The Interpreter's Responsibility

A woman on one of my walks watched a green hermit disappear into the understory and said, mildly interested, that it was probably going back to its nest. It wasn't. The female of that species builds and tends the nest alone; the bird she had just watched was a male, and a male hermit spends its day doing something considerably stranger than nesting. I asked her to wait, and pointed at the same drooping clump of Heliconia the bird had just left. Eleven minutes later, almost to the second, it reappeared, worked the same six bracts in the same order it had worked before, and vanished again through the same gap in the foliage. Twenty minutes after that, it did it a third time. She watched the third pass with a different quality of attention than she had given the first. Nothing in the forest had changed between the two. The bird was doing exactly what it had been doing all along. What had changed was her.

She had just watched, without quite knowing it, a demonstration of a foraging strategy ethologists call traplining: a fixed, repeated circuit between widely dispersed flowers, timed not to convenience but to the rate at which each flower's nectar renews. The term was coined in the early 1970s to describe exactly this kind of behavior in tropical bees and hummingbirds, and the hermits are its textbook case, nonterritorial, long-billed foragers that cannot afford to defend a single flower patch because their preferred flowers, unlike the dense, easily guarded blossoms favored by territorial hummingbirds, are sparse and scattered through the forest understory. A hermit that tried to hold territory over such a resource would spend more energy defending it than the flowers could return. Instead it memorizes a route, visits its flowers in a fixed sequence, and returns to each one only after enough time has passed for the nectar to be worth collecting again a foraging economy documented in careful field studies of Costa Rican Heliconia stands, where researchers found that hermit hummingbirds behave as nonterritorial, traplining specialists while other hummingbird species at the same flowers behave as territorial defenders, two coexisting solutions to the same problem of finding sugar in a forest that does not offer it for free.

None of this was invisible to my visitor before I spoke. The bird had always been flying its circuit; the flowers had always been renewing their nectar at whatever pace they renew it. What she lacked was not access to the pattern but a reason to hold three separate sightings together as a single fact rather than three unconnected ones. The retinal information reaching her eyes on the second and third passes was, for all practical purposes, identical to what had reached them on the first. I had not given her anything to see. I had given her a reason to keep watching, and a category (circuit, not visit) into which the watching could accumulate. This raises a question worth taking seriously on its own terms, independent of any particular forest or any particular bird: if two observations of an unchanged scene can differ this much depending on what happened in between them, what exactly does an interpreter do, and what is that person actually responsible for doing well?

The conventional answer treats interpretation as a delivery problem. An interpreter possesses facts; the facts are true or false, complete or partial, well or poorly explained; the interpreter's job is to move accurate facts from their own head into someone else's with minimal loss and maximal clarity. This model is not absurd a wrong bird name is still a wrong bird name, and clear delivery is a real skill but it cannot account for what happened on that trail, because nothing was delivered in the conventional sense at the moment the transformation occurred. The information that mattered arrived in the eleven minutes of watching, not in the sentence that preceded them. What I supplied was closer to an instruction for organizing forthcoming observations than to an observation itself, and an honest account of interpretation has to explain what that kind of supply consists of and what obligations come attached to it.

Philosophy of science offers a useful, if initially unlikely, place to start. In the middle of the twentieth century, a body of work on the nature of scientific observation converged on the claim that seeing is not a passive registration of light on a retina but an act already shaped by the concepts an observer brings to it that two people, or the same person at two different times, can look at an identical physical scene and correctly be said to see different things, because what counts as a meaningful unit of the scene depends on a conceptual scheme that is doing organizational work before conscious attention even begins. This idea, developed most influentially in the philosophy of science under the label theory-ladenness of observation, was originally aimed at scientists reading the same instrument dial or the same X-ray plate and drawing different conclusions depending on the theory each already held. It generalizes further than its authors probably intended. A trained radiologist and a layperson looking at the same film are not having the same experience with different amounts of interpretation added afterward; the trained eye is organizing the image differently from the moment it is registered, because years of exposure to diagnostically relevant patterns have changed what counts, perceptually, as a unit worth noticing.

Cognitive psychology supplies a controlled demonstration of the same principle, in a domain distant enough from natural history that the parallel should be drawn carefully rather than assumed. A classic study of chess players found that masters could glance at a real, mid-game position for a few seconds and reconstruct it from memory with near-perfect accuracy, while novices managed only a handful of pieces but when the same players were shown positions with pieces arranged randomly, arrangements that could never arise from an actual game, the master's advantage nearly disappeared entirely. The finding ruled out the obvious explanation, superior visual memory, and pointed instead toward something more specific: masters had spent thousands of hours encountering meaningful configurations of pieces, and had come to perceive the board not as thirty-two individual objects but as a much smaller number of larger, familiar chunks, each one a unit their memory could store as a single item rather than many. Strip the position of the structure that chess theory actually recognizes, and the chunks have nothing to attach to; the master's advantage collapses because expertise was never a matter of sharper eyes, only of a richer scheme for organizing what the eyes already deliver. Chess and birdwatching are different enough activities that the finding should not be imported wholesale a forest is not a board, and a hummingbird's circuit is not a legal position in a known game but the underlying claim travels reasonably well: what looks like superior perception in an expert is very often superior organization applied to perceptual input that a novice already possesses in full.

Ethology offers a third, older parallel, closer to the forest than either philosophy of science or chess. Researchers studying predators that hunt cryptic, camouflaged prey have long documented a phenomenon in which the predator's success rate against a particular prey type improves the more recently and frequently it has encountered that type, as though the animal were building, through repeated exposure, an internal template a search image that makes an otherwise well-hidden shape easier to detect on subsequent encounters, at some cost to its ability to notice other prey types it has not been recently primed for. A predator with a search image for one species of moth is not seeing better in any general sense; it has narrowed and sharpened what it is prepared to notice, at the price of a corresponding blindness elsewhere. The concept was developed to explain hunting behavior, not human perception, and the analogy should not be pushed further than it supports. But it names, from within biology itself, the same basic trade-off at work in the trail I have been describing: attention that has been shaped to notice one kind of pattern reliably becomes, by the same act of shaping, less available for noticing others. Whatever an interpreter installs, they are also foreclosing something, and a full account of the interpreter's responsibility has to reckon with the foreclosure as much as with the gain.

The field of environmental interpretation arrived at a version of this same distinction on its own, decades before either literature existed in its current form, through nothing more sophisticated than one man's careful attention to what worked and what didn't in front of actual visitors. Writing in the 1950s after years spent watching guided tours and museum talks across the American park system, Freeman Tilden concluded that information, however accurate, is not the same thing as interpretation that interpretation is something closer to revelation based upon information, a distinct act that uses facts as raw material without being reducible to them and that the interpreter's real aim is not instruction but a kind of productive disturbance of the visitor's attention. Tilden had no access to theory-ladenness or perceptual chunking as formal concepts, and it would be a mistake to credit him with anticipating either literature in any technical sense. But he had noticed, from the practitioner's side of the same problem, exactly what the woman on the trail demonstrated: that handing someone a correct fact and changing what they are capable of noticing afterward are not the same accomplishment, and that a profession built around the second one needs its own theory of what it is actually doing, separate from whatever theory covers accurate reporting.

An earlier chapter in this collection argued that ecological meaning is carried disproportionately by relationships rather than by the species that participate in them that a forest read as a set of named organisms and a forest read as a set of interactions among those organisms are not the same forest, epistemically, even when every organism present is identical in both readings. That argument was made at the level of what is true about ecosystems. This one is about what follows for the person whose job is to make a relationship perceivable to someone who cannot yet see it, and the two arguments turn out to be harder to separate than they first appear. A relationship, unlike a species, is very rarely visible in a single instant. Nobody sees traplining in one glance at one hummingbird; the pattern exists only across a sequence of moments, held together by whoever is doing the watching. An army ant swarm and its retinue of specialist birds present the same demand in a more compressed form: a single glance shows a bird perched near moving ants, which is not yet the relationship, only its momentary cross-section. Seeing the actual relationship the size-sorted hierarchy of species along the advancing edge, the dependence of the smallest followers on scraps the larger ones miss, the collapse of the whole arrangement when the ant colony itself cannot be sustained requires holding together observations separated by minutes, days, or in the case of population-level effects, years, in a way no unaided glance provides. If relationships are where ecological meaning concentrates, then the perceptual work of holding separate moments together into a single relational fact is not a minor addition to interpretation it is close to the whole of what interpretation, understood as something other than naming, actually has to accomplish. An interpreter who wants a visitor to perceive relationships rather than merely learn about them is not in the business of expanding a list. They are in the business of installing, in someone else's head, the organizing scheme that turns an otherwise arbitrary sequence of observations into a pattern that continues to register as a pattern once the interpreter has gone home.

It is worth pausing on how close this description sounds to persuasion, and being precise about where the two part ways, because an interpreter who is comfortable with the phrase installing a frame in someone else's head ought to be uncomfortable with how easily that phrase could describe indoctrination instead. Persuasion, in its more troubling forms, aims at a fixed conclusion and treats the audience's continued agreement as the measure of success, regardless of what the audience goes on to independently observe; a persuaded belief that survives contact with disconfirming evidence by reinterpreting the evidence rather than revising itself is functioning exactly as its author intended. Responsible interpretation has to aim at something structurally different: not a conclusion the visitor will continue to hold, but a capacity the visitor can continue to exercise, including against the interpreter's own claims. A frame that cannot be revised by the visitor's own subsequent observation is not doing the job this essay has been describing, however true the specific claim attached to it happens to be at the moment of delivery. This is a demanding standard, harder to meet than simple accuracy, because it requires building the possibility of one's own correction into the thing being taught.

That last clause is where responsibility enters, and it enters with a structure unlike the responsibility attached to an ordinary factual claim. A wrong fact, stated plainly, is available for correction the moment someone checks it against a field guide, a specialist, or a second source; its error is bounded to the specific claim and dies quickly once exposed. A perceptual frame does not work this way, because it is not a claim the visitor holds and can test it is a lens the visitor now looks through, and lenses are not, in the moment of looking, aware of themselves as lenses. Teach someone to notice a hermit's fixed circuit and they will, correctly, start noticing timed returns in other contexts for years afterward, on trails I will never walk with them, applying a category I supplied without any further input or correction from me. This is precisely the mechanism that makes interpretation powerful, and precisely the mechanism that makes an inaccurate or overconfident version of it costly in a way an inaccurate fact is not. A visitor told, incorrectly, that a given bird is twelve inches long will simply be wrong about a number. A visitor taught to see connection, memory, or intention where the evidence does not yet support it will keep finding confirming instances of that frame indefinitely, because ambiguous future evidence is not neutral once a frame is already installed a documented tendency in the psychology of reasoning, in which people systematically favor evidence that fits an existing belief and discount or reinterpret evidence that does not, operating on relational beliefs about nature exactly as readily as on beliefs about anything else. The earlier chapter's caution about the more inflated claims made for underground fungal networks in forests belongs here as a case in point: once popular writing had installed the frame of trees deliberately nurturing their offspring through hidden connections, ordinary, ambiguous observations one seedling thriving near a mature tree, another failing farther away became, for readers already carrying that frame, evidence for a story the primary literature had not actually established. The frame did the work of manufacturing its own confirmation, long after the specific claims behind it had come under serious question.

This suggests that the interpreter's responsibility cannot be discharged simply by making sure the relational claim offered in the moment is true. It has to extend to the durability and the portability of the frame being installed, which means it has to include, as part of the interpretation itself rather than as an afterthought, some signal of how tightly the frame ought to be held. A hermit's traplining route is about as secure a relational claim as field ecology currently offers on this subject it rests on decades of foraging-economics research and a coherent explanation of why the strategy exists and can reasonably be handed to a visitor with confidence. The idea that a forest's fungal network functions as a communication and provisioning system between trees is, at the time of writing, considerably less secure, resting on real underlying biology but stretched by popular treatment well past what controlled studies currently support. Teaching the first without qualification is responsible interpretation. Teaching the second without qualification installs, in a visitor who has no independent way to check it later, a frame calibrated to a confidence the evidence does not warrant and because frames generate their own future confirmation once installed, the visitor is unlikely ever to discover the miscalibration on their own. Calibrating confidence, in other words, is not a hedge tacked onto the content of an interpretation. It is part of the perceptual instrument being handed over, as consequential as the content itself, because it determines whether the instrument can later be corrected or only ever confirmed.

None of this makes the transmission model of interpretation obsolete; facts remain the raw material Tilden insisted they always were, and no amount of skillful reorganizing rescues an interpretation built on facts that are simply wrong. But it does mean that accuracy about the specific claim in front of an interpreter, at the specific moment it is delivered, is a necessary condition for responsible interpretation and not a sufficient one. The sufficient condition includes what happens to that claim once it leaves the interpreter's control and becomes a standing feature of someone else's attention whether it was handed over with the internal machinery for its own future revision intact, or whether it was handed over sealed, primed to manufacture agreement with itself indefinitely regardless of what the visitor goes on to actually observe.

This has a further consequence for the interpreter rather than the visitor, one that is easy to overlook because it points inward instead of outward. If a frame installed years ago keeps generating its own confirmation in someone else's mind long after the interpreter has moved on, the same mechanism should be expected to operate on the interpreter's own frames, including the ones acquired earliest and rehearsed most often on the most trails. An interpreter who has explained a relationship the same way for a decade is, by the logic of this essay, in exactly the position most susceptible to mistaking a well-worn frame for a settled fact, since every ambiguous encounter since the frame was formed has had the same opportunity to be absorbed as confirmation rather than tested as a claim.

The mycorrhizal case discussed in the previous chapter is instructive here too: it was not members of the public who first advanced the strongest versions of that claim, but interpreters, some of them scientifically trained, repeating an account that had begun to function as settled knowledge well before the evidence supporting it had been checked with the same rigor applied to newer claims.

Responsible interpretation, on this account, cannot be a credential earned once and then exercised indefinitely from the same fixed stock of frames. It requires the interpreter to hold their own explanatory habits to the same revisability they are obligated to build into everyone else's, which is a harder and less finished task than acquiring expertise is usually assumed to be.

There is also a structural difference between interpretation and classroom teaching worth making explicit, because the two are often collapsed together and the collapse obscures where the responsibility actually falls heaviest. A classroom relationship is typically sustained: a curriculum can be revisited, a mistaken lesson corrected the following week, a student's understanding checked against a test that gives the teacher some feedback about what actually landed. A guided walk, a museum placard, a single encounter on a trail is very often the entire contact the interpreter will ever have with that particular mind. There is no following week in which to correct an overstated claim, no test result reporting back what frame actually took hold. The single-contact structure of most environmental interpretation does not make the responsibility smaller than a teacher's; if anything it makes it more concentrated, because whatever frame is installed in that one encounter has to be built correctly and calibrated correctly on the first and only attempt, with no institutional mechanism downstream to catch what went wrong.

I don't know what that woman is doing with the frame I gave her. I don't know whether, on some other trail, in some other country, she has already watched something return to the same spot at what first looked like an arbitrary interval and thought, unprompted, of a hermit's circuit or whether she has since watched something that resembled a pattern and was not one, and extended the frame past where it should have stopped, with nobody present to tell her so. What I handed her that morning was not a fact she could check against a book. It was a way of holding her attention that is presumably still operating, correctly or incorrectly, on scenes I will never see and can no longer revise. I am no longer sure interpretation, understood this way, is a responsibility that ends when the walk does and I don't yet know what an interpreter is supposed to do about a responsibility that outlasts their own presence in the story.

References

Tilden, F. (1957). Interpreting Our Heritage. University of North Carolina Press.
Hanson, N. R. (1958). Patterns of Discovery: An Inquiry into the Conceptual Foundations of Science. Cambridge University Press.
Chase, W. G., & Simon, H. A. (1973). Perception in chess. Cognitive Psychology, 4(1), 55–81.
Nickerson, R. S. (1998). Confirmation bias: A ubiquitous phenomenon in many guises. Review of General Psychology, 2(2), 175–220.
Janzen, D. H. (1971). Euglossine bees as long-distance pollinators of tropical plants. Science, 171(3967), 203–205.
Gill, F. B. (1988). Trapline foraging by hermit hummingbirds: competition for an undefended, renewable resource. Ecology, 69(6), 1933–1942.
Stiles, F. G. (1975). Ecology, flowering phenology, and hummingbird pollination of some Costa Rican Heliconia species. Ecology, 56(2), 285–301.
Karst, J., Jones, M. D., & Hoeksema, J. D. (2023). Positive citation bias and overinterpreted results lead to misinformation on common mycorrhizal networks in forests. Nature Ecology & Evolution, 7, 501–511.

Author's note

The scientific content of this essay — theory-ladenness of observation, the chess-expertise findings on perceptual chunking, confirmation bias, and the specific natural history of hermit hummingbird traplining — reflects established research, cited above to its primary sources. The extension of the chess-perception findings to naturalist expertise is offered as an analogy grounded in a shared underlying principle, not as a claim that the two skills have been directly compared in the literature; this essay has tried to flag that extension rather than let it pass as settled fact.

What is not an established finding, and belongs instead to The Neotropic Philosophy of Interpretation as a proposed framework, is the essay's central and load-bearing claim: that interpretation is best understood as an act of reorganizing perception rather than delivering information, and that this reframing generates a distinct form of responsibility — extending to the durability and calibration of the perceptual frames an interpreter installs, beyond the accuracy of any single fact stated in the moment. Tilden's founding distinction between information and interpretation supports this claim without fully entailing it; he was describing what makes interpretation effective, not building an explicit account of what makes an interpreter accountable for its downstream effects. That account is this essay's own contribution, offered for what it is: a philosophical argument built on established science, not a summary of a settled research consensus.

The Interpreter's Responsibility

​

References

Suggested further reading

Author's note