NEXUS ACADEMY · GUIDED LEARNING
How Models Help Us Think
A guided introduction to reasoning, neurons and AI evaluation.
Work through three already-published guides and labs to learn how different fields test a model, interpret results and recognize limits.
Use your browser’s Print command to keep a worksheet with an explained answer key.
Your learning objective
Compare what logical forms, toy neuron thresholds and classifier evaluations reveal, then explain what each simplified model cannot establish.
Modules
Philosophy & Meaning · MODULE 1
1 · Test the structure of a claim
Distinguish a formally valid inference from the factual truth of its premises.
Read the guide, then open the logic lab and compare modus ponens with affirming the consequent.
Check your understanding
Can an argument's structure be valid when one of its premises is false?
Reveal the explained answer
Yes. Formal validity asks whether the conclusion would have to follow if the premises were true. A separate factual inquiry is needed to establish whether the premises are actually true.
Answer key: Yes. Formal validity asks whether the conclusion would have to follow if the premises were true. A separate factual inquiry is needed to establish whether the premises are actually true.
Why the next world connects: A truth table offers a precise formal test. The next world shows how a transparent threshold model can teach a concept without representing every detail of a biological system.
Life, Body & Mind · MODULE 2
2 · Identify what a simple model leaves out
Explain why a toy input threshold is a useful educational model but not a real neuron.
Read the neuron guide and change the invented excitatory and inhibitory inputs around the lab's threshold.
Check your understanding
Does the sandbox calculate actual biological voltage or predict an individual neuron's activity?
Reveal the explained answer
No. The sandbox applies a deliberately invented arithmetic rule to invented units. Actual neural signaling involves voltage, ion channels, timing and many interacting biological properties.
Answer key: No. The sandbox applies a deliberately invented arithmetic rule to invented units. Actual neural signaling involves voltage, ion channels, timing and many interacting biological properties.
Why the next world connects: Both the neuron exercise and a software classifier can use thresholds. That similarity does not mean the biological cell and the computing model work in the same way.
Technology & Futures · MODULE 3
3 · Choose measures that answer the question
Use a confusion matrix to explain the difference between accuracy, precision and recall.
Read the model-evaluation guide, then move the fictional classifier threshold and compare the resulting errors.
Check your understanding
Can a high accuracy percentage alone tell you whether the model detects most positive cases?
Reveal the explained answer
No. Accuracy combines all correct predictions. Recall describes how many actual positives were found; the confusion counts also show missed cases and false alarms. Context and the costs of different errors matter.
Answer key: No. Accuracy combines all correct predictions. Recall describes how many actual positives were found; the confusion counts also show missed cases and false alarms. Context and the costs of different errors matter.
Why the next world connects: The three activities illustrate a broader habit: state the precise question, choose a method with transparent assumptions, then name what the result does not prove.
Bring the worlds together
Choose one example from the three labs. State its question, assumptions, output and limitation, then explain which different discipline could help examine what the model leaves unanswered.
No grading or profile: reflect privately, revisit earlier steps and compare your answers with the source-linked explanations.
Continue: the consciousness trail ↗
Other complete Academy courses
- Evidence Across Worlds: How to Test a Claim — A complete six-module inquiry course using all six published Nexus Humano guides and three original Labs to distinguish observations, interpretations, validity, model assumptions and evaluation evidence.
Research references
These are the references already listed by the linked published guides and labs; review the original resources for context.
- Stanford Encyclopedia of Philosophy — Argument and Argumentation
- Stanford Encyclopedia of Philosophy — Classical Logic
- OpenStax — Biology 2e: How Neurons Communicate
- OpenStax — Anatomy and Physiology: The Action Potential
- Google Machine Learning Crash Course — Overfitting and Generalization
- Google Machine Learning Crash Course — Classification Metrics
- NIST — Artificial Intelligence Risk Management Framework
Previously published source-linked Nexus Humano guides and original deterministic labs; no independent specialist sign-off. This guided sequence was compiled with AI assistance from published Nexus Humano materials; no external academic endorsement or course accreditation is implied.