EFTA02658453.pdf

From: jeffrey E. <[email protected]> Sent: Wednesday, March 1, 2017 8:09 PM To: Joscha Bach Subject: Re: yes , confidentiL On Wed, Mar 1, 2017 at 2:29 PM, Joscha Bach < > wrote: Thank you, Jeffrey! This is from Noam, right? I would be very interes=ed in reading the responses of linguists and computational language modele=s to this. May I forward it to a friend at Google X? Some notes: > basic assumptions about human language that should I think be uncontro=ersial, extensively discussed elsewhere, then turning to a sample of chall=nges. A person's language is an internal system, part of hum=n biology, based on configurations of the brain, developing in each person=through interaction of specific biological endowment (the topic of UG =80 universal grammar in contemporary terminology), external environment= and general properties and principles of growth and development that are =ndependent of language. As far as I understand, there is not yet an agreement among linguists wrt. =G, i.e. how much is innate vs. do humans just converge on the simplest typ= 3 grammar that is consistent with the constraints they observe in their kcal environment. I think Noam argues that we have very specific circuitry =or language, whereas the other camp would suggest that we are general lear=ers, with specific rewards that bias us towards compositionality and syste=aticity. OTOH, this might also be read as a variant of Noam's =80 Strong Minimalist Thesis" (SMT). The controversy will be eventually resolved by progress in building systems=that learn natural language. > The acquired system is an "internal language" (l-langu=ge), a computational system that yields an infinite array of hierarchicall= structured expressions that are interpreted at the conceptual-intentional=Cl interface as, in effect, a "language of thought" (LOT),=and that can be externalized to one or another sensorymotor system, typica=ly sound. Also relevant are some considerations about evolution of I=nguage. > Little is known about the evolution of cognitive faculties, a matter d=scussed in an important article by Richard Lewontin, whose own view for th= prospects was dim Most folks in cognitive science would probably agree that most cortical act=vity is devoted to building a generative simulation of the outside world b= a process of hierarchical learning. These simulations can be mapped on a =onceptual manifold, something like an address space of our sensory motor r=presentations of the world, which we can use to evoke and shape our mental=simulations. Language is our interface to that conceptual space, and exter=al language allows us to synchronize concepts even in the absence of match=ng sensory motor representations, i.e. we can build mental simulations of =hings that we never experienced, by interpolating between concepts that ad=ress mental simulations we know. It seems that Noam's approach is unique in that he focuses entirely=on language and concepts, while treating the understanding of the underlyi=g cognitive faculties as hopeless, while many others would argue that understanding language without first understanding pre-linguistic mental repres=ntations might be impossible. EFTA_R1_01905083 EFTA02658453  That said, Noam's characterization of I-language and LOT at the =80 conceptual-intentional" interface, with an externalization th=ough generative mechanisms, is probably a useful basis, regardless of wher= individual researchers come from. > [i] Anatomically modern humans (AMH) appear about 200 thousand years a=o. > [ii] The faculty of language FL appears to be a true species property:=shared among human groups (with limited individual differences) and in all=essential respects, unique to humans. In particular, there is no mea=ingful evidence for existence of language prior to AMH. > [iiiJ Recent genomic studies indicate that some human groups (San peop=e) separated from other humans about 150kya. As far as we know, they=share FL with other human groups. > [iv] The San languages are all and only those that have the curious pr=perty of phonetic clicks, and there may be some articulatory adaptation to=producing them (See Huijbregts, forthcoming). Nguni languages have clicks, too, but they seem to have imported them from =hoisan. > [vi The first (very limited) indication of some form of symbolic behav=or appears at about 75kya. Not long after that, we have rich evidenc= of quite extraordinary creative achievements (Lascaux, etc.). This is consistent with another observation: Modern humans had a population=bottleneck of 2000-3000 individuals ca 75000 years ago, which coincides wi=h the Tonga eruption. This does not necessarily mean that the volcano kill=d off almost all hominids, but it increased the evolutionary pressure, and=it is possible that our ancestors evolved a mutation that enabled them to =utcompete and kill most of the hominid competition (including Neanderthals=. What if that mutation is something that roughly translates into =9Csymbolic behavior"? I currently think that much of our civilization might be the result of a se=ies of quite specific mutations. Our ancestors went from 3000 individuals =o one million and remained there until they developed religions. Religion =nd other ideologies are based on a need for conformance to internalized no=ms, i.e. an innate desire to serve as part of a system that is larger than=the individual's reputation based group. They were also based on a sha=ed conceptual space. Challenge 1 seems mostly to amount to: verify that 1. all human groups have=language, and 2. there is no grammatical non-human language. One of the in=eresting questions might be if dolphins have grammatical language, another=one concerns the limits of learning in non-human primates. The challenge i= completely empirical. Challenge 2 seems very exciting to me; I read it as: has language intrinsic=lly linear order, or is that only imposed by the sequentialization of arti=ulation? Grammatical language has a tree structure, and the tree seems to =e created probabilistically in the listener, from a string of discrete sym=ols. Would natural language be learnable without the constraints of sequen=iality and discreteness? Challenge 3: do we need externalization to learn and process language? I wo=ld suspect that an individual can play a language game against itself unti= it converges on its own language, but it is not clear that humans are amo=g the class of individuals that can do that from scratch. Most research su=gests that there is a critical window in which we must pick up our first l=nguage for perfect fluency, and there seems to be no evidence of entirely =ndividualistic acquisition/formation of a first language. If that is true=is that a constraint of the way language learning is implemented in the hu=an brain, or a complexity constraint within language itself? It seems to be clear that learning a programming language changes the way w= think, i.e. it provides evidence for a weak version of the Sapir Whorf hy=othesis. But that is not so much a constraint of externalization, but of t=e semantic structures addressed by the language. 2 EFTA_R1_01905084 EFTA02658454  I imagine that pure work in a computer science lab can make some interestin= progress on challenges 2 and 3. Challenge 4: I don't understand enough about the context to see the=significance yet; I would think that once we have an SMT model of language=formation, we can learn additional operations that perform operations on t=e generated mental representation, based on arbitrary signals. This may re=uire us to leave an approach that attempts to sandbox language from genera= cognition, but why would we want to constrain SMT based models by such a =andbox? Challenge 5: Again, I don't understand enough of the context to und=rstand why probabilistic interpretation cannot fill in the gaps. A probabi=istic model will weight alternatives, and the binary Merge is the simplest= preferred case? Challenge 6: The question of the structure of individual lexical items migh= require a perspective that integrates mental representations beyond langurge/SMT. Challenge 7: Do semantic atoms refer to the external world ("refere=tial doctrine")? — This seems to be quite clearly false; t=ey refer to representations in the neocortex that are mutable and acquired=through learning (structure or reinforcement) and inference. Challenge 8: Noam seems to agree with my take on 7. How are semantic items =cquired? — This challenge comes down to the general problems of le=rning and perception, i.e. pretty much everything in cognitive science out=ide of language! Challenge 8 seems to be designed by a rocket scientist wh= specializes in combustion chambers and leaves all other parts of getting =he rocket to fly as an exercise to their grad student... Challenge 9: Noam suggests that meaning must be derived from innate informa=ion, and wants to study universals between language to identify the innate=bits. However, it is not clear if they do not stem from the properties of =athematics, i.e. there is a limited space of "useful simple axioma=ic systems" that can be individually explored by learning systems.=Kant attempted to describe this space, identified it as apriori and synthe=ic, and listed the basic structural categories that we would use to charac=erize the world. Sowa and a few others have made contributions to basic on=ologies, and perhaps it is time to revisit Kant's project? Challenge 10: Do music, planning, arithmetic stem from language, or do all =esult from a shared innovation of modern hominid brains? — Obvious=y, different answers in that space might be possible, for instance music c=uld be a parasitic byproduct of rewards for discovering compositional repr=sentations that our brain needs to make us interested in learning grammar,=while basic planning is independent, and complex planning needs language f=r structuring and operating on the conceptual space. This makes the questi=n extremely general. It also gives rise to the more general question of what exactly makes homo =apiens different from the other chimpanzees. I suspect that our brains are=trained layer by layer, whereby each layer has a time of high plasticity d=ring its primary training phase, then undergoes synaptic pruning, and has =ow plasticity later on. The duration of the training phases is regulated b= genetic switches. Increasing the duration will extend infancy and childho=d (i.e. increase the cost of upbringing), but give each layer more trainin= data. Perhaps humans outperform other apes because they get a magnitude m=re training data before their brains lose infant plasticity, which results=in dramatically better ability to generalize and abstract? Challenge 11: Rare constructions can be understood by children, and thus th=re should be a mechanism to derive them from more simple rules, despite ap=arent evidence to the contrary, which should be explained [away). Challenge 12: Noam suggests that the complexity of most constructions in th= face of "poverty of stimuli" means that I-languages are 1= very similar, 2. differences result from externalization, 3. should there=ore stem from UG. He wants this shown, or an alternative. 3 EFTA_R1_01905085 EFTA02658455  An alternative explanation might be that the space of possible human gramma=s is small enough to allow rapid convergence, and in polyglots even allow =or a complete mapping. That would not be a property of an evolutionary- eng=neering UG, but an apriori of the mathematics of human grammars. Challenge 13: What small change in a brain could lead to the unique cogniti=e abilities of homo sapiens, including language? — There are a lot=of different hypotheses of this, among them what I suggest in (10), and di=ferential attention/reward for learning compositional structures, or sever=l successive modifications in the reward system. I think that Noam suspect= that the culprit is a new connective pathway, perhaps somewhat similar to=Julian Jayne's Bicameral Mind hypothesis? These challenges are extremely inspiring food for thought! Bests, Joscha > Am Mar 1, 2017 um 7:01 AM schrieb jeffrey E. <[email protected] <mailto:[email protected]»: > <Challenges Language 2-17.docx> =AO please note The information contained i= this communication is confidential, may be attorney-client privileged,=may constitute inside information, and is intended only for the use 4 the addressee. It is the property of JEE Unauthorized use, disclos=re or copying of this communication or any part thereof is strictly pro=ibited and may be unlawful. If you have received this communication =n error, please notify us immediately by return e-mail or by e-mail to =a href="[email protected]" target="_blank">jeevacation@gmai=.com, and destroy this communication and all copies thereof, inc=uding all attachments. copyright -all rights reserved 4 EFTA_R1_01905086 EFTA02658456