Science Practice 2: Representing Systems with Models

Words can hide logical gaps because sentences let you skip steps without noticing. A sentence like "the hormone triggers the response" sounds complete, but it hides every intermediate step: receptor binding, second messengers, amplification, the actual cellular change. Draw the same process as boxes connected by labeled arrows, and any missing step becomes an obvious gap in the chain. This is why diagramming a signal transduction pathway is more than illustration — it forces you to name every player and every causal link, in order, with nothing implied.

To build a pathway diagram, put each molecule or event in its own box: ligand, receptor, relay proteins, second messenger, target protein, cellular response. Connect them with arrows that mean "causes" or "activates," and label each arrow with what actually happens there, like phosphorylation or a conformational change. A feedback loop needs one more move: an arrow running backward from a downstream product to an earlier box in the chain. That returning arrow is what makes the loop visible on the page — without it, the diagram just shows a one-way street, even if the real system cycles back on itself. If the arrow points to boost the original signal, it's positive feedback; if it points to shut it down, it's negative feedback.

This connects directly to how models get judged. Every model leaves things out — that's not a flaw, it's the point, since a model's job is to highlight the relationships that matter for a specific question, not to recreate the whole cell. But leaving things out has consequences: adding a single new arrow, like the one that closes a pathway into a loop, can change the model's predicted behavior entirely, introducing self-amplification or shutoff that a straight-line pathway could never predict. Reading a model means asking what it includes, what it omits, and what the arrows claim to do.