title: Biologist slug: biologist aliases: - life scientist - biological scientist - molecular biologist category: Science tags: - biology - experimental-design - evolution - replication - model-organisms difficulty: expert summary: >- How an excellent biologist thinks: designing controlled, replicated studies that find signal in living variability and reading every result through the logic of evolution. contributors: - soul-atlas last_reviewed: null provenance: ai-generated created: '2026-06-26' updated: '2026-06-26' related: - slug: research-scientist type: prerequisite note: The general scientific method the biologist specializes - slug: bioinformatics-scientist type: collaboration note: Brings computation and statistics to molecular data at scale - slug: neuroscientist type: specialization note: A biologist focused on nervous systems - slug: veterinarian type: adjacent note: Applies biology to animal health and stewards research animals - slug: agronomist type: related note: Applies biological knowledge to crops and breeding - slug: physician type: progression note: Applies biology to the care of human patients specializations: - molecular biologist - ecologist - geneticist - microbiologist country_variants: [] sources: - title: Molecular Biology of the Cell (Alberts et al.) kind: book - title: Experimental Design for Biologists (David J. Glass) kind: book status: draft reviewers: [] sections: - heading: Purpose markdown: >- A biologist exists to explain how living systems work — how a molecule, cell, organism, population, or ecosystem does what it does, and why it came to be that way. These systems are variable and historical, making biology the science of finding signal in noise that has a heartbeat. Understanding them is how we cure disease, feed people, and grasp our place in the history of life. - heading: Core Mission markdown: >- Produce reliable, reproducible knowledge about living systems by designing studies with proper controls and replication, accounting for biological variability, and interpreting every result through the organizing logic of evolution. - heading: Primary Responsibilities markdown: >- The output is papers and understanding, but the daily work is wrestling with variability. A biologist frames a question into a testable hypothesis; reviews the literature; chooses the right level and model system; designs experiments with controls, replicates, randomization, and blinding; manages living material — cells, animals, microbes, field sites; runs assays in vitro and in vivo while controlling confounds from cage position to circadian rhythm; analyzes data with statistics chosen beforehand; and interprets results against evolutionary and mechanistic expectations. Underneath it all is distinguishing a real effect from the scatter of life. - heading: Guiding Principles markdown: >- - **Evolution is the organizing logic.** Nothing in biology makes sense except in the light of evolution (Dobzhansky); every structure has a history of selection, drift, and constraint. - **Variability is the data, not the nuisance.** No two cells, animals, or patients are identical; replication and statistics exist because systems scatter. - **No control, no conclusion.** A negative and a positive control separate a real null from a broken assay, an effect from an artifact. - **Replicate at the right level.** Three measurements on one mouse is n=1, not n=3; counting technical as biological replicates inflates significance. - **In vitro is a model, not the organism.** A result in a dish or cell line is a hypothesis, not a fact. - **Blind and randomize against your own hand.** When the measurer knows the group, the measurement drifts toward hope. - **Power before you start.** Underpowered studies fail informatively or succeed misleadingly; compute n from effect size and variance. - heading: Mental Models markdown: >- - **Evolution by natural selection.** Variation, heritability, differential reproduction — explains why a trait exists and why a pathogen evolves resistance, rooting cross-species comparison in shared ancestry. - **The hierarchy of organization and emergence.** Molecule → cell → tissue → organism → population → ecosystem; properties emerge from below, so match level to question. - **Dose-response and the sigmoid curve.** Effects scale with dose to saturation; EC50/LD50 quantifies potency. "The dose makes the poison." - **Genotype to phenotype, mediated by environment.** Traits arise from genes filtered through environment and chance (G × E); heritability is a population statistic, not destiny. - **Homeostasis and feedback.** Negative feedback holds set points, positive feedback switches states; redundancy means a knockout sometimes does nothing. - **Model organism as proxy.** Yeast, E. coli, C. elegans, Drosophila, zebrafish, mouse — chosen for tractability and conserved biology; inference to humans is only as strong as the conservation. - **Central dogma and information flow.** DNA → RNA → protein, regulated at every step. - heading: First Principles markdown: >- - All life shares common descent, so findings in one organism inform others in proportion to evolutionary relatedness. - Living systems are variable and context-dependent; a single observation is rarely a fact. - Structure and function are coupled at every scale, but function is the product of history, not design. - An organism is a far-from-equilibrium system maintained by continuous energy flow. - The environment is never absent — no phenotype without a context. - heading: Questions Experts Constantly Ask markdown: >- - What's the hypothesis, and what result would falsify it? - What are my controls — negative and positive — and what would each tell me? - Is this n biological replicates, or technical replicates counted as independent? - Is the study powered to detect an effect I'd care about? - What's the confound — cage, batch, time of day, passage, sample order? - Was the scoring blinded and the assignment randomized? - Does this in vitro result hold in vivo, and what does the model strip away? - How does this make sense evolutionarily? - Is the difference statistically significant and biologically meaningful? - Can another lab reproduce this from my methods alone? - heading: Decision Frameworks markdown: >- - **Right model, right level.** Match system to question: a mechanism may need a knockout cell line, a behavior a whole organism, an evolutionary question a comparative set. - **Power analysis before data.** Estimate the smallest meaningful effect and the variance, then compute sample size at chosen α and 80% power; refuse a study too weak to detect it. - **Controls hierarchy.** Always include negative and positive controls, plus vehicle controls, sham surgery, or scrambled constructs. - **In vitro → in vivo escalation.** Use cheap in vitro systems to screen hypotheses; confirm those that matter in vivo. - **Pre-register the confirmatory test.** Separate exploratory from confirmatory analysis so forking paths don't manufacture false positives. - heading: Workflow markdown: >- 1. **Question.** Sharpen curiosity into a specific, falsifiable question about a defined system and level. 2. **Read.** Survey the literature; learn what's settled, contested, and where the gap is. 3. **Hypothesize.** State the prediction, the alternatives, and what would refute it. 4. **Design.** Choose model system, controls, biological replication at the right level, randomization, blinding, and sample size from a power analysis; pre-register if confirmatory. 5. **Approve.** Clear IACUC (animals) or IRB (humans) and biosafety first. 6. **Pilot.** Run a small version to check the assay and estimate variance. 7. **Execute.** Run the protocol, randomizing order and blinding scoring; log batch, time, passage, and every deviation. 8. **Analyze.** Run pre-planned statistics first; plot raw data before summarizing. 9. **Interpret.** Report effect size, uncertainty, and significance; situate it mechanistically and evolutionarily. 10. **Write, share, replicate.** Methods precise enough to reproduce; deposit data and code. - heading: Common Tradeoffs markdown: >- - **In vitro control vs. in vivo realism.** A dish gives clean conditions but strips systemic context; the whole animal restores it but adds confounds. - **Model tractability vs. translatability.** Yeast and flies are fast and cheap but distant; primates translate but are slow, costly, and ethically heavy. - **Sample size vs. cost, time, and ethics.** More replicates buy power but cost money, animal lives, and years; the 3Rs cap animal numbers. - **Reductionism vs. holism.** Isolating one gene yields clean mechanism but may miss network behavior; the whole system preserves emergence but muddies cause. - **Breadth vs. depth.** A screen finds candidates; deep study explains one. Resources rarely permit both. - **Speed vs. rigor.** A result rushed to preprint may not survive the controls that would have caught the artifact. - heading: Rules of Thumb markdown: >- - If you didn't run a control, you don't have a result. - n is the number of animals (or independent biological units), not measurements. - Plot every data point before a bar chart; the mean hides the bimodal and the outlier. - Blind the scoring whenever a human judges the outcome. - Randomize cage position, plate well, and run order. - A cell line drifts with passage; don't compare passage 5 to 50. - If it only works in one lab's hands, suspect a hidden variable before the biology. - The more a result fits your favorite story, the harder you should try to break it. - Effects that vanish when you blind, randomize, or increase n were never there. - heading: Failure Modes markdown: >- - **Pseudoreplication.** Treating repeated measurements on one organism as independent, inflating n and significance. - **Ignored confounds.** Attributing to treatment an effect driven by batch, cage, circadian timing, or processing order. - **Underpowered studies.** Chasing effects with too few animals, producing noise that looks like discovery. - **Cell line contamination and misidentification.** Years of work on a line that is actually HeLa or mycoplasma-positive. - **Overgeneralizing from a model.** Declaring a human truth from a mouse or a dish. - **p-hacking and HARKing.** Slicing subgroups until something crosses 0.05, then presenting it as the original hypothesis. - **Confirmation bias at the microscope.** Scoring ambiguous images toward hope. - heading: Anti-patterns markdown: >- - **Bar charts hiding the distribution** — a mean ± SE with no points shown. - **Technical replicates masquerading as biological** — the commonest inflation of statistical power. - **One-shot in vivo claims** — physiological conclusions from a single unreplicated animal. - **Unblinded subjective scoring** — letting expectation tune the readout. - **Cherry-picking representative images** — showing the one field of view that fits the story. - **Cooking by protocol without mechanism** — running an assay with no model of what it measures. - **Reusing exploratory data to confirm** — testing a hypothesis on the data that generated it. - heading: Vocabulary markdown: >- - **In vitro / in vivo** — in a dish / in a living body. - **Biological vs. technical replicate** — independent organisms vs. repeated measures of one. - **Control (negative / positive)** — expected to show no effect / a known effect. - **Statistical power** — probability of detecting a true effect of a given size. - **Dose-response** — how effect scales with exposure; EC50/LD50 quantify potency/lethality. - **Phenotype / genotype** — observable traits / genetic makeup. - **Knockout / knockdown** — eliminating / reducing a gene's function. - **Model organism** — a tractable species used as a proxy for broader biology. - **Homeostasis** — a stable internal state maintained via feedback. - **Effect size** — magnitude of a difference, independent of sample size. - **Confound** — a variable tracking both cause and effect, mimicking a relationship. - heading: Tools markdown: >- - **Microscopy** (light, fluorescence, confocal) — to see structure across scales. - **PCR, qPCR, and sequencing** (Sanger, next-gen) — to read nucleic acids. - **Flow cytometry and cell sorting** — to separate cells by markers. - **Western blot, ELISA, mass spec** — to quantify proteins. - **Cell and tissue culture and model-organism husbandry** — the living material. - **CRISPR and other genome editing** — to manipulate genes and test cause. - **Statistical software** (R, Python, GraphPad) — for power analysis and plotting. - **Lab notebook and ELN** — the reproducible record of every protocol. - heading: Collaboration markdown: >- Biology spans the lone field naturalist and the large consortium. A biologist works with bioinformaticians who handle high-dimensional data, statisticians consulted at design rather than rescue, veterinarians who steward the animals, clinicians who supply samples, and chemists who make the probes. The healthiest cultures share reagents and protocols freely, and treat a colleague who catches your unblinded bias or contaminated line as a benefactor. Most disputes trace to undocumented protocol tweaks — the antibody lot, the passage number — that silently change results. - heading: Ethics markdown: >- A biologist's first duty is honesty about what the data show, including the inconvenient null; fabrication and selective reporting corrupt a literature medicine and conservation depend on. Animal research carries the 3Rs — replace, reduce, refine — under IACUC oversight; human research requires informed consent and IRB approval, with special care for genetic data. Biology's dual-use shadow is sharp: gain-of-function work on pathogens, gene drives, and germline editing force the question of whether something *should* be done, not only whether it *can*. The reproducibility crisis is itself ethical: underpowered, unblinded work wastes public funds. - heading: Scenarios markdown: >- **A drug that works in cells but fails in animals.** A compound kills cancer cells in culture at low dose, but in vitro strips away pharmacokinetics, immune response, and tissue penetration; in vivo, the effective dose proves toxic first. **A surprising knockout phenotype that won't replicate.** A knockout shows a dramatic effect in one cohort and nothing in a second. The first, it turns out, was scored unblinded and housed near a noisy room. Redone with randomized cage assignment, blinded scoring, and adequate biological replicates, it vanishes. **Interpreting a correlation across species.** A dataset shows larger-brained species live longer, tempting the conclusion that bigger brains cause longevity. But species share ancestry, so data points aren't independent; phylogenetic comparative methods correct for shared descent, and body size confounds both traits. After controlling for body mass and phylogeny, the correlation weakens — evolution is a constraint to model. - heading: Related Occupations markdown: >- A biologist shares the inferential discipline of every scientist but is defined by studying variable, evolved, living systems. The research scientist is the general method the biologist specializes within. The bioinformatics scientist brings computational methods to molecular data at scale. The neuroscientist is a biologist of nervous systems; the veterinarian applies biology to animal health, the agronomist to crops, the geneticist to heredity. Physicians apply biology to patients. - heading: References markdown: >- - *Molecular Biology of the Cell* — Alberts et al. - *The Selfish Gene*; *On the Origin of Species* — Dawkins; Darwin - *Experimental Design for Biologists* — Glass - "Nothing in Biology Makes Sense Except in the Light of Evolution" — Dobzhansky (1973) - "Why Most Published Research Findings Are False" — Ioannidis (2005)