Hume vände sig emot spekulativ metafysik i Leibniz och Spinozas anda. Citatet ovan gör en skillnad mellan vad som senare kom att kallas "analytiska" och "syntetiska" sanningar. Analytiska sanningar är matematiska och logiska sanningar, som enligt en uppfattning är sanna enbart i kraft av de ingående tecknens betydelse. Syntetiska sanningar är (natur)-vetenskapliga sanningar, som man får reda på genom observationer och experiment. Enligt Hume, och många av hans efterföljare (inte minst de s.k. logiska positivisterna), intar spekulativ metafysik en mellanställning: den innehåller inga experiment, men försöker ändå, till skillnad från matematiken, att uttala sig om förhållanden i världen. Enligt Hume och positivisterna är det felaktigt.
En intressant kritik mot positivisterna, som kanske kunde tillämpas mot Hume själv, är följande:
http://www.mtholyoke.edu/~ebarnes/python/python.htm
There's a wonderful irony behind this revolt against metaphysics. It's that the logical positivists, whose basic complaint against metaphysics was that it was all irretrievably confused and fuzzy, themselves had a notion of metaphysics that was, you guessed it, somewhat confused and fuzzy. The fuzziness was manifested in a couple of now-famous technical glitches in the positivist program. Whenever anyone came up with a means of sorting out the good philosophy from the metaphysical, some young upstart logic whiz always pointed out that the proposed means either ruled out some clearly good philosophy or ruled in some ghastly bit of metaphysics.
Even worse, nobody could ever find an acceptable way to defend the main sail on the positivist's ship, the verifiability criterion. The verifiability criterion said that the meaning of a meaningful statement was conveyed completely by the means by which it is verified. The criterion was enormously useful in accomplishing sort of an end-run around metaphysics; since metaphysical statements couldn't be verified, the criterion told you they were meaningless.
The positivists were pretty happy about all this until it sank in that the verifiability criterion itself couldn't be verified. Think about it-to verify the verifiability criterion, you'd have to sort out all the meaningful statements beforehand, and that you couldn't do without first assuming the verifiability criterion! So the verifiability criterion is not verifiable. If it's not verifiable, then, according to itself, it's meaningless. So the verifiability criterion might as well be a bit of metaphysics. In the possible world in which Homer Simpson is not only a real person but a logical positivist living in Vienna in the 1930s, Homer just said "Dhoe!"
Kort sagt; är Humes filosofi syntetisk eller analytisk? Man ser ju inte mycket experiment i hans verk, men ändå uttalar han sig om saker som knappast kan hänföras till "abstrakta resonemang om kvantitet och tal".