Henri Bergson is an early 20th century French philosopher that PEL listeners may recall from our philosophy of humor episode, and we'll be tackling his philosophy proper via the entrance drug "An Introduction to Metaphysics," a short essay from 1903 (freely available online) that is essentially pheonomenology without the jargon (Husserl, Heidegger, and Sartre are all in his debt, as is on other grounds Whitehead's "process philosophy"). Deleuze explicitly identified him as a key influence.
Bergson argues that science proceeds by analysis: you apply pre-existing concepts to something and thereby compare it to other things for some practical purpose. But in doing this, you've abstracted away from the original phenomena, and you shouldn't think that by then playing around with such concepts, you can in any way reconstruct it. So for instance, empiricist philosophers like Hume analyzed the self as just a series of appearances, of states, like right now I'm having this experience of typing, and in the next moment I have a different experience, etc. So what is the self? Why, just a collection of those states. Rationalists (like Descartes, and even Kant has a version of this) grant that this may be all we experience, but there must be something that unifies these experiences, some spirit that has them, and that's the self.
Bergson thinks both sides are wrong. Our primary experience of the self is of a continuity, gradually and constantly changing, and not of a series of pointilistic states. Analysis can usefully isolate one or more states, but that's an abstraction, and no matter how many states you abstract from the primary experience, you can never reconstruct that experience. If you tried to teach someone what the self was by describing such states, you could convey some kind of approximation, but never the real thing.
Bergson takes this account to be generalizable, and distinguishes science, which does properly perform analysis using concepts in the way described, from metaphysics, which is not supposed to import pre-existing concepts and abstract away from the real phenomena as encountered in experience. So, for instance, ancient Parmedian philosopher Zeno argued that motion is impossible, using the example of an arrow: At any one durationless moment in time, the arrow is at a location; it's not moving. You can't then add together all these motionless moments and get movement. For Bergson, this obviously misses the point that there's no such real thing as a durationless moment: what is real is the whole duration of the movement, and it's just our concepts that chop it up.
What may be surprising here is taking what is experientially primary as "metaphysics," i.e. as what's really real, where science is instead presenting models, creating representations. From intuitive contemplation of the contents of consciousness, we can come up with these models, and in fact all the really ground breaking (paradigm shifting) science does involve metaphysics in this way; however, we can't get from the models back to the intuition, just like you can't really explain to a blind-from-birth individual what color is really like.
So the real world is essentially defined as that which we experience; reality is not some Thing-In-Itself "out there" which our senses bring to us (and maybe distort).
There was a Not School Group on this topic.