TO NELSON GOODMAN In my earlier paper1
have indicated some requirements which
believemust be fulfilled in any application of a system of inductive logic to a givenknowledge situation in order to lead to adequate results. In his discussion*Goodman regards these requirements as quite unacceptable; in particularhe regards the simplicity of properties as meaningful only with respect toa sphere of reference.
must confess that
too have a rather uneasy feel-ing concerning the concepts of absolute simplicity and absolute complete-ness referred to in the requirements.
hope very much that it will bepossible to find a way of avoiding these problematic concepts and replacingthem by the kind of relative concepts with which we usually work. But
the present moment
do not see whether or how this can be done.
though those absolute concepts involve problems and difficulties,
do notthink that they are meaningless. The question: "Are all properties ofindividuals in a given universe expressible in a certain language?" is for-mulated in what
at an earlier time: called the material mode of speech.After the appearance of the semantical method it became clear that ques-tions of this kind can be formulated and dealt with in an exact way. Weshould certainly always look out for the dangers involved
the materialmode, also in the present case; but it is not necessary to prohibit this modecompletely.
as the task of deductive logic to supply not only positive butalso negative ansnTers o questions of logical truth and logical implication(e.g.,
It is with respect to these negative resultsthat the requirement of simplicity becomes relevant, as explained in mypaper.
feel, as Goodman does, that questions concerning the intuitive adequacyof any proposed system of inductive logic are of greatest importance, and
shall discuss in my book questions of this kind
detail nith respect toother systems and to my o~m.Of course, this examination may centerupon many different points.
have found that an examination of the sub-sequent two points, which are closely related, seems especially fruitful,because most methods proposed make it easily possible to calculate valuesfor at least one of the two cases and me have often a fairly clear intuitive
"On the Application of Inductive Logic," Vol. VIII, So. 1, pp. 133-148.
"On Infirmities of Confirmation Theory," Vol. VIII, So. 1, pp. 149-151.