I am skeptical of the idea that there is a sharp distinction between module and non-module. Over the weekend, a friend of mine who is a child psychologist told me how he did something that "is supposed to be impossible" - he taught a young adult (~14) who couldn't read faces to do so (not in a mechanistic fashion). It seemed to me that there is a false equation of "pick up naturally without being taught" = "module". And yet, very few people pick up reading naturally, they have to be taught, but for most of us reading becomes a module after a while. I just look at written English and "see" the words, in contrast to other languages, which I have to "read" as a conscious act. It seems to me that there is a range of modularity, or perhaps the modules act on sub-tasks which are glued together by non-module thought.

he did something that "is supposed to be impossible" - he taught a young adult (~14) who couldn't read faces to do so

can you offer any more details on this? i can't imagine, frankly, how you would learn this non-mechanistically if you didn't have an "instinct" for it. and of course, yes, you are surely correct about the hard-and-fast dichotomies. there is i think a range of 'modularity,' instead of domain-general vs. specialized set ups.

can you offer any more details on this?

This friend of mine has had some remarkable successes of this nature. Unfortunately, I can't give more details of the kind I suspect you want. He's not a very theoretically-oriented psychologist, which I think accounts for a lot of his success. He uses the kitchen-sink method: just keeps trying different things until something works. He readily admits that he often doesn't know "why" things work. However, as a general observation, he says that a lot of people who think they "just can't" do things really can, just not in the way most people do them.

My theory in this case is that the boy could learn to read faces the way we learn to read language, but not the way we learn to speak language, i.e. that some step of the learning process was non-intuitive, but that the final act was.

Another example: He had a client that "just couldn't" do math. After working with him for a while, he realized that in fact he "just couldn't" conceptualize division. So, in fact, a lot of math became doable with some workarounds.

david, i think i understand exactly what you are saying. there are many things like that with me, actually. i need to be shown, but then i have normal human fluency.

Sounds interesting Razib. Care to elaborate? I'm actually often uncertain even of what "normal human fluency" is in a given task, especially because people use
work-arounds to compensate for their weaknesses so well.

"He had a client that "just couldn't" do math. After working with him for a while, he realized that in fact he "just couldn't" conceptualize division."


Is your friend a learning disabilities specialist?

You see, we don't have to disagree every time. I find your take on the book (Chapter 1, at least so far) very interesting.

BTW, you wrote Zora instead of Bora at the very end.... ;-)

That was an unusually interesting and informative post, razib. In fact I think I'll just rely on it for my take on the whole Tomosello (sp?) phenomenon, since I don't have the time for morel.

Intersubjectivity, btw, is an important concept, at least in its Popperian sense. It is the idea of a objective fact, a perception of the external world that two or more people (subjects) agree on. That you have just written a long post, for example, is something your readers can "intersubjectively" agree on. That you are right on the main issues, however, is a matter of individual judgment, about which consensus is unlikely.

Glad to see you find children interesting. Just wait til you have one of your own, cause you ain't seen nothing yet!

David,

Do you mean the boy suffered from Face Blindness- Prosopagnosia??

If so that's interesting, as my Mom IMO is partially on the Autistic Spectrum, as are a good number of relatives, and I'd say my brother has Asperger's Syndrome. He has an inability to remember peoples faces very well also - he could talk to someone for 2 hours and the next day not be sure what color hair, eyes, skintone etc. Yet, he is gifted at Math and Logic, has a degree in Electronic Engineering and Masters in both Philopophy and Film Making.

One additional comment. That tool use preceeded the imputation of intentionality to the behavior of others seems plausible in that tool production is an outstanding example of intentional behavior in its own right. Making a stone tool is a very complex undertaking which, because it does not involve a lot of locomotion, like hunting and gathering, even young children would be in a position to watch and study.

razib, this is a great post on Tomasello. It's exactly the sort of thing I hoped for when I first stole the idea of a reading group from Clark.

I agree with some of your criticisms, particularly the point that it's not the whole story. While I do think that language, for instance, developed largely culturally, there have clearly been some significant morphological changes in humans that have made that development possible. They include the co-option of a lot of the motor cortex by the muscles in and around the mouth. Speaking requires movements that are more complex than chewing, but that are built largely upon the sorts of movements that we use for chewing. It seems clear, then, that communication systems in our evolutionary anscestors were evolving in a way that involved more complex vocal abilities.

Another obvious example is cortex size. This too has changed dramatically, mostly within the last 2 million years. And much of the change in size involves the frontal lobe. However, the neuroscientific work on theory of mind implicates primarily pre-cortical areas. If theory of mind advances drove cognitive advances, it would appear that they occurred long enough ago to cause some significant brain growth in areas that aren't directly involved in theory of mind.

So like you, I tend to think that Tomasello has hit on something very important, and I also think that cultural transmission is one of the main reasons why we can use many of our shared (with nonhuman animals) cognitive abilities in a way that chimps cannot, even with extensive training, but I don't think it's the whole story, and I think a good part of the story requires going back to changes that occurred at least 2 million years ago (perhaps when theory of mind abilities really began to take off, evolutionarily), and understanding that those changes must have radically changed our evolutionary landscape in ways that we can't fully understand today.

By the way, if there's any other jargon besides "perspectival" and "intersubjectivity" that you were annoyed by, or simply weren't comfortable with, let me know. I'm trying to keep a list of words and concepts that non-cognitive scientists aren't comfortable with, so that when I write posts on the book that include them, I can give a simple explanation.

"there is basically one species-unique feature of our cognitive phenotype which is at the root of much of what differentiates us from other beasts, and that is our ability to conceive of conspecifics (other humans) as intentional agents like ourselves."

The connection to religion seems to be that if human thought is in some way fundamentally dependent on social and "theory of mind" capacities, then religious interpretations of the world would naturally emerge because in some sense all natural phenomena would be viewed as "intentional" phenomena. But it seems that the problem with that idea is that the better a distinction the organism could make between physical phenomena and intentional phenomena, the more highly adapted it would be. Religious explanations seem to treat the physical world as if it were some sort of social world with things like punishment and reward, Gods punishing evil-doers with floods and rewarding the virtuous with a good harvest, but to the extent that we can form an accurate picture of the world we tend to view natural phenomena as unconscious and impersonal. Evolution would want the distinction between the mental and physical worlds to be as sharp and clean as possible. The only thing I can think of that might help explain the connection between understanding of intentional causation and religiosity would be that the capacity to concieve of intentional causation shares something in common with the capacity to concieve of physical causation and the tendency to see intentional causation everywhere emerges as a by-product of this commonality. Something similiar may be going on when we describe a physical system as "knowing" something or "wanting" to do something when what we really mean is that something else caused the system to behave in a certain way or the system having a certain tendency to behave in a certain way. Even the word "behave" seems to come naturally when describing both physical and intentional causation.

sorry bora. it was a bit late :)

mort,

1) evolution is often sub-optimal, you reproduce enough (or more) than conspecifics.

2) you don't know what the correlated response to greater sensitivity to distinctions between social and physical would be. perhaps there are structural issues with how this ability is generated biologically which prevents it from ameliorating the problems with false positives.

3) i am not convinced that the false positives would have been that fitness reducing in the pre-modern world.

4) remember that the social matrix which is generated by religion might have a) functional purposes b) create its own selection coefficient because of "group" effects which prevents a hyper-accurate strategy from evolving (conforming is better than non-conforming).

Initial disclaimer: I haven't read the Tomasello book, just the overview provided at gnxp and mixingmemory. But the points seem strong enough to discuss w/o knowing the subtle nuances.

Second disclaimer: I majored in linguistics as an undergrad, so I may be biased. But the idea that central properties of the language faculty are not innate runs into big problems when you look at language acquisition. The state-of-the-art in all linguistics departments is that there is lots of innate archeticture specific to language. You all know the nativists like Chomsky and Pinker, but even the most radical empiricists w/in cogsci and linguistics -- the connectionists -- posit *some* kind of language-specific learning algorithm. Don't mind a lot of the PR from them: although it is the most domain-general/plastic model on the market, it's not actually *that* domain-general or that plastic when the details are spelled out.

The other thing is that linguistic symbols aren't just any old symbols, like Prada suits signifying high status, that you can learn by mere observation and not a priori bias. Razib already hit on the "critical window" of language acquisition, but that deserves to be emphasized. I can learn what symbols signify status at any point in life, even if you threw me into an alien culture, my broad symbol-detector would pick up on who's the Big Man. Also, you can teach people how to read & write at any point in their life; ditto for basic math algorithms like long division. Stick the average person in a foreign language community after a certain age, though, and they're SOL -- not just halfway between native-speaker and helpless, but damn near helpless.

The blurb @ Amazon for Tomasello's acquisition book says that all that's needed is symbolic thinking and pattern-recognition... but again, why do children abstract *some* patterns from what they hear, but not others? E.g., that syntax pays attention to both phrase-structure (the "trees") and left-to-right order, not just l-to-r order alone. And why do they never make *some* types of errors, but do others? E.g., they don't mistakingly say: "What did you drink juice and?" intending to ask what is the thing X such that you drank juice and X? But they do say: "What did you drink juice *with*?" If the little tikes have deduced on-the-fly that the former contains a "complex" noun-phrase (two atomic nouns joined by a conjunction: "juice and milk," say), while in the latter the second noun is part of a modifying prepositional phrase ("drank juice with bread," say)... then they're geniuses. Both of these hold cross-linguisticly.

In semantics (that is, how the meaning of the whole is computed from the meaning of the parts and their configuration), children readily use set conjunction -- "black dog" for the set of things that are both black and canine. They never use set union -- "blue toy" for things either blue, toy, or both. It isn't that the latter is meaningless, or even not useful. You can surely imagine situations where you'd want to refer to the set of things this-or-that, say "lion rhinoceros" for the set of things that pose a danger if you venture too far into those woods, or "babe weapons" for the set of things you want to steal on the next raid.

...anyway, I feel I've rambled too long. Pinker gives a pretty good overview in _LI_, _HTMW_, and _Words and Rules_. Once you study in detail the different parts of human language -- syntax, morphology, phonology, and semantics -- it becomes increasingly less easy to believe we do this stuff on-the-fly using a broad symbolic capacity. I don't want to sound unfair, but to my ears it sounds like suggesting that the supposedly "innate" richness of our visual system is actually a bunch of emergent properties of our more basic Intuitive Physics. We surely have Intuitive Physics and Theory-of-Mind, but it's awfully difficult to derive on-the-fly the richness of human vision and language from those two. You hit on the Baldwin Effect, again a point that deserves to be emphasized.

Sorry, but another point about language change. The "cultural selection and pruning" model of how the language faculty arrived at its present state seems suspect as well. It implies that, say, long long ago different individuals or perhaps different speech communities varied w.r.t allowing questioning of nouns inside "complex" noun-phrases ("What did you drink juice and?") -- some did, some didn't, and cultural pruning sided with those who didn't. If it's cultural selection in a strong sense, meaning decided by the folks then-and-there w/o an innate bias, then it should resemble fashion. Some groups decided men would wear their hair this way and women that way, while other groups decided vice versa.

But in reality these are universals: in all languages "moving" nouns from inside complex noun-phrases is bad, and all syntax uses both phrase-structure and left-to-right order. Pretty bizarre, then, that everyone ended up with the same linguistic fashion. Plus, if they're not innate but crafted on-the-fly, wouldn't at least someone in the group (perhaps just one old sage) be able to explain what's so bad about "What did you drink juice and?" or why they don't like using set union?

Sound structure in languages also doesn't change consciously. Take the "flapping" in American English -- a /t/ or /d/ between vowels, the first of which is stressed, results in a weird flapping sound. Compare British and American pronunciation of "butter," "ladder," etc. It's not as if one group decided to do one, the other group the other, perhaps some third historical relic group which decided to change the /t/ or /d/ to /babababababa/, and so on. Nor do these things have much to do with Theory-of-Mind.

OK, I should shut up now, but I want to defend a weaker version of T-o-M driving *some* *narrow* properties of language, even if it doesn't emerge from T-o-M. Again, haven't read Tomasello, and truthfully I never really explored the language evolution literature much in college, so I may have been scooped on this, but here goes.

Languages (English especially) pay attention to the use of force by intentional agents in verbs, particularly how much force is imparted by the intentional agent. An English example is the well known case of paint getting on the wall or floor.

1) I sprayed paint onto the wall
2) I sprayed the wall with paint

You can switch the order of the paint and it's destination. But you can't for

3) I dripped paint onto the floor
4) *I dripped the floor with paint

Asterisk means it's bad. There are lots of similar examples, but suffice to say that a bunch of nerds at MIT in the 80s discovered that the difference is that "spray" (and other words in its class like "smear") refers to an intentional agent imparting force, while "drip" (and other words in its class like "pour") refers to the force of gravity, not the force imparted by an intentional agent. Similarly:

5) Paint dripped onto the floor
6) *Paint sprayed onto the wall

So, quirky facts like this surely derive from T-o-M concepts like intentional agent. No clue what autistic preferences are for the above sentences, though...

We have words that mean essentially the same thing but differ in degree of force imparted by the intentional agent. For striking something in English: "chip," "chink," "tap," "bang," "whack," "hack," "chop," "thwomp," etc. Why not just "strike hard / harder / really hard / etc."? If you're trying to teach or learn the making of tools, the use of tools, or the preparing of food, having fine-grained distinctions among separate words is actually useful. If you want to get the job done, you don't want to sit around all day saying "no wait, not that hard, maybe a little lighter." You want to say "hack the wood," "chip the arrowhead," or "mince the root" and get on with it. Again, T-o-M and culture playing a role.

So, I think in any strong form, Tomasello's views must be off, but a weaker for where they only *influence* properties of language seems right. Again, where Tomasello himself stands on this continuum, I don't know.

"To be human is to be social...."

Are you saying that people who are not social (autists, for instance) are not human? (I'm trying not to take offense here....) ;)

Seriously -- what I do have a problem with is "there is basically one species-unique feature of our cognitive phenotype which is at the root of much of what differentiates us from other beasts, and that is our ability to conceive of conspecifics (other humans) as intentional agents like ourselves." I just have a little trouble with the word conceive because, to my mind, it conjures up the idea of people being consciously aware and in conscious control of their knowledge that other humans are intentional agents, although perhaps neither you nor Toamsello meant that.

I think it's true that some of the time some people are consciously aware of their knowledge; but most of the time most people appear to run on auto-pilot in their interactions with others -- and certainly I don't think 9-month-old infants are aware of their "knowledge" of other people as intentional agents.

I may be nit-picking, but this really gets to the heart of the whole "theory-of-mind" issue -- most people, I think, have an innate, instinctive theory-of-mind that they don't have to consciously think about much (if ever). Other people (on the autistic spectrum, for instance) have a more Vulcan-like theory-of-mind, really trying to work out logically the fact that other people are intentional agents and what their motivations might be. And, still other people, of course, don't seem to have any sort of theory-of-mind whatsoever.

Mirror-neurons appear to have a lot to do with the innate sort-of theory-of-mind -- mirror neurons fire both when individuals perform an action and when they observe an action. They were first observed in macaques in 1996 (1) -- and some research has suggested that faulty or missing mirror neuron systems in autists may lead to the very different autistic-spectrum theory-of-mind (where it exists). (2)

So, I'm left with these questions: How can one say that the typical human theory-of-mind is conceived? Why call it a theory at all when it seems more to be an extrememly fascinating biological function that enables most members of our species to live and function in highly complex groups? And, although this may sound like a ridiculous question, I ask it in all earnestness: how is it that people know where they 'themselves' leave off and the 'other' begins (or what are their own vs. what are others' actions) when mirror neurons fire both when an action is performed AS WELL AS when it is observed (or do they know)?

(1) http://www.msnbc.msn.com/id/7656021/

(2) http://www.sciencedaily.com/ rele...50411204511.htm

agnostic, i used the wrong phrase to express 'pruning' and 'selection.' deacon's idea, if i recall correctly, is that some language structures for whatever abtruse reason are easier for the pre-fab structure of the brain to handle. in other words, language does not emerge out of universal grammar encoded into the brain, but universal grammar was shaped to fit the brain over time as people experimented with languages. there was no protolanguage that was universal grammatical, over time, the many less universal grammaresque language forms and structures simply died off, leaving the most 'fit' mode to parasitize the brain.

btw, good points theresa. though i wouldn't get too stuck on my rather sloppy verbiage. i don't recall if tomasello used the phrase conceive. but the mirror neurons deserve a post here...though i haven't followed cognitive neuroscience closely. stanislas dehaene has been doing research on them.

agnostic, you may be interested in Tomasello's usage-based theory of language acquisition, which he presents in an entirely separate book. His theory of language acquisition is actually quite complex, and is meant to answer the questions you ask, along with others.

Mirror neurons are a tricky subject. For one, there's currently no direct evidence of a mirror neuron system in humans. Sure, part of the mirror neuron system in monkeys is in a structure that is analogous to Broca's area in humans. And sure, the mirror neurons most closely associated with imitation in monkeys are in an area analogous to our left medial frontal cortex (in Brodmann's area F5), and lesions in that area in humans are associated with difficulties in imitating the actions of others. But again, there's no direct evidence that mirror neurons are doing the work in Broca's area or the left medial frontal cortex, and it's definitely premature to say that if there is a human mirror neuron system, it's doing the same things it does in monkeys.

The connection between language and the mirror neuron system (if it exists) in humans is complete speculation. There was a good book chapter on the topic a couple years ago. The central point of the chapter was that the mirror neuron system can't do all of the things that some seem to be claiming it does. The chapter is online. You can read it here.

Chris >> For one, there's currently no direct evidence of a mirror neuron system in humans.

You sure about that? (I haven't done all that much reading on the subject, so I am actually asking if this is the case.)

Iacoboni, et. al., seem pretty confident about the existence of a mirror neuron system in humans:

"Inferior frontal and posterior parietal human areas with mirror properties have also been described with different techniques in several labs [they offer 13 citations]."

http://biology.plosjournals.org/ ...al.pbio.0030079

Their conclusion in this study:

An important clue for clarifying the intentions behind the actions of others is given by the context in which these actions are performed. The same action done in two different contexts acquires different meanings and may reflect two different intentions. Thus, what we aimed to investigate was whether the observation of the same grasping action, either embedded in contexts that cued the intention associated with the action or in the absence of a context cueing the observer, elicited the same or differential activity in mirror neuron areas for grasping in the human brain. If the mirror neuron system simply codes the type of observed action and its immediate goal, then the activity in mirror neuron areas should not be influenced by the presence or the absence of context. If, in contrast, the mirror neuron system codes the global intention associated with the observed action, then the presence of a context that cues the observer should modulate activity in mirror neuron areas. To test these competing hypotheses, we studied normal volunteers using functional magnetic resonance imaging, which allows in vivo monitoring of brain activity. We found that observing grasping actions embedded in contexts yielded greater activity in mirror neuron areas in the inferior frontal cortex than observing grasping actions in the absence of contexts or while observing contexts only. This suggests that the human mirror neuron system does not simply provide an action recognition mechanism, but also constitutes a neural system for coding the intentions of others.

Theresa, yeah, the 13 different techniques are highly indirect. It's interesting, but premature to say for certain that there is a human miror neuron system.

Ah, then Tomasello must be refering to the what the Hauser, Chomsky, and Fitch paper are calling the FLB, the faculty of language in the broad sense, including its interfaces with the conceptual-intentional systems... which could conceivably have properties determined by how the T-o-M, which it interacts with, is structured. Still, though, that leaves the properties of the FLB that reflect its interface with the sensory-motor systems.

Surely T-o-M has little to say about the lower levels of phonology and phonetics -- ease of articulation, phonotactics, and so on. However, intonational phonology could reflect T-o-M influences -- "_I'LL_ do it" indicating that other intentional agents are slacking off, and this one will have to pick up the slack.

About the Universal Grammar being something that's just best suited to pre-fab circuits, there is confusion b/w "Universal Grammar," which its proponents use to refer to the properties all languages have, not the Language Acquisition Device which individuals have that acquires language. So UG is like a description of how light behaves, while the LAD is like a description of how the human visual system hones in on the properties in order to develop "visual competence."

Again, something must be innate to acquire the knowledge, so Tomasello and Deacon seem to be saying that while there is innate cognitive architecture, it's only at the interfaces -- that is, there's nothing *unique* to language (FL in the Narrow sense, in the HCF paper) that's innate, only in the systems of T-o-M, the sensory-motor systems, etc. So it'll be interesting to see how researchers resolve the question of what's part of FLN, since then it would imply innate *uniquely* linguistic circuits. We'll have to wait and see.