Chapter 4



This chapter is mainly on Lamarckism and has critical and constructive messages. We focus on the meaning of Lamarckism as “the inheritance of acquired characters”.

On the critical side, the chapter rebuts the idea that social evolution is Lamarckian rather than Darwinian. This is mistaken for a number of reasons. First, Darwinism (suitably defined) and Lamarckism are not mutually exclusive. In fact, Darwin believed in Lamarckian inheritance in the biological world. Subsequent biologists have thought otherwise, but that is beyond the point. Second, even if Lamarckian inheritance did occur (as on our imaginary Planet Lamarck) it would still require some process of selection to explain evolution in populations. Consequently, if Lamarckism were true it would require Darwinism. (This point was made previously by Richard Dawkins.)

Lamarckism raises a number of other questions and problems. As the late David Hull pointed out, it is important to be clear about the meaning of “inheritance” in the “inheritance of acquired characters”. Inheritance must be distinguished from contagion. Otherwise the transmission of fleas from one dog to another would be the “inheritance of acquired characters”. It is clearly not.

In contrast to contagion, inheritance means the copying of developmental instructions. It is not good enough for a Lamarckian to say that if we develop strong arm muscles then our children will have strong arms too. The Lamarckian must uphold that we have passed on the developmental capacity to develop strong arms through our genes, which carry these (new) developmental instructions.

Now here comes the clinch, which signals the constructive part of this chapter. Hereditable “developmental instructions” (with a bit more definitional stuff added), are REPLICATORS. Consequently, simply to make sense of Lamarckian inheritance (leaving aside whether it is true or not) we have to adopt something like the REPLICATOR-INTERACTOR distinction. Those that reject the replicator concept while saying that (social) evolution is Lamarckian (e.g. Richard Nelson and Peter Richerson) cannot have it both ways.

Finally, and also on the constructive side, we consider possible social replicators (above the level of genes), namely habits and routines. When we address these replicators and consider possible Lamarckian inheritance in social evolution, further problems arise. Something (very vaguely) like Lamarckism may exist in social evolution but calling it Lamarckian is highly misleading.


1. Are Darwinism and Lamarckism mutually exclusive?
2. What is the difference between inheritance and contagion?
3. Why does coherent Lamarckism require the replicator concept?
4. What are possible social replicators?


5 thoughts on “Chapter 4

  1. schmid2013

    I know the authors prefer to stay at the abstract level but I would appreciated a few examples. I observe in the business world, that new employees who express sympathy for the Democratic Partly and empathy for the poor are scorned and laughed at. The employee gets the message that if you are not one of the boys, chances for promotion are nil. (survival is at stake) . Identification with the Republican Party may not necessarily be passed to the employee’s children, but children who admire their father may adopt his behavior. Does scorn/approval function as a replicator? I once heard a 10 year old say “If Obama wins, it will cost my father money.”

  2. Geoffrey M Hodgson Post author

    It is not true that we “prefer to stay at the abstract level”. Our book has several illustrative examples. It is good to think of an example and reflect on its implications for the abstract framework, as Allan Schmid does here. The example he gives is one where a habit of thought (of a political nature) gets transmitted from one person to another, by party coercive means (scorn for contrary views). There are many examples of this kind of thing and it has been studied widely by social psychologists (see Solomon Asch and others). The key question is whether this is replication or contagion. (The question is important, at least to determine whether the inheritance of acquired characters is involed.)

    To answer this question one has to look at the criteria laid out on page 77, concerning causality, similarity and information transfer. (These criteria were developed by Kim Sterelny, Kelly Smith, Michael Dickison, Peter Godfrey-Smith and Dan Sperber.) The answer with Allan’s example is affirmative. So this is a case of replication/inheritance. (We treat replication and inheritance as synonymns.)

    What would be an example of social copying that did not involve replication/inheritance? An example we give is contagious laughter. Here the capacity to produce the laughter is already there – it is simply triggered by a stimulus. If some Americans inherited genetically the disposition to vote Republican then an employee may simply be undergoing a stimulus to realise his or her cogenital views, even if sanctions are imposed to prevent deviation, and this itself would not be a case of social replication or inheritance.

    Some people do argue that some of our ideological dispositions are traceable to our genes. But if this were true, we still would not inherit a fully-fledged Republican disposition. At least a major part of that must be cultural. Just as a “language instinct” does not mean that we inherit a fully-formed capacity to speak and understand a language, political habits of thought require immersion in a particular culture. Lines are difficult to draw here, but that is always the case with multiple, variegated, complex phenomena.

  3. len wallast

    This chapter 4 about the definition and significance of the replicator and Lamarckism has quite a number of points to address, indeed critically as well as constructively. H&K seek to replace the gene of biological evolution by a closely similar unit of selection, the replicator, for social evolution. In this “gene selective” context Lamarckism is dealt with in a systematic way. Replacing the gene by a suitable unit of selection is undoubtedly the compelling thing to do, not only for defining a unit of social selection but also to provide for the explanatory imperfections of the gene as a unit of biological selection. After all the gene is far from perfect as a unit of selection. If the connection of replicator and gene is too close, the imperfections will be inherited within the explanatory framework of social evolution. I prefer here not to go into the many interesting details of this chapter but to confine my contribution to the real problem that plays and H&K wish to deal with. That problem is how information/(Shannon entropy) is transferred from the past into the future. To address the matter I need first to make several explanatory notes. The conclusion I will eventually arrive at is that some (essential) repairment of H&K’s replicator definition is necessary in order to get a full fledged universal and consistent theory of evolution in which the transference of information from the past into the future simply always occurs and cannot be disputed. Lamarckism, irrespective of how it is defined and interpreted, is then of no significance with respect to that transference of information.
    In biology genes are the carriers for the transportation of information in the course of time. The present generation of the evolving population passes its genes to the next generation by sexual procreation. This is called “inheritance” in a biological evolutionary context. But that term is one-sided as it addresses only one side/aspect of the transfer of information. The phenotype/population is not the only interactor/sector subjected to input selection (by which it looses information) and to output selection (by which it gains information). Also the environment is subordinate to similar statistical processes of adaptation through input and output selection. For instance, woodland is converted into farmland and permanently cultivated for raising crops to the benefit of men. The information content of the land is constantly increasing/decreasing by this process (from an economic point of view). We have here two problems. The first is: what is the common unit of selection for the evolution of population and environment? I think that habits, routines, customs etcetera, rudimentarily specified as they are by H&K, are too less accurately defined to contribute to the increasing or decreasing value of the land. Neither genes suffice here. The second problem is: how does the environment (here: the land) transfer its content of information of the past into its content of information in the future? If we try to run away from answering these questions, we are definitely on the wrong track. We cannot deny that the habitat of the environment and the population exploiting that environment evolve concurrently and affect one another. So there must be mutual exchange of information.
    These delicate questions can remarkably be done with (only) by a balanced mathematical argumentation. An evolutionary system S0 consists of two subsystems S1 and S2. It must be. Else selection (which requires two different states of selection) is impossible. In set terminology: S0 is the union of S1 and S2. As a convention I take S1 to be the population and S2 to be the environment. Both S1 and S2 stock information/entropy that consists only of very many small units of information. Each unit of information stocked in S0 has been selected in the past for output creation at an earlier time and will be selected for input annihilation in the future at a later time. Before it was created it did not exist and after it gets annihilated it does not exist anymore. And anything that does not exist contains no information. Hence each unit of information stocked in S0 has a finite total lifetime stretching from the moment it was selected as output in the past (at its initial time) until the moment it will be selected as input in the future (at its final time). Moreover, either it is presently stocked in S1 (and carries the state 1) or it is presently stocked in S2 (and carries the state 2). Keeping that in mind, it follows that the described units of selection that are hosted in S1 (the population) and in S2 (the environment) are the equivalent of replicators of unit entropy carrying either the state 1 or the state 2. Note I have reintroduced a revised replicator here as a unit of selection. To warrant that those revised replicators are similar (for probabilistic selection requirements), it is necessary that each replicator carries the same single unit of information/entropy during the time it exists: i.e. from the initial time it originated (when it was selected as output) until the final time it ceases to exist (when it will get selected as input).
    Mark that this scheme warrants that information/entropy is always transferred from the past into the future as the simultaneous execution of the input selection experiment and the output selection experiment over a particular time-interval reduces the input uncertainty from k to (k–X) and the output uncertainty from k to (k–Y). This results in an information/entropy gain of (k – X) – (k – Y) = (Y – X), which is the surplus of output over input over the time-interval of selection (See for the more detailed explanation my earlier comments with respect to Chapter 3). This surplus of output Y over input X is equal to the difference between the number of all the newly created replicators within S0 and the number of all the utilized replicators within S0 over the time-interval of selection Thus S0 does not only keep stock of the current set of replicators but is constantly adding new replicators to its stock and removing old replicators from its stock. If (Y – X) is positive the systems grows, if (Y – X) is negative the system contracts.
    The above conception of the selection process offers a general explanation of the transfer of information/entropy within an evolutionary system S0 from the past into the future. Once we adopt this explanation of the selection process, it is no longer necessary to deal with many nasty and still unanswered or insufficiently answered questions connected with genetic or other forms of inheritance such as: Are genes the only biological carriers of information? What is the information content of each of the genes? And in what way are different genes, each with varying content of information, similar objects of selection? Do different genes operating together express more phenotypic information than the genetic information of each of the genes separately? What are the details of the process by which the content of information of the population grows? Etcetera, etcetera. I can list many more and I am sure we will never arrive at satisfactory answers without the proposed revision of the replicator concept.
    The theory of selection by replicators that carry each one bit of entropy during the time they exist from their random initial time until their random final time provides for a perfect explanation where all these nasty questions have been answered or been done with. The theory applies to all biological systems of evolution and equally well to all evolutionary systems other than biological. It offers a really general solution of the passage of information/entropy in the course of time and its associated net growth. This is also the reason why Lamarckism (the inheritance of acquired characteristics) is superfluous to explain the transfer of information from the past into the future. We don’t need such explanation anymore. Moreover Lamarckism is incapable of giving any better explanation. Thus why considering it further?
    I wish yet to devote a few words with respect to what a replicator actually must be to get it all straight. It is often assumed that the units of selection (the replicators) are random sequences of alternating states (1’s and 2’s) extending over time such as the binary sequence 12122111221212. However that supposition is wrong. As I argued in the above, replicators can only reside in a single state (1 or 2) during the finite time they exist. However, over the very small time-interval of selection, a statistical experiment of selection is executed k times with k being very, very large. (See for the notation and a more detailed explanation my comments in Chapter 3). Each time we draw a sample of a replicator from S0 we get a replicator either in state 1 or in state 2. After executing this k = 14 times we have selected 14 replicators consecutively and we might for example get the rearranged random sequence 12122111221212. This sequence is what I call a variation. Apparently a variation holds k replicators. Thus a replicator is not the equivalent of a variation, it is a single element (one draw) of a variation of k draws. While replicators have a finite random lifetime, a variation has a “lifetime” dt that must tend to zero.
    This discernment of replicator and variation demands that a variation extends over an infinitesimally small length of the time-interval of selection: let us for instance assume a time-interval (t,t+dt) of selection stretching from time t to time t+dt, which is an infinitesimally small instant dt of time later than t. Only then it is warranted that all the draws of samples (that constitute the selected input variation and output variation as the outcomes of the input and output selection experiments) occur on the same time-interval (t,t+dt) during which the selected output replicators originate and the selected input replicators cease existence. This is as it should be because the sacrifice of input to produce output must occur at the same instant of time at which that output is produced.

  4. Geoffrey M Hodgson Post author


    In his comment posted on the Chapter 1 blog page on 21st April, Peter Richerson wrote; “I don’t think the replicator concept works for cultural variation. It is easy to come up with inheritance schemes that do not involve literal replication.” In response I asked Pete to raise his criticism of the replicator concept in the blog page devoted to the chapter of the book where the replicator concept first appears, which is chapter 4. Hence the time and place for this discussion is now and here.

    If I understand Pete correctly, he suggests that genes are the only true replicators and they are not strictly or directly relevant for cultural inheritance.

    To claim that genes are the only possible replicators one would have to identify the defining features of a replicator and show that other candidates do not comply. Thorbjoern and I adopted the abstract definition of the replicator developed by philosophers of biology and show that habits, customs and routines can be treated as replicators.

    Furthermore, I would argue that the inspiring models of dual (genetic plus cultural) inheritance developed by Pete and his colleagues involve both cultural and genetic replicators. Essentially, replicators are not “things” but special bits of information. Cultural transmission means the transmission of information. It means replication (or inheritance – these words we treat as synomyms).

    Some accounts see a replicator as a “thing” that “makes copies of itself”. We think that this is mistaken, misleading and vulnerable to criticism. Instead our concept of the replicator is essentially informational.

    1. Harrison Searles

      Myself, I do not think that we can make sense of history without a concept of a cultural replicator. History is clearly evolutionary with forces, whatever those forces may be, selecting for certain institutions, and social structures. There is also the clear process of learning within all human societies by which youth learn the skills and knowledge necessary not only to adapt to their environment, but also, more importantly (especially in modern times), to the conventions that make social cooperation possible. If we are to make sense of all this, it seems to me that there must be some mechanism by which these cultural traits, then we we need a concept of a replicator that can ensure the inheritance of cultural traits across the generations and between individuals.

      Habits, customs, and routines fill in this needed explanatory hole. Even better, they actually have real presences in the world rather than simply being theoretical constructs to make the lives of academics easier. Yes, they may much more messy than biological replicators, that messiness does not detract from the fundamental point that habits, customs, and routines have made possible the transmission of cultural adaptations across people.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s