[4B2-GS-1] Fundamental AI, theory (1)

[4B2-GS-1-02] Do Japanese indicative conditionals have the Jeffrey truth table?

〇Ayako Sato¹, Hideki Yoshizawa¹, Tatsuji Takahashi¹ (1. Tokyo Denki University)

Keywords:indicative conditionals, subjective probability theory

Conditional like "If p then, q" is defined material conditional by binary logic, but it's different to Human's interpretation.
About Human's interpretation, in inferential psychology, conditional statements have a stochastic property consistent with the conditional probability P(q|p), and the antecedent is false, the truth value of the conditional statement is "neither true nor false", it was established to take new third values.
The Jeffrey table predicts what third value will take as probability value, and the value (probabilistic truth value) of the conditional statement when the antecedent p is false is always P(q|p).
We researched Jeffrey table by experiment, both when specified probability population and, when unspecified probability population.
In result, when the population was specified, responses were obtained according to a certain number of Jeffrey table patterns or P(p&q), and when the population was not specified, many responses were interpreted as equivalence.