Differentiation and Integration
Differentiation and Integration
According to standard embryology, the process of ontogenesis is characterized by organic differentiation. From an initially homogeneous collection of cells tissues of various kinds are formed. This is no doubt powered by genetic instructions from within the original uniform cells. Maturation is thus a transformation from sameness to diversity as tissues develop in the body according to a fixed schedule. The end result is an organism composed of many organs each equipped with a characteristic type of cell and associated physiology—heart, lungs, bones, kidneys, etc. As differentiation occurs there is a need for coordination between the new types of tissue and the organs that tissue serves, and the adult organism clearly contains components that need to be integrated into a functioning whole. Thus there is no differentiation without integration (and vice versa). These are the twin pillars of biological development: genetically driven diversification and concomitant integration of the elements thereby generated.
This basic picture can be applied to language development. Initially there are just undifferentiated cooing sounds existing alongside the inarticulate sounds of crying, but these soon come to be replaced by language-like sounds (consonants and vowels) without structural complexity. These in turn are replaced by one-word sentences (“dada”, “milk”) that display the rudiments of language. Only later do these come to be combined into two-word strings, and subsequently into the full range of syntactic and semantic categories. The details don’t matter for present purposes: what concerns us is that language development follows a pattern of differentiation analogous to that undergone by the body.[1]A relatively formless initial state is gradually transformed into a highly structured system of elements that combine together. Anatomy develops by differentiation, but so does grammar. When the initial unstructured sentences are transformed into noun phrases and verb phrases there has been a process of differentiation comparable to the formation of heart and lungs (or the internal anatomy of each). This is not surprising once we accept that language is itself a biological phenomenon—an aspect of the human organism. And it makes logical sense: you derive an intricately structured organism from unstructured beginnings by a process of differentiation (it would be difficult to implement such differentiation in the sperm and egg). Language doesn’t emerge fully formed in the child but matures in the brain by a process of increasing structure and complexity. It grows by splitting into different functional units—that must nevertheless be integrated if they are to achieve their purpose. We are accustomed to the idea that language is a system built for integration—producing sentences by combinations of words—but we must also recognize that it is a system that arises by differentiation from something more primitive.[2]It was once a growing thing in the child and only reaches stasis after a lengthy sequence of differentiating stages—just like other human organs. Language is a product of gradual ontogenesis, which only later achieves its full combinatorial power. The adult language faculty enables integration ad infinitum, but at one time it was without much in the way of internal structure (relatively speaking). So let us add to the productive power of language its origins in a simpler form of living tissue. Grammar is the form that linguistic differentiation takes in human ontogenesis—just like anatomy and physiology. Linguistic differentiation is biologically at one with histological differentiation. Language thus follows the same basic pattern of organic growth. How else could the architecture of language arise?
But if this is true of language, isn’t it also true of the mind more generally? The various components (“organs”) of the mind must be integrated in order to function as a unity, but first they had to develop by some sort of maturational process. Perception, cognition, emotion, will—all need to emerge as distinct systems during ontogenesis, to be integrated later (or pari passu). But what was the initial state? Some may speak darkly of a blank state, but that would need to be supplemented by an account of how such a state could be progressively differentiated. Something has to turn into the various faculties of mind—some state of the pre-natal brain. About this process we know little to nothing, yet it must be so in some way. What is the analogue to the one word sentence or those even more primitive cooing sounds? And how did concepts emerge by differentiation? They were not present fully formed in the fetus’s brain but arose by a maturational process to become the vast combinatorial system we now deploy with such consummate ease. Presumably they arose by a process of differentiation: from William James’ “blooming, buzzing confusion” (whatever that means) to an articulated array of combinable elements. This maturational differentiation must be genetically driven, like other biological growth, but it results in a psychological faculty far removed from the initial state of the organism. Again, we know little to nothing about how this works, but we have good grounds for supposing a distinct path of differentiation and integration. One can certainly imagine that unstructured thoughts might with time transform into thoughts with a subject-predicate structure—and thence into more complex forms. First there were feature-placing thoughts (the mental counterpart of “It’s raining”) and later they turned into structured thoughts (along the lines of “There is heavy rain in London now”). Conceptual differentiation created the panoply of concepts we now take for granted.
Coordinating conceptual differentiation and integration is a highly non-trivial task, as it is in the case of the body and language. Once you have the plurality you need to keep it under control. The brain is the ringmaster here. Grammar is encoded in the brain and it sets the rules for combining words; in the case of concepts something analogous must be true—rules that combine concepts in certain acceptable ways (not just arbitrary lists or jumbles). So differentiation combined with integration requires rules or principles of coordination. The heart must be coordinated with the lungs to produce satisfactory aerobic performance; similarly concepts must be coordinated in the right way to produce intelligible thoughts. Integration doesn’t happen by magic. The more differentiation there is the greater the demands of integration. An organism with very simple thoughts doesn’t need much apparatus to keep its thoughts on track, but an organism like us relies on mechanisms that prevent thoughts from forming defectively or randomly. In aphasia these mechanisms can fail, leaving words to fail to link up correctly; in principle the same thing could happen to thoughts—concepts fail to link up to form coherent thoughts. It is surprising that more breakdowns of this kind don’t occur.[3]If there is a language of thought, there ought to be aphasia in that language if the brain is suitably damaged, which produces aphasic thought. The differentiation and integration of concepts will be tied to linguistic differentiation and integration in the language of thought.
The innate language faculty thus has two basic properties: (a) it permits unbounded productivity in its mature form, and (b) it enables a stupendous feat of differentiation as it guides the maturation of language in the child’s brain. It is as creative in the latter respect as in the former (though it doesn’t get as much credit for the latter). The adult lexicon is the product of maturational differentiation (how, we don’t know); sentence production is the outcome of integration rules. Both are built into the genetic blueprint for language. Nor does differentiation cease at normal linguistic maturity, since we continue to make linguistic and conceptual distinctions. The differentiation machine doesn’t go completely offline, its job done; it allows us to make ever-finer distinctions that aid thought. So it isn’t that one kind of creativity completely ends to be replaced by another; we are still able creatively to generate distinctions (though it doesn’t come as naturally as during childhood). Distinction making is as crucial to language development as the growth of the ability to combine existing elements. So I propose conceiving of the language faculty (and the conceptual faculty) as a union of differentiation and integration: it allows the combination of pre-formed elements, to be sure, but it also generates those elements by a (mysterious) process of differentiation. When abnormalities arrest language development we see in sharper outline the maturational stages speakers go through—we see how the differentiation process can be blocked (the same is true of human bodily growth). As adults we tend to forget this early history, but it is as essential to our mastery of language as the growth of the heart is to our survival. The language faculty is as much a creative product as it is a creative producer.
I said that differentiation and integration are the basic laws of biology (so far as concerns ontogenesis), but they are also relevant to evolutionary change. For what is species evolution but biological differentiation? Natural selection causes species differentiation (along with other factors), though mechanically not by pre-set program. There is no predetermined evolutionary schedule like the maturational schedule. Thus simpler forms evolve into more complex forms, splitting off to make a new kind of biological entity. Phylogeny recapitulates ontogeny. However, there is no real analogue to integration, since the different species don’t operate together to form a larger whole. True, it used to be thought that this might be so, as if each species had its functional role in the super-entity called Nature (“the biosphere”); but these days we tend to think that the entities that have arisen by evolutionary differentiation are independent entities subject to no coordinating principles. They are not like organs in a body or words in a language. There is no “grammar of nature”. The evolution of species is differentiation withoutintegration.[4]
Colin McGinn
[1]This point of view is defended by Eric Lenneberg in Biological Foundations of Language(1967), esp. Chapter Seven.
[2]This is in no way incompatible with the nativist account of language acquisition: the genetic instructionsfor generating full-fledged language are present at birth, but that is consistent with the existence of a maturational schedule that involves cellular and cognitive differentiation. Similarly, instructions for building a heart at a certain maturational stage are present at birth, but it takes time for the actual organ to be constructed by a process of differentiation. Nouns and verbs emerge at a certain maturational stage (the second year of life), but the program for making them was written in the genes.
[3]Drunkenness can cause breakdowns of motor coordination analogous to earlier stages of motor development (a kind of regression), including speech difficulties; but it doesn’t appear that drunkenness can derail the performance of the language of thought—we keep thinking coherently as we stagger and trip, slur and mumble.
[4]Suppose there was a symbiotic parasite that entered the brains of host species and conferred language on the host—the parasite contains grammar. The parasite might instigate a series of maturational stages of language development in the host just like those of humans. It is functioning as an organ of the host’s body/mind—much as symbiosis in general has this character. This would provide a sense in which different species might operate as a unity.
Paul Bloom has some interesting ideas about how the lexicon develops in children—grasp of the communicative intent of ambient speakers, given a prior or simultaneous differentiation of the concepts underlying the use of their words. But whence the discrimination of particular kinds of communicative intent? That’s easier: there’s a plethora of behavioural cues of speakers reliably linked to intent, upon which aspirant language learners might seize. This is even more true, kind of paradoxically, in mature second-language acquisition.
But the child must grow his own lexicon independently of knowing the words of a specific spoken language, just like grammar. We don’t get our concepts from others.
There’s probably a distinction to be made between pathological misfires in the exercise of one’s intelligence (as in the case of retardation, e.g. Downs syndrome) and the kind of conceptual aphasia we see in cases of Alzheimer’s or schizophrenia. Curious that the latter conditions are much more debilitating than the former. “We don’t get our concepts from others”? There surely are innate concepts, but just as surely are there culturally elaborated concepts. (By the way, thanks for cleaning-up my last Comment). See, I always suspected you had talent as an editor.
The important point is that the child doesn’t acquire its basic lexical competence by extracting it from adult speech, just as for grammar.
Another point, equally worth remarking, is Nadal’s sub-par performance so far in the clay court season. I predicted it, but I take no pleasure in it. I’ve mentioned Ray Jackendoff’s work on semantics and language development several times in my Comments, but you haven’t said whether or not you find him conducive..
I don’t know much about him.
Before reading this post, I hadn’t thought explicitly about the breadth of applicability of the principle of differentiation and integration. The pair do provide a richer and more dynamic framework than pure reductionism: they still explain/interpret a whole in terms of its parts; but they also explain how those parts themselves arose, and give the context for the integration (and for why the whole can be greater than the mere sum of its parts). Presumably, the integration is made possible by the fact that the parts arose through the differentiation of one thing (a unity) that provided the seed/potential for the integration. The cells can come together to form organs, and the organs to form a body, because the cell boundaries are compatible (a shared physical feature inherited from the original cell) and because the interactions between cells are appropriately coordinated (a shared logical feature inherited from the original cell). We could also look at the evolution of the internet for an example like this.
Knowledge can definitely involve a process of differentiation and integration. Though physically embedded rather than purely conceptual, one can really see this process up close when learning a new sport as an adult. I am not clear, however, on what is the “seed” (analogous to the original cell) in the case of knowledge, or if there is one. Is there an original or seed concept, out of which others arise through differentiation; and which, through its inner structure, creates the potential for the integration of its children? I’d be keen to hear your thoughts on this.
I understand what you say about species not integrating, but there is another perspective from which life does involve integration. Nearly all forms of life are either explicitly or implicitly dependent on other forms of life – life is highly integrated. There is a sense in which all life forms here on earth are part of the one experiment that started with some form of self replication and has now, through repeated ongoing instances of differentiation and integration, turned into a complex dynamic self-modifying and (partially) self-sustaining system.
Though life seems to display differentiation and integration in a number of ways and at a number of levels, I suspect the principle is more fundamental, and the story is older.
It is indeed a puzzle what the original of concepts is: it would be marvellous if there was a single seed concept. What would it be? World? Reality? Nothingness?
The trinity of the seed, differentiation and integration itself? Or is that the concept of Reality?
On Nothingness, I wonder if that is at or beyond the limits of what we (with our biological limitations, our innate seed concepts) can conceive. We can certainly know the concept of Nothing as ‘no thing’, I.e. the absence of a determinate thing (be that a person, a planet, a universe). But that concept has as its starting point a thing to be negated. Pure nothingness that does not even have something to be negated, no potential to become something, and not even a subject that can stand in relation to this Nothingness, is hard to really conceive. We can talk about a possible extrapolation of no-thing, of course. Or maybe some people do have the seed concept for pure Nothingness?
It may well be that pure nothingness is inconceivable by us. What would be the point of giving us such a concept?
I’m sorry to keep harping on—or referring to, the works of Ray Jackendoff. But to my mind, his stuff on conceptual primitives and on the combinatorial relations among concepts in our “language of thought”, surpasses in explanatory depth all previous psycho-linguistic inquiry (even over and above Fodor in his more empirically minded moments) Colin ofcourse has it right that there’s a philosophical gloss to be put upon the matter. Who knows, “differentiation and integration” might be the “meta-” way to go. Philosophy, after all, is exploration of the wider explanatory space not merely of the plausible but of the possible. Nadal over Tsitsipas tomorrow. Theim over Djokovich. I’ve layed down a five-dollar bet—the limit of my credit.
The mind can internalise and externalise (as per your previous post), and it can differentiate and integrate (concepts being an act of integration). These are quite different, probably orthogonal, activities: I-E describes how the mind interacts with the world (and maybe in some cases how it interacts with itself), while D-I describes how it comes to make sense of what it can phenomenologically know about the world (which includes itself).
Is there anything to say about how I-E and D-I interact or complement one another?
It’s a very good question: one would think these basic pairs of activity would mesh in some way. One would think there is a I-E-D-I structure. As you say, I-E is about the mind-world interface while D-I is about the mind’s internal operations. Maybe it does D-I so that it can do I-E.
Yes, I see what you mean: D-I would seem to be required for an effective I-E. But I wonder if I-E, specifically the boundary that facilitates I-E, is necessary for, or underpins, D-I. In any case, boundary as an operation seems to be relevant to both.