The Grammar of Interaction
Interactions, Ordinary and Specific
Imagine a boulder rolling down a mountainside and crashing into another boulder. To explain how two big rocks interact with one another, we turn to the laws of physics. The outcome of a collision between two rocks depends on their initial positions and velocities, their masses, their compositions, and the forces impinging on them. If a chunk of hard rock like granite collides with a chunk of soft rock like sandstone, we expect the sandstone to be damaged more than the granite. Why? Because the internal forces holding the sandstone together are not as strong.
Now imagine a wolf charging down a mountainside to attack a moose. To explain how two mammals interact with one another, we might also try the laws of physics, but the results will be unsatisfactory. These interactions take on specific stylized forms that cannot be predicted from the usual laws and initial conditions. We expect the moose to be damaged more than the wolf, but this is not because the internal forces holding the moose together are not as strong. Furthermore, when a wolf interacts with a moose the outcome is nothing like what happens when a wolf interacts with other wolves, or a moose with other moose.
Wolves and moose are intelligent compared to rocks, but their behavioral repertoire pales in comparison to the many ways in which mammals of the human species can interact with one another. Moose do not get together to build airplanes or microprocessors, and wolves do not file lawsuits against their neighbors, or expect other wolves to bring them food in exchange for bits of paper.
For fundamental questions like how life differs from the ordinary physical world or how human civilization differs from ordinary biology, much of the answer has to do with situational specificity. Living things react to the world with a specificity not predicted by the underlying physical law, and people react to the world, and to each other, with a specificity not predicted by the underlying biology. “Biological specificity is fundamentally the same,” says Carl Woese, “no matter where it is encountered” (emphasis his).1 The laws of physics cannot tell us how a wolf will react to a moose, and our knowledge of biology cannot tell us how to open a bank account.
Biology and civilization depend on this remarkable situational precision, which is guided by complex constraints, sequences of DNA in the case of biology and sequences of text in the case of our literate technological civilization. If you want a generic interaction like a collision between rocks, the laws of physics are fine; if you want special interactions, you need special constraints or, as Richard Dawkins says, “those blind physical forces are going to have to be deployed in a very peculiar way.”2
We have seen that sequences have no direct, meaningful effect on their environments; in essence they are job descriptions without anyone to do the job. Absent Peter, “Please pass the pepper” accomplishes nothing. The pepper will not move on its own, no matter how stridently you demand. Nonetheless, sequences do constrain and coordinate vast amounts of matter and energy on our planet. Symbolic information actually does get control of physical systems. It seems that sequences have not only posted the job description, but filled the vacant position, and trained someone or something to do the job for them.
To constrain the behavior of objects in the world, sequences rely on a workforce of third-party mechanisms. They engage the services of physical agents, of interactors, to function on their behalf. The duties of an interactor, says Kim Sterelny, are “to link the registration of a salient feature of the world to an appropriate response,” in other words to couple perception to behavior* Sequences can describe alternatives, but they need real-world interactors to perform the classification and execute the decision.4
What are these interactors, where do they come from, and how do they fulfill the job descriptions of their governing sequences? Interactor as a biological idea was coined by David Hull, who considers the individual organism, the biological phenotype, to be the quintessential example.5 But he also realizes that his idea can be applied much more broadly.6 '‘Interactor is defined with sufficient generality that it is not necessarily limited to one common-sense level of organization,” Hull writes. “Certainly organisms are paradigm interactors, but entities at other levels of the traditional organizational hierarchy can also function as interactors” (emphasis his).7 In this book we will follow Hull’s lead and treat interactor as a fundamental concept with wide applicability.
Interactors are constraints, but far more specialized and discriminating than tabletops or inclined planes. They are constructed or configured to respond in improbable ways to patterns in their environment and, if they are good at what they do, they increase the likelihood that their corresponding sequences will be replicated to construct or configure future interactors. An interactor can be anything from a single cell to a brain surgeon to a criminal justice system. Each links salient features to appropriate responses.
Setting whole organisms aside, what is the simplest possible example of an interactor? Within the cell, the fundamental unit of interaction is the protein molecule whose construction is constrained by the DNA sequence of its gene. “Proteins generate most of the selectable traits in contemporary organisms,” write biochemist Steven Benner and colleagues, “from structure to motion to catalysis.”8 The catalytic protein known as an enzyme is the most primitive mechanism that couples perception to behavior.
In his book The Selfish Gene, Dawkins promulgates a gene-centric view of natural selection, looking at the evolution in the living world from the perspective of the individual gene rather than the animal.9 I am proposing a complementary enzyme-centric view of interaction, looking at perception and behavior from the perspective of the individual enzyme molecule. “We could call the enzymes the ‘verbs’ of the molecular language,” say Nobel laureate chemist Manfred Eigen and colleague Ruthild Winkler, “because they convey what is to be done.”10
Before we get started on the micro scale, however, remember that our macro-scale dining companion Peter is also an interactor. Interactors are physical entities that perform the jobs described by sequences, that link the registration of a salient feature to an appropriate response. Passing the pepper is such a job. Unlike a gene, however, “please pass the pepper” does not so much construct a pepper-passing interactor as configure a general-purpose interactor who is already present. We will revisit the distinction between construction and configuration in the chapters ahead.