shytone  books  music  essays  home  exploratories  new this month

book reviews

Henry Plotkin: The Nature of Knowledge:
concerning adaptations, instinct, and the evolution of intelligence
(Allen Lane: 1994)

“To know something is to incorporate the thing known into ourselves. Not literally, of course, but the knower is changed by the knowledge, and that change represents, even if very indirectly, the thing known. This is ancient folklore, but also a commonplace assumption about knowledge that we all make without really thinking about it.... However, it is also something that bears a very close relationship to a much more widespread property of living things, namely the organizational and structural harmony that exists between life and the world in which it has its being. First exposure to this idea leads many to judge it rather weird, and perhaps difficult to understand.... [But] the apparent fit, the matching, of living things to the features and conditions of their a readily observed characteristic of life forms, that immediately impresses itself upon us humans, with our special talent for detecting correlated patterns in the world about us. [And that] matching is a result of living creatures somehow incorporating into themselves those aspects of the world that are matched. This is the source of the sense of harmony between the organization of living things and the world about them.”
(Plotkin, pp.ix-xiii)

And this rarely-noted harmony - between biological adaptations and knowledge as such - is the keynote insight that informs this crucial book which, in its careful and pragmatic fashion, to my mind makes redundant the entirety of philosophical ontology, as well as much of our conventional epistemology to boot... To be sure, Plotkin has hardly invented the approach he argues for here - “evolutionary epistemology” - but since his is the only book available which outlines same for a general audience, as well as being in itself a masterpiece of clear thinking and insightful reasoning which substantially advances the area as a whole, we may perhaps be forgiven for treating it as sui generis...

“[Moreover,] if adaptations are knowledge, and what we commonly call knowledge (or better, our ability to gain knowledge) is an adaptation, then what in ordinary everyday life we call knowledge is actually a special form of this wider phenomenon, what I am here calling biological knowledge.... And science itself is a very special kind of human knowledge. A science of knowledge, then, is a particular kind of knowledge about a special case, human knowledge, that is part of a wider form of knowledge, biological knowledge. If these claims elicit the image of wheels within wheels, that, as we shall see, is entirely appropriate. What follows, then, is the unpacking of this seemingly contorted argument.”
(Plotkin, p.xvi)

Interestingly enough, the basic ideas noted above are not actually that alien to many today, more familiar perhaps with natural history than with the convolutions of philosophical reasonings upon this topic. And yet, such a perspective has evidently come hard, since there is virtually no sign of it prior to Darwin, whilst the traditional philosophical perspectives on knowledge - rationalist and empiricist - both shipwrecked upon the elusive (and frankly delusionary) notion of absolute proof, albeit they did make some useful analyses at certain points along the way. A good sampling of Plotkin’s incisive intelligence can be gained from examining his summary of this enormous well as his unusual  ability to cut through the crap and highlight the exact point under discussion, shorn of all unnecessary baggage:

“To summarize, knowledge is always something that comes in two parts. There is the ‘knower’s end’ of knowledge, comprising feelings, brain states and, of course, the means of expressing the knowledge; and there is the ‘world’s end’ of knowledge, which is that aspect of the world that is known. All knowledge is a relationship between the knower and the known. It is also a matter of common experience, as well as useful to this book, to make certain distinctions, among the many that can be made, between different forms of knowledge and knowing. One is between knowledge that is emotionally laden and that which is not. The second is between knowledge of events that are temporally coincident with the act of knowing (knowing by the senses) and knowledge of events that are temporally dislocated from the act of knowing (knowing by the mind). The third distinction is between knowledge that is gained by the direct experience of the knower, and knowledge that is gained through the experience of others, that is, shared vs non-shared knowledge.”
(Plotkin, pp.10-11)

Having thus culled the philosophical literature - although retaining a surpassing fondness for careful argumentation and clear definitions/explanations - Plotkin then introduces his readers to the various abstract versions of Darwinian logic which will be crucial to the upcoming argument. Firstly, however, he also disposes of non-Darwinian evolution, by explaining that - logically - the set of possible explanations is already full...and, that all the evidence so far supports Darwin rather than Lamarck in those areas where we have some real understanding and that, contra much pessimistic verbiage, this is undoubtedly a very good thing indeed:

“At an abstract level of description, Darwinian theory and Lamarckian theory offer the only two forms of evolution that have ever been presented. All else derives from these fundamental types. Darwin’s was a selectionist theory of evolution, whereas Lamarck’s was instructionist.... In terms of mechanisms, what Lamarckian evolution requires is a highly malleable substrate that can be modified by the environment and then transmitted to offspring - hence the inheritance of acquired characteristics. Darwinian evolution is driven by mechanisms that generate great diversity of characteristics, by some means of selecting between them and then by a device for propagating the selected variants to offspring.”
(Plotkin, pp.32-3)

“Put crudely, in the case of instruction the environment rules; in the case of selection, internal or organismic states lead.... [And, furthermore,] while selection can mimic instruction, the reverse is never true. Instructional processes can never lead to creativity. Instructional intelligence comprises only what has been actually experienced. To go beyond experience requires the generation of something from inside the knower, and only an intelligence driven by selectional machinery can do that. Indeed, according to D.T. Campbell, the father of modern evolutionary epistemology, selectional processes are required for the acquisition of any truly new knowledge about the world: ‘In going beyond what is already known, one cannot but go blindly.’”
(Plotkin, pp.166-72)

However, once we have been properly introduced to evolutionary processes, it is their inherent logic - and minimum embodiment requirements - which are of most relevance here. This is quite abstract territory for a work aimed at a general readership, but Plotkin handles the issues with his usual clarity:

“In a classic article on evolutionary theory published over twenty years ago, the American biologist R.C. Lewontin described Darwin’s theory as comprising three [abstract] principles...[which Donald Campbell later] labelled collectively a ‘blind-variation-selective-retention process’. Although there is some mismatch with Lewontin’s three principles, the three essential ingredients of Campbell’s overall process are similar to those of Lewontin. These are ‘mechanisms for inducing variation’ (Lewontin’s phenotypic variation); a ‘consistent selection process’ (this is a combination of Lewontin’s differential fitness and its heritability); and ‘mechanism for preserving and/or propagating the selected variants’ (Lewontin’s principle that fitness be heritable). There is also a more economic, and for some purposes more convenient, way of stating [this]...which goes by the name of a ‘g-t-r heuristic’ (g-t-r standing for generate-test-regenerate).... It is in the universal processes of variation, differential fitness and heritability, transmission of selected variants and their combination with new variants that we have ‘universal Darwinism’.... So the next question is, given these general principles of evolution, what is it in nature that we would like to put our fingers on and measure? What is it that provides the causal mechanisms for these principles? What are the significant units of evolution?”
(Plotkin, pp.82-6)

And here, again, we find an abstract approach - generated, in different forms, by Waddington, Dawkins & Hull - which sets out the minimum range of necessary entities required to generate a g-t-r heuristic...which, following Hull’s formulation, is summarised as follows:

“If entities that can make copies of themselves (replicators) are propagated in space and conserved in time because of the differential extinction and proliferation of interactors, these will in turn lead to historical change in lineages, and evolution will have occurred.”
(Plotkin, p.97)

Plotkin’s treatment of the variety of models on offer here - in attempts to strip down evolutionary processes to their functional core - is an exemplary piece of work, and an essential plank in his argument. For, what makes understanding of Darwinian evolution upon such a level of abstraction truly essential, he explains, is the need to deploy it hierarchically, given the complexity of the processes involved. For Plotkin’s arguments re knowledge and adaptation are no simple analogy. Instead, they stem directly from the very centre of modern evolutionary theory...albeit they also insist that this theory will need to be extended...

“A hierarchy is an ordering of entities, at least a part of the ordering being dependent upon scale, that is, some dimension such as size, energy level or frequency. Now, a source of great confusion is the existence of two seemingly quite different kinds of hierarchy. The one, called a structural hierarchy, is characterized by a feature known as containment, or the Chinese-puzzle characteristic.... The most obvious example is ourselves. Open us up and you find organs, inside of which are tissues, which contain cells, in which are to be found organelles, which are made up of macromolecules.... The second kind of hierarchy, a control hierarchy, is much closer in meaning and form to the original conception of hierarchical structure, which described the relationships of the angels to the Lord.... Here the scaling is one of authority, and the characteristic of containment is absent.... Another important feature of control hierarchies is that they are much more dynamic than structural hierarchies...[and] are characterized by complex causal interactions between levels, whereas for structural hierarchies the principal interactions occur within levels.... The relationship between structural and control hierarchies is an extremely important theoretical issue. It is one of the central conceptual puzzles that have to be solved if the biological, cognitive, and social sciences are to be married within the kind of grand synthesis that is occurring in the physical sciences...[for] Neo-Darwinism has never been built upon a foundation of hierarchy theory.”
(Plotkin, pp.44-6)

There have been many proposals for a hierarchical extension of neo-Darwinism - Eldredge’s perhaps coming to mind most readily - but all have tended to suffer from the intractable complexity of the task...rather as attempts to reconsider “economic man” have also suffered. Because, put simply, once a discipline becomes wedded to mathematical modelling - as have population genetics & microeconomic theory - any proposal for reform is doomed if it cannot be rendered matter how clearly such reform is needed. However, Plotkin’s does have the advantage in addressing an area - the relation between adaptation and knowledge - where, thankfully, no psuedo-mathematical standard exists. This, in combination with the clarity of his argument, and the insight it delivers into the origins of behaviour & intelligence, should have made it the standard in this area...and still might, if the book could only become better-known?

Anyway, after the lengthy preparatory argument - setting out the key ingredients to be combined in his work - Plotkin then makes the crucial move which, intriguingly, emerges from the basic nature of adaptations:

“Two particular features of adaptations [may be] considered to be of special importance. One [is] their goal- or end-directed nature.... [And] it is...being ‘for’ something, this purposefulness, that gives biology its teleological character, and makes it so easy to talk about adaptations [and genes] as having goals. It also leads directly to the second characteristic of adaptations. This is their relational quality. Every adaptation comprises organization of an organism relative to some feature of environmental order.”
(Plotkin, pp.116-17)

But, as we know all too well from personal experience, environmental order is also inextricably mixed with environmental disorder...which is the point at which Plotkin’s insistence that adaptations comprise a form of knowledge starts to pay real dividends. For it is only what he terms the “uncertain futures problem” which makes any degree of flexibility at all necessary - and, viewed from this perspective, the continuity between developmental plasticity, fixed-action behaviours, simple & complex learning, and intelligence-driven knowledge acquisition/action appears compelling, as all are in essence tools evolved in order to deal with such disorder. And that continuity is most clearly displayed in development:

“Given the essential causal links between behaviour, genetics, and development, the claim that behaviour is a form of knowledge should properly be extended to include the genetic and developmental factors that determine behaviour.... [Moreover,] development is not an automatic, pre-ordained unfolding process which, once initiated, proceeds to the completed state of the adult organism. Rather, each individual is, in a real sense, created anew, the unique outcome of an immensely complex series of interactions between the different parts of the genetic constitution of that individual; and also between its genes, its developing parts and its environment.... [Therefore,] developmental plasticity can itself be seen as an adaptive device, that is, as a knowledge-gaining device.... At conception, the knowledge relationship is a potential one between the genes representing the internal end of the knowledge relationship, the external end being the range of possible developmental environments that have been the conditions exerting the selection pressures which led to those genes being selected over long periods of time. As development of the individual proceeds, actual environments and a reduced selection of genes in the total gene complement of that creature make up the external and internal components of the knowledge-gaining process.”
(Plotkin, pp.122-5)

And, the end result is an organism tailored - from the much wider range of choices present in the genome itself - by developmental “choices” triggered by environmental inputs throughout the process. So much for “nature vs nurture”, eh? However, perhaps the key lesson to be learnt from such processes is the sheer continuity between adaptation and learning - the key insight behind this book. For, after all, such contingent developmental processes can easily be understood as learning - or adaptation - it merely depends which set of blinkers you prefer, since such processes are clearly poised on the dividing line we usually assume between the two. But, to realise this is to truly understand Plotkin’s point - that such a divide is in some ways more an artefact of our thinking than of nature itself...and that a broader view may teach us lessons we could learn in no other way...

“There are three main groups of multicellular organisms. Two of them are the plants and fungi. Neither has has the equivalent of a nervous system and, of course, their members do not have actively to move through space in order to earn their own living.... [But,] subject to the same uncertain futures problem as all other living forms, they combat it with extraordinary reproductive cycles, in the case of fungi, and very extensive developmental flexibility, particularly in plants.... Such developmental plasticity is itself a form of tracking of environmental conditions, and is undoubtedly the principle answer that plants and fungi have to the uncertain futures problem.... [On the other hand,] it is significant that the phyla containing the known intelligent species are mostly those characterized by vigorous activity. Animals that move are subjected to self-induced high-frequency change, and I believe such change has been an essential selection pressure in the evolution of intelligence.”
(Plotkin, pp.154-6)

But this argument, fine as it is, clearly does not go far enough, given the sheer cost of intelligence - in terms of very expensive to operate brain tissue - not to mention the lengthy training it requires before becoming fully functional. Much cheaper to rely on fixed-action sequences. However, these too have their disadvantages. Instead, we need to conceive of intelligence, as you might expect, as one possibility within the range of adaptive responses to that selfsame uncertain futures problem. Because the g-t-r or primary heuristic cannot “track” somewhat irregular - non-periodic - changes, what we might term “predictable unpredictability” needs to generate more flexible systems in order to do so, albeit few of these are intelligent. The bulk of such systems, in fact, simply serve to iron-out instabilities which may be viewed as being superimposed upon longer-term stabilities, or cyclical patterns - such as homeostatic mechanisms deal w/temperature changes. Longer-term uncertainties, such as may effect an entire life, can be dealt with, as noted earlier, by developmental plasticity. However, they may also be evaded, as it were, by choosing the dominant reproductive strategy.

That is what MacArthur & Wilson dubbed the r-strategy: very large numbers of offspring, very little parental investment & short life spans - usually one-shot reproduction - thus allowing the g-t-r heuristic ample material from which to choose, and minimizing the chance that the genetic instructions to be passed on have been rendered out-of-date by environmental changes over time. And although some r-strategists have developed intelligence - the cephalopods, in particular - by far the bulk of species showing clear learning abilities (as good a definition of “intelligence” as any) come from lineages marked by what are termed k-strategies for reproduction: small numbers of offspring, high levels of parental investment & long life spans...thus having to cope w/change rather than simply being selected by it...

“The situation with regard to the brain can be set out as follows. Think of the world as comprising sets of features, some unchanging and others changing at different rates and with different degrees of regularity. Irregular and infrequent change is unpredictable, and only sheer luck...will govern survival. What of more regular and frequent change? Well, if the frequency of change is less than that set by generational deadtime for extracting genetic information from the gene pool and then returning them to it, then the conservative component of the primary heuristic will be able to ‘see’ these changes, and will furnish adaptations to match them. But, if the frequency of change is faster than the frequency set by generational deadtime, then though the primary heuristic will be able to see the long-term stabilities upon which these changes are superimposed...[it] will have to evolve devices that operate at a much higher frequency - at a frequency high enough to track these values. If the high-frequency changes are unstable, the tracking device itself does not need to change state. Indeed, it would be maladaptive to do so. All it need do is command an immediate compensatory response. However, if these changes also have temporal characteristics that make them short-term stabilities, then these tracking devices must comprise a secondary heuristic that is able to change and maintain new states that match those features of the world that are being tracked.”
(Plotkin, pp.149-50)

And that, in a somewhat compressed and inelegant form (compared to the original and extended explanation) is how & why Henry Plotkin is so insistent upon the broader notion of knowledge he so effectively champions in this book. For, viewed through the temporal lens of the uncertain futures problem, it does indeed seem clear that there are a whole variety of adaptive means used to cope with this, and that - as far as the nature of knowledge goes - our human knowledge is simply one highly specialized offshoot of same.

“Intelligence is an adaptation that allows animals, including ourselves, to track and accommodate to change that occurs at a certain frequency. Slower rates of change are adapted to by the genetical and developmental machinery. Higher rates of change are adapted to by tracking mechanisms whose own states are not altered by the transient events to which they respond. Intelligence, rationality, is an adaptation that has evolved to deal with changes that occur at rates somewhere between these two. [But] it is very important to remember that...the same features of the world changing at some fixed absolute rate will constitute different forms of change for individuals of different species. The change from autumn to winter will be a gradual linear change for a bug that lives only a matter of weeks, part of an unrepeating sequence of change for a bug that lives for one year, and part of an oft-repeated cycle for a bug that lives for many years.”
(Plotkin, pp.150-1)

Furthermore, as Plotkin goes on to argue, this perspective offers some distinct advantages when it comes to some very traditional problems, in particular, our perennial wrangles over  nature vs nurture:

“Intelligence is part of the secondary heuristic that evolved because of the temporal sampling limitations of the primary heuristic. Because intelligence can only operate as a device for dealing with predictable unpredictability, where the unpredictable element is a rather extended wobble (a short-term stability) on the predictable element, this means that the secondary heuristic is functionally under the wing, so to speak, of the primary heuristic. And this means that it is directed by the primary heuristic - the primary heuristic tells it, roughly, what to do. There is a technical phrase for this condition, which is that the two heuristics are in a relationship towards one another known as a nested hierarchy. Specifically, it is a control hierarchy where the scaling factor is the frequency of change in the world to which each is sensitive, and the primary heuristic is the more fundamental level.... [This] resolves the old nature-nurture problem, that goes back at least to the writings of Plato, in a manner different from the currently widely-accepted resolution...such claims going, in general, as follows. ‘The instinct-intelligence and genes-experience dichotomies, or any other of the dichotomies that characterize the nature-nurture distinction, are false dichotomies. This is because all genes require an environment in which to develop."
(Plotkin, pp.161-4)

"In my view, this is a fundamentally wrong approach to the matter. The solution it provides is ‘horizontal’ in that it still maintains the separation of the genes, coming as it were from one side, and the environment, coming from the other, with some kind of developmental integration in the middle. It is called interactionism, with the interaction purporting to get rid of the separateness of the two components. However, the nested hierarchy a quite different kind of resolution, because the scheme is ‘vertical’. Intelligence, the secondary heuristic, is subsumed under, enclosed by, the primary heuristic. The horizontal components of interactionism, the internal and external causes, remain present at both levels. They are not the issue. The issue is some superordinate concept that subsumes all levels of the hierarchy, and that concept, I am arguing, is adaptation: knowledge. Intelligence nested under development and the genetic process does not allow any claims about such-and-such being caused either only by nature or only by nurture or by both. Such language and imagery are wrong. Intelligence is an adaptation, and the required integration can only be achieved vertically, not horizontally.”
(Plotkin, p.164)

However, lest my readers think that all is sweetness & light in this work, it is worth pointing out that its penultimate chapter is both seriously misleading & sadly out of date. Because, writing here over a decade ago, Plotkin unfortunately mistook some (now) very dated modular theories of brain function for the only real alternative to tabula rasa approaches - and totally ignored (as most did then) the very existence of that sophisticated modern developmental work which is increasingly being vindicated. However, given that he has since devoted two entire books to his main concerns in said chapter - evolutionary psychology & a prospective science of culture - and corrected his most significant mistakes in doing so, it would be uncharitable of me to dwell upon this further, particularly given excellence of the rest of this book.

For, to my mind, Henry Plotkin’s The Nature of Knowledge is a crucially important work, and its approach to the problem of knowledge is exactly what the doctor ordered...after millennia of inconclusive philosophical debates. Because...from the perspective of evolution as a process and, looking at the uncertain futures problem in detail, we can now see just why (and how) intelligence turned up...and just what its basic nature appears to be, at least from this privileged vantage-point.

“A science of knowledge is the first and most essential part of a more general project to write a proper science of human beings. We are not able to do that yet, but some time in the future we surely will be. Central to such a science will be a proper understanding of our extraordinary capacity for gaining and communicating knowledge; knowledge that must be understood first as a part of our nature, and only after that as an issue in nurture. Something like [this] theory...will be part of that science.”
(Plotkin, p.xviii)

John Henry Calvinist