Friday, July 10, 2009

'Monkey language' update (guest post)

Below is an invited guest post from Team Verb undercover operative "√v". — Mr. V

I've been disturbed from my summertime slacking and had my attention directed to a recent study by Ansgar Endress and colleagues which shows that cotton-top tamarins can tell the difference between stimuli that have an affix either as a prefix, i.e. shoy-breast, or as a suffix, i.e. breast-shoy. Yes, those are some of the stimuli that the tamarins got to listen to. This kick in the pants from Mr. Verb reminded me of something else that roused my attention a few weeks ago also on Mr. Verb …

To begin, the study by Endress et al. is very cool although at a certain level we should be more surprised if the tamarins could not do the task. This is because the task they were tested on consisted of being familiarized with a bunch of stimuli of the form shoy-X (prefix version) or X-shoy (suffix version) where the X consisted of a bunch of different monosyllabic syllables. The test then consisted of taking appropriately familiarized tamarins into their little test room at Harvard and listening to tapes of either shoy-Y or Y-shoy where Y is a list of different monosyllables. The tamarins behaved in a way that demonstrated they could differentiate between the prefix and suffix versions of the test. One very interesting thing about this particular experiment is that the tamarins were not trained in this task. The little tamarins just got to hang out and listen to strange sounds and just flipped their head to a side when the stimuli didn't match what they had heard previously.

Endress et al. make the correct claim that this task is not exactly like human language but demonstrates that tamarins are capable of parsing and processing auditory stimuli in a way that has to be done in order for human language to work. Little infant humans very easily pick up on similar type of prefixing or suffixing patterns while they learn whatever language they are exposed to. The real issue that we want to pay attention to is to not conflate specific tasks that occur in human language with the overall complexity of human language. The task that the tamarins did consists of two basic (although controversial for some people out there but I don't know why) skills. One is inducing from experience an abstract equivalence category and the other is using said abstract equivalence category to learn an abstract pattern. The abstract equivalence category that the tamarins learned on the fly is the X in the paragraph above which consisted of an equivalence class of different monosyllables. The abstract pattern that the tamarins learned was either shoy-X or X-shoy where the abstract equivalence class either precedes or follows the syllable shoy.

On the one hand, when we break down this behavior, this task is rather trivial since all you need to do is to be able to create an abstract equivalence class and know how to keep track of the order of two things, one concrete and one abstract. With all of the work on primate intelligence that we have (see Seyfarth and Cheney's and also work by Marc Hauser, who is a coauthor on Endress' paper and let's not forget Randy Gallistel's work that shows ants have abstract representations…) showing that primates have social structure, distinct vocal calls, cheat, steal and group hunt and the like, the surprise that we should have from work like this is that the tamarins got bored enough in the lab to sit down and pay attention to the tapes and produced the behavior needed to convince confused people about whether tamarins are 'smart' or not. Don't get me wrong, I'm very happy that Endress and colleagues invested the time and effort to produce results like this but I would be really, really, really surprised if the tamarins couldn't do this type of task…

The problem with people (all kinds) and studies like this is that human language is involved (Endress et al. are extremely reasonable about the conclusions to be drawn from the study so kudos to them but we'll see what the media and bloggers do with it). For some odd reason, people just go all kinds of crazy when we start asking questions about human language. People also get a huge case of the expert's illusion too. Since we all use human language everyday without thought, we must understand how human language works. Throw in some animals to twist things further and then things really start to go crazy…which leads me to my foggy remembrance from a previous Mr. Verb blog…

Mr. Verb talked about Cherokee graffiti but included a link to a NY Times article about vision and hearing that he suggested everyone read. Since I'm a lackey of Mr. Verb I read it. As usual, it had a whole bunch of really cool information but the journalist didn't really evaluate any of the claims based on the cool information. If you didn't follow the jump link, a very small potentially true claim from the article was that visual sampling frequency and auditory sampling frequency differ with audition having a higher sampling rate. The wicked cool example that showed this was that the McGurk effect shows that vision trumps audition in one case and in another case where vision and audition conflict (7 flashes of a light along with 8 beeps), vision can win out there. Excellent examples which clearly show that we construct reality and thus should always question what kinds of tricks our brain/mind are playing on us continually.

After the bait of the cool illusions has been taken, craziness begins to ensue. One of the claims that drove me bonkers was that human language evolved because of changes in human audition. Now, this sounds like a really good idea because there doesn't have to be anything special about the human brain for language (some people really like that idea for some reason). Our ears got special and this allowed us to have human language. The thing that really blows my mind about this idea is that it completely ignores manual languages and cochlear implants.

The basic idea that was suggested is that we have human language because our auditory system works/samples/does something faster than our visual system does. The first big problem with this is that from what we can tell, manual languages (e.g. sign languages) are just as complex and beautiful as spoken languages. Hearing babies manually babble at the same time vocal babbling commences even if they are not exposed to a signed language. If human language is somehow inherently connected to how our ear works, we would expect there to be drastic differences in the structures in manual languages vs. spoken languages and possibly the non-existence of manual languages. This prediction does not pan out at all.

Another questionable aspect of this view is the existence of cochlear implants. Ignoring the contested cultural questions about cochlear implants , the fact that we can slap a rather poor microphone on the outside of a child head, embed wires from said microphone into the auditory cortex and children can apparently acquire language rather well again calls into question that the main kick to human language was the evolution of a faster auditory sampling rate. To put it another way, we can work around an ear that doesn't work with our new-fangled technology and children can still acquire language. We don't have a better electronic ear but we can make one that will apparently do.

It’s the evolution angle on the 'more speed' idea that really got me going though. My bet is that general primate (and probably mammalian, possibly all animals with two eyes and ears) auditory systems are faster than their visual systems. This is likely the reason that we are this way. A second whiskey-foxtrot-tango moment occurred when the article talked with John Hawks who is investigating 'eight genes involved in shaping the human ear over the last 40,000 years'. Since the genes are somehow shaping the human ear, they must be affecting human language of course and Hawks suggests just this idea. Maybe so, maybe not… maybe those tamarins are a lot smarter than we ever thought and doing studies on them to show homologues of human language processing are just wrongheaded. Maybe we need to sit Mr. Endress and Mr. Hawks in a room together to hash this out because one of them has to be misguided on this front. If human language is being rapidly shaped and changed over the last 40,000 years then tamarins can't tell us anything about human language. If tamarins can tell us something about human language then the processing of human language has to remain rather unchanged for a good deal time longer than 40,000 years… I smell an academic cage match coming on…

In any event I'd love to see actual evidence to back up this claim about ears, genes and China. Figuring out what kind of evidence is necessary to support claims about evolution is where it all goes kinda meshugenah… if you read what people who know what they are doing (i.e. Sean Carroll and other biologists) you can see that the evidence they traffic in usually involves manipulating actual genes in lab animals (flies get the worst of this stuff) to see exactly what happens. This way you can actually see how different regulatory networks affect the expression of a particular gene and then what observable difference (either in morphology or behavior or some other characteristic) happens. There is basically no way we can conduct research on human beings like this so we have to be very careful when thinking about humans, evolution and genes. All I know is that biologists just start shaking their heads at people who do the 'I've found the X gene' shtick. As a good friend once chided me, genes code for proteins which should remind us that no one has a good story for a protein to behavior link yet.

[Don't know the source of the image, but anything that has a member of the Damned is worth posting. — Mr. V]

4 comments:

Anonymous said...

Thank you, √v. I had read a news report of this and couldn't figure out what to make of it.

Anonymous said...

The whole Hawks thing makes me very nervous.

Anonymous said...
This comment has been removed by a blog administrator.
Anonymous said...
This comment has been removed by a blog administrator.