More recently objective Bayesianism has taken on the traditional problem of the justification of universal inference.
Further, what do we relate in inference: universals with universals, universals with a particular, or a particular with a particular?
The three of these come together (though the latter two are predominant) within the epistemological traditions in an almost universal regard of inference as a knowledge source.
Methodological incommensurability (§4.1 above) denies that there are universal methods for making inferences from the data.
Since universals do not exist, inferences which are based on such notions are intrinsically flawed.
By acting on the basis of the inference that there is "fire" on the hill when one perceives smoke on the hill, although the object of this inference is the universal "fire," not a real fire, when one acts on the basis of this inference one obtains a particular that fulfills the expected function of burning, as the previously experienced particulars that one excluded from non-fire did.
This paper defines a formal inference rule of universal attachment and discusses the properties of soundness, completeness and correctness for this rule.
The object of the former is the pure particular (svalakshana), and the object of the latter (he regarded judgments as containing elements of inference) is the universal (samanyalakshana).
The validity of inferential knowledge was challenged on the ground that all inference requires a universal major premise ("All that possesses smoke possesses fire") whereas there is no means of arriving at a certainty about such a proposition.
This insensitivity suggests that the fitness ranking depends primarily on a more universal quantity on which the inference algorithm builds.
If this turns out to be difficult, the inference of a universal gene regulatory network for a broad range of cell types from just, say, expression data without detailed knowledge of actual TF binding in each cell type might be impossible.
