My response: I think it's a question of computational complexity, not the presence or absence of any particular piece of bio-ware.

Intentional Attunement. The Mirror Neuron system and its role in interpersonal relations by http://www.interdisciplines.org/mirror/papers/1 back to the macaque Dan Sperber Nov 15, 2004 22:54 UT Assuming that everything that Gallese is suggesting is true, how much can we infer back regarding the social understanding of macaques monkeys, among whom mirror neurons were first discovered? Do they too engage in Intentional Attunement? And if they don t, or at least if it is possible that they don t, two questions: - What is their mirror neuron system about? - What do humans have -- and macaque monkeys do not have -- over and above mirror neurons, that allows them to base interpersonal relations on Intentional Attunement? making sense of presence Douglas Galbi Nov 16, 2004 0:01 UT dan sperber writes: What do humans have -- and macaque monkeys do not have -- over and above mirror neurons, that allows them to base interpersonal relations on Intentional Attunement? ----- My response: I think it's a question of computational complexity, not the presence or absence of any particular piece of bio-ware. Consider the general biological problem of making sense of presence of another like oneself. One approach is to have a signal that indicates the category of the other. An example of this is a chick using an "imprint" to identify its mother. Another approach is to use a history of observations to characterize the other. But as the relevant state space and decision capabilities of the organism expands, the optimal strategy shifts to attunement. For further discussion, see Section II.A in my work "Sense in Communication," freely available at http://www.galbithink.org Evolutionary Extensions Michael Arbib Nov 16, 2004 5:20 UT Sperber s questions are related to the following from the end of Gallese s fascinating position paper: Intentional attunement produces the peculiar quality of familiarity we entertain with other individuals. This is what being empathic is about. [But] social stimuli are also understood on the basis of the explicit cognitive elaboration of their contextual aspects and of previous information. Embodied simulation scaffolds the propositional, more cognitively sophisticated mind reading abilities. In the Mirror System Hypothesis (MSH) for the evolution of the brain mechanisms underlying language (e.g., Arbib 2002), there are multiple steps from the mirror system for grasping to the language-ready brain. Presumably a similarly complex evolutionary path must be charted from the putative mirror system for emotional expression of the monkey-human common ancestor to the rich emotional system of the human. MSH argues 1 sur 5 10-05-26 13:48

that the call system of nonhuman primates is not the direct ancestor of the language supporting mechanisms. Instead, the affective content of language may build upon the monkey vocalization system, while the information content of language has a complementary evolutionary history. Imprecations survive damage to the human brain that blocks normal speech(e.g., Hughlings Jackson 1879-80). Thus (Arbib, 2002) the language-ready brain may integrate action-oriented and affect-oriented systems in a pattern of cooperative computation. In any case, the suggestion is that the basic mirror systems for grasping and emotional expression are separate, that we need to unravel the complex evolutionary history for each (Arbib, 2005; Arbib & Fellous, 2004), and that there are intriguing things to be learned about their interactions. Indeed, STSa neurons fire during movement observation though perhaps not during movement execution (Carey et al., 1997), and STSa and F5 may be indirectly connected via area PF which, like F5, contains mirror neurons. STSa is also part of a circuit that includes the amygdala and the orbitofrontal cortex and so may be involved in the elaboration of affective aspects of social behavior (Allison et al., 2000). Allison, T., Puce, A. & McCarthy, G. (2000) Social perception from visual cues: role of the STS region. Trends Cogn. Sci. 4:267 278. Arbib, M.A., 2002, The Mirror System, Imitation, and the Evolution of Language, in Imitation in Animals and Artifacts, (Chrystopher Nehaniv and Kerstin Dautenhahn, Editors), The MIT Press, pp. 229-280. Arbib, M.A., 2005, Beware the Passionate Robot, in Who needs emotions? The brain meets the robot, (J.-M. Fellous and M.A. Arbib, Eds.), New York: Oxford University Press. (In press.) Arbib, M.A., and Fellous, J.-M., 2004, Emotions: from brain to robot, Trends in Cognitive Science, December (in press). Carey, D.P., Perrett, D.I. & Oram, M.W. (1997) Recognizing, Understanding, and Producing Action, in Handbook of Neuropsychology: Action and Cognition Vol. 11 (Jeannerod, M. & Grafman, J., Eds.) Amsterdam: Elsevier, pp.111 130 Jackson, J.H., 1878-79, On affections of speech from disease of the brain. Brain 1: 304-330; 2: 203-222, 323-356. Of Monkeys and Men (1) Nov 16, 2004 10:12 UT Sperber's reply focusses on two main issues: (1) Do monkeys engage in intentional attunement? (2) If not, what makes humans different? Accoording to my hypothesis, "intentional attunement" is a basic requisite for social identity. In that respect, I think that monkeys too exploit the mirror neuron system to otpimize their social interactions. At least, the evidence we have collected so far (which doesn't include emotions-related mirroring neural activity) seems to suggest that the mirror neuron system for actions is enough sophisticated to enable its exploitation for social purposes. Recent results by Csibra and Kalaska (2004)show that neurons in the dorsal premotor cortex can covertly simulate observed behaviors of others, like a cursor moved to a target on a computer screen, even when the relation between the observed sensory event and the unseen motor behavior producing it is learned through stimulus-response associations. But Sperber is right in pointing out that we do not exactly know which social cognitive skills are indeed supported by these neural mechanisms in the monkey. My hypothesis is that monkeys might entertain a rudimentary form of "telelogical stance", a likely precursor of a full-blown intentional stance. This hypothesis extends to the phylogenetic domain the ontogenetic scenario proposed by Gergely and Csibra (2003)for infants. Indeed we are just starting a series of behavioral experiments on monkeys based on a modified version of the Gergely&Csibra's habituation/dishabituation paradigm to empirically test the hypothesis. Hopefully, in few months we'll know more about it. 2 sur 5 10-05-26 13:48

But monkeys certainly do not entertain full-blown mentalization. Thus, what makes humans different? First of all, from a behavioral point of view human infants for years heavily rely on interactions with their caregivers and with other individuals to learn how to cope with the world. This is an important difference between humans and other species that may play a major role in bootstrapping more sophisticated cognitive social skills (we are back to the nature/nurture debate). At present we can only make hypotheses about the relevant neural mechanisms underpinning the still poorly understood (from a functional point of view) mentalizing abilities of humans. In a recent fmri study presented at the last Neuroscience Meeting in San Diego (Iacoboni et al. 2004) we showed that premotor mirror neuronsrelated areas not only code the "what" of an action but also its "why", that is, the intention promoting it. We hypothesize that this could be accomplished by chaining different populations of mirror neurons coding not only the observed motor act, but also those that in a given context would normally follow. Ascribing intentions could therefore consist in predicting a forthcoming new goal. If this is true, then one important difference between humans and monkeys could be the level of recursivity attained by the mirror neuron system in our species. A similar proposal has been recently put forward in relation to the faculty of language (see Hauser et al. 2002; Fitch and Hauser 2004). But I am fully aware that this is still a "just-so" story. Gergely, G. and Csibra, G. (2003) Teleological reasoning in infancy: the naive theory of rational action. TICS 7: 287-292. Cisek, P and kalaska, J.(2004)Neural correlates of mental rehearsal in dorsal premotor cortex. Nature 431:993-996. Iacoboni, M., Molnar-Szakacs, I., Gallese, V., Buccino, G., Mazziotta, J., and Rizzolatti, G. (2004) Grasping intentions with mirror neurons. Soc. for Neurosci Abs. 254.11. Hauser, MD, Chomsky, N., and Fitch, WT. (2002) The faculty of language: What is it, Who has it, and How did it evolve? Science 298: 1569-1579. Hauser, MD and Fitch, WT.(2004) Computational constraints on sysntactic processing in a non human primate. Science 303: 377-380. Meta-mirror neurons? Andre Sytnyk Nov 16, 2004 10:18 UT Perhaps human cognitive systems are endowed with higher-order meta-mirror neurons, which enable us, unlike monkeys, to metarepresent metarepresentations (ad infinitum), e.g. study and discuss mirror neurons on forums like this? The teleological monkey Gergely Csibra Nov 16, 2004 11:49 UT [This is a reply to Gallese's reply above ('Of Monkeys and Men')] I would, indeed, love to see any evidence (or otherwise) on monkeys' teleological stance. However, that would not answer Sperber's question on the functional role of mirror neurons in macaques. It would not because one does not need mirror neurons to take the teleological stance, as it is based not on 'embodied simulation' or 'intentional attunement' but on evaluation of efficiency. This evaluation may rely on information from one's own motor system, but it does not have to do so. This was shown in our experiments in infants (Gergely et al., 1995; Csibra et al., 1999, 2003), who were able to attribute goals to geometric figures on a computer screen. I found it unlikely that they managed to do this via attuning to moving circles. Thus, if Gallese's hypothesis about teleological stance in monkeys was confirmed, it would contradict his assertion that human mind reading abilities are just further additions to an evolutionarily more ancient simulation mechanism. Or at least it would demonstrate that simulation is not the only way to understand actions, even in monkeys. 3 sur 5 10-05-26 13:48

[Let me also mention that Gallese's reference to Csibra and Kalaska (2004) would correctly be Cisek and Kalaska (2004).] Csibra, G., Gergely, G., Bíró, S., Koós, O., & Brockbank, M. (1999). Goal attribution without agency cues: The perception of 'pure reason' in infancy. Cognition, 72, 237-267. Csibra, G., Bíró, S., Koós, S., & Gergely, G. (2003). One-year-old infants use teleological representations of actions productively. Cognitive Science, 27, 111-133. Gergely, G., Nádasdy, Z., Csibra, G., & Bíró, S. (1995). Taking the intentional stance at 12 months of age. Cognition, 56, 165-193. The Teleological Monkey (reply to Csibra) Nov 16, 2004 13:31 UT If "evaluation of efficiency" relied on the motor system, this would still qualify in my opinion as the possible outcome of a simulation routine. The evidence on the attribution of goals to geometric figures in infants is at best neutral on the hypothesis at stake, at least until one can show unambigously that this capacity exclusively relies on domain-specific neural mechanisms not involving the sensory-motor and affective systems. A recently published paper by Heberlein and Adolphs (2004)shows that a patient with bilateral lesion of the amygdala completely lost the capacity to interpret in intentional terms the abovementioned geometric figures, while preserving intact perception and social knowledge. If the attribution of intentionality to geometrical shapes entirely depended on the activity of a ToM module, it is hard to explain why this capacity should be lost after a damage to the amygdala. This evidence casts serious doubts on the hypothesis that the capacity of anthropomorphizing in intentional terms the motion of geometric figures should entirely rely on domain-specific, mentalizing-related neural centers. Heberlein, AS, and Adolphs, R. (2004) Impaired spontaneous anthropomorphizing despite intact perception and social knowledge. PNAS 101: 7487-7491. From Mirror Neurons to Understanding Michael Arbib Nov 16, 2004 5:34 UT I do not deny that the monkey s recognition of action may be quite rich (though I would argue that human language and other cognitive abilities make human awareness very different from the monkey s; Arbib, 2001); but I do deny that the mere activity of F5 mirror neurons alone suffices to provide such richness, or to constitute understanding the action. Rather, I hypothesize that the adaptive pressure for the initial evolution of mirror neurons was to provide feedback for dexterous manual control - thus yielding extraction of hand-object relations (Oztop & Arbib, 2002). But once a neural representation that could extract such relations existed, it could provide the opportunity for social circuitry to evolve to extend the basic mirror system. Consider a pattern recognition device that can be trained to classify pixel patterns from its camera into those which resemble a line drawing of a circle and those which do not - this device does not understand circles. However, to the extent that this recognition could be linked to circuitry for drawing a circle, or for forming associations like the outline of the sun or an orthogonal cut through a cone yields an appropriate stimulus, to that extent could one say that the system of which the pattern recognizer is part does exhibit understanding. Understanding is thus not a binary concept but rather a matter of degree; some things may be encoded appropriately yet not understood at all, others 4 sur 5 10-05-26 13:48

may be understood in great richness because their neural encoding is linked to many other behaviors and perceptions. I thus suggest that to mediate understanding, a system that can recognize an action A, recall (perhaps implicitly) that action A is most likely to have consequences B or C, determine from the context that B is more likely to occur, and then use the expectation of B s occurrence to speed the choice of a course of action appropriate to B. Thus the mirror neurons of PF and F5 must be embedded in a larger neural context if understanding is to occur, extending "knowing" from the individual action to the social context. Alas, I don t know of recordings of mirror neurons while monkeys engage in natural social interactions. My current position is that the mirror systems for action and systems for understanding action overlap but that a number of different subsystems had to develop before each gained the capability seen in humans. Arbib, M.A., 2001, Co-Evolution of Human Consciousness and Language, in Cajal and Consciousness, (Pedro C. Marijuan, Ed.), Ann NY Acad Sci, 929: 195-220. Oztop, E., and Arbib, M.A., 2002, Schema Design and Implementation of the Grasp-Related Mirror Neuron System, Biological Cybernetics, 87:116-140. Of Monkeys and Men (2) Nov 16, 2004 10:28 UT I think that we should distinguish between the putative cognitive social skills underpinned by MNs in monkeys and humans (see my reply to Sperber). My target article was mainly focusing on how the MNs system and related mirroring neural clusters outside the domain of action can be relevant for social cognition in humans. Arbib's evolutionary scenario is mostly tackling a different topic: how and why the MNs might have evolved. As other evolutionary scenarios, it can be plausible (indeed I think it is highly plausible), but not empirically testable. That said, I fully agree with Arbib that the MNs systems per se are not sufficient to explain the sophisticated mentalizing abilities of humans, as I pointed out at the conclusion of my target article. dog, cats, and others Douglas Galbi Nov 16, 2004 14:59 UT Consider Freedman et.al. (2001). They showed that monkeys quickly categorized blended images of dogs and cats through to 60/40 blends. A monkey can't make sense of a dog or a cat through embodied simulation, because, of course, they have rather different bodies. Now consider a white racist being asked for a dollar by a black man. Is the white racist going to make sense of the black man like the monkeys made sense of the images of cats and dogs, or will the white racist make sense of him like (with the same neural process) he would a personal friend? I hypothesize that the process is the former, not because the white racist is incapable of the latter, but because that kind of response has been downloaded to set up a simple signal-response loop. What do you think? The patterns of ordinary communication behavior among humans suggest to me that making sense of presence of another like oneself is typical and highly valued behavior, but it's not the only possible physiological structure of interpersonal relations. Michael Arbib seems to make a similar point, "Understanding is thus not a binary concept but rather a matter of degree; some things may be encoded appropriately yet not understood at all, others may be understood in great richness because their neural encoding is linked to many other behaviors and perceptions." Reference: Freedman, D.J., M. Riesenhuber, T. Poggio, and E.K. Miller (2001), "Categorical Representation of Visual Stimuli in the Primate Prefontal Cortex," Science 291, pp. 312-16. 5 sur 5 10-05-26 13:48