Twin Earth, brains in vats and Goodman’s grue: three puzzles about knowing

Twin Earth separates internal similarity from reference

The Twin Earth thought experiment stipulates a counterpart whose internal life resembles yours, while the superficially familiar liquid around them has a different underlying composition. The question concerns what the word water refers to in each environment. Externalist arguments suggest that facts beyond an individual’s internal state can partly determine the content of a thought or expression.

Use the Twin-Earth interactive scenario to compare an internalist judgment with a judgment that gives the environment a role. Stanford’s article on externalism explains the philosophical setting. The scenario is not an empirical demonstration that chemistry alone fixes every word’s meaning. It isolates a contested claim about reference under a stipulation that would be difficult to reproduce as an ordinary experiment.

The brain-in-a-vat scenario changes the skeptical question

A brain-in-a-vat story imagines experiences produced by a radically different environment from the one the subject takes themselves to inhabit. The initial skeptical challenge asks what could distinguish those experiences from ordinary perception. Semantic arguments add another question: if a subject’s words develop entirely within the simulated environment, what do those words refer to?

The brain-in-a-vat activity records how you assess a scenario and its evidence; it does not reveal a hidden metaphysical answer. Stanford’s treatment of content externalism and skepticism helps distinguish the skeptical challenge from proposed semantic replies. A philosophical argument about what a sentence could refer to is not a physical test that rules out every possible simulation story.

Worked example: compare two prediction rules

Goodman’s puzzle supplies a different difficulty. Use the activity’s stipulated observation cutoff and definitions; these are invented classifications chosen to expose a problem about induction.

1. Observe the sample of green objects shown before the cutoff. Write the ordinary rule: future objects of this kind will be green. Note the observed cases supporting it.
2. Define grue as green before the cutoff and blue afterward. The same observed pre-cutoff cases fit a rule predicting grue objects, yet that rule predicts blue after the cutoff. Record where the predictions first diverge.
3. Ask why you regard green as projectible and grue as contrived. Consider established use, explanatory relationships and background knowledge. Simply counting past matches cannot distinguish the two rules on the supplied observations.

Goodman’s riddle asks which predicates support projection

The new-riddle-of-induction activity is not claiming that any prediction is as good as any other. It asks why some ways of describing a pattern seem suitable for extending observations into unobserved cases. An equivalent fit to old examples does not by itself make two descriptions equally useful within a broader scientific practice.

This is a useful prompt when evaluating a rule constructed after inspecting data. Ask what was fixed before the observations, how many alternatives were considered and what independent test could distinguish the candidate explanations. That practical concern connects with scientific overfitting, but it does not dissolve Goodman’s philosophical problem into one statistical metric. The standards for selecting concepts and the performance of a fitted predictor are related questions at different levels.

Read a philosophical puzzle without losing the question

A short written record helps prevent the discussion from moving between meaning, evidence and reality without noticing the transition. Use these checks before defending a conclusion.

• Restate the stipulations in ordinary language. Separate facts assumed for the scenario from claims the argument is trying to establish.
• Identify the target: whether a belief is true, whether it is justified, what a term refers to or what future cases a rule predicts.
• Vary one assumption and state whether your answer changes. A different environmental history can matter even when the present experience is described as similar.
• Give the strongest reason for an alternative judgment. A recorded disagreement can make the underlying theory clearer without requiring a correctness score.

Carry the method into ethical puzzles carefully

The same habit helps with the nonidentity problem, where a policy can affect which people exist, and the utility-monster allocation game, where stipulated utility numbers put pressure on an aggregative rule. Those puzzles concern ethical reasons rather than the reference of water or the projection of colour predicates.

Do not collapse them into a contest for the most surprising answer. State the principle being tested, explain why a case challenges it and identify what a revision would preserve or sacrifice. Modic keeps these activities open-ended because their educational value comes from a reasoned account of your commitments, including what would make you change them.