There is something almost irresistible about the phrase “the selfish gene.”
It is vivid.
It is memorable.
It gives an abstract process a protagonist.
A gene is no longer merely something that participates in a complex biological system. It appears to have an interest. It wants to survive. It wants to reproduce. It competes with other genes. Organisms become the vehicles through which these ambitions are pursued.
The metaphor is extraordinarily effective.
And that is precisely why we should pause over it.
For the danger of a successful metaphor is not that we mistake it for a metaphor immediately.
It is that, after it has become familiar, we forget that it was one.
The gene did not suddenly acquire desires when Dawkins called it selfish. Nor did it become an agent. What changed was the construal through which we were invited to understand the processes in which genes participate.
This distinction matters.
Because once we have made the gene the protagonist of evolution, it becomes surprisingly difficult to remember that the gene does not, in any ordinary sense, do the things the grammar of the metaphor says it does.
The gene does not want to survive.
It does not decide to reproduce.
It does not compete.
It does not make plans.
And yet it can be extraordinarily useful to speak as though it does.
The question, then, is not whether the metaphor is true.
It is:
What does the metaphor allow us to see, and what does its grammar encourage us to overlook?
The seduction of agency
We understand events partly by assigning roles to participants.
Someone does something.
Something happens to something else.
Someone causes something.
Someone benefits.
Someone suffers.
Human language is extraordinarily good at construing experience in these terms.
And there is good reason for that.
Much of human experience actually does involve agents.
People act.
Animals act.
Institutions act.
Machines can be designed to perform actions.
When we encounter a process whose organisation is complicated, one of the easiest ways to understand it is to imagine an entity standing behind the process and making it happen.
The metaphor of the selfish gene exploits this cognitive and linguistic affordance.
Instead of saying that certain patterns of genetic inheritance tend to persist because of their consequences within a system of reproduction and selection, we can say:
the gene is selfish.
A vast network of processes is compressed into the behaviour of a single participant.
The result is conceptually economical.
It is also rhetorically brilliant.
But economy has a price.
The more readily the metaphor explains the process, the easier it becomes to forget what has been compressed.
From process to participant
There is a deeper linguistic transformation here.
We can begin with a congruent process:
genes are replicated through biological reproduction.
Here replicated is a verb: the process is construed congruently as a Process, with genes participating in it.
We can then reconstrue that process nominally:
the replication of genes through biological reproduction.
Now replication, the process expressed congruently by the verb replicate, has been turned into a noun. It can therefore function grammatically as an entity-like participant in another process.
a gene is involved in replication.
The word replicator makes a further move. Through the agentive suffix -or, an entity is construed in relation to the process of replication as the one that performs it. The language therefore makes it remarkably easy to move from replication to replicator, and from there to the idea that the replicator itself is the agent of replication.
The grammar has begun to do some of the philosophical work for us.
The gene as actor
This is especially visible in the language surrounding the selfish gene.
Genes are said to:
compete;
survive;
reproduce;
cooperate;
manipulate;
suppress;
favour;
promote;
protect.
Again, there is nothing illegitimate about such language.
Scientists routinely use metaphor to make complex processes intelligible. The problem begins only when the metaphor's construal of agency becomes invisible.
The gene is not an actor in the same sense as an organism.
It does not initiate behaviour in pursuit of a goal.
It is a participant within a network of biochemical and reproductive processes.
Those processes include interactions among genes, cells, organisms and environments.
The gene's persistence is a consequence of how it participates in that organisation.
The gene does not need to possess a desire for survival in order for some genetic variants to become more prevalent.
This seems obvious when stated explicitly.
Yet the language of selfishness makes the opposite construal remarkably easy to inhabit.
The metaphor does not merely decorate the explanation.
It gives the explanation a dramatic architecture.
There is a protagonist.
There are competitors.
There are strategies.
There are winners.
There are losers.
There is a struggle for survival.
Evolution becomes a story.
And stories are powerful ways of organising experience.
The advantage of the metaphor
We should therefore resist the temptation to dismiss the metaphor as simply misleading.
It does something genuinely valuable.
The gene-centred perspective draws attention to an important fact about evolutionary processes: the persistence of inherited patterns can be analysed independently of the intentions or interests of individual organisms.
An organism may reproduce without “wanting” its genes to survive.
Indeed, the organism does not need to be interpreted as the ultimate beneficiary of every evolutionary process.
The metaphor of selfishness helps us see that evolutionary persistence can be considered from another level of organisation.
It asks us to follow the inheritance.
Which inherited patterns persist?
Under what conditions?
What consequences do they have for their own continued representation?
How do interactions among inherited variants affect their differential persistence?
These are serious questions.
The metaphor can make them easier to ask.
So the problem is not:
“Selfishness is false, therefore the theory is false.”
The problem is subtler:
A metaphor can illuminate a pattern while simultaneously encouraging us to construe that pattern as the action of an entity.
Its usefulness and its danger are therefore connected.
The same conceptual compression that makes the theory memorable can make its ontology seem more straightforward than it is.
Agency without an agent
This raises a more general question.
Can a process exhibit what looks like agency without there being an agent?
The answer appears to be yes.
A whirlpool moves.
A flame spreads.
A crystal grows.
A market adjusts.
An ecosystem changes.
None of these necessarily requires us to posit an inner subject directing the process.
There can be organised, directional behaviour without intention.
This is important because evolution itself is full of such patterns.
Natural selection can produce structures that look designed without requiring a designer.
Evolution can produce apparent purpose without foresight.
Adaptation can emerge without anticipation.
The existence of organisation does not by itself establish an organising agent.
That distinction is easily obscured when we use strongly agentive language.
If a gene “tries” to survive, we may begin to imagine that its persistence is explained by its attempt.
But the explanatory direction is the other way around.
The persistence is what makes the metaphor of selfishness useful.
The metaphor does not explain the persistence by adding an actual desire.
This is a crucial distinction.
Agency can be a way of construing an organised process without being the mechanism that produces it.
The grammar of evolutionary explanation
From an SFL perspective, this becomes especially interesting because grammar provides different ways of construing the same experiential domain.
Compare:
The gene replicates.
with:
The gene is replicated.
Or:
The gene produces a phenotype.
with:
Genetic processes contribute to the development of a phenotype.
These clauses are not equivalent in their construal.
The first foregrounds the gene as Actor.
The second places the gene in a different participant role (Goal) within a process that happens to it.
Neither clause is necessarily inappropriate in every context.
The point is that they organise our attention differently.
The grammar tells us where to look.
It tells us what counts as the process and what counts as the participant.
It makes some causal relations more salient and others less so.
This is one of the reasons grammatical metaphor deserves our attention.
Nominalisation allows us to package processes into entities.
That is enormously useful for scientific reasoning because it enables processes to become stable objects of discourse.
We can talk about selection, inheritance, replication, adaptation and fitness as though they were things.
This allows us to build increasingly elaborate conceptual structures.
But the gain comes with a possible loss.
Once a process has been nominalised, we can forget that it was ever a process.
And once the resulting noun enters an agentive construction, we can forget that its apparent agency is a grammatical construal.
The language becomes so convenient that the metaphor becomes invisible.
The little theatre of evolution
There is something almost theatrical about the resulting picture.
Imagine the evolutionary stage.
The genes enter.
Some are selfish.
Some cooperate.
Some suppress their competitors.
Some manipulate their organisms.
Some are successful.
Others disappear.
The organisms become vehicles.
The environment becomes the arena.
Selection becomes the judge.
The genes are the players.
It is an extraordinarily effective drama.
And there is genuine explanatory insight within it.
But a drama is not the same thing as the process it represents.
The actors in the drama are linguistic constructions.
The underlying evolutionary processes involve biochemical interactions, developmental processes, reproduction, ecological relations, inheritance and differential persistence.
No little gene stands backstage deciding what to do next.
The metaphor gives us a compressed representation of a distributed process.
The danger is not that the representation is compressed.
Science depends upon compression.
The danger is that the compression can become mistaken for the underlying organisation.
From selfish gene to selfish meme
This matters even more when the metaphor is extended beyond biology.
The gene was described as selfish.
Then the meme was proposed as its cultural analogue.
The meme, too, could now be imagined as something that wants to spread.
A catchy slogan seeks attention.
A religious belief seeks converts.
A political idea seeks adherents.
A tune seeks to be remembered.
An internet meme seeks to go viral.
The language is familiar because it is already built into the metaphor.
But now the danger increases.
At least with the gene, the metaphor is anchored in a well-developed biological theory.
With cultural forms, the underlying processes of inheritance are far less straightforward.
There is no obvious equivalent of DNA replication.
There is no universally agreed cultural unit.
There is no single mechanism by which cultural forms are transmitted.
There is learning, imitation, communication, teaching, ritual, institutions, technologies, language, memory and interpretation.
The metaphor of the selfish meme can conceal all this complexity.
A cultural form appears to be an actor whose only concern is reproduction.
But the form may persist because people find it useful.
Or because an institution preserves it.
Or because it fits existing categories.
Or because it is aesthetically compelling.
Or because it is embedded in a ritual.
Or because a technology makes it easy to transmit.
Or because it gives people a way of understanding their circumstances.
Or because it has become part of the infrastructure of a social practice.
The “meme” does not need to want any of this.
Indeed, perhaps nothing needs to want it.
The problem of the hidden subject
There is a recurrent pattern here.
We encounter an organised process.
We identify a persistent pattern within it.
We give the pattern a name.
The name becomes a noun.
The noun is placed into an agentive grammatical role.
And suddenly the pattern appears to be doing the organising.
This is a powerful form of conceptual compression.
It is also a recurring source of confusion.
We might call this the hidden subject problem.
Who is really acting?
The gene?
The organism?
The population?
The environment?
The developmental system?
The institution?
The individual?
The linguistic system?
The cultural practice?
Sometimes the question itself may be badly posed.
Complex processes may have no single privileged subject.
Their organisation may be distributed across relations among participants.
That possibility is particularly important for evolutionary thinking.
Perhaps evolution does not require an agent at all.
Perhaps it requires only a system in which inherited differences participate in processes that affect their differential persistence.
Once put that way, the selfishness metaphor becomes optional.
It can remain as a vivid way of seeing the pattern.
But it no longer needs to be mistaken for an entity's motive.
What happens when the metaphor becomes literal?
This is where our broader approach to conceptual naturalism becomes useful.
We do not need to ask whether a metaphor is literally true.
That is often the wrong question.
A better sequence is:
What phenomenon does the metaphor help us construe?
Then:
What relations does that construal make salient?
Then:
What does it background?
And finally:
What happens if we begin to treat the metaphorical construal as though it were the structure of the phenomenon itself?
The selfish gene is an excellent case study.
The metaphor foregrounds differential persistence.
It backgrounds the distributed processes through which that persistence occurs.
It makes the gene salient as a participant.
It can make the gene appear to be an agent.
It encourages a language of competition, strategy and interest.
It can therefore reveal something important while simultaneously narrowing the field of vision.
This is not a defect unique to Dawkins.
It is what metaphors do.
Their power comes precisely from the fact that they allow us to see one thing through the organisation of another.
The problem begins when the borrowed organisation becomes invisible.
The organism disappears
There is another consequence worth noticing.
Once the gene becomes the protagonist, the organism can begin to disappear from view.
Yet genes are not reproduced in isolation.
They participate in organisms.
Organisms develop.
They behave.
They interact with environments.
They form populations.
They modify their environments.
Their developmental histories matter.
The causal organisation of biological inheritance is therefore distributed across multiple levels.
A gene-centred perspective can reveal a genuine pattern at one level without making the other levels unreal.
This is an important principle for what follows.
Finding a useful level of description does not make other levels disappear.
The mistake would be to infer:
because genes matter, only genes matter.
The same mistake could later be made with memes:
because cultural forms persist differentially, therefore memes are the fundamental actors of cultural evolution.
The conceptual history of memetics may therefore repeat, at a different scale, a more general temptation:
to mistake a productive level of analysis for the level of reality itself.
From agents back to relations
Perhaps this is the first place where our investigation begins to turn.
The selfish-gene metaphor directs our attention toward entities.
But the processes it describes may be better understood through relations.
A gene has effects because of the biological system in which it participates.
A phenotype emerges through developmental relations.
A variant persists through reproductive and ecological relations.
A cultural form persists through communicative, institutional, technological and semiotic relations.
A concept becomes powerful through its relations with other concepts and practices.
In each case, what matters is not simply what the entity is, but what it can do within a system of relations.
This is very close to the orientation that has guided much of our recent thinking.
Meaning is relational.
Function is relational.
Possibility is relational.
And perhaps evolutionary persistence is relational too.
The metaphor of selfishness gives us a convenient shorthand for one consequence of those relations.
But the relations themselves remain.
The gene did not want to live
We can now return to our title.
The gene that wanted to live never actually wanted anything.
That is not a criticism.
It is the point.
The gene did not need a desire for survival for evolutionary processes to generate differential persistence.
The appearance of purpose can emerge from processes that contain no intending subject.
The metaphor of selfishness gives that emergent pattern a vivid form.
It makes evolution narratable.
It makes the abstract concrete.
It makes the logic memorable.
And it gave Dawkins a way of forcing us to reconsider where evolutionary explanation might properly begin.
But once we have learned what the metaphor can do, we can also learn to see its limits.
The gene is not the protagonist of evolution in the ordinary sense.
It is a participant in processes whose organisation extends far beyond it.
And the same caution will become even more important when we move from genes to memes.
For if the gene can be made to look like an agent by grammar, then the meme can be made to look like an agent by exactly the same machinery.
The meme will spread.
The meme will compete.
The meme will seek hosts.
The meme will reproduce.
The meme will survive.
And before long, we may find ourselves talking about it as though it were a little creature inhabiting the cultural world.
That is the next step in our investigation.
Before we can ask whether memes really exist, however, we need to examine something more fundamental:
what happens when a process becomes a thing, and a thing becomes an agent?
The answer may be hiding in a single word.
Replicator.