NEXUS START HERE
Three authentic entrances into the Nexus, built entirely from chapters and experiences already published—not unfinished content or a promised course that does not exist.
No account required. Read at your own pace, use our original labs, and save published pages privately in your own browser.
PATH 01 · GUIDED ENTRY
See how philosophy, neuroscience and AI approach a shared question without confusing their methods.
PATH 02 · GUIDED ENTRY
Begin with formal reasoning, inspect its limitations, and compare how a model is evaluated.
PATH 03 · GUIDED ENTRY
Compare a simple neuron-threshold teaching model with broader research on cognition and predictive systems.
NEXUS ACADEMY · TWO COMPLETE SOURCE-LINKED COURSES
Choose between the three-module introduction and a six-module investigation course. Both work without an account or grades and use only our current published guides and original teaching Labs.
M31 · SIX ORIGINAL GUIDED INVESTIGATIONS
Six complete four-stage missions turn the published guide collection into real learning activities. Each has a task, something concrete to produce, a related discipline, a working published Lab and an explained model response. You can keep your notes on paper; this site does not collect answers, award certificates or record scores.
These are original learning proposals based on our existing published materials. Fictional cases and model answers are educational, not independent specialist research, personal assessment or verified field observations.
MISSION 01 · Philosophy & Meaning · 25–35 minutes
A persuasive conclusion can fail even when its premise sounds plausible. Find the missing logical step rather than relying on confidence.
Read: Deductive Arguments: Validity and Soundness · Try the adjacent teaching model: Deductive Validity Lab
Invent an everyday conditional statement, such as ‘if an object was stored in this cabinet, it has a green label’. Separate the conditional premise, the observed label and the conclusion in three clearly written lines.
Keep: A numbered list of premises and conclusion, without adding any facts you have not been given.
Imagine that a second cabinet also uses green labels. Could the observed label be true even when the cabinet-origin conclusion is false? State exactly which assignment challenges the inference.
Keep: One possible world in which the written premises hold but the conclusion does not.
Open the published Deductive Arguments guide and the truth-table Lab. Compare affirming the consequent with modus ponens, then record why the two forms produce different logical results.
Keep: A two-sentence explanation of what validity checks, and what it cannot establish about the premises.
Read the related Model Evaluation guide. Explain why a classifier false positive and a formal counterexample are not interchangeable, even if both show that a tempting conclusion can fail.
Keep: A comparison naming the object of each test, the kind of evidence, and the limit of the result.
One green label cannot prove that a sample came from a particular cabinet when other cabinets use the same label. A validity counterexample assigns true premises and a false conclusion; a classifier false positive instead occurs in a dataset under a decision rule. Neither procedure independently verifies an unexamined claim about actual cabinet provenance.
Example explanation: One green label cannot prove that a sample came from a particular cabinet when other cabinets use the same label. A validity counterexample assigns true premises and a false conclusion; a classifier false positive instead occurs in a dataset under a decision rule. Neither procedure independently verifies an unexamined claim about actual cabinet provenance.
Boundary: The cabinet is fictional; this reasoning exercise cannot establish a historical object's actual provenance or prove an empirical statistical result.
Compare perspectives: Evaluating AI Models Beyond Accuracy
Both use explicit rules to test whether conclusions follow from information, though formal validity and predictive performance are different concepts.
MISSION 02 · Philosophy & Meaning · 30–40 minutes
A signal recorded at the same time as a report can matter scientifically without being a complete answer to what experience means.
Read: Consciousness · Try the adjacent teaching model: Deductive Validity Lab
Imagine a volunteer describing a blue shape while a researcher records an electrical signal. Write one question about an observable association and a different question about what it means to have a subjective experience.
Keep: Two questions with deliberately different evidence standards.
Record what the invented situation actually contains: a verbal report, a recording and their timing. Mark every further statement about causes or first-person experience as an interpretation rather than a direct observation.
Keep: An observation column and a separate inference column.
Read the Consciousness guide alongside Brain and Cognition. Identify one claim that scientific methods might help investigate and one philosophical issue that a neural correlation alone would leave unresolved.
Keep: A pair of explanatory sentences identifying different disciplinary purposes.
Use the adjacent logical-argument Lab to ask whether a proposed conclusion follows from the described premises. Could another explanation remain compatible with the same observation?
Keep: An alternative explanation and a clear statement about the limits of inference.
An association between a report and measured neural activity can contribute evidence about cognition, provided the actual methods and controls support it. It does not automatically settle the definition or nature of subjective experience. A further causal interpretation would require more evidence than coincident timing.
Example explanation: An association between a report and measured neural activity can contribute evidence about cognition, provided the actual methods and controls support it. It does not automatically settle the definition or nature of subjective experience. A further causal interpretation would require more evidence than coincident timing.
Boundary: No real volunteer or neural recording is involved. This is not a neuroscientific finding, clinical assessment or measurement of consciousness.
Compare perspectives: Brain and Cognition
Scientific study of cognition offers a different perspective on questions of subjective experience.
MISSION 03 · Life, Body & Mind · 25–35 minutes
Recognising a familiar explanation is different from using it independently. Put that distinction to work in an original study task.
Read: Brain and Cognition · Try the adjacent teaching model: Neuron Threshold Sandbox
Choose a harmless concept from the existing Brain and Cognition guide, then write an objective that demands an explanation of an unfamiliar example rather than repetition of a memorised sentence.
Keep: One observable task showing what successful application would look like.
Design a recognition question using a phrase that appeared in the reading, and a separate open-response question involving a new scenario. Make the intended differences in difficulty and prior context explicit.
Keep: Two small study questions, with a note about what each can and cannot show.
Compare the broader cognition chapter with the Neuron Signals Lab. Notice that the Lab's invented signal threshold addresses a much narrower mechanism than a person's ability to explain a concept.
Keep: A limitation statement that does not mistake invented neuron units for a measure of learning.
Describe what evidence a response would provide, what a delayed question could add, and what you could never infer about a person's intelligence or health from one task.
Keep: A small feedback rubric that avoids diagnosis or personal ranking.
A delayed explanation of a novel example may give more direct evidence of applying a concept than recognising a repeated phrase. Either result depends on the task and circumstances; a simplified neural threshold is not an individual's learning score.
Example explanation: A delayed explanation of a novel example may give more direct evidence of applying a concept than recognising a repeated phrase. Either result depends on the task and circumstances; a simplified neural threshold is not an individual's learning score.
Boundary: This task is an educational design exercise, not a validated psychological test, diagnosis or universal study prescription.
Compare perspectives: How Neurons Send Signals
Neural signalling is one of the biological mechanisms studied when investigating cognition and brain activity.
MISSION 04 · Life, Body & Mind · 20–30 minutes
A classroom equation can be correct without describing a real living system. Trace the border between demonstration and evidence.
Read: How Neurons Send Signals · Try the adjacent teaching model: Neuron Threshold Sandbox
Read the toy neuron-threshold Lab and write its two input variables, the subtraction rule and the chosen firing boundary in arbitrary units. Label the units as illustrative rather than physiological.
Keep: A transparent definition of the invented computation.
Try excitation six and inhibition two, then excitation five and inhibition two. Before pressing the controls, predict which side of the four-unit boundary each invented result will lie.
Keep: Two predictions and the corresponding model outputs.
Use the Neuron Signals guide to identify physical processes, timing or cell interactions that the subtraction rule deliberately leaves out.
Keep: A list of at least three scientific questions the toy rule has not answered.
Read the related Brain and Cognition guide. Decide whether a toy cell-level illustration is enough to infer what an entire person understands, remembers or experiences.
Keep: A cautious conclusion about what an illustrative model can show.
For the chosen invented boundary of four units, six minus two reaches the threshold while five minus two does not. This is a property of the classroom rule. Biological neurons depend on real physical mechanisms and dynamic context; the arithmetic alone predicts no person's brain activity.
Example explanation: For the chosen invented boundary of four units, six minus two reaches the threshold while five minus two does not. This is a property of the classroom rule. Biological neurons depend on real physical mechanisms and dynamic context; the arithmetic alone predicts no person's brain activity.
Boundary: This educational sandbox uses invented quantities and does not measure membrane voltage, provide medical guidance or simulate a particular neuron.
Compare perspectives: Brain and Cognition
Neural signalling is one of the biological mechanisms studied when investigating cognition and brain activity.
MISSION 05 · Technology & Futures · 30–40 minutes
A single high score can hide the mistakes that matter for a particular question. Learn to ask what the summary leaves out.
Read: Evaluating AI Models Beyond Accuracy · Try the adjacent teaching model: Classifier Trade-offs Lab
Imagine a fictional classifier deciding whether invented library items belong to a marked collection. Specify what counts as an actual positive and why a missed item and a wrongly flagged item would matter differently.
Keep: A definition of the task and two possible error costs, without making real deployment claims.
Open the existing classifier-tradeoffs Lab. Change the decision threshold and record TP, FP, TN and FN for two settings of the fictional examples.
Keep: A four-cell confusion matrix for each chosen threshold.
Compare accuracy and recall for the two settings. State which error is better represented by each metric and whether positive prediction precision is defined when no positive cases are predicted.
Keep: A short table of metrics and a note about the undefined case.
Read Model Evaluation and the connected AI and Consciousness guide. Explain why benchmark success can inform a performance claim without establishing an artificial system's experience or safety in every context.
Keep: A conclusion that matches the evidence's scope and names an unresolved question.
Changing a threshold can alter false negatives and false positives; accuracy aggregates correct predictions while recall focuses on detecting actual positives. With no predicted positives, precision is undefined rather than a factual zero. Performance scores cannot independently establish consciousness or universal safety.
Example explanation: Changing a threshold can alter false negatives and false positives; accuracy aggregates correct predictions while recall focuses on detecting actual positives. With no predicted positives, precision is undefined rather than a factual zero. Performance scores cannot independently establish consciousness or universal safety.
Boundary: All observations and consequences in the Lab are fictional. No evaluation here certifies a real AI system, protected population or deployment.
Compare perspectives: AI and Consciousness
Measured prediction accuracy does not on its own establish whether a system has subjective experience.
MISSION 06 · Technology & Futures · 25–35 minutes
A compelling answer invites interpretation; a responsible investigation distinguishes what was observed from what we infer.
Read: AI and Consciousness · Try the adjacent teaching model: Classifier Trade-offs Lab
Imagine a fictional program responding fluently to a question about fear. Record only the visible input, the generated text and its apparent relevance, avoiding any assumption about the program's inner life.
Keep: A three-line observation record.
Formulate two different hypotheses about what the fluent output might mean. Identify what evidence would be needed to distinguish the hypotheses rather than simply choosing the one that sounds more persuasive.
Keep: Two interpretations and a missing-evidence note for each.
Read the AI and Consciousness guide with the philosophical Consciousness chapter. Compare the concepts at issue and make clear which claims are conceptual and which concern observable outputs.
Keep: A comparison of definitions and available evidence.
Examine the fictional classifier-tradeoffs Lab as an adjacent example of measurable task evaluation. Explain why success on its chosen labels still does not directly answer the subjective-experience question.
Keep: A concise evidence boundary statement and one thoughtful unanswered question.
The fictional machine produced fluent, apparently relevant text. This is an observation about output, not direct evidence of a particular subjective state. Evaluating task outcomes and debating possible experience require different definitions and arguments; neither should silently substitute for the other.
Example explanation: The fictional machine produced fluent, apparently relevant text. This is an observation about output, not direct evidence of a particular subjective state. Evaluating task outcomes and debating possible experience require different definitions and arguments; neither should silently substitute for the other.
Boundary: The program is invented; the exercise makes no claim about any real AI provider's systems or their internal experience.
Compare perspectives: Consciousness
Debates about artificial consciousness draw on philosophical definitions and theories of mind.