Module 4 – Lesson 2: Controls and Null Models

A severe test cannot exist without controls and null models. These are the tools that show what happens when the claimed mechanism is absent. Without them, results can look impressive while resting on hidden assumptions. This lesson explains why controls and null models are essential for Design Biology and how they protect explanations from storytelling.

A control is a comparison condition. It answers the question: what happens when the key feature is removed or altered? If the same outcome appears in both the test and the control, then the claimed mechanism is not doing the work. Controls indicate whether a result depends on the feature being claimed or arises from background processes, contamination, or the experimental setup.

Null models serve a similar role at a higher level. A null model represents what would be expected if no special mechanism were operating. It is the baseline against which a claim must be measured. If the observed result does not differ meaningfully from the null model, then the claim has not passed a severe test.

Design Biology treats both controls and null models as non-negotiable. They are not optional add-ons. They define whether an experiment is actually testing a claim or merely illustrating a possibility.

There are several kinds of controls commonly used in biological evaluation.

Negative controls remove the claimed cause. If a sequence is said to produce a function, a scrambled or altered sequence should be tested. If a pathway is said to generate an effect, the pathway should be disabled. If the effect remains, the claim is weakened.

Positive controls confirm that the system can produce an effect under known conditions. They show that the test apparatus itself is capable of detecting the phenomenon. Without positive controls, a failure could be due to poor measurement rather than a false claim.

No-template and no-input controls test whether results arise without the proposed source. These are especially important in claims involving information, replication, or assembly. If something appears when the supposed source is missing, then another explanation must be considered.

Null models go beyond individual experiments. They represent expectations based on chance, noise, or simple physical processes. A null model might predict what random sequences should produce, how often weak binding should occur, or how structures should form without guidance. The claim must exceed the null model in a way that cannot be dismissed as a coincidence.

This is where many explanations fail. They show that something happens but do not show that it happens more often, more specifically, or more reliably than the null expectation. A severe test requires demonstrating the difference.

Design Biology also emphasizes that controls must be realistic. A control that is too artificial can hide essential dependencies. For example, using a highly purified system as a control may remove the very noise and interference that the claim must overcome in real conditions. The goal is not to make the system succeed but to reveal what it truly depends on.

Controls and null models also protect against narrative drift. When results are ambiguous, stories rush in to explain them. But a well-designed control limits how far interpretation can go. Either the claimed mechanism shows unique effects, or it does not. This keeps evaluation grounded in observation rather than imagination.

When you write your audits, include a control and null model section. Ask three questions.

What control removes the claimed cause?
What null model represents the baseline expectation?
Does the result exceed both in a meaningful way?

If any of these are missing, the test is not severe.

Controls and null models do not weaken science. They strengthen it. They make claims accountable. They turn possibilities into evaluations. They separate what could be happening from what must be happening.

In the next lesson, we will examine alternative explanations. Once controls and null models are in place, we can ask whether other mechanisms could account for the same observation and whether the evidence truly rules them out.

Lesson Summary

Module 4 – Lesson 2: Controls and Null Models emphasizes the critical role of controls and null models in conducting severe tests in Design Biology. Without these tools, results may appear convincing but rest on hidden assumptions and fail to rigorously test proposed mechanisms.

Key Concepts:

• Controls: These are comparison conditions designed to answer “what happens when the key feature is removed or altered?” If the outcome persists in controls, the claimed mechanism may not be responsible.
• Null Models: These represent baseline expectations assuming no special mechanism. Claims must surpass these baselines significantly to be validated.

Types of Controls Used in Biological Evaluation:

• Negative Controls: Remove or alter the claimed cause (e.g., scrambled sequence, disabled pathway) to test if the effect persists, weakening the claim if it does.
• Positive Controls: Confirm the system can produce an effect under known conditions, ensuring that failure to observe an effect isn’t due to faulty measurement.
• No-Template or No-Input Controls: Test if results arise without the proposed source, especially important when information or replication is claimed.

Importance of Null Models:

• Predict results expected by chance, noise, or simple processes (e.g., random sequences, weak binding frequency).
• Claims must meaningfully exceed null predictions, showing effects occur more often, specifically, or reliably.

Guiding Principles for Controls and Null Models:

• They are mandatory, not optional, defining whether experiments truly test claims or just illustrate possibilities.
• Controls must be realistic, avoiding artificial setups that remove realistic challenges the mechanism must overcome.
• They prevent narrative drift by limiting over-interpretation and grounding conclusions in observable effects.

Practical Audit Questions for Experiments:

• Which control removes the claimed cause?
• What null model represents the baseline expectation?
• Does the observed result exceed both meaningfully?

Summary:

• Controls and null models enhance scientific rigor by making claims accountable and separating what might be happening from what must be happening.
• They enable severe tests by ensuring results depend on the proposed mechanism beyond background effects.
• Next steps involve exploring alternative explanations once controls and null models establish a strong foundation.