II 237
Knowledge/Riddles/Kripke/Nozick: conundrum: why would you seek evidence against something that you know. - You know then that the evidence must be wrong.
Nozick: a theory of knowledge must be able to handle it.
>
Knowledge, >
Theory of knowledge, >
Recognition, >
Certainty.
Solution: conversely, if one does not know that the evidence is misleading, one should not ignore it.
>
Method.
II 250
Evidence/hypothesis/Nozick: often evidence can apply, even if the hypothesis is false.
>
Truth, >
Hypotheses.
Test: search for data that would not apply if the hypothesis was true, but the evidence is not. - Then, the hypothesis has not passed the test.
II 254f
Evidence/hypotheses/Nozick/(s): the initial probability (P0) of the hypothesis must be considered. One cannot just put up any hypothesis. Therefore conclusion from P (evidence e I Hypo h)> = 0.95, P (e,~h) <= 0.05 not sure if e is more likely to follow from h-h or not, depends on which of the two weighted conditional probabilities is greater, P (el h) times P0(h) or P(e l ~ h) times P0(not-h).
>
Bayesianism, >
Conditional probability.
II 261
Evidence/hypothesis/theory/Nozick: if e is evidence for hypothesis h, depends on what other theories we have that connects e and h .
Problem: the other theories could in turn be embedded in a wider context, etc. - regress.
>
Regress, >
Context, >
Dependence.
PutnamVsTradition: therefore "evidence for" is not a formal logical relation. - It is rather dependent on other theories.
Cf. >
Ontological Relativity, >
Internal Realism.
II 262
Induction/evidence/logic/Nozick: the inductive logic is twofold relative
1. probability is relative to the evidence
2. There must be a principle of total evidence, which is applied to the probability statements.
>
Induction.
Some authors: Solution: an evidence is an evidence for what it explains.
>
Explanation, >
Causal explanation.
NozickVs: much evidence is not explanatory - e.g. lightning/thunder do not explain themselves mutually.
E.g. a symptom makes probably more, but they do not explain mutually.
Perhaps there are quite general statistical relations between statements - e.g. principles of the uniformity of nature.
>
Symptoms, >
Uniformity, >
Regularity.
