan.  Pretending and Believing: Issues in the TOMM. Cognition 50: 211-238. [18 of 19]
Lewis, David.  Psychophysical and Theoretical Identifications. Australasian Journal of Philosophy 50: 249258. .  Lewis, David: Reduction of Mind. A Companion to the Philosophy of Mind. Samuel Guttenplan, editor. Oxford: Blackwell Publishers, Ltd. Nichols, S., Stich, S., Leslie, Alan, and Klein, David.  Varieties of Off-Line Simulation. In P. Carruthers and P.K. Smith, editors, 1996. Cambridge: Cambridge University Press. Ravenscroft, Ian.  Folk Psychology as a Theory. Stanford Internet Encyclopedia of Philosophy. Edward N. Zalta, editor. http://plato.stanford.edu/archives/win1997/entries/folkpsych-theory. Rizzolatti et al.  Localizations of grasp representations in humans by PET: 1. Observation versus Execution. Experimental. Brain Research 111: 246-52. Stich, Stephen.  Deconstructing the Mind. Oxford: Oxford University Press. Stich, S. and Nichols, S.  Folk Psychology: Simulation or tacit Theory? Mind and Language 7: 35-71. Reprinted in Davies and Stone, editors, 1995a. .  Second Thoughts on Simulation. In Davies and Stone, editors, 1995b. .  How Do Minds Understand Minds? Mental Simulation versus Tacit Theory. In S. Stich (1996), Ch. 4. .  A Cognitive Theory of Pretense. Cognition 74: 115-147. Stich S., and Ravenscroft, Ian.  What is Folk Psychology? Cognition 50: 447-468. Reprinted in S. Stich (1996), Ch. 3. Wellman H. . The Child’s Theory of Mind. MIT Press.
[19 of 19]